diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32634 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 5434,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2195121951219514e-07,
+      "loss": 1.3236,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.439024390243903e-07,
+      "loss": 1.2216,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6585365853658536e-07,
+      "loss": 1.2443,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 1.2701,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.097560975609757e-07,
+      "loss": 1.2991,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.317073170731707e-07,
+      "loss": 0.2083,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.53658536585366e-07,
+      "loss": 1.2723,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.2176,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0975609756097562e-06,
+      "loss": 1.2974,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 1.2702,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3414634146341465e-06,
+      "loss": 1.2249,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 1.2025,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5853658536585368e-06,
+      "loss": 1.201,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.707317073170732e-06,
+      "loss": 1.1718,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8292682926829268e-06,
+      "loss": 1.1334,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.1178,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.073170731707317e-06,
+      "loss": 1.0853,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1951219512195125e-06,
+      "loss": 1.127,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.317073170731708e-06,
+      "loss": 1.0583,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 1.1085,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5609756097560977e-06,
+      "loss": 1.0942,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.682926829268293e-06,
+      "loss": 1.0832,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8048780487804884e-06,
+      "loss": 1.1144,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 1.1446,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 1.0794,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1707317073170736e-06,
+      "loss": 1.0424,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.292682926829269e-06,
+      "loss": 1.0914,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 0.2118,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5365853658536588e-06,
+      "loss": 1.0634,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 1.0736,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.780487804878049e-06,
+      "loss": 1.0923,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 1.0468,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.024390243902439e-06,
+      "loss": 0.2316,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.146341463414634e-06,
+      "loss": 1.04,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.268292682926829e-06,
+      "loss": 1.0456,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 0.2199,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.51219512195122e-06,
+      "loss": 1.0306,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 0.9704,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75609756097561e-06,
+      "loss": 1.0261,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.9683,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.9507,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 0.9837,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.243902439024391e-06,
+      "loss": 0.9667,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.0394,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 0.2215,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.609756097560977e-06,
+      "loss": 0.9826,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.731707317073171e-06,
+      "loss": 1.0305,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.969,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9756097560975615e-06,
+      "loss": 0.9924,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 1.0021,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.219512195121951e-06,
+      "loss": 1.0219,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 0.9401,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.463414634146342e-06,
+      "loss": 0.9621,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.585365853658538e-06,
+      "loss": 0.9689,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 0.8648,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 1.0098,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.951219512195122e-06,
+      "loss": 0.9834,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0731707317073175e-06,
+      "loss": 0.959,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1951219512195125e-06,
+      "loss": 0.9587,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.9837,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.439024390243903e-06,
+      "loss": 1.0277,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.560975609756098e-06,
+      "loss": 0.9393,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.682926829268293e-06,
+      "loss": 0.9335,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.9526,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.926829268292685e-06,
+      "loss": 0.9157,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.048780487804879e-06,
+      "loss": 0.9924,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.170731707317073e-06,
+      "loss": 0.9368,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 0.9007,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.414634146341464e-06,
+      "loss": 0.9445,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 0.9476,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.658536585365854e-06,
+      "loss": 0.8984,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.9335,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.902439024390244e-06,
+      "loss": 0.9171,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.02439024390244e-06,
+      "loss": 0.9602,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.146341463414635e-06,
+      "loss": 0.8766,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 0.9234,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.390243902439025e-06,
+      "loss": 0.9534,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.51219512195122e-06,
+      "loss": 0.9447,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.634146341463415e-06,
+      "loss": 0.9647,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.9712,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.878048780487805e-06,
+      "loss": 0.9357,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9292,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0121951219512197e-05,
+      "loss": 0.9141,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.024390243902439e-05,
+      "loss": 0.9722,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0365853658536586e-05,
+      "loss": 0.9115,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0487804878048782e-05,
+      "loss": 0.9519,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0609756097560976e-05,
+      "loss": 0.9504,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.9392,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0853658536585368e-05,
+      "loss": 0.9014,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0975609756097562e-05,
+      "loss": 0.9496,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1097560975609758e-05,
+      "loss": 0.9214,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1219512195121953e-05,
+      "loss": 0.8613,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1341463414634146e-05,
+      "loss": 0.8732,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1463414634146342e-05,
+      "loss": 0.9085,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1585365853658537e-05,
+      "loss": 0.9483,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.988,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1829268292682927e-05,
+      "loss": 0.9643,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1951219512195123e-05,
+      "loss": 0.9065,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2073170731707317e-05,
+      "loss": 0.9262,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.9315,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2317073170731709e-05,
+      "loss": 0.8939,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2439024390243903e-05,
+      "loss": 0.1821,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2560975609756098e-05,
+      "loss": 0.885,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.9983,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2804878048780488e-05,
+      "loss": 0.8889,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2926829268292684e-05,
+      "loss": 0.2473,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.304878048780488e-05,
+      "loss": 0.9402,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3170731707317076e-05,
+      "loss": 0.9181,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.329268292682927e-05,
+      "loss": 0.8877,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 1.004,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3536585365853661e-05,
+      "loss": 0.942,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.9161,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.378048780487805e-05,
+      "loss": 0.9003,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3902439024390244e-05,
+      "loss": 0.9051,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.402439024390244e-05,
+      "loss": 0.9432,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4146341463414635e-05,
+      "loss": 0.9153,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4268292682926829e-05,
+      "loss": 0.8582,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4390243902439025e-05,
+      "loss": 0.2654,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.451219512195122e-05,
+      "loss": 0.9379,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.9341,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.475609756097561e-05,
+      "loss": 0.9218,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4878048780487806e-05,
+      "loss": 0.922,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8894,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5121951219512196e-05,
+      "loss": 0.9167,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5243902439024392e-05,
+      "loss": 0.9334,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5365853658536586e-05,
+      "loss": 0.9052,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5487804878048782e-05,
+      "loss": 0.9027,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.9362,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5731707317073173e-05,
+      "loss": 0.9242,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.585365853658537e-05,
+      "loss": 0.9085,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.597560975609756e-05,
+      "loss": 0.9546,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 0.9414,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6219512195121953e-05,
+      "loss": 0.9152,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6341463414634145e-05,
+      "loss": 0.8972,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.646341463414634e-05,
+      "loss": 0.9054,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.9537,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6707317073170733e-05,
+      "loss": 0.8722,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.682926829268293e-05,
+      "loss": 0.9441,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6951219512195124e-05,
+      "loss": 0.8701,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 0.9287,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7195121951219512e-05,
+      "loss": 0.94,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7317073170731708e-05,
+      "loss": 0.9062,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7439024390243904e-05,
+      "loss": 0.9566,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.9136,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7682926829268296e-05,
+      "loss": 0.924,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7804878048780488e-05,
+      "loss": 0.9376,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7926829268292684e-05,
+      "loss": 0.9084,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.804878048780488e-05,
+      "loss": 0.9202,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8170731707317075e-05,
+      "loss": 0.9727,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.829268292682927e-05,
+      "loss": 0.9622,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8414634146341467e-05,
+      "loss": 0.9327,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.8808,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8658536585365855e-05,
+      "loss": 0.9432,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.878048780487805e-05,
+      "loss": 0.8727,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8902439024390243e-05,
+      "loss": 0.9155,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.902439024390244e-05,
+      "loss": 0.887,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9146341463414635e-05,
+      "loss": 0.955,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.926829268292683e-05,
+      "loss": 0.2666,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9390243902439026e-05,
+      "loss": 1.0024,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.2725,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9634146341463414e-05,
+      "loss": 0.9108,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.975609756097561e-05,
+      "loss": 0.8873,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9878048780487806e-05,
+      "loss": 0.9023,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9023,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999822315925e-05,
+      "loss": 0.9117,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992892637624e-05,
+      "loss": 0.9544,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999984008437015e-05,
+      "loss": 0.9049,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999971570560586e-05,
+      "loss": 0.9458,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999955579012754e-05,
+      "loss": 0.938,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999360337992e-05,
+      "loss": 0.9363,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999912934926874e-05,
+      "loss": 0.8952,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999886282403983e-05,
+      "loss": 0.9792,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999856076239997e-05,
+      "loss": 0.9283,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999822316445652e-05,
+      "loss": 0.9258,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978500303294e-05,
+      "loss": 0.9154,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999744136015132e-05,
+      "loss": 0.9117,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999699715406743e-05,
+      "loss": 0.9167,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999651741223557e-05,
+      "loss": 0.9103,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999600213482627e-05,
+      "loss": 0.894,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999545132202263e-05,
+      "loss": 0.912,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999948649740204e-05,
+      "loss": 0.9236,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999942430910279e-05,
+      "loss": 0.9376,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999358567326617e-05,
+      "loss": 0.899,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999289272096886e-05,
+      "loss": 0.9245,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999216423438218e-05,
+      "loss": 0.922,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999140021376505e-05,
+      "loss": 0.8788,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999060065938893e-05,
+      "loss": 0.928,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998976557153797e-05,
+      "loss": 0.939,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99988894950509e-05,
+      "loss": 0.879,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998798879661128e-05,
+      "loss": 0.3118,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998704711016694e-05,
+      "loss": 0.9001,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998606989151057e-05,
+      "loss": 0.2242,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998505714098948e-05,
+      "loss": 0.9488,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998400885896355e-05,
+      "loss": 0.9062,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998292504580528e-05,
+      "loss": 0.2252,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998180570189986e-05,
+      "loss": 0.8636,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999806508276451e-05,
+      "loss": 0.8871,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997946042345128e-05,
+      "loss": 0.8747,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997823448974157e-05,
+      "loss": 0.8954,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997697302695157e-05,
+      "loss": 0.9287,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997567603552957e-05,
+      "loss": 0.9471,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997434351593648e-05,
+      "loss": 0.9121,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999729754686458e-05,
+      "loss": 0.8469,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997157189414373e-05,
+      "loss": 0.978,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997013279292906e-05,
+      "loss": 0.9145,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996865816551317e-05,
+      "loss": 0.9241,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996714801242014e-05,
+      "loss": 0.9647,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999656023341866e-05,
+      "loss": 0.2539,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999640211313618e-05,
+      "loss": 0.2908,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996240440450773e-05,
+      "loss": 0.9252,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996075215419887e-05,
+      "loss": 0.8976,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995906438102238e-05,
+      "loss": 0.9382,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995734108557808e-05,
+      "loss": 0.912,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999555822684783e-05,
+      "loss": 0.8965,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995378793034815e-05,
+      "loss": 0.9482,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999519580718252e-05,
+      "loss": 0.8995,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995009269355978e-05,
+      "loss": 0.9453,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994819179621477e-05,
+      "loss": 0.8813,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999462553804657e-05,
+      "loss": 0.9128,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999442834470007e-05,
+      "loss": 0.8754,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999422759965205e-05,
+      "loss": 0.9446,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994023302973854e-05,
+      "loss": 0.9006,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999381545473808e-05,
+      "loss": 0.9421,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999360405501859e-05,
+      "loss": 0.9973,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993389103890513e-05,
+      "loss": 0.8687,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993170601430233e-05,
+      "loss": 0.9828,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992948547715394e-05,
+      "loss": 0.9024,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999272294282491e-05,
+      "loss": 0.9119,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992493786838957e-05,
+      "loss": 0.8236,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992261079838966e-05,
+      "loss": 0.9136,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992024821907636e-05,
+      "loss": 0.957,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991785013128922e-05,
+      "loss": 0.9579,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991541653588053e-05,
+      "loss": 0.9489,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99912947433715e-05,
+      "loss": 0.9209,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991044282567016e-05,
+      "loss": 0.9042,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99907902712636e-05,
+      "loss": 0.8922,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990532709551526e-05,
+      "loss": 0.9341,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990271597522318e-05,
+      "loss": 0.927,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990006935268766e-05,
+      "loss": 0.9188,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998973872288493e-05,
+      "loss": 0.9357,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989466960466118e-05,
+      "loss": 1.005,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989191648108907e-05,
+      "loss": 0.8837,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998891278591113e-05,
+      "loss": 0.8789,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988630373971896e-05,
+      "loss": 0.8617,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988344412391557e-05,
+      "loss": 0.8434,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998805490127174e-05,
+      "loss": 0.8657,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998776184071532e-05,
+      "loss": 0.897,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998746523082645e-05,
+      "loss": 0.9088,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998716507171053e-05,
+      "loss": 0.9233,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998686136347423e-05,
+      "loss": 0.9162,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986554106225474e-05,
+      "loss": 0.9528,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998624330007346e-05,
+      "loss": 0.8537,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985928945128634e-05,
+      "loss": 0.9187,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985611041502704e-05,
+      "loss": 0.9268,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985289589308645e-05,
+      "loss": 0.9355,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984964588660692e-05,
+      "loss": 0.8731,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984636039674342e-05,
+      "loss": 0.9481,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984303942466346e-05,
+      "loss": 0.8622,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983968297154726e-05,
+      "loss": 0.9317,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983629103858754e-05,
+      "loss": 0.9454,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983286362698973e-05,
+      "loss": 0.8791,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998294007379718e-05,
+      "loss": 0.8504,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998259023727644e-05,
+      "loss": 0.8814,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982236853261067e-05,
+      "loss": 0.9106,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998187992187664e-05,
+      "loss": 0.8899,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998151944325001e-05,
+      "loss": 0.8835,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981155417509277e-05,
+      "loss": 0.8933,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99807878447838e-05,
+      "loss": 0.8712,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998041672520421e-05,
+      "loss": 0.8617,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980042058902383e-05,
+      "loss": 0.9263,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979663846011466e-05,
+      "loss": 0.9282,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979282086665864e-05,
+      "loss": 0.9269,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978896781001245e-05,
+      "loss": 0.8527,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978507929154534e-05,
+      "loss": 0.9065,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997811553126391e-05,
+      "loss": 0.9183,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997771958746883e-05,
+      "loss": 0.9169,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997732009790999e-05,
+      "loss": 0.9066,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997691706272936e-05,
+      "loss": 0.8482,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976510482070162e-05,
+      "loss": 0.9218,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997610035607689e-05,
+      "loss": 0.9161,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997568668489528e-05,
+      "loss": 0.922,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975269468672342e-05,
+      "loss": 0.9201,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974848707556343e-05,
+      "loss": 0.2675,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997442440169681e-05,
+      "loss": 0.8946,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997399655124452e-05,
+      "loss": 0.8814,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973565156351524e-05,
+      "loss": 0.8934,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973130217171125e-05,
+      "loss": 0.9091,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997269173385788e-05,
+      "loss": 0.8835,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972249706567623e-05,
+      "loss": 0.885,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997180413545743e-05,
+      "loss": 0.8911,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971355020685642e-05,
+      "loss": 0.9179,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997090236241186e-05,
+      "loss": 0.9227,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997044616079695e-05,
+      "loss": 0.9177,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969986416003026e-05,
+      "loss": 0.9683,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969523128193468e-05,
+      "loss": 0.8814,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969056297532914e-05,
+      "loss": 0.9157,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968585924187257e-05,
+      "loss": 0.8768,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996811200832366e-05,
+      "loss": 0.8651,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967634550110536e-05,
+      "loss": 0.8991,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996715354971755e-05,
+      "loss": 0.8903,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996666900731565e-05,
+      "loss": 0.9146,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996618092307701e-05,
+      "loss": 0.9166,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965689297175093e-05,
+      "loss": 0.8792,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965194129784597e-05,
+      "loss": 0.2322,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964695421081496e-05,
+      "loss": 0.9284,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964193171243006e-05,
+      "loss": 0.2521,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963687380447624e-05,
+      "loss": 0.8811,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996317804887508e-05,
+      "loss": 0.9095,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962665176706382e-05,
+      "loss": 0.9694,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962148764123785e-05,
+      "loss": 0.9263,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961628811310805e-05,
+      "loss": 0.9258,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996110531845222e-05,
+      "loss": 0.8843,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960578285734056e-05,
+      "loss": 0.962,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996004771334361e-05,
+      "loss": 0.8772,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959513601469432e-05,
+      "loss": 0.9055,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958975950301322e-05,
+      "loss": 0.8975,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958434760030346e-05,
+      "loss": 0.9406,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957890030848828e-05,
+      "loss": 0.8958,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957341762950346e-05,
+      "loss": 0.9413,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956789956529738e-05,
+      "loss": 0.8824,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956234611783095e-05,
+      "loss": 0.9504,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955675728907776e-05,
+      "loss": 0.8742,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955113308102384e-05,
+      "loss": 0.9028,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954547349566783e-05,
+      "loss": 0.8741,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953977853502103e-05,
+      "loss": 0.9482,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953404820110725e-05,
+      "loss": 0.9247,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952828249596282e-05,
+      "loss": 0.9724,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952248142163672e-05,
+      "loss": 0.97,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951664498019044e-05,
+      "loss": 0.2728,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995107731736981e-05,
+      "loss": 0.8876,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950486600424634e-05,
+      "loss": 0.9213,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949892347393438e-05,
+      "loss": 0.939,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99492945584874e-05,
+      "loss": 0.8975,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994869323391895e-05,
+      "loss": 0.2671,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948088373901793e-05,
+      "loss": 0.8865,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947479978650867e-05,
+      "loss": 0.2454,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946868048382372e-05,
+      "loss": 0.9548,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994625258331378e-05,
+      "loss": 0.8984,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99456335836638e-05,
+      "loss": 0.8614,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994501104965241e-05,
+      "loss": 0.8966,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994438498150084e-05,
+      "loss": 0.8561,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994375537943156e-05,
+      "loss": 0.8789,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994312224366833e-05,
+      "loss": 0.8943,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994248557443613e-05,
+      "loss": 0.8714,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941845371961226e-05,
+      "loss": 0.8352,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941201636471115e-05,
+      "loss": 0.8993,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940554368194564e-05,
+      "loss": 0.9256,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939903567361594e-05,
+      "loss": 0.8865,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939249234203476e-05,
+      "loss": 0.8965,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993859136895274e-05,
+      "loss": 0.9331,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937929971843168e-05,
+      "loss": 0.913,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99372650431098e-05,
+      "loss": 0.9006,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936596582988936e-05,
+      "loss": 0.8818,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993592459171812e-05,
+      "loss": 0.9149,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993524906953616e-05,
+      "loss": 0.2463,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993457001668311e-05,
+      "loss": 0.8985,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933887433400284e-05,
+      "loss": 0.9692,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933201319930257e-05,
+      "loss": 0.7742,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932511676516846e-05,
+      "loss": 0.8995,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931818503405132e-05,
+      "loss": 0.8651,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931121800841442e-05,
+      "loss": 1.0156,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930421569073365e-05,
+      "loss": 0.9545,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992971780834974e-05,
+      "loss": 0.8726,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929010518920667e-05,
+      "loss": 0.9064,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928299701037482e-05,
+      "loss": 0.9016,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99275853549528e-05,
+      "loss": 0.923,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926867480920466e-05,
+      "loss": 0.9557,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926146079195597e-05,
+      "loss": 0.8618,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992542115003455e-05,
+      "loss": 0.8884,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924692693694953e-05,
+      "loss": 0.9038,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923960710435662e-05,
+      "loss": 0.8946,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992322520051681e-05,
+      "loss": 0.8774,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992248616419977e-05,
+      "loss": 0.9356,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992174360174717e-05,
+      "loss": 0.8664,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920997513422898e-05,
+      "loss": 0.9146,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992024789949209e-05,
+      "loss": 0.8888,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991949476022113e-05,
+      "loss": 0.9167,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918738095877665e-05,
+      "loss": 0.8948,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917977906730588e-05,
+      "loss": 0.8578,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991721419305004e-05,
+      "loss": 0.9205,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991644695510743e-05,
+      "loss": 0.8684,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915676193175405e-05,
+      "loss": 0.9134,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914901907527865e-05,
+      "loss": 0.9277,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914124098439976e-05,
+      "loss": 0.8418,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991334276618814e-05,
+      "loss": 0.2767,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991255791105002e-05,
+      "loss": 0.9044,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911769533304527e-05,
+      "loss": 0.8614,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910977633231826e-05,
+      "loss": 0.9448,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991018221111333e-05,
+      "loss": 0.8849,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909383267231715e-05,
+      "loss": 0.9082,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908580801870895e-05,
+      "loss": 0.8752,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907774815316037e-05,
+      "loss": 0.2071,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990696530785357e-05,
+      "loss": 0.8488,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906152279771162e-05,
+      "loss": 0.8679,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905335731357737e-05,
+      "loss": 0.9176,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990451566290348e-05,
+      "loss": 0.9063,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903692074699804e-05,
+      "loss": 0.8652,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990286496703939e-05,
+      "loss": 0.2356,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990203434021617e-05,
+      "loss": 0.8938,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990120019452532e-05,
+      "loss": 0.9052,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900362530263265e-05,
+      "loss": 0.8915,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989952134772769e-05,
+      "loss": 0.8607,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898676647217527e-05,
+      "loss": 0.848,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897828429032946e-05,
+      "loss": 0.9122,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989697669347538e-05,
+      "loss": 0.9402,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896121440847515e-05,
+      "loss": 0.8692,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989526267145327e-05,
+      "loss": 0.8963,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894400385597835e-05,
+      "loss": 0.9021,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893534583587633e-05,
+      "loss": 0.8448,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892665265730344e-05,
+      "loss": 0.836,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891792432334893e-05,
+      "loss": 0.884,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890916083711463e-05,
+      "loss": 0.9012,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890036220171472e-05,
+      "loss": 0.8832,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889152842027607e-05,
+      "loss": 0.271,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988826594959378e-05,
+      "loss": 0.8804,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988737554318517e-05,
+      "loss": 0.8515,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886481623118203e-05,
+      "loss": 0.2746,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885584189710546e-05,
+      "loss": 0.9174,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884683243281117e-05,
+      "loss": 0.8764,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883778784150083e-05,
+      "loss": 0.8844,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882870812638863e-05,
+      "loss": 0.9413,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881959329070123e-05,
+      "loss": 0.868,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881044333767768e-05,
+      "loss": 0.8555,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880125827056967e-05,
+      "loss": 0.8755,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879203809264117e-05,
+      "loss": 0.8694,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878278280716885e-05,
+      "loss": 0.903,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987734924174417e-05,
+      "loss": 0.8639,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876416692676123e-05,
+      "loss": 0.8529,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875480633844138e-05,
+      "loss": 0.9381,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874541065580865e-05,
+      "loss": 0.9357,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873597988220198e-05,
+      "loss": 0.9209,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872651402097273e-05,
+      "loss": 0.9115,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987170130754848e-05,
+      "loss": 0.8493,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987074770491145e-05,
+      "loss": 0.8991,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986979059452506e-05,
+      "loss": 0.8236,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868829976729444e-05,
+      "loss": 0.9217,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986786585186597e-05,
+      "loss": 0.895,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866898220277256e-05,
+      "loss": 0.8667,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865927082307174e-05,
+      "loss": 0.8438,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864952438300826e-05,
+      "loss": 0.8838,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863974288604576e-05,
+      "loss": 0.9049,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862992633566026e-05,
+      "loss": 0.9075,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862007473534026e-05,
+      "loss": 0.9051,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986101880885867e-05,
+      "loss": 0.8931,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860026639891298e-05,
+      "loss": 0.9365,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985903096698449e-05,
+      "loss": 0.9391,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858031790492082e-05,
+      "loss": 0.9352,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857029110769154e-05,
+      "loss": 0.872,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856022928172013e-05,
+      "loss": 0.8806,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985501324305824e-05,
+      "loss": 0.8899,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985400005578663e-05,
+      "loss": 0.2216,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852983366717248e-05,
+      "loss": 0.8286,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985196317621139e-05,
+      "loss": 0.8813,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850939484631598e-05,
+      "loss": 0.8464,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984991229234166e-05,
+      "loss": 0.8893,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848881599706605e-05,
+      "loss": 0.8884,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847847407092713e-05,
+      "loss": 0.9714,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98468097148675e-05,
+      "loss": 0.9178,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845768523399732e-05,
+      "loss": 0.8671,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984472383305941e-05,
+      "loss": 0.9083,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843675644217785e-05,
+      "loss": 0.9182,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842623957247355e-05,
+      "loss": 0.8742,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841568772521853e-05,
+      "loss": 0.8714,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984051009041626e-05,
+      "loss": 0.8965,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983944791130679e-05,
+      "loss": 0.8449,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838382235570915e-05,
+      "loss": 0.841,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983731306358734e-05,
+      "loss": 0.9298,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836240395736017e-05,
+      "loss": 0.8738,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835164232398134e-05,
+      "loss": 0.8833,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983408457395613e-05,
+      "loss": 0.2335,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833001420793673e-05,
+      "loss": 0.8707,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983191477329569e-05,
+      "loss": 0.9726,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983082463184834e-05,
+      "loss": 0.8918,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 0.9073,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828633868656376e-05,
+      "loss": 0.8421,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982753324769029e-05,
+      "loss": 0.8888,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826429134331887e-05,
+      "loss": 0.88,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982532152897354e-05,
+      "loss": 0.886,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982421043200885e-05,
+      "loss": 0.9091,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982309584383267e-05,
+      "loss": 0.9576,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821977764841086e-05,
+      "loss": 0.8966,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820856195431428e-05,
+      "loss": 0.8439,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819731136002267e-05,
+      "loss": 0.8706,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818602586953414e-05,
+      "loss": 0.8349,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981747054868592e-05,
+      "loss": 0.9024,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816335021602072e-05,
+      "loss": 0.9387,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815196006105402e-05,
+      "loss": 0.9262,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814053502600683e-05,
+      "loss": 0.9018,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812907511493917e-05,
+      "loss": 0.9148,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981175803319236e-05,
+      "loss": 0.9628,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810605068104495e-05,
+      "loss": 0.8713,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809448616640052e-05,
+      "loss": 0.9006,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808288679209996e-05,
+      "loss": 0.8771,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807125256226532e-05,
+      "loss": 0.873,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98059583481031e-05,
+      "loss": 0.9271,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980478795525439e-05,
+      "loss": 0.8959,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803614078096315e-05,
+      "loss": 0.9042,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980243671704604e-05,
+      "loss": 0.9256,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980125587252195e-05,
+      "loss": 0.9135,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800071544943696e-05,
+      "loss": 0.88,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798883734732138e-05,
+      "loss": 0.9055,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797692442309387e-05,
+      "loss": 0.9195,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796497668098797e-05,
+      "loss": 0.8397,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795299412524948e-05,
+      "loss": 0.9488,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979409767601366e-05,
+      "loss": 0.8758,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792892458991995e-05,
+      "loss": 0.9087,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791683761888253e-05,
+      "loss": 0.8814,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790471585131956e-05,
+      "loss": 0.8981,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978925592915388e-05,
+      "loss": 0.8982,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978803679438603e-05,
+      "loss": 0.9176,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786814181261646e-05,
+      "loss": 0.8937,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785588090215205e-05,
+      "loss": 0.8954,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784358521682424e-05,
+      "loss": 0.8955,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783125476100254e-05,
+      "loss": 0.2724,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781888953906874e-05,
+      "loss": 0.901,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780648955541706e-05,
+      "loss": 0.8297,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977940548144541e-05,
+      "loss": 0.8782,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977815853205987e-05,
+      "loss": 0.8804,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776908107828218e-05,
+      "loss": 0.8508,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775654209194816e-05,
+      "loss": 0.9048,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774396836605257e-05,
+      "loss": 0.8372,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977313599050637e-05,
+      "loss": 0.8708,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771871671346222e-05,
+      "loss": 0.9208,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770603879574108e-05,
+      "loss": 0.8856,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769332615640566e-05,
+      "loss": 0.9122,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976805787999736e-05,
+      "loss": 0.8943,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976677967309749e-05,
+      "loss": 0.8013,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976549799539519e-05,
+      "loss": 0.9169,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976421284734593e-05,
+      "loss": 0.8937,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762924229406405e-05,
+      "loss": 0.889,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761632142034555e-05,
+      "loss": 0.9158,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760336585689544e-05,
+      "loss": 0.8861,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975903756083177e-05,
+      "loss": 0.9053,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975773506792287e-05,
+      "loss": 0.9216,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756429107425704e-05,
+      "loss": 0.8598,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975511967980437e-05,
+      "loss": 0.9088,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753806785524193e-05,
+      "loss": 0.8863,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975249042505174e-05,
+      "loss": 0.882,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751170598854804e-05,
+      "loss": 0.8847,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749847307402406e-05,
+      "loss": 0.8765,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748520551164797e-05,
+      "loss": 0.8426,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747190330613475e-05,
+      "loss": 0.8895,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974585664622115e-05,
+      "loss": 0.9174,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974451949846177e-05,
+      "loss": 0.9208,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743178887810524e-05,
+      "loss": 0.8175,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741834814743812e-05,
+      "loss": 0.8837,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740487279739278e-05,
+      "loss": 0.8428,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739136283275794e-05,
+      "loss": 0.2698,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737781825833465e-05,
+      "loss": 0.8169,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736423907893612e-05,
+      "loss": 0.9268,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735062529938807e-05,
+      "loss": 0.9162,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973369769245283e-05,
+      "loss": 0.8746,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973232939592071e-05,
+      "loss": 0.8862,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973095764082869e-05,
+      "loss": 0.9322,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972958242766425e-05,
+      "loss": 0.8741,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97282037569161e-05,
+      "loss": 0.9317,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726821629074168e-05,
+      "loss": 0.904,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725436044629625e-05,
+      "loss": 0.8771,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972404700407486e-05,
+      "loss": 0.9733,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722654507903497e-05,
+      "loss": 0.839,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721258556610383e-05,
+      "loss": 0.8608,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719859150691595e-05,
+      "loss": 0.8753,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718456290644435e-05,
+      "loss": 0.8774,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717049976967437e-05,
+      "loss": 0.2251,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971564021016036e-05,
+      "loss": 0.2824,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971422699072419e-05,
+      "loss": 0.8676,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971281031916114e-05,
+      "loss": 0.913,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971139019597465e-05,
+      "loss": 0.8772,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970996662166939e-05,
+      "loss": 0.8358,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708539596751245e-05,
+      "loss": 0.9022,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970710912172734e-05,
+      "loss": 0.894,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970567519710602e-05,
+      "loss": 0.8719,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704237823396853e-05,
+      "loss": 0.8967,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702797001110642e-05,
+      "loss": 0.9095,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701352730759403e-05,
+      "loss": 0.7983,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699905012856383e-05,
+      "loss": 0.9453,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698453847916063e-05,
+      "loss": 0.9179,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696999236454135e-05,
+      "loss": 0.8979,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695541178987518e-05,
+      "loss": 0.8928,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969407967603437e-05,
+      "loss": 0.8096,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969261472811405e-05,
+      "loss": 0.9331,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691146335747165e-05,
+      "loss": 0.8429,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689674499455526e-05,
+      "loss": 0.9041,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688199219762183e-05,
+      "loss": 0.8712,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96867204971914e-05,
+      "loss": 0.8764,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968523833226867e-05,
+      "loss": 0.9072,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683752725520702e-05,
+      "loss": 0.9017,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682263677475442e-05,
+      "loss": 0.8471,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680771188662044e-05,
+      "loss": 0.8898,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679275259610894e-05,
+      "loss": 0.8565,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677775890853595e-05,
+      "loss": 0.8325,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676273082922973e-05,
+      "loss": 0.9109,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674766836353085e-05,
+      "loss": 0.8285,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96732571516792e-05,
+      "loss": 0.8824,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967174402943781e-05,
+      "loss": 0.8999,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967022747016663e-05,
+      "loss": 0.877,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96687074744046e-05,
+      "loss": 0.8364,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667184042691877e-05,
+      "loss": 0.8586,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665657175569838e-05,
+      "loss": 0.901,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664126873581086e-05,
+      "loss": 0.8799,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662593137269436e-05,
+      "loss": 0.87,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966105596717994e-05,
+      "loss": 0.8272,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659515363858847e-05,
+      "loss": 0.9021,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657971327853644e-05,
+      "loss": 0.8582,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656423859713034e-05,
+      "loss": 0.8441,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654872959986936e-05,
+      "loss": 0.913,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965331862922649e-05,
+      "loss": 0.8967,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651760867984054e-05,
+      "loss": 0.8674,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650199676813214e-05,
+      "loss": 0.8599,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648635056268757e-05,
+      "loss": 0.8993,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647067006906707e-05,
+      "loss": 0.8592,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645495529284292e-05,
+      "loss": 0.8596,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643920623959975e-05,
+      "loss": 0.864,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964234229149342e-05,
+      "loss": 0.8829,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640760532445518e-05,
+      "loss": 0.872,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639175347378378e-05,
+      "loss": 0.889,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637586736855314e-05,
+      "loss": 0.8687,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635994701440882e-05,
+      "loss": 0.9213,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963439924170083e-05,
+      "loss": 0.8084,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632800358202142e-05,
+      "loss": 0.8934,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631198051513004e-05,
+      "loss": 0.8418,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962959232220283e-05,
+      "loss": 0.9155,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627983170842235e-05,
+      "loss": 0.8927,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962637059800307e-05,
+      "loss": 0.8483,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962475460425839e-05,
+      "loss": 0.8539,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962313519018247e-05,
+      "loss": 0.8586,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621512356350787e-05,
+      "loss": 0.3097,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619886103340056e-05,
+      "loss": 0.862,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961825643172819e-05,
+      "loss": 0.8425,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616623342094328e-05,
+      "loss": 0.9071,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961498683501881e-05,
+      "loss": 0.8518,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961334691108321e-05,
+      "loss": 0.944,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961170357087029e-05,
+      "loss": 0.8226,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610056814964053e-05,
+      "loss": 0.864,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608406643949697e-05,
+      "loss": 0.8521,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960675305841364e-05,
+      "loss": 0.8762,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605096058943517e-05,
+      "loss": 0.9261,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603435646128172e-05,
+      "loss": 0.878,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601771820557662e-05,
+      "loss": 0.8916,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960010458282326e-05,
+      "loss": 0.8789,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598433933517446e-05,
+      "loss": 0.9126,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596759873233916e-05,
+      "loss": 0.9026,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959508240256758e-05,
+      "loss": 0.8948,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959340152211455e-05,
+      "loss": 0.8626,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959171723247217e-05,
+      "loss": 0.8794,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590029534238977e-05,
+      "loss": 0.879,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958833842801472e-05,
+      "loss": 0.8584,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586643914400372e-05,
+      "loss": 0.8608,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584945993998103e-05,
+      "loss": 0.8517,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583244667411306e-05,
+      "loss": 0.8252,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581539935244573e-05,
+      "loss": 0.8078,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579831798103716e-05,
+      "loss": 0.908,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578120256595746e-05,
+      "loss": 0.8831,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95764053113289e-05,
+      "loss": 0.9057,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574686962912607e-05,
+      "loss": 0.853,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572965211957515e-05,
+      "loss": 0.9045,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571240059075482e-05,
+      "loss": 0.922,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956951150487957e-05,
+      "loss": 0.8775,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956777954998405e-05,
+      "loss": 0.8972,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956604419500441e-05,
+      "loss": 0.8392,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564305440557335e-05,
+      "loss": 0.8844,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562563287260724e-05,
+      "loss": 0.8713,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560817735733683e-05,
+      "loss": 0.8557,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559068786596526e-05,
+      "loss": 0.8926,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955731644047077e-05,
+      "loss": 0.8391,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555560697979147e-05,
+      "loss": 0.8193,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553801559745594e-05,
+      "loss": 0.8996,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955203902639525e-05,
+      "loss": 0.8695,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550273098554463e-05,
+      "loss": 0.8789,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548503776850785e-05,
+      "loss": 0.8959,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546731061912977e-05,
+      "loss": 0.8789,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544954954371013e-05,
+      "loss": 0.9201,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543175454856057e-05,
+      "loss": 0.8885,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954139256400049e-05,
+      "loss": 0.8345,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539606282437893e-05,
+      "loss": 0.8709,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537816610803056e-05,
+      "loss": 0.8267,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953602354973197e-05,
+      "loss": 0.8858,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534227099861827e-05,
+      "loss": 0.9161,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532427261831033e-05,
+      "loss": 0.9056,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95306240362792e-05,
+      "loss": 0.87,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528817423847123e-05,
+      "loss": 0.8372,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952700742517682e-05,
+      "loss": 0.9361,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952519404091151e-05,
+      "loss": 0.862,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952337727169561e-05,
+      "loss": 0.8605,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521557118174742e-05,
+      "loss": 0.9078,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951973358099573e-05,
+      "loss": 0.9217,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517906660806602e-05,
+      "loss": 0.8644,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516076358256585e-05,
+      "loss": 0.305,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514242673996115e-05,
+      "loss": 0.8699,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512405608676822e-05,
+      "loss": 0.8709,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.8867,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95087213374743e-05,
+      "loss": 0.8331,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950687413290035e-05,
+      "loss": 0.2777,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950502354988612e-05,
+      "loss": 0.2441,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503169589089248e-05,
+      "loss": 0.2797,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501312251168574e-05,
+      "loss": 0.8964,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499451536784143e-05,
+      "loss": 0.903,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497587446597185e-05,
+      "loss": 0.8836,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949571998127014e-05,
+      "loss": 0.863,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949384914146665e-05,
+      "loss": 0.8799,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491974927851544e-05,
+      "loss": 0.2767,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490097341090868e-05,
+      "loss": 0.2409,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948821638185185e-05,
+      "loss": 0.9153,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948633205080292e-05,
+      "loss": 0.8357,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948444434861372e-05,
+      "loss": 0.8789,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482553275955067e-05,
+      "loss": 0.9232,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480658833498997e-05,
+      "loss": 0.8795,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947876102191873e-05,
+      "loss": 0.8989,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476859841888687e-05,
+      "loss": 0.8775,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474955294084492e-05,
+      "loss": 0.8058,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473047379182955e-05,
+      "loss": 0.9017,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471136097862093e-05,
+      "loss": 0.8799,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946922145080111e-05,
+      "loss": 0.864,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467303438680414e-05,
+      "loss": 0.9001,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465382062181603e-05,
+      "loss": 0.9144,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946345732198748e-05,
+      "loss": 0.9057,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461529218782026e-05,
+      "loss": 0.8762,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945959775325043e-05,
+      "loss": 0.8758,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457662926079076e-05,
+      "loss": 0.8632,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945572473795554e-05,
+      "loss": 0.8796,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453783189568593e-05,
+      "loss": 0.8799,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94518382816082e-05,
+      "loss": 0.9016,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944989001476551e-05,
+      "loss": 0.8622,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944793838973289e-05,
+      "loss": 0.2811,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445983407203873e-05,
+      "loss": 0.8853,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444025067873205e-05,
+      "loss": 0.9102,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442063372436816e-05,
+      "loss": 0.9232,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440098321591825e-05,
+      "loss": 0.8128,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438129916036553e-05,
+      "loss": 0.928,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943615815647051e-05,
+      "loss": 0.8416,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943418304359439e-05,
+      "loss": 0.8625,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432204578110094e-05,
+      "loss": 0.9175,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430222760720703e-05,
+      "loss": 0.8795,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428237592130487e-05,
+      "loss": 0.9008,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426249073044917e-05,
+      "loss": 0.8615,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424257204170643e-05,
+      "loss": 0.8048,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942226198621552e-05,
+      "loss": 0.8849,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420263419888577e-05,
+      "loss": 0.8206,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941826150590005e-05,
+      "loss": 0.8675,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416256244961348e-05,
+      "loss": 0.8322,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414247637785075e-05,
+      "loss": 0.8573,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412235685085034e-05,
+      "loss": 0.877,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410220387576206e-05,
+      "loss": 0.8304,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940820174597476e-05,
+      "loss": 0.8695,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940617976099806e-05,
+      "loss": 0.2619,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404154433364652e-05,
+      "loss": 0.8718,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402125763794278e-05,
+      "loss": 0.8367,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940009375300786e-05,
+      "loss": 0.8972,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398058401727508e-05,
+      "loss": 0.9181,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396019710676527e-05,
+      "loss": 0.8518,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393977680579394e-05,
+      "loss": 0.93,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391932312161786e-05,
+      "loss": 0.8648,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938988360615057e-05,
+      "loss": 0.902,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387831563273775e-05,
+      "loss": 0.8179,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938577618426064e-05,
+      "loss": 0.8616,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938371746984158e-05,
+      "loss": 0.8527,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381655420748205e-05,
+      "loss": 0.8759,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379590037713287e-05,
+      "loss": 0.8556,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377521321470806e-05,
+      "loss": 0.795,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375449272755917e-05,
+      "loss": 0.9023,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373373892304957e-05,
+      "loss": 0.8541,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371295180855454e-05,
+      "loss": 0.8824,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369213139146114e-05,
+      "loss": 0.8928,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367127767916828e-05,
+      "loss": 0.2702,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936503906790867e-05,
+      "loss": 0.8623,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93629470398639e-05,
+      "loss": 0.849,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936085168452596e-05,
+      "loss": 0.8741,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358753002639466e-05,
+      "loss": 0.8733,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356650994950223e-05,
+      "loss": 0.8865,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935454566220522e-05,
+      "loss": 0.8388,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352437005152633e-05,
+      "loss": 0.8764,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93503250245418e-05,
+      "loss": 0.261,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348209721123255e-05,
+      "loss": 0.8664,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346091095648712e-05,
+      "loss": 0.8992,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343969148871057e-05,
+      "loss": 0.8826,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341843881544372e-05,
+      "loss": 0.8946,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93397152944239e-05,
+      "loss": 0.8699,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933758338826608e-05,
+      "loss": 0.9112,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933544816382852e-05,
+      "loss": 0.8884,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333309621870012e-05,
+      "loss": 0.8038,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933116776315052e-05,
+      "loss": 0.9593,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329022588431204e-05,
+      "loss": 0.8777,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932687409847438e-05,
+      "loss": 0.8979,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 0.8974,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932256717590342e-05,
+      "loss": 0.8373,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932040874481983e-05,
+      "loss": 0.8904,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318247001559814e-05,
+      "loss": 0.8245,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9316081946891604e-05,
+      "loss": 0.8355,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313913581584572e-05,
+      "loss": 0.8828,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311741906409302e-05,
+      "loss": 0.8266,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309566922137532e-05,
+      "loss": 0.8645,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307388629542183e-05,
+      "loss": 0.8806,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305207029397348e-05,
+      "loss": 0.8382,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303022122478303e-05,
+      "loss": 0.899,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930083390956149e-05,
+      "loss": 0.861,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929864239142453e-05,
+      "loss": 0.8511,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296447568846218e-05,
+      "loss": 0.865,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294249442606533e-05,
+      "loss": 0.8857,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929204801348661e-05,
+      "loss": 0.8841,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289843282268773e-05,
+      "loss": 0.903,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287635249736505e-05,
+      "loss": 0.8708,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285423916674478e-05,
+      "loss": 0.2505,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283209283868523e-05,
+      "loss": 0.896,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280991352105656e-05,
+      "loss": 0.8663,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278770122174057e-05,
+      "loss": 0.8148,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927654559486308e-05,
+      "loss": 0.8415,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274317770963252e-05,
+      "loss": 0.8848,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927208665126627e-05,
+      "loss": 0.8646,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269852236564998e-05,
+      "loss": 0.9125,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926761452765349e-05,
+      "loss": 0.8815,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926537352532694e-05,
+      "loss": 0.8846,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263129230381735e-05,
+      "loss": 0.8958,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260881643615432e-05,
+      "loss": 0.8662,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925863076582674e-05,
+      "loss": 0.8584,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256376597815565e-05,
+      "loss": 0.8731,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254119140382952e-05,
+      "loss": 0.8778,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251858394331136e-05,
+      "loss": 0.7707,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249594360463514e-05,
+      "loss": 0.8031,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924732703958465e-05,
+      "loss": 0.8509,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245056432500277e-05,
+      "loss": 0.8856,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92427825400173e-05,
+      "loss": 0.8712,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240505362943782e-05,
+      "loss": 0.834,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238224902088968e-05,
+      "loss": 0.8317,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235941158263253e-05,
+      "loss": 0.8239,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923365413227821e-05,
+      "loss": 0.873,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231363824946572e-05,
+      "loss": 0.8538,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229070237082252e-05,
+      "loss": 0.8698,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226773369500305e-05,
+      "loss": 0.9148,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922447322301697e-05,
+      "loss": 0.8615,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222169798449647e-05,
+      "loss": 0.7885,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92198630966169e-05,
+      "loss": 0.8275,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217553118338453e-05,
+      "loss": 0.8616,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215239864435204e-05,
+      "loss": 0.8969,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212923335729206e-05,
+      "loss": 0.7921,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210603533043682e-05,
+      "loss": 0.8691,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208280457203014e-05,
+      "loss": 0.8994,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205954109032747e-05,
+      "loss": 0.9293,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92036244893596e-05,
+      "loss": 0.9141,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201291599011435e-05,
+      "loss": 0.8938,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198955438817294e-05,
+      "loss": 0.8584,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196616009607372e-05,
+      "loss": 0.8582,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194273312213027e-05,
+      "loss": 0.8175,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191927347466778e-05,
+      "loss": 0.8845,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918957811620231e-05,
+      "loss": 0.8801,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187225619254458e-05,
+      "loss": 0.8056,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184869857459233e-05,
+      "loss": 0.9269,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918251083165379e-05,
+      "loss": 0.2396,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180148542676456e-05,
+      "loss": 0.9418,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177782991366716e-05,
+      "loss": 0.9117,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175414178565205e-05,
+      "loss": 0.7952,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173042105113725e-05,
+      "loss": 0.8482,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170666771855242e-05,
+      "loss": 0.8824,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168288179633866e-05,
+      "loss": 0.8333,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165906329294875e-05,
+      "loss": 0.8239,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163521221684705e-05,
+      "loss": 0.8315,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161132857650945e-05,
+      "loss": 0.8556,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158741238042346e-05,
+      "loss": 0.8198,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915634636370881e-05,
+      "loss": 0.8893,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91539482355014e-05,
+      "loss": 0.9217,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151546854272335e-05,
+      "loss": 0.8509,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149142220874992e-05,
+      "loss": 0.8705,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146734336163898e-05,
+      "loss": 0.8864,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914432320099474e-05,
+      "loss": 0.8979,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141908816224356e-05,
+      "loss": 0.9176,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139491182710747e-05,
+      "loss": 0.8702,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137070301313053e-05,
+      "loss": 0.8588,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913464617289159e-05,
+      "loss": 0.8634,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132218798307806e-05,
+      "loss": 0.8254,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912978817842432e-05,
+      "loss": 0.8414,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127354314104892e-05,
+      "loss": 0.8924,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124917206214447e-05,
+      "loss": 0.8816,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122476855619045e-05,
+      "loss": 0.8698,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120033263185917e-05,
+      "loss": 0.819,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117586429783433e-05,
+      "loss": 0.2958,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115136356281124e-05,
+      "loss": 0.8631,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112683043549665e-05,
+      "loss": 0.8911,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110226492460886e-05,
+      "loss": 0.2374,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107766703887764e-05,
+      "loss": 0.8305,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105303678704437e-05,
+      "loss": 0.8954,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910283741778618e-05,
+      "loss": 0.9121,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100367922009428e-05,
+      "loss": 0.8593,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097895192251754e-05,
+      "loss": 0.8463,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095419229391893e-05,
+      "loss": 0.9095,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909294003430972e-05,
+      "loss": 0.8652,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090457607886268e-05,
+      "loss": 0.9085,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087971951003705e-05,
+      "loss": 0.8823,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908548306454536e-05,
+      "loss": 0.253,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082990949395696e-05,
+      "loss": 0.8326,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908049560644034e-05,
+      "loss": 0.8772,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077997036566054e-05,
+      "loss": 0.8338,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907549524066075e-05,
+      "loss": 0.8185,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072990219613486e-05,
+      "loss": 0.8356,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907048197431447e-05,
+      "loss": 0.9388,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906797050565505e-05,
+      "loss": 0.8717,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065455814527722e-05,
+      "loss": 0.7884,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062937901826128e-05,
+      "loss": 0.886,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9060416768445046e-05,
+      "loss": 0.8041,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905789241528042e-05,
+      "loss": 0.891,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055364843229317e-05,
+      "loss": 0.8428,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052834053189957e-05,
+      "loss": 0.875,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050300046061702e-05,
+      "loss": 0.877,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904776282274506e-05,
+      "loss": 0.2555,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045222384141675e-05,
+      "loss": 0.2426,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042678731154337e-05,
+      "loss": 0.9233,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040131864686986e-05,
+      "loss": 0.9083,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037581785644695e-05,
+      "loss": 0.2467,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035028494933674e-05,
+      "loss": 0.8241,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903247199346129e-05,
+      "loss": 0.9337,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029912282136038e-05,
+      "loss": 0.8466,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902734936186756e-05,
+      "loss": 0.7893,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024783233566633e-05,
+      "loss": 0.891,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022213898145176e-05,
+      "loss": 0.8705,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019641356516258e-05,
+      "loss": 0.8179,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901706560959407e-05,
+      "loss": 0.8965,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014486658293953e-05,
+      "loss": 0.8495,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901190450353238e-05,
+      "loss": 0.9096,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009319146226974e-05,
+      "loss": 0.9001,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900673058729649e-05,
+      "loss": 0.8301,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004138827660806e-05,
+      "loss": 0.8553,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900154386824096e-05,
+      "loss": 0.8475,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899894570995912e-05,
+      "loss": 0.9259,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996344353738587e-05,
+      "loss": 0.8533,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89937398005038e-05,
+      "loss": 0.8531,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991132051180332e-05,
+      "loss": 0.8783,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988521106694896e-05,
+      "loss": 0.8698,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985906967975337e-05,
+      "loss": 0.8594,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983289635950638e-05,
+      "loss": 0.8824,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980669111550917e-05,
+      "loss": 0.9013,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897804539570742e-05,
+      "loss": 0.8568,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975418489352534e-05,
+      "loss": 0.8949,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897278839341978e-05,
+      "loss": 0.8795,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970155108843812e-05,
+      "loss": 0.8447,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967518636560412e-05,
+      "loss": 0.9038,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964878977506496e-05,
+      "loss": 0.868,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896223613262012e-05,
+      "loss": 0.267,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959590102840465e-05,
+      "loss": 0.8825,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956940889107843e-05,
+      "loss": 0.8603,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954288492363708e-05,
+      "loss": 0.2871,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951632913550625e-05,
+      "loss": 0.8699,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894897415361231e-05,
+      "loss": 0.9178,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946312213493602e-05,
+      "loss": 0.8601,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943647094140473e-05,
+      "loss": 0.8429,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894097879650001e-05,
+      "loss": 0.9022,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938307321520453e-05,
+      "loss": 0.803,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893563267015115e-05,
+      "loss": 0.878,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893295484334259e-05,
+      "loss": 0.8242,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930273842046393e-05,
+      "loss": 0.843,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927589667215294e-05,
+      "loss": 0.8516,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924902319803165e-05,
+      "loss": 0.8405,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922211800765006e-05,
+      "loss": 0.8746,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8919518111056943e-05,
+      "loss": 0.8546,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891682125163622e-05,
+      "loss": 0.922,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914121223461223e-05,
+      "loss": 0.8148,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911418027491453e-05,
+      "loss": 0.9257,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908711664687537e-05,
+      "loss": 0.8471,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906002136011235e-05,
+      "loss": 0.8543,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903289442425426e-05,
+      "loss": 0.7855,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900573584894112e-05,
+      "loss": 0.9004,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897854564382425e-05,
+      "loss": 0.9024,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895132381856617e-05,
+      "loss": 0.8371,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892407038284064e-05,
+      "loss": 0.2467,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888967853463327e-05,
+      "loss": 0.8181,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888694687187386e-05,
+      "loss": 0.8648,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884212050976568e-05,
+      "loss": 0.7483,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881474072913275e-05,
+      "loss": 0.8601,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878732938656962e-05,
+      "loss": 0.8512,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875988649181752e-05,
+      "loss": 0.8397,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873241205462864e-05,
+      "loss": 0.845,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870490608476663e-05,
+      "loss": 0.8772,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886773685920062e-05,
+      "loss": 0.8826,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864979958613328e-05,
+      "loss": 0.8977,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862219907694505e-05,
+      "loss": 0.8998,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859456707424985e-05,
+      "loss": 0.8741,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885669035878672e-05,
+      "loss": 0.2778,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853920862762777e-05,
+      "loss": 0.8677,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851148220337357e-05,
+      "loss": 0.8391,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848372432495762e-05,
+      "loss": 0.8136,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845593500224416e-05,
+      "loss": 0.8922,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842811424510872e-05,
+      "loss": 0.8533,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840026206343786e-05,
+      "loss": 0.8461,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837237846712936e-05,
+      "loss": 0.8688,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834446346609216e-05,
+      "loss": 0.8865,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831651707024632e-05,
+      "loss": 0.8674,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882885392895232e-05,
+      "loss": 0.8436,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826053013386513e-05,
+      "loss": 0.855,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823248961322572e-05,
+      "loss": 0.9217,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820441773756966e-05,
+      "loss": 0.8679,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817631451687277e-05,
+      "loss": 0.9018,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814817996112208e-05,
+      "loss": 0.8363,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881200140803157e-05,
+      "loss": 0.8328,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880918168844629e-05,
+      "loss": 0.8822,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806358838358403e-05,
+      "loss": 0.8676,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803532858771067e-05,
+      "loss": 0.8472,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800703750688536e-05,
+      "loss": 0.9008,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797871515116194e-05,
+      "loss": 0.8047,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795036153060518e-05,
+      "loss": 0.8682,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879219766552911e-05,
+      "loss": 0.8845,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878935605353068e-05,
+      "loss": 0.8477,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786511318075045e-05,
+      "loss": 0.8222,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783663460173135e-05,
+      "loss": 0.8704,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878081248083698e-05,
+      "loss": 0.2515,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777958381079737e-05,
+      "loss": 0.8739,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877510116191566e-05,
+      "loss": 0.9502,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877224082436011e-05,
+      "loss": 0.8717,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876937736942956e-05,
+      "loss": 0.7664,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766510798141597e-05,
+      "loss": 0.8858,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87636411115149e-05,
+      "loss": 0.8692,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760768310569273e-05,
+      "loss": 0.8839,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757892396325612e-05,
+      "loss": 0.8754,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755013369805926e-05,
+      "loss": 0.8477,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752131232033334e-05,
+      "loss": 0.898,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749245984032053e-05,
+      "loss": 0.8714,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746357626827403e-05,
+      "loss": 0.8526,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743466161445823e-05,
+      "loss": 0.8439,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874057158891484e-05,
+      "loss": 0.8546,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87376739102631e-05,
+      "loss": 0.8424,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734773126520344e-05,
+      "loss": 0.9124,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731869238717413e-05,
+      "loss": 0.8433,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872896224788626e-05,
+      "loss": 0.8788,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726052155059937e-05,
+      "loss": 0.8491,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723138961272597e-05,
+      "loss": 0.2388,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720222667559498e-05,
+      "loss": 0.8493,
+      "step": 1022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717303274956997e-05,
+      "loss": 0.8588,
+      "step": 1023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714380784502553e-05,
+      "loss": 0.8949,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711455197234728e-05,
+      "loss": 0.907,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870852651419318e-05,
+      "loss": 0.9196,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705594736418667e-05,
+      "loss": 0.7968,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870265986495306e-05,
+      "loss": 0.8472,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699721900839306e-05,
+      "loss": 0.8548,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869678084512147e-05,
+      "loss": 0.8635,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869383669884471e-05,
+      "loss": 0.821,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690889463055285e-05,
+      "loss": 0.8718,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687939138800542e-05,
+      "loss": 0.9276,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684985727128936e-05,
+      "loss": 0.7652,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868202922909001e-05,
+      "loss": 0.8772,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867906964573442e-05,
+      "loss": 0.8971,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676106978113897e-05,
+      "loss": 0.8304,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673141227281282e-05,
+      "loss": 0.2251,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867017239429051e-05,
+      "loss": 0.8492,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866720048019661e-05,
+      "loss": 0.2452,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866422548605571e-05,
+      "loss": 0.8315,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866124741292502e-05,
+      "loss": 0.8903,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658266261862852e-05,
+      "loss": 0.8261,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655282033928618e-05,
+      "loss": 0.8289,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652294730182815e-05,
+      "loss": 0.8576,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864930435168703e-05,
+      "loss": 0.8098,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646310899503964e-05,
+      "loss": 0.8669,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643314374697377e-05,
+      "loss": 0.8346,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640314778332148e-05,
+      "loss": 0.8685,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637312111474237e-05,
+      "loss": 0.8304,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634306375190695e-05,
+      "loss": 0.8377,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863129757054966e-05,
+      "loss": 0.8518,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628285698620376e-05,
+      "loss": 0.8483,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625270760473164e-05,
+      "loss": 0.8281,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622252757179423e-05,
+      "loss": 0.3027,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619231689811673e-05,
+      "loss": 0.8639,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86162075594435e-05,
+      "loss": 0.8546,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613180367149577e-05,
+      "loss": 0.2453,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610150114005678e-05,
+      "loss": 0.8545,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8607116801088658e-05,
+      "loss": 0.8906,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604080429476457e-05,
+      "loss": 0.8991,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860104100024811e-05,
+      "loss": 0.876,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597998514483724e-05,
+      "loss": 0.2744,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594952973264512e-05,
+      "loss": 0.8676,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859190437767276e-05,
+      "loss": 0.8186,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588852728791833e-05,
+      "loss": 0.8905,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585798027706196e-05,
+      "loss": 0.8888,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85827402755014e-05,
+      "loss": 0.8209,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579679473264057e-05,
+      "loss": 0.8542,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576615622081893e-05,
+      "loss": 0.8686,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573548723043698e-05,
+      "loss": 0.8469,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570478777239348e-05,
+      "loss": 0.8779,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567405785759802e-05,
+      "loss": 0.8589,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856432974969711e-05,
+      "loss": 0.8506,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561250670144395e-05,
+      "loss": 0.8981,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558168548195858e-05,
+      "loss": 0.8174,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555083384946796e-05,
+      "loss": 0.8236,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551995181493574e-05,
+      "loss": 0.8837,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854890393893364e-05,
+      "loss": 0.8766,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545809658365522e-05,
+      "loss": 0.8525,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542712340888834e-05,
+      "loss": 0.8033,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853961198760426e-05,
+      "loss": 0.8956,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536508599613566e-05,
+      "loss": 0.8203,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533402178019596e-05,
+      "loss": 0.8298,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853029272392628e-05,
+      "loss": 0.804,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527180238438612e-05,
+      "loss": 0.8226,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524064722662672e-05,
+      "loss": 0.8703,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520946177705622e-05,
+      "loss": 0.968,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517824604675686e-05,
+      "loss": 0.8643,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514700004682172e-05,
+      "loss": 0.8548,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511572378835466e-05,
+      "loss": 0.8846,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850844172824702e-05,
+      "loss": 0.858,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505308054029378e-05,
+      "loss": 0.8884,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 0.9097,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499031639161997e-05,
+      "loss": 0.2708,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495888900742694e-05,
+      "loss": 0.8783,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492743143155064e-05,
+      "loss": 0.8804,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489594367517015e-05,
+      "loss": 0.9101,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848644257494751e-05,
+      "loss": 0.8883,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483287766566607e-05,
+      "loss": 0.8493,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480129943495416e-05,
+      "loss": 0.886,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847696910685613e-05,
+      "loss": 0.8927,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473805257772017e-05,
+      "loss": 0.914,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470638397367397e-05,
+      "loss": 0.8558,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467468526767677e-05,
+      "loss": 0.8772,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464295647099323e-05,
+      "loss": 0.9143,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461119759489884e-05,
+      "loss": 0.8479,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845794086506796e-05,
+      "loss": 0.865,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454758964963234e-05,
+      "loss": 0.8906,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451574060306452e-05,
+      "loss": 0.8079,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448386152229428e-05,
+      "loss": 0.8348,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844519524186504e-05,
+      "loss": 0.8932,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844200133034724e-05,
+      "loss": 0.8344,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8438804418811038e-05,
+      "loss": 0.8587,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435604508392517e-05,
+      "loss": 0.8231,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432401600228823e-05,
+      "loss": 0.8338,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429195695458166e-05,
+      "loss": 0.8289,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425986795219828e-05,
+      "loss": 0.8743,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422774900654142e-05,
+      "loss": 0.8635,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841956001290252e-05,
+      "loss": 0.8755,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416342133107426e-05,
+      "loss": 0.8388,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413121262412395e-05,
+      "loss": 0.8734,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840989740196202e-05,
+      "loss": 0.8038,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406670552901958e-05,
+      "loss": 0.8649,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840344071637893e-05,
+      "loss": 0.8851,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400207893540714e-05,
+      "loss": 0.8781,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396972085536155e-05,
+      "loss": 0.8187,
+      "step": 1127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393733293515158e-05,
+      "loss": 0.8767,
+      "step": 1128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390491518628683e-05,
+      "loss": 0.8904,
+      "step": 1129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387246762028758e-05,
+      "loss": 0.8459,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383999024868456e-05,
+      "loss": 0.8458,
+      "step": 1131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838074830830193e-05,
+      "loss": 0.8799,
+      "step": 1132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837749461348438e-05,
+      "loss": 0.8884,
+      "step": 1133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837423794157206e-05,
+      "loss": 0.9257,
+      "step": 1134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837097829372229e-05,
+      "loss": 0.8639,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836771567109344e-05,
+      "loss": 0.8001,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364450074844956e-05,
+      "loss": 0.8949,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836118150613732e-05,
+      "loss": 0.8805,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357909966132065e-05,
+      "loss": 0.8759,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835463545599181e-05,
+      "loss": 0.9088,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351357976880205e-05,
+      "loss": 0.8414,
+      "step": 1141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348077529961957e-05,
+      "loss": 0.8143,
+      "step": 1142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834479411640284e-05,
+      "loss": 0.8318,
+      "step": 1143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834150773736967e-05,
+      "loss": 0.8562,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833821839403032e-05,
+      "loss": 0.8463,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334926087553727e-05,
+      "loss": 0.9075,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331630819109857e-05,
+      "loss": 0.8375,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328332589869756e-05,
+      "loss": 0.8477,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325031401005505e-05,
+      "loss": 0.8232,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832172725369024e-05,
+      "loss": 0.2767,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318420149098153e-05,
+      "loss": 0.8157,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315110088404484e-05,
+      "loss": 0.8665,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311797072785516e-05,
+      "loss": 0.8268,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308481103418597e-05,
+      "loss": 0.2693,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305162181482113e-05,
+      "loss": 0.8573,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301840308155507e-05,
+      "loss": 0.8489,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298515484619264e-05,
+      "loss": 0.84,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295187712054922e-05,
+      "loss": 0.8191,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829185699164506e-05,
+      "loss": 0.8431,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288523324573314e-05,
+      "loss": 0.8356,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285186712024367e-05,
+      "loss": 0.8464,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281847155183944e-05,
+      "loss": 0.8595,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278504655238808e-05,
+      "loss": 0.9108,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275159213376783e-05,
+      "loss": 0.8475,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271810830786738e-05,
+      "loss": 0.8813,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826845950865857e-05,
+      "loss": 0.8694,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826510524818324e-05,
+      "loss": 0.8761,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261748050552745e-05,
+      "loss": 0.8264,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258387916960124e-05,
+      "loss": 0.8314,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255024848599466e-05,
+      "loss": 0.8931,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251658846665893e-05,
+      "loss": 0.8573,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248289912355575e-05,
+      "loss": 0.8555,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244918046865724e-05,
+      "loss": 0.8101,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82415432513946e-05,
+      "loss": 0.8508,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823816552714149e-05,
+      "loss": 0.7404,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823478487530673e-05,
+      "loss": 0.8127,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231401297091703e-05,
+      "loss": 0.8227,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228014793698817e-05,
+      "loss": 0.8849,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8224625366331534e-05,
+      "loss": 0.8336,
+      "step": 1179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221233016194343e-05,
+      "loss": 0.8287,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821783774449278e-05,
+      "loss": 0.8232,
+      "step": 1181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214439552433417e-05,
+      "loss": 0.894,
+      "step": 1182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211038441223866e-05,
+      "loss": 0.909,
+      "step": 1183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207634412072765e-05,
+      "loss": 0.8695,
+      "step": 1184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204227466189802e-05,
+      "loss": 0.8315,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200817604785702e-05,
+      "loss": 0.8209,
+      "step": 1186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197404829072214e-05,
+      "loss": 0.8717,
+      "step": 1187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193989140262133e-05,
+      "loss": 0.8473,
+      "step": 1188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190570539569284e-05,
+      "loss": 0.7708,
+      "step": 1189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818714902820853e-05,
+      "loss": 0.8467,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818372460739577e-05,
+      "loss": 0.8656,
+      "step": 1191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818029727834793e-05,
+      "loss": 0.8711,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176867042282972e-05,
+      "loss": 0.8534,
+      "step": 1193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81734339004199e-05,
+      "loss": 0.8541,
+      "step": 1194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169997853978742e-05,
+      "loss": 0.8357,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166558904180547e-05,
+      "loss": 0.8393,
+      "step": 1196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163117052247427e-05,
+      "loss": 0.2731,
+      "step": 1197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159672299402493e-05,
+      "loss": 0.8274,
+      "step": 1198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156224646869907e-05,
+      "loss": 0.8559,
+      "step": 1199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815277409587485e-05,
+      "loss": 0.8552,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149320647643544e-05,
+      "loss": 0.8633,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145864303403224e-05,
+      "loss": 0.8527,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814240506438218e-05,
+      "loss": 0.8381,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138942931809702e-05,
+      "loss": 0.9041,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135477906916127e-05,
+      "loss": 0.8249,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132009990932816e-05,
+      "loss": 0.914,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8128539185092157e-05,
+      "loss": 0.8464,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125065490627556e-05,
+      "loss": 0.8632,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812158890877346e-05,
+      "loss": 0.8322,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118109440765334e-05,
+      "loss": 0.8039,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811462708783967e-05,
+      "loss": 0.8733,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111141851233986e-05,
+      "loss": 0.8271,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810765373218682e-05,
+      "loss": 0.8832,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104162731937746e-05,
+      "loss": 0.9038,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100668851727346e-05,
+      "loss": 0.8767,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097172092797244e-05,
+      "loss": 0.2406,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093672456390066e-05,
+      "loss": 0.8453,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.8674,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086664556120155e-05,
+      "loss": 0.8879,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083156294747807e-05,
+      "loss": 0.835,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807964516087916e-05,
+      "loss": 0.9288,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807613115576195e-05,
+      "loss": 0.8602,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807261428064495e-05,
+      "loss": 0.8371,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069094536777938e-05,
+      "loss": 0.8987,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065571925411727e-05,
+      "loss": 0.2533,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806204644779814e-05,
+      "loss": 0.8198,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058518105190016e-05,
+      "loss": 0.9051,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054986898841217e-05,
+      "loss": 0.2573,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051452830006616e-05,
+      "loss": 0.8377,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804791589994212e-05,
+      "loss": 0.8644,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804437610990463e-05,
+      "loss": 0.8597,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804083346115208e-05,
+      "loss": 0.2519,
+      "step": 1232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037287954943415e-05,
+      "loss": 0.8266,
+      "step": 1233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033739592538598e-05,
+      "loss": 0.8693,
+      "step": 1234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030188375198593e-05,
+      "loss": 0.8319,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80266343041854e-05,
+      "loss": 0.8537,
+      "step": 1236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802307738076202e-05,
+      "loss": 0.8376,
+      "step": 1237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019517606192467e-05,
+      "loss": 0.8387,
+      "step": 1238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015954981741773e-05,
+      "loss": 0.8714,
+      "step": 1239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012389508675984e-05,
+      "loss": 0.8644,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008821188262153e-05,
+      "loss": 0.8207,
+      "step": 1241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800525002176835e-05,
+      "loss": 0.2662,
+      "step": 1242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800167601046365e-05,
+      "loss": 0.896,
+      "step": 1243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998099155618147e-05,
+      "loss": 0.8242,
+      "step": 1244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994519458502936e-05,
+      "loss": 0.816,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990936920390133e-05,
+      "loss": 0.2739,
+      "step": 1246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987351542552852e-05,
+      "loss": 0.8645,
+      "step": 1247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798376332626523e-05,
+      "loss": 0.3032,
+      "step": 1248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980172272802398e-05,
+      "loss": 0.8843,
+      "step": 1249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976578383440502e-05,
+      "loss": 0.8438,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972981659456702e-05,
+      "loss": 0.8722,
+      "step": 1251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969382102129153e-05,
+      "loss": 0.8188,
+      "step": 1252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965779712737025e-05,
+      "loss": 0.8981,
+      "step": 1253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962174492560492e-05,
+      "loss": 0.8182,
+      "step": 1254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958566442880736e-05,
+      "loss": 0.8087,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954955564979944e-05,
+      "loss": 0.8812,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79513418601413e-05,
+      "loss": 0.8547,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794772532964901e-05,
+      "loss": 0.822,
+      "step": 1258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944105974788267e-05,
+      "loss": 0.8853,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940483796845275e-05,
+      "loss": 0.8448,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793685879710724e-05,
+      "loss": 0.8601,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933230976862375e-05,
+      "loss": 0.8357,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792960033739989e-05,
+      "loss": 0.8309,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925966880009998e-05,
+      "loss": 0.8015,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922330605983913e-05,
+      "loss": 0.8415,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918691516613855e-05,
+      "loss": 0.8426,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915049613193037e-05,
+      "loss": 0.8154,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911404897015674e-05,
+      "loss": 0.2565,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907757369376984e-05,
+      "loss": 0.8837,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904107031573186e-05,
+      "loss": 0.865,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790045388490149e-05,
+      "loss": 0.7747,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896797930660107e-05,
+      "loss": 0.8377,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789313917014825e-05,
+      "loss": 0.8995,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889477604666124e-05,
+      "loss": 0.8207,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788581323551493e-05,
+      "loss": 0.2458,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788214606399687e-05,
+      "loss": 0.8504,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7878476091415146e-05,
+      "loss": 0.8257,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787480331907394e-05,
+      "loss": 0.811,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871127748278443e-05,
+      "loss": 0.8164,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867449380334834e-05,
+      "loss": 0.8419,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786376821655029e-05,
+      "loss": 0.8511,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860084258232978e-05,
+      "loss": 0.8359,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856397506692062e-05,
+      "loss": 0.756,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785270796323769e-05,
+      "loss": 0.8305,
+      "step": 1284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849015629181016e-05,
+      "loss": 0.8687,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845320505834176e-05,
+      "loss": 0.876,
+      "step": 1286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841622594510293e-05,
+      "loss": 0.8655,
+      "step": 1287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783792189652349e-05,
+      "loss": 0.849,
+      "step": 1288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783421841318888e-05,
+      "loss": 0.8271,
+      "step": 1289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830512145822564e-05,
+      "loss": 0.8381,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782680309574163e-05,
+      "loss": 0.845,
+      "step": 1291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782309126426415e-05,
+      "loss": 0.8247,
+      "step": 1292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819376652709198e-05,
+      "loss": 0.8787,
+      "step": 1293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815659262396825e-05,
+      "loss": 0.8108,
+      "step": 1294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811939094648076e-05,
+      "loss": 0.8111,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808216150784977e-05,
+      "loss": 0.8205,
+      "step": 1296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780449043213055e-05,
+      "loss": 0.8256,
+      "step": 1297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780076194000879e-05,
+      "loss": 0.8204,
+      "step": 1298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797030675744682e-05,
+      "loss": 0.8181,
+      "step": 1299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793296640664205e-05,
+      "loss": 0.832,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789559836094314e-05,
+      "loss": 0.8533,
+      "step": 1301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785820263362953e-05,
+      "loss": 0.8105,
+      "step": 1302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782077923799043e-05,
+      "loss": 0.8046,
+      "step": 1303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778332818732492e-05,
+      "loss": 0.8554,
+      "step": 1304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777458494949419e-05,
+      "loss": 0.8249,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770834317416018e-05,
+      "loss": 0.9287,
+      "step": 1306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776708092383083e-05,
+      "loss": 0.8348,
+      "step": 1307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776332477007245e-05,
+      "loss": 0.8687,
+      "step": 1308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759565857475708e-05,
+      "loss": 0.8241,
+      "step": 1309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755804187376398e-05,
+      "loss": 0.8802,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77520397611113e-05,
+      "loss": 0.8942,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748272580018168e-05,
+      "loss": 0.885,
+      "step": 1312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744502645435736e-05,
+      "loss": 0.8394,
+      "step": 1313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740729958703725e-05,
+      "loss": 0.8052,
+      "step": 1314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736954521162822e-05,
+      "loss": 0.8548,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733176334154704e-05,
+      "loss": 0.8723,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729395399022012e-05,
+      "loss": 0.8119,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772561171710837e-05,
+      "loss": 0.8358,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721825289758383e-05,
+      "loss": 0.871,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771803611831762e-05,
+      "loss": 0.8202,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771424420413264e-05,
+      "loss": 0.7881,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710449548550964e-05,
+      "loss": 0.8306,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770665215292109e-05,
+      "loss": 0.8102,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702852018592493e-05,
+      "loss": 0.8777,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769904914691562e-05,
+      "loss": 0.825,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695243539241893e-05,
+      "loss": 0.853,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76914351969237e-05,
+      "loss": 0.8487,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687624121314406e-05,
+      "loss": 0.8014,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683810313768345e-05,
+      "loss": 0.834,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679993775640824e-05,
+      "loss": 0.8401,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676174508288118e-05,
+      "loss": 0.8942,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672352513067476e-05,
+      "loss": 0.8643,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668527791337108e-05,
+      "loss": 0.8014,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664700344456198e-05,
+      "loss": 0.8024,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660870173784906e-05,
+      "loss": 0.8252,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657037280684352e-05,
+      "loss": 0.9172,
+      "step": 1336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653201666516614e-05,
+      "loss": 0.8512,
+      "step": 1337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764936333264476e-05,
+      "loss": 0.8093,
+      "step": 1338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76455222804328e-05,
+      "loss": 0.8391,
+      "step": 1339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641678511245733e-05,
+      "loss": 0.8759,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637832026449505e-05,
+      "loss": 0.8717,
+      "step": 1341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763398282741103e-05,
+      "loss": 0.8312,
+      "step": 1342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76301309154982e-05,
+      "loss": 0.2507,
+      "step": 1343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762627629207986e-05,
+      "loss": 0.8651,
+      "step": 1344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622418958525816e-05,
+      "loss": 0.9049,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761855891620684e-05,
+      "loss": 0.8401,
+      "step": 1346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761469616649468e-05,
+      "loss": 0.8208,
+      "step": 1347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610830710762022e-05,
+      "loss": 0.8439,
+      "step": 1348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606962550382533e-05,
+      "loss": 0.2684,
+      "step": 1349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603091686730827e-05,
+      "loss": 0.7875,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759921812118249e-05,
+      "loss": 0.8623,
+      "step": 1351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595341855114065e-05,
+      "loss": 0.7944,
+      "step": 1352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591462889903052e-05,
+      "loss": 0.8894,
+      "step": 1353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758758122692791e-05,
+      "loss": 0.8006,
+      "step": 1354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583696867568063e-05,
+      "loss": 0.8295,
+      "step": 1355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579809813203883e-05,
+      "loss": 0.8302,
+      "step": 1356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757592006521671e-05,
+      "loss": 0.8219,
+      "step": 1357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757202762498883e-05,
+      "loss": 0.2878,
+      "step": 1358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568132493903505e-05,
+      "loss": 0.7712,
+      "step": 1359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564234673344928e-05,
+      "loss": 0.7756,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560334164698264e-05,
+      "loss": 0.895,
+      "step": 1361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556430969349633e-05,
+      "loss": 0.8163,
+      "step": 1362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75525250886861e-05,
+      "loss": 0.8993,
+      "step": 1363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548616524095697e-05,
+      "loss": 0.8869,
+      "step": 1364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75447052769674e-05,
+      "loss": 0.8667,
+      "step": 1365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540791348691144e-05,
+      "loss": 0.8426,
+      "step": 1366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536874740657807e-05,
+      "loss": 0.823,
+      "step": 1367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532955454259236e-05,
+      "loss": 0.8337,
+      "step": 1368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529033490888215e-05,
+      "loss": 0.8124,
+      "step": 1369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752510885193849e-05,
+      "loss": 0.9377,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521181538804745e-05,
+      "loss": 0.8598,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751725155288263e-05,
+      "loss": 0.8898,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513318895568734e-05,
+      "loss": 0.8289,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509383568260597e-05,
+      "loss": 0.8454,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505445572356707e-05,
+      "loss": 0.8956,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750150490925651e-05,
+      "loss": 0.8183,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497561580360382e-05,
+      "loss": 0.8553,
+      "step": 1377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749361558706966e-05,
+      "loss": 0.8517,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748966693078663e-05,
+      "loss": 0.8424,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485715612914513e-05,
+      "loss": 0.8475,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748176163485748e-05,
+      "loss": 0.7713,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477804998020658e-05,
+      "loss": 0.8996,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74738457038101e-05,
+      "loss": 0.8832,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469883753632817e-05,
+      "loss": 0.8865,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746591914889676e-05,
+      "loss": 0.8955,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7461951891010822e-05,
+      "loss": 0.8449,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745798198138484e-05,
+      "loss": 0.8515,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74540094214296e-05,
+      "loss": 0.8015,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450034212556813e-05,
+      "loss": 0.8869,
+      "step": 1389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446056356179145e-05,
+      "loss": 0.8032,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74420758537102e-05,
+      "loss": 0.8334,
+      "step": 1391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438092706564527e-05,
+      "loss": 0.8341,
+      "step": 1392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74341069161576e-05,
+      "loss": 0.8309,
+      "step": 1393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743011848390585e-05,
+      "loss": 0.8219,
+      "step": 1394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742612741122664e-05,
+      "loss": 0.815,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742213369953826e-05,
+      "loss": 0.8327,
+      "step": 1396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418137350259955e-05,
+      "loss": 0.8641,
+      "step": 1397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74141383648119e-05,
+      "loss": 0.8721,
+      "step": 1398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741013674461521e-05,
+      "loss": 0.8418,
+      "step": 1399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406132491091922e-05,
+      "loss": 0.8739,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402125605665033e-05,
+      "loss": 0.7913,
+      "step": 1401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398116089758454e-05,
+      "loss": 0.8131,
+      "step": 1402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394103944797044e-05,
+      "loss": 0.9063,
+      "step": 1403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 0.8218,
+      "step": 1404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386071773413817e-05,
+      "loss": 0.87,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382051749846376e-05,
+      "loss": 0.8059,
+      "step": 1406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378029102932858e-05,
+      "loss": 0.8058,
+      "step": 1407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737400383410278e-05,
+      "loss": 0.8301,
+      "step": 1408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73699759447866e-05,
+      "loss": 0.8305,
+      "step": 1409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73659454364157e-05,
+      "loss": 0.2848,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736191231042239e-05,
+      "loss": 0.8818,
+      "step": 1411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735787656823992e-05,
+      "loss": 0.8045,
+      "step": 1412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735383821130246e-05,
+      "loss": 0.856,
+      "step": 1413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349797241045115e-05,
+      "loss": 0.8389,
+      "step": 1414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345753658903918e-05,
+      "loss": 0.3074,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734170746631583e-05,
+      "loss": 0.8764,
+      "step": 1416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337658664718738e-05,
+      "loss": 0.7921,
+      "step": 1417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333607255551455e-05,
+      "loss": 0.8391,
+      "step": 1418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329553240253725e-05,
+      "loss": 0.8466,
+      "step": 1419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325496620266216e-05,
+      "loss": 0.789,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732143739703052e-05,
+      "loss": 0.8476,
+      "step": 1421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317375571989158e-05,
+      "loss": 0.8206,
+      "step": 1422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313311146585573e-05,
+      "loss": 0.8682,
+      "step": 1423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730924412226413e-05,
+      "loss": 0.8195,
+      "step": 1424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305174500470122e-05,
+      "loss": 0.8571,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730110228264976e-05,
+      "loss": 0.8169,
+      "step": 1426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297027470250186e-05,
+      "loss": 0.82,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729295006471945e-05,
+      "loss": 0.9227,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728887006750654e-05,
+      "loss": 0.8573,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728478748006136e-05,
+      "loss": 0.2719,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280702303834717e-05,
+      "loss": 0.7878,
+      "step": 1431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276614540278368e-05,
+      "loss": 0.7833,
+      "step": 1432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272524190844966e-05,
+      "loss": 0.8545,
+      "step": 1433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726843125698809e-05,
+      "loss": 0.8401,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264335740162244e-05,
+      "loss": 0.8407,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726023764182284e-05,
+      "loss": 0.8308,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256136963426212e-05,
+      "loss": 0.8607,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725203370642961e-05,
+      "loss": 0.7964,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72479278722912e-05,
+      "loss": 0.86,
+      "step": 1439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724381946247007e-05,
+      "loss": 0.8737,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239708478426217e-05,
+      "loss": 0.8351,
+      "step": 1441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235594921620546e-05,
+      "loss": 0.8695,
+      "step": 1442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723147879351489e-05,
+      "loss": 0.843,
+      "step": 1443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227360095571992e-05,
+      "loss": 0.8713,
+      "step": 1444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223238829255503e-05,
+      "loss": 0.8231,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721911499602999e-05,
+      "loss": 0.8595,
+      "step": 1446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721498859736093e-05,
+      "loss": 0.8229,
+      "step": 1447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721085963471472e-05,
+      "loss": 0.8617,
+      "step": 1448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206728109558652e-05,
+      "loss": 0.8022,
+      "step": 1449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720259402336095e-05,
+      "loss": 0.8199,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198457377590727e-05,
+      "loss": 0.864,
+      "step": 1451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719431817371802e-05,
+      "loss": 0.8646,
+      "step": 1452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719017641321377e-05,
+      "loss": 0.8508,
+      "step": 1453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186032097549822e-05,
+      "loss": 0.8025,
+      "step": 1454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718188522819894e-05,
+      "loss": 0.8693,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717773580663479e-05,
+      "loss": 0.8008,
+      "step": 1456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173583834331937e-05,
+      "loss": 0.8376,
+      "step": 1457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169429312765863e-05,
+      "loss": 0.7662,
+      "step": 1458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716527224341296e-05,
+      "loss": 0.9109,
+      "step": 1459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161112627750503e-05,
+      "loss": 0.8436,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71569504672567e-05,
+      "loss": 0.8456,
+      "step": 1461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152785763410648e-05,
+      "loss": 0.8778,
+      "step": 1462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148618517692344e-05,
+      "loss": 0.8326,
+      "step": 1463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144448731582698e-05,
+      "loss": 0.8975,
+      "step": 1464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140276406563523e-05,
+      "loss": 0.8336,
+      "step": 1465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136101544117526e-05,
+      "loss": 0.8478,
+      "step": 1466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713192414572832e-05,
+      "loss": 0.8787,
+      "step": 1467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712774421288042e-05,
+      "loss": 0.8206,
+      "step": 1468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712356174705924e-05,
+      "loss": 0.8134,
+      "step": 1469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71193767497511e-05,
+      "loss": 0.8343,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711518922244321e-05,
+      "loss": 0.8616,
+      "step": 1471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110999166623683e-05,
+      "loss": 0.838,
+      "step": 1472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106806583781535e-05,
+      "loss": 0.8163,
+      "step": 1473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102611475406676e-05,
+      "loss": 0.8571,
+      "step": 1474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709841384298991e-05,
+      "loss": 0.8269,
+      "step": 1475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094213688022947e-05,
+      "loss": 0.8167,
+      "step": 1476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090011011998385e-05,
+      "loss": 0.8478,
+      "step": 1477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085805816409723e-05,
+      "loss": 0.8773,
+      "step": 1478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081598102751355e-05,
+      "loss": 0.7809,
+      "step": 1479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707738787251856e-05,
+      "loss": 0.8568,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707317512720753e-05,
+      "loss": 0.8223,
+      "step": 1481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068959868315334e-05,
+      "loss": 0.8496,
+      "step": 1482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7064742097339947e-05,
+      "loss": 0.8599,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060521815780225e-05,
+      "loss": 0.8755,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056299025135925e-05,
+      "loss": 0.8418,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705207372690769e-05,
+      "loss": 0.748,
+      "step": 1486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047845922597054e-05,
+      "loss": 0.8715,
+      "step": 1487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043615613706448e-05,
+      "loss": 0.8409,
+      "step": 1488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039382801739187e-05,
+      "loss": 0.8372,
+      "step": 1489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703514748819948e-05,
+      "loss": 0.9214,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030909674592422e-05,
+      "loss": 0.8464,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026669362423995e-05,
+      "loss": 0.8808,
+      "step": 1492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022426553201076e-05,
+      "loss": 0.8472,
+      "step": 1493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7018181248431416e-05,
+      "loss": 0.8816,
+      "step": 1494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701393344962367e-05,
+      "loss": 0.8337,
+      "step": 1495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700968315828736e-05,
+      "loss": 0.8327,
+      "step": 1496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700543037593291e-05,
+      "loss": 0.7865,
+      "step": 1497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001175104071625e-05,
+      "loss": 0.788,
+      "step": 1498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996917344215697e-05,
+      "loss": 0.8306,
+      "step": 1499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992657097878184e-05,
+      "loss": 0.8992,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988394366573056e-05,
+      "loss": 0.8223,
+      "step": 1501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984129151815147e-05,
+      "loss": 0.8048,
+      "step": 1502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979861455120174e-05,
+      "loss": 0.8168,
+      "step": 1503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975591278004747e-05,
+      "loss": 0.7551,
+      "step": 1504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971318621986346e-05,
+      "loss": 0.9002,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967043488583342e-05,
+      "loss": 0.8774,
+      "step": 1506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696276587931498e-05,
+      "loss": 0.8212,
+      "step": 1507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695848579570138e-05,
+      "loss": 0.8755,
+      "step": 1508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954203239263557e-05,
+      "loss": 0.8283,
+      "step": 1509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949918211523386e-05,
+      "loss": 0.836,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945630714003636e-05,
+      "loss": 0.8438,
+      "step": 1511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941340748227942e-05,
+      "loss": 0.8548,
+      "step": 1512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937048315720827e-05,
+      "loss": 0.8739,
+      "step": 1513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932753418007683e-05,
+      "loss": 0.8326,
+      "step": 1514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928456056614776e-05,
+      "loss": 0.8779,
+      "step": 1515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924156233069253e-05,
+      "loss": 0.8269,
+      "step": 1516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919853948899137e-05,
+      "loss": 0.8332,
+      "step": 1517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691554920563332e-05,
+      "loss": 0.8799,
+      "step": 1518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911242004801572e-05,
+      "loss": 0.8656,
+      "step": 1519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690693234793453e-05,
+      "loss": 0.8519,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902620236563723e-05,
+      "loss": 0.2561,
+      "step": 1521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898305672221523e-05,
+      "loss": 0.8172,
+      "step": 1522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6893988656441195e-05,
+      "loss": 0.8543,
+      "step": 1523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688966919075687e-05,
+      "loss": 0.8811,
+      "step": 1524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688534727670354e-05,
+      "loss": 0.8441,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881022915817088e-05,
+      "loss": 0.855,
+      "step": 1526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687669610963425e-05,
+      "loss": 0.8938,
+      "step": 1527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687236685969263e-05,
+      "loss": 0.8836,
+      "step": 1528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686803516753071e-05,
+      "loss": 0.8618,
+      "step": 1529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686370103468783e-05,
+      "loss": 0.8475,
+      "step": 1530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859364462704214e-05,
+      "loss": 0.284,
+      "step": 1531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855025453120935e-05,
+      "loss": 0.8347,
+      "step": 1532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850684007479935e-05,
+      "loss": 0.8739,
+      "step": 1533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684634012732403e-05,
+      "loss": 0.8557,
+      "step": 1534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68419938141969e-05,
+      "loss": 0.8568,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837645069643075e-05,
+      "loss": 0.8488,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833293895207968e-05,
+      "loss": 0.8541,
+      "step": 1537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682894029243785e-05,
+      "loss": 0.8607,
+      "step": 1538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682458426287985e-05,
+      "loss": 0.8357,
+      "step": 1539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682022580808196e-05,
+      "loss": 0.8165,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6815864929593043e-05,
+      "loss": 0.7737,
+      "step": 1541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811501628962807e-05,
+      "loss": 0.809,
+      "step": 1542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807135907741838e-05,
+      "loss": 0.7911,
+      "step": 1543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680276776748157e-05,
+      "loss": 0.8436,
+      "step": 1544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67983972097343e-05,
+      "loss": 0.8969,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794024236053186e-05,
+      "loss": 0.8388,
+      "step": 1546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789648847992247e-05,
+      "loss": 0.8531,
+      "step": 1547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785271047106352e-05,
+      "loss": 0.8986,
+      "step": 1548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6780890834951235e-05,
+      "loss": 0.8453,
+      "step": 1549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776508213083484e-05,
+      "loss": 0.8211,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772123183060538e-05,
+      "loss": 0.8208,
+      "step": 1551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767735746440705e-05,
+      "loss": 0.8975,
+      "step": 1552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763345904783134e-05,
+      "loss": 0.8461,
+      "step": 1553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758953659647838e-05,
+      "loss": 0.7714,
+      "step": 1554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675455901259568e-05,
+      "loss": 0.8578,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750161965188376e-05,
+      "loss": 0.8503,
+      "step": 1556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67457625189885e-05,
+      "loss": 0.894,
+      "step": 1557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741360675559475e-05,
+      "loss": 0.261,
+      "step": 1558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736956436465573e-05,
+      "loss": 0.823,
+      "step": 1559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732549803271922e-05,
+      "loss": 0.8436,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728140777544498e-05,
+      "loss": 0.8821,
+      "step": 1561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672372936085013e-05,
+      "loss": 0.832,
+      "step": 1562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719315554756496e-05,
+      "loss": 0.2654,
+      "step": 1563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714899360832118e-05,
+      "loss": 0.8737,
+      "step": 1564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710480780646375e-05,
+      "loss": 0.8764,
+      "step": 1565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706059815769483e-05,
+      "loss": 0.8167,
+      "step": 1566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670163646777252e-05,
+      "loss": 0.7958,
+      "step": 1567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697210738227402e-05,
+      "loss": 0.8309,
+      "step": 1568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6692782628706886e-05,
+      "loss": 0.845,
+      "step": 1569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688352140784587e-05,
+      "loss": 0.8285,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683919276034956e-05,
+      "loss": 0.8509,
+      "step": 1571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679484036033295e-05,
+      "loss": 0.8902,
+      "step": 1572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667504642235574e-05,
+      "loss": 0.8231,
+      "step": 1573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667060643657929e-05,
+      "loss": 0.8639,
+      "step": 1574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666164080281767e-05,
+      "loss": 0.8426,
+      "step": 1575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661719355041838e-05,
+      "loss": 0.8098,
+      "step": 1576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657272262439027e-05,
+      "loss": 0.8132,
+      "step": 1577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652822804053683e-05,
+      "loss": 0.7839,
+      "step": 1578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648370981467006e-05,
+      "loss": 0.8549,
+      "step": 1579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643916796261025e-05,
+      "loss": 0.8372,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663946025001862e-05,
+      "loss": 0.8367,
+      "step": 1581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635001344323506e-05,
+      "loss": 0.8514,
+      "step": 1582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630540080760233e-05,
+      "loss": 0.8214,
+      "step": 1583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66260764609142e-05,
+      "loss": 0.8084,
+      "step": 1584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621610486371625e-05,
+      "loss": 0.7493,
+      "step": 1585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617142158719577e-05,
+      "loss": 0.8617,
+      "step": 1586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661267147954596e-05,
+      "loss": 0.8728,
+      "step": 1587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660819845043951e-05,
+      "loss": 0.8658,
+      "step": 1588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603723072989796e-05,
+      "loss": 0.8167,
+      "step": 1589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659924534878723e-05,
+      "loss": 0.8529,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594765279423044e-05,
+      "loss": 0.8694,
+      "step": 1591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659028286648932e-05,
+      "loss": 0.8813,
+      "step": 1592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658579811157896e-05,
+      "loss": 0.8823,
+      "step": 1593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658131101628571e-05,
+      "loss": 0.8261,
+      "step": 1594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576821582204133e-05,
+      "loss": 0.8141,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572329810929635e-05,
+      "loss": 0.3202,
+      "step": 1596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567835704058444e-05,
+      "loss": 0.8267,
+      "step": 1597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656333926318763e-05,
+      "loss": 0.8515,
+      "step": 1598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655884048991508e-05,
+      "loss": 0.8751,
+      "step": 1599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655433938583951e-05,
+      "loss": 0.8364,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6549835952560477e-05,
+      "loss": 0.8505,
+      "step": 1601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545330191678356e-05,
+      "loss": 0.8427,
+      "step": 1602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654082210479435e-05,
+      "loss": 0.8432,
+      "step": 1603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653631169351049e-05,
+      "loss": 0.8438,
+      "step": 1604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531798959429633e-05,
+      "loss": 0.7996,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527283904155457e-05,
+      "loss": 0.831,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652276652929247e-05,
+      "loss": 0.8068,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651824683644601e-05,
+      "loss": 0.8205,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513724827222225e-05,
+      "loss": 0.8532,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509200503228092e-05,
+      "loss": 0.8864,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504673866071416e-05,
+      "loss": 0.8599,
+      "step": 1611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650014491736082e-05,
+      "loss": 0.8479,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495613658705746e-05,
+      "loss": 0.8364,
+      "step": 1613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491080091716457e-05,
+      "loss": 0.7907,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486544218004047e-05,
+      "loss": 0.8075,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482006039180406e-05,
+      "loss": 0.8558,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477465556858274e-05,
+      "loss": 0.8083,
+      "step": 1617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472922772651182e-05,
+      "loss": 0.8817,
+      "step": 1618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468377688173497e-05,
+      "loss": 0.7914,
+      "step": 1619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463830305040395e-05,
+      "loss": 0.8526,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459280624867876e-05,
+      "loss": 0.8604,
+      "step": 1621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454728649272743e-05,
+      "loss": 0.8512,
+      "step": 1622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450174379872626e-05,
+      "loss": 0.7886,
+      "step": 1623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445617818285974e-05,
+      "loss": 0.9039,
+      "step": 1624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441058966132037e-05,
+      "loss": 0.8811,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436497825030886e-05,
+      "loss": 0.8541,
+      "step": 1626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643193439660341e-05,
+      "loss": 0.8025,
+      "step": 1627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427368682471302e-05,
+      "loss": 0.8713,
+      "step": 1628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642280068425707e-05,
+      "loss": 0.8008,
+      "step": 1629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418230403584034e-05,
+      "loss": 0.8608,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413657842076335e-05,
+      "loss": 0.79,
+      "step": 1631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640908300135891e-05,
+      "loss": 0.8281,
+      "step": 1632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404505883057515e-05,
+      "loss": 0.286,
+      "step": 1633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399926488798702e-05,
+      "loss": 0.8759,
+      "step": 1634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395344820209852e-05,
+      "loss": 0.9083,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639076087891914e-05,
+      "loss": 0.8626,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386174666555557e-05,
+      "loss": 0.8434,
+      "step": 1637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638158618474889e-05,
+      "loss": 0.8341,
+      "step": 1638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376995435129745e-05,
+      "loss": 0.8613,
+      "step": 1639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372402419329523e-05,
+      "loss": 0.889,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6367807138980443e-05,
+      "loss": 0.3022,
+      "step": 1641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363209595715518e-05,
+      "loss": 0.8468,
+      "step": 1642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358609791168564e-05,
+      "loss": 0.8558,
+      "step": 1643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354007726974205e-05,
+      "loss": 0.8745,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349403404767875e-05,
+      "loss": 0.8479,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63447968261858e-05,
+      "loss": 0.8319,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634018799286501e-05,
+      "loss": 0.7911,
+      "step": 1647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633557690644334e-05,
+      "loss": 0.879,
+      "step": 1648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330963568559423e-05,
+      "loss": 0.8645,
+      "step": 1649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632634798085269e-05,
+      "loss": 0.8722,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321730144963375e-05,
+      "loss": 0.8677,
+      "step": 1651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631711006253251e-05,
+      "loss": 0.8605,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312487735201926e-05,
+      "loss": 0.29,
+      "step": 1653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630786316461425e-05,
+      "loss": 0.815,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303236352412908e-05,
+      "loss": 0.8175,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298607300242117e-05,
+      "loss": 0.8714,
+      "step": 1656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.62939760097469e-05,
+      "loss": 0.8765,
+      "step": 1657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6289342482573073e-05,
+      "loss": 0.7542,
+      "step": 1658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284706720367237e-05,
+      "loss": 0.8175,
+      "step": 1659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280068724776795e-05,
+      "loss": 0.8994,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275428497449948e-05,
+      "loss": 0.8251,
+      "step": 1661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6270786040035678e-05,
+      "loss": 0.783,
+      "step": 1662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266141354183776e-05,
+      "loss": 0.7985,
+      "step": 1663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261494441544805e-05,
+      "loss": 0.8359,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256845303770138e-05,
+      "loss": 0.8128,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625219394251192e-05,
+      "loss": 0.884,
+      "step": 1666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624754035942311e-05,
+      "loss": 0.8199,
+      "step": 1667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242884556157438e-05,
+      "loss": 0.8486,
+      "step": 1668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6238226534369425e-05,
+      "loss": 0.8394,
+      "step": 1669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233566295714384e-05,
+      "loss": 0.8084,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622890384184842e-05,
+      "loss": 0.8323,
+      "step": 1671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224239174428414e-05,
+      "loss": 0.8846,
+      "step": 1672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621957229511205e-05,
+      "loss": 0.8521,
+      "step": 1673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6214903205557774e-05,
+      "loss": 0.8382,
+      "step": 1674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210231907424846e-05,
+      "loss": 0.8206,
+      "step": 1675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205558402373286e-05,
+      "loss": 0.8229,
+      "step": 1676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200882692063916e-05,
+      "loss": 0.8753,
+      "step": 1677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196204778158334e-05,
+      "loss": 0.8156,
+      "step": 1678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191524662318915e-05,
+      "loss": 0.8666,
+      "step": 1679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618684234620883e-05,
+      "loss": 0.2992,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182157831492022e-05,
+      "loss": 0.8225,
+      "step": 1681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617747111983322e-05,
+      "loss": 0.8223,
+      "step": 1682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617278221289793e-05,
+      "loss": 0.8599,
+      "step": 1683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6168091112352443e-05,
+      "loss": 0.8485,
+      "step": 1684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163397819863824e-05,
+      "loss": 0.8324,
+      "step": 1685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615870233709992e-05,
+      "loss": 0.8494,
+      "step": 1686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154004665729356e-05,
+      "loss": 0.8232,
+      "step": 1687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6149304807421535e-05,
+      "loss": 0.849,
+      "step": 1688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614460276384664e-05,
+      "loss": 0.7972,
+      "step": 1689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139898536675622e-05,
+      "loss": 0.8445,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135192127580217e-05,
+      "loss": 0.838,
+      "step": 1691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130483538232932e-05,
+      "loss": 0.8105,
+      "step": 1692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612577277030705e-05,
+      "loss": 0.832,
+      "step": 1693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612105982547663e-05,
+      "loss": 0.8358,
+      "step": 1694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116344705416496e-05,
+      "loss": 0.8623,
+      "step": 1695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111627411802263e-05,
+      "loss": 0.866,
+      "step": 1696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106907946310297e-05,
+      "loss": 0.866,
+      "step": 1697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102186310617744e-05,
+      "loss": 0.8246,
+      "step": 1698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609746250640253e-05,
+      "loss": 0.8652,
+      "step": 1699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092736535343343e-05,
+      "loss": 0.8551,
+      "step": 1700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088008399119642e-05,
+      "loss": 0.7892,
+      "step": 1701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608327809941165e-05,
+      "loss": 0.8871,
+      "step": 1702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078545637900375e-05,
+      "loss": 0.8341,
+      "step": 1703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607381101626758e-05,
+      "loss": 0.2734,
+      "step": 1704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069074236195795e-05,
+      "loss": 0.2823,
+      "step": 1705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064335299368322e-05,
+      "loss": 0.865,
+      "step": 1706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059594207469233e-05,
+      "loss": 0.8217,
+      "step": 1707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054850962183354e-05,
+      "loss": 0.8354,
+      "step": 1708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605010556519629e-05,
+      "loss": 0.8042,
+      "step": 1709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60453580181944e-05,
+      "loss": 0.7945,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040608322864808e-05,
+      "loss": 0.8523,
+      "step": 1711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603585648089541e-05,
+      "loss": 0.8802,
+      "step": 1712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031102493974858e-05,
+      "loss": 0.8545,
+      "step": 1713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 0.8004,
+      "step": 1714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602158809203871e-05,
+      "loss": 0.7907,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016827680404236e-05,
+      "loss": 0.8267,
+      "step": 1716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601206513058083e-05,
+      "loss": 0.8469,
+      "step": 1717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007300444260963e-05,
+      "loss": 0.8697,
+      "step": 1718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002533623137845e-05,
+      "loss": 0.8494,
+      "step": 1719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997764668905455e-05,
+      "loss": 0.8606,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5992993583258524e-05,
+      "loss": 0.8622,
+      "step": 1721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598822036789255e-05,
+      "loss": 0.8339,
+      "step": 1722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983445024503782e-05,
+      "loss": 0.8405,
+      "step": 1723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978667554789216e-05,
+      "loss": 0.8438,
+      "step": 1724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597388796044662e-05,
+      "loss": 0.8078,
+      "step": 1725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596910624317451e-05,
+      "loss": 0.8263,
+      "step": 1726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596432240467215e-05,
+      "loss": 0.8795,
+      "step": 1727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5959536446639572e-05,
+      "loss": 0.8051,
+      "step": 1728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954748370777546e-05,
+      "loss": 0.8249,
+      "step": 1729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949958178787605e-05,
+      "loss": 0.8477,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594516587237203e-05,
+      "loss": 0.8385,
+      "step": 1731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940371453233853e-05,
+      "loss": 0.8561,
+      "step": 1732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593557492307686e-05,
+      "loss": 0.8362,
+      "step": 1733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930776283605585e-05,
+      "loss": 0.7991,
+      "step": 1734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592597553652531e-05,
+      "loss": 0.9115,
+      "step": 1735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592117268354207e-05,
+      "loss": 0.8496,
+      "step": 1736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591636772636264e-05,
+      "loss": 0.8143,
+      "step": 1737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5911560666694557e-05,
+      "loss": 0.2834,
+      "step": 1738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590675150624609e-05,
+      "loss": 0.8296,
+      "step": 1739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901940246726268e-05,
+      "loss": 0.8283,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589712688984485e-05,
+      "loss": 0.8398,
+      "step": 1741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589231143731236e-05,
+      "loss": 0.9062,
+      "step": 1742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887493890840052e-05,
+      "loss": 0.8452,
+      "step": 1743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882674252139928e-05,
+      "loss": 0.8227,
+      "step": 1744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.2675,
+      "step": 1745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587302870490796e-05,
+      "loss": 0.8011,
+      "step": 1746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868202799803837e-05,
+      "loss": 0.8217,
+      "step": 1747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5863374809327337e-05,
+      "loss": 0.8327,
+      "step": 1748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858544735194174e-05,
+      "loss": 0.7929,
+      "step": 1749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853712579120807e-05,
+      "loss": 0.8091,
+      "step": 1750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848878342824422e-05,
+      "loss": 0.8812,
+      "step": 1751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584404202802296e-05,
+      "loss": 0.828,
+      "step": 1752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839203636435092e-05,
+      "loss": 0.9073,
+      "step": 1753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834363169780227e-05,
+      "loss": 0.7801,
+      "step": 1754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829520629778513e-05,
+      "loss": 0.8686,
+      "step": 1755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582467601815083e-05,
+      "loss": 0.8372,
+      "step": 1756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819829336618804e-05,
+      "loss": 0.8768,
+      "step": 1757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814980586904795e-05,
+      "loss": 0.8755,
+      "step": 1758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810129770731886e-05,
+      "loss": 0.8425,
+      "step": 1759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805276889823903e-05,
+      "loss": 0.8484,
+      "step": 1760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800421945905412e-05,
+      "loss": 0.275,
+      "step": 1761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57955649407017e-05,
+      "loss": 0.2835,
+      "step": 1762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790705875938795e-05,
+      "loss": 0.8319,
+      "step": 1763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578584475334345e-05,
+      "loss": 0.8495,
+      "step": 1764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578098157464316e-05,
+      "loss": 0.8889,
+      "step": 1765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776116341566134e-05,
+      "loss": 0.8235,
+      "step": 1766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771249055841328e-05,
+      "loss": 0.8186,
+      "step": 1767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766379719198418e-05,
+      "loss": 0.84,
+      "step": 1768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761508333367814e-05,
+      "loss": 0.8703,
+      "step": 1769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575663490008065e-05,
+      "loss": 0.8273,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751759421068784e-05,
+      "loss": 0.8279,
+      "step": 1771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746881898064813e-05,
+      "loss": 0.8463,
+      "step": 1772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574200233280205e-05,
+      "loss": 0.8319,
+      "step": 1773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737120727014535e-05,
+      "loss": 0.8329,
+      "step": 1774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573223708243704e-05,
+      "loss": 0.8541,
+      "step": 1775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727351400805054e-05,
+      "loss": 0.7862,
+      "step": 1776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572246368385479e-05,
+      "loss": 0.7885,
+      "step": 1777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717573933323195e-05,
+      "loss": 0.7934,
+      "step": 1778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712682150947926e-05,
+      "loss": 0.8456,
+      "step": 1779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707788338467362e-05,
+      "loss": 0.8616,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702892497620615e-05,
+      "loss": 0.8155,
+      "step": 1781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569799463014751e-05,
+      "loss": 0.8424,
+      "step": 1782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693094737788585e-05,
+      "loss": 0.8484,
+      "step": 1783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688192822285116e-05,
+      "loss": 0.8165,
+      "step": 1784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5683288885379084e-05,
+      "loss": 0.8426,
+      "step": 1785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567838292881319e-05,
+      "loss": 0.8658,
+      "step": 1786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567347495433086e-05,
+      "loss": 0.8483,
+      "step": 1787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668564963676224e-05,
+      "loss": 0.8456,
+      "step": 1788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566365295859414e-05,
+      "loss": 0.816,
+      "step": 1789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658738940830185e-05,
+      "loss": 0.858,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653822912130634e-05,
+      "loss": 0.8724,
+      "step": 1791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564890487424249e-05,
+      "loss": 0.8597,
+      "step": 1792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643984828913473e-05,
+      "loss": 0.8326,
+      "step": 1793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639062777892e-05,
+      "loss": 0.8781,
+      "step": 1794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634138722927215e-05,
+      "loss": 0.8688,
+      "step": 1795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562921266576898e-05,
+      "loss": 0.7902,
+      "step": 1796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5624284608167842e-05,
+      "loss": 0.8144,
+      "step": 1797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619354551875093e-05,
+      "loss": 0.8389,
+      "step": 1798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614422498642703e-05,
+      "loss": 0.8138,
+      "step": 1799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560948845022338e-05,
+      "loss": 0.7905,
+      "step": 1800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604552408370513e-05,
+      "loss": 0.8013,
+      "step": 1801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599614374838226e-05,
+      "loss": 0.7936,
+      "step": 1802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5594674351381332e-05,
+      "loss": 0.8385,
+      "step": 1803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5589732339755362e-05,
+      "loss": 0.8428,
+      "step": 1804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558478834171655e-05,
+      "loss": 0.7677,
+      "step": 1805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579842359021834e-05,
+      "loss": 0.889,
+      "step": 1806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5574894393428856e-05,
+      "loss": 0.8736,
+      "step": 1807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556994444669597e-05,
+      "loss": 0.7966,
+      "step": 1808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564992520582223e-05,
+      "loss": 0.7901,
+      "step": 1809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560038616847384e-05,
+      "loss": 0.8181,
+      "step": 1810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555082737251897e-05,
+      "loss": 0.8714,
+      "step": 1811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550124883556938e-05,
+      "loss": 0.8514,
+      "step": 1812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545165057524358e-05,
+      "loss": 0.8456,
+      "step": 1813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540203260916728e-05,
+      "loss": 0.8161,
+      "step": 1814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535239495497313e-05,
+      "loss": 0.8351,
+      "step": 1815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530273763030076e-05,
+      "loss": 0.8604,
+      "step": 1816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552530606527968e-05,
+      "loss": 0.8791,
+      "step": 1817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5520336404011483e-05,
+      "loss": 0.854,
+      "step": 1818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515364780991552e-05,
+      "loss": 0.8901,
+      "step": 1819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510391197986635e-05,
+      "loss": 0.8298,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550541565676419e-05,
+      "loss": 0.8321,
+      "step": 1821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550043815909237e-05,
+      "loss": 0.8398,
+      "step": 1822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495458706740008e-05,
+      "loss": 0.837,
+      "step": 1823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490477301476648e-05,
+      "loss": 0.8796,
+      "step": 1824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548549394507253e-05,
+      "loss": 0.7978,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548050863929857e-05,
+      "loss": 0.8673,
+      "step": 1826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547552138592639e-05,
+      "loss": 0.8231,
+      "step": 1827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470532186728303e-05,
+      "loss": 0.8413,
+      "step": 1828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546554104347731e-05,
+      "loss": 0.8404,
+      "step": 1829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5460547957947105e-05,
+      "loss": 0.8949,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545555293191207e-05,
+      "loss": 0.8678,
+      "step": 1831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450555967147282e-05,
+      "loss": 0.8716,
+      "step": 1832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445557065428497e-05,
+      "loss": 0.8838,
+      "step": 1833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440556228532168e-05,
+      "loss": 0.843,
+      "step": 1834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435553458235436e-05,
+      "loss": 0.8652,
+      "step": 1835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430548756316127e-05,
+      "loss": 0.8348,
+      "step": 1836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425542124552747e-05,
+      "loss": 0.8475,
+      "step": 1837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420533564724495e-05,
+      "loss": 0.2803,
+      "step": 1838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541552307861126e-05,
+      "loss": 0.8013,
+      "step": 1839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410510667993596e-05,
+      "loss": 0.7931,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540549633465277e-05,
+      "loss": 0.8207,
+      "step": 1841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400480080370702e-05,
+      "loss": 0.8491,
+      "step": 1842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5395461906930023e-05,
+      "loss": 0.8463,
+      "step": 1843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5390441816114022e-05,
+      "loss": 0.8624,
+      "step": 1844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385419809706677e-05,
+      "loss": 0.8298,
+      "step": 1845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538039588949266e-05,
+      "loss": 0.847,
+      "step": 1846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375370057257306e-05,
+      "loss": 0.8411,
+      "step": 1847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370342314786638e-05,
+      "loss": 0.8281,
+      "step": 1848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365312663867347e-05,
+      "loss": 0.8625,
+      "step": 1849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360281106286823e-05,
+      "loss": 0.8419,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355247643833115e-05,
+      "loss": 0.8449,
+      "step": 1851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350212278294952e-05,
+      "loss": 0.8192,
+      "step": 1852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534517501146175e-05,
+      "loss": 0.92,
+      "step": 1853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534013584512359e-05,
+      "loss": 0.8144,
+      "step": 1854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335094781071234e-05,
+      "loss": 0.8265,
+      "step": 1855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330051821096107e-05,
+      "loss": 0.7477,
+      "step": 1856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325006966990325e-05,
+      "loss": 0.7977,
+      "step": 1857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531996022054666e-05,
+      "loss": 0.8563,
+      "step": 1858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314911583558577e-05,
+      "loss": 0.8672,
+      "step": 1859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530986105782019e-05,
+      "loss": 0.8255,
+      "step": 1860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304808645126305e-05,
+      "loss": 0.837,
+      "step": 1861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299754347272375e-05,
+      "loss": 0.2898,
+      "step": 1862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529469816605455e-05,
+      "loss": 0.8039,
+      "step": 1863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5289640103269626e-05,
+      "loss": 0.8076,
+      "step": 1864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284580160715084e-05,
+      "loss": 0.7337,
+      "step": 1865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279518340189063e-05,
+      "loss": 0.8085,
+      "step": 1866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5274454643490373e-05,
+      "loss": 0.2749,
+      "step": 1867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269389072418488e-05,
+      "loss": 0.8012,
+      "step": 1868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526432162877356e-05,
+      "loss": 0.8507,
+      "step": 1869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259252314356387e-05,
+      "loss": 0.8196,
+      "step": 1870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254181130968446e-05,
+      "loss": 0.8166,
+      "step": 1871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249108080411876e-05,
+      "loss": 0.7911,
+      "step": 1872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244033164489472e-05,
+      "loss": 0.8647,
+      "step": 1873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238956385004703e-05,
+      "loss": 0.7567,
+      "step": 1874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5233877743761696e-05,
+      "loss": 0.8579,
+      "step": 1875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5228797242565229e-05,
+      "loss": 0.772,
+      "step": 1876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223714883220757e-05,
+      "loss": 0.8773,
+      "step": 1877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218630667534391e-05,
+      "loss": 0.7947,
+      "step": 1878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213544597312892e-05,
+      "loss": 0.8964,
+      "step": 1879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208456674363693e-05,
+      "loss": 0.8443,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203366900494883e-05,
+      "loss": 0.8592,
+      "step": 1881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198275277515195e-05,
+      "loss": 0.7736,
+      "step": 1882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193181807234039e-05,
+      "loss": 0.8269,
+      "step": 1883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188086491461467e-05,
+      "loss": 0.292,
+      "step": 1884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182989332008194e-05,
+      "loss": 0.2727,
+      "step": 1885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177890330685588e-05,
+      "loss": 0.8049,
+      "step": 1886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172789489305668e-05,
+      "loss": 0.8336,
+      "step": 1887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167686809681117e-05,
+      "loss": 0.8235,
+      "step": 1888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162582293625262e-05,
+      "loss": 0.8632,
+      "step": 1889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157475942952085e-05,
+      "loss": 0.7976,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152367759476221e-05,
+      "loss": 0.8168,
+      "step": 1891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147257745012956e-05,
+      "loss": 0.8546,
+      "step": 1892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142145901378222e-05,
+      "loss": 0.7937,
+      "step": 1893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137032230388613e-05,
+      "loss": 0.8563,
+      "step": 1894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131916733861358e-05,
+      "loss": 0.8657,
+      "step": 1895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126799413614346e-05,
+      "loss": 0.8592,
+      "step": 1896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121680271466108e-05,
+      "loss": 0.8385,
+      "step": 1897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116559309235825e-05,
+      "loss": 0.7777,
+      "step": 1898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511143652874332e-05,
+      "loss": 0.849,
+      "step": 1899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510631193180907e-05,
+      "loss": 0.8467,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101185520254195e-05,
+      "loss": 0.8596,
+      "step": 1901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096057295900455e-05,
+      "loss": 0.8632,
+      "step": 1902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090927260570259e-05,
+      "loss": 0.7961,
+      "step": 1903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085795416086655e-05,
+      "loss": 0.8325,
+      "step": 1904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080661764273342e-05,
+      "loss": 0.8156,
+      "step": 1905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075526306954653e-05,
+      "loss": 0.8554,
+      "step": 1906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507038904595557e-05,
+      "loss": 0.8338,
+      "step": 1907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065249983101706e-05,
+      "loss": 0.8808,
+      "step": 1908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060109120219327e-05,
+      "loss": 0.8424,
+      "step": 1909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054966459135323e-05,
+      "loss": 0.7398,
+      "step": 1910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049822001677242e-05,
+      "loss": 0.835,
+      "step": 1911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044675749673254e-05,
+      "loss": 0.8529,
+      "step": 1912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5039527704952172e-05,
+      "loss": 0.8372,
+      "step": 1913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034377869343453e-05,
+      "loss": 0.8032,
+      "step": 1914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029226244677179e-05,
+      "loss": 0.8528,
+      "step": 1915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024072832784078e-05,
+      "loss": 0.7942,
+      "step": 1916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018917635495504e-05,
+      "loss": 0.8486,
+      "step": 1917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013760654643456e-05,
+      "loss": 0.904,
+      "step": 1918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008601892060553e-05,
+      "loss": 0.8235,
+      "step": 1919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003441349580059e-05,
+      "loss": 0.8471,
+      "step": 1920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998279029035867e-05,
+      "loss": 0.825,
+      "step": 1921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49931149322625e-05,
+      "loss": 0.7861,
+      "step": 1922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987949061095113e-05,
+      "loss": 0.8034,
+      "step": 1923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982781417369496e-05,
+      "loss": 0.8002,
+      "step": 1924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497761200292206e-05,
+      "loss": 0.2739,
+      "step": 1925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972440819589853e-05,
+      "loss": 0.8738,
+      "step": 1926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967267869210546e-05,
+      "loss": 0.7693,
+      "step": 1927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962093153622445e-05,
+      "loss": 0.9031,
+      "step": 1928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495691667466448e-05,
+      "loss": 0.8526,
+      "step": 1929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951738434176197e-05,
+      "loss": 0.8332,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946558433997792e-05,
+      "loss": 0.8259,
+      "step": 1931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941376675970058e-05,
+      "loss": 0.8913,
+      "step": 1932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936193161934436e-05,
+      "loss": 0.8642,
+      "step": 1933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931007893732981e-05,
+      "loss": 0.8148,
+      "step": 1934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492582087320837e-05,
+      "loss": 0.8544,
+      "step": 1935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920632102203902e-05,
+      "loss": 0.8019,
+      "step": 1936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915441582563508e-05,
+      "loss": 0.7733,
+      "step": 1937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910249316131728e-05,
+      "loss": 0.7976,
+      "step": 1938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905055304753727e-05,
+      "loss": 0.7692,
+      "step": 1939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48998595502753e-05,
+      "loss": 0.8087,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894662054542841e-05,
+      "loss": 0.8593,
+      "step": 1941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4889462819403381e-05,
+      "loss": 0.8651,
+      "step": 1942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884261846704564e-05,
+      "loss": 0.8772,
+      "step": 1943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879059138294647e-05,
+      "loss": 0.7958,
+      "step": 1944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873854696022507e-05,
+      "loss": 0.8089,
+      "step": 1945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868648521737638e-05,
+      "loss": 0.8141,
+      "step": 1946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863440617290148e-05,
+      "loss": 0.81,
+      "step": 1947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858230984530759e-05,
+      "loss": 0.8763,
+      "step": 1948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853019625310813e-05,
+      "loss": 0.7413,
+      "step": 1949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847806541482253e-05,
+      "loss": 0.8358,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842591734897653e-05,
+      "loss": 0.8264,
+      "step": 1951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837375207410182e-05,
+      "loss": 0.2619,
+      "step": 1952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483215696087363e-05,
+      "loss": 0.8548,
+      "step": 1953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826936997142399e-05,
+      "loss": 0.8467,
+      "step": 1954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821715318071492e-05,
+      "loss": 0.7998,
+      "step": 1955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816491925516528e-05,
+      "loss": 0.8696,
+      "step": 1956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481126682133374e-05,
+      "loss": 0.7977,
+      "step": 1957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806040007379955e-05,
+      "loss": 0.8199,
+      "step": 1958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800811485512623e-05,
+      "loss": 0.8946,
+      "step": 1959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795581257589793e-05,
+      "loss": 0.8619,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790349325470122e-05,
+      "loss": 0.2816,
+      "step": 1961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785115691012866e-05,
+      "loss": 0.8081,
+      "step": 1962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47798803560779e-05,
+      "loss": 0.8352,
+      "step": 1963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774643322525691e-05,
+      "loss": 0.8306,
+      "step": 1964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769404592217317e-05,
+      "loss": 0.8435,
+      "step": 1965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764164167014451e-05,
+      "loss": 0.8588,
+      "step": 1966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758922048779378e-05,
+      "loss": 0.8261,
+      "step": 1967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753678239374977e-05,
+      "loss": 0.2768,
+      "step": 1968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474843274066473e-05,
+      "loss": 0.2815,
+      "step": 1969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743185554512725e-05,
+      "loss": 0.7834,
+      "step": 1970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473793668278364e-05,
+      "loss": 0.8019,
+      "step": 1971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473268612734276e-05,
+      "loss": 0.2848,
+      "step": 1972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727433890055962e-05,
+      "loss": 0.8007,
+      "step": 1973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722179972789725e-05,
+      "loss": 0.7966,
+      "step": 1974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716924377411129e-05,
+      "loss": 0.8313,
+      "step": 1975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711667105787835e-05,
+      "loss": 0.812,
+      "step": 1976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706408159788118e-05,
+      "loss": 0.8127,
+      "step": 1977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701147541280836e-05,
+      "loss": 0.8293,
+      "step": 1978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695885252135447e-05,
+      "loss": 0.7999,
+      "step": 1979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690621294222002e-05,
+      "loss": 0.8508,
+      "step": 1980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685355669411143e-05,
+      "loss": 0.8454,
+      "step": 1981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680088379574104e-05,
+      "loss": 0.8491,
+      "step": 1982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674819426582713e-05,
+      "loss": 0.2844,
+      "step": 1983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466954881230939e-05,
+      "loss": 0.801,
+      "step": 1984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466427653862714e-05,
+      "loss": 0.8506,
+      "step": 1985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659002607409565e-05,
+      "loss": 0.8319,
+      "step": 1986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653727020530846e-05,
+      "loss": 0.8149,
+      "step": 1987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648449779865763e-05,
+      "loss": 0.8329,
+      "step": 1988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464317088728968e-05,
+      "loss": 0.8358,
+      "step": 1989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637890344678547e-05,
+      "loss": 0.8344,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632608153908898e-05,
+      "loss": 0.8492,
+      "step": 1991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627324316857858e-05,
+      "loss": 0.8438,
+      "step": 1992
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622038835403135e-05,
+      "loss": 0.8488,
+      "step": 1993
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616751711423016e-05,
+      "loss": 0.7794,
+      "step": 1994
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611462946796383e-05,
+      "loss": 0.8719,
+      "step": 1995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606172543402685e-05,
+      "loss": 0.8526,
+      "step": 1996
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600880503121975e-05,
+      "loss": 0.8123,
+      "step": 1997
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595586827834869e-05,
+      "loss": 0.8389,
+      "step": 1998
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459029151942257e-05,
+      "loss": 0.8114,
+      "step": 1999
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584994579766865e-05,
+      "loss": 0.8736,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579696010750115e-05,
+      "loss": 0.8241,
+      "step": 2001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457439581425526e-05,
+      "loss": 0.8176,
+      "step": 2002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456909399216583e-05,
+      "loss": 0.7443,
+      "step": 2003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563790546365914e-05,
+      "loss": 0.8246,
+      "step": 2004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558485478740191e-05,
+      "loss": 0.7724,
+      "step": 2005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455317879117392e-05,
+      "loss": 0.8283,
+      "step": 2006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547870485552917e-05,
+      "loss": 0.8204,
+      "step": 2007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542560563763594e-05,
+      "loss": 0.8391,
+      "step": 2008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537249027692925e-05,
+      "loss": 0.7716,
+      "step": 2009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531935879228457e-05,
+      "loss": 0.8799,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526621120258319e-05,
+      "loss": 0.8364,
+      "step": 2011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521304752671209e-05,
+      "loss": 0.8149,
+      "step": 2012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515986778356386e-05,
+      "loss": 0.845,
+      "step": 2013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510667199203697e-05,
+      "loss": 0.8446,
+      "step": 2014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505346017103542e-05,
+      "loss": 0.7569,
+      "step": 2015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500023233946907e-05,
+      "loss": 0.2728,
+      "step": 2016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494698851625341e-05,
+      "loss": 0.7984,
+      "step": 2017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489372872030955e-05,
+      "loss": 0.8055,
+      "step": 2018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484045297056435e-05,
+      "loss": 0.8735,
+      "step": 2019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478716128595032e-05,
+      "loss": 0.8358,
+      "step": 2020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473385368540557e-05,
+      "loss": 0.8489,
+      "step": 2021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4468053018787402e-05,
+      "loss": 0.8235,
+      "step": 2022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462719081230507e-05,
+      "loss": 0.8599,
+      "step": 2023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 0.8938,
+      "step": 2024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452046450288111e-05,
+      "loss": 0.8723,
+      "step": 2025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446707760695324e-05,
+      "loss": 0.8251,
+      "step": 2026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441367490884224e-05,
+      "loss": 0.865,
+      "step": 2027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436025642752575e-05,
+      "loss": 0.8327,
+      "step": 2028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430682218198697e-05,
+      "loss": 0.8341,
+      "step": 2029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442533721912147e-05,
+      "loss": 0.8798,
+      "step": 2030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419990647420345e-05,
+      "loss": 0.8215,
+      "step": 2031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414642504995315e-05,
+      "loss": 0.7747,
+      "step": 2032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440929279374694e-05,
+      "loss": 0.8339,
+      "step": 2033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403941515576344e-05,
+      "loss": 0.8195,
+      "step": 2034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4398588672385196e-05,
+      "loss": 0.8345,
+      "step": 2035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393234266075725e-05,
+      "loss": 0.84,
+      "step": 2036
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438787829855072e-05,
+      "loss": 0.8202,
+      "step": 2037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4382520771713517e-05,
+      "loss": 0.8727,
+      "step": 2038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437716168746801e-05,
+      "loss": 0.778,
+      "step": 2039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371801047718654e-05,
+      "loss": 0.8608,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4366438854370444e-05,
+      "loss": 0.8534,
+      "step": 2041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436107510932893e-05,
+      "loss": 0.2766,
+      "step": 2042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355709814500224e-05,
+      "loss": 0.8093,
+      "step": 2043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350342971790979e-05,
+      "loss": 0.7724,
+      "step": 2044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434497458310839e-05,
+      "loss": 0.8753,
+      "step": 2045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339604650360227e-05,
+      "loss": 0.8047,
+      "step": 2046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334233175454782e-05,
+      "loss": 0.8107,
+      "step": 2047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328860160300909e-05,
+      "loss": 0.8251,
+      "step": 2048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323485606808008e-05,
+      "loss": 0.8879,
+      "step": 2049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431810951688602e-05,
+      "loss": 0.8223,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312731892445442e-05,
+      "loss": 0.8556,
+      "step": 2051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307352735397307e-05,
+      "loss": 0.892,
+      "step": 2052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4301972047653194e-05,
+      "loss": 0.8023,
+      "step": 2053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296589831125234e-05,
+      "loss": 0.8784,
+      "step": 2054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429120608772609e-05,
+      "loss": 0.7745,
+      "step": 2055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285820819368976e-05,
+      "loss": 0.2722,
+      "step": 2056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280434027967645e-05,
+      "loss": 0.8266,
+      "step": 2057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275045715436387e-05,
+      "loss": 0.7382,
+      "step": 2058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269655883690041e-05,
+      "loss": 0.7893,
+      "step": 2059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264264534643979e-05,
+      "loss": 0.8025,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258871670214119e-05,
+      "loss": 0.8656,
+      "step": 2061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253477292316907e-05,
+      "loss": 0.8126,
+      "step": 2062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248081402869339e-05,
+      "loss": 0.7345,
+      "step": 2063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242684003788934e-05,
+      "loss": 0.8504,
+      "step": 2064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237285096993767e-05,
+      "loss": 0.8655,
+      "step": 2065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231884684402428e-05,
+      "loss": 0.8225,
+      "step": 2066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226482767934057e-05,
+      "loss": 0.8504,
+      "step": 2067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422107934950832e-05,
+      "loss": 0.7818,
+      "step": 2068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215674431045425e-05,
+      "loss": 0.8042,
+      "step": 2069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210268014466097e-05,
+      "loss": 0.8055,
+      "step": 2070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204860101691615e-05,
+      "loss": 0.7954,
+      "step": 2071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199450694643777e-05,
+      "loss": 0.8594,
+      "step": 2072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194039795244912e-05,
+      "loss": 0.8222,
+      "step": 2073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418862740541788e-05,
+      "loss": 0.8052,
+      "step": 2074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183213527086074e-05,
+      "loss": 0.8075,
+      "step": 2075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177798162173412e-05,
+      "loss": 0.8209,
+      "step": 2076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172381312604348e-05,
+      "loss": 0.8244,
+      "step": 2077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166962980303849e-05,
+      "loss": 0.776,
+      "step": 2078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161543167197422e-05,
+      "loss": 0.7526,
+      "step": 2079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156121875211101e-05,
+      "loss": 0.8216,
+      "step": 2080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150699106271432e-05,
+      "loss": 0.8357,
+      "step": 2081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145274862305497e-05,
+      "loss": 0.7725,
+      "step": 2082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139849145240904e-05,
+      "loss": 0.2511,
+      "step": 2083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134421957005775e-05,
+      "loss": 0.8386,
+      "step": 2084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128993299528762e-05,
+      "loss": 0.8499,
+      "step": 2085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123563174739036e-05,
+      "loss": 0.8212,
+      "step": 2086
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118131584566293e-05,
+      "loss": 0.8037,
+      "step": 2087
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411269853094074e-05,
+      "loss": 0.8393,
+      "step": 2088
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410726401579312e-05,
+      "loss": 0.8577,
+      "step": 2089
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101828041054682e-05,
+      "loss": 0.8541,
+      "step": 2090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.40963906086572e-05,
+      "loss": 0.8397,
+      "step": 2091
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090951720532965e-05,
+      "loss": 0.8342,
+      "step": 2092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085511378614779e-05,
+      "loss": 0.8395,
+      "step": 2093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080069584835971e-05,
+      "loss": 0.7963,
+      "step": 2094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4074626341130382e-05,
+      "loss": 0.818,
+      "step": 2095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069181649432363e-05,
+      "loss": 0.7836,
+      "step": 2096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063735511676789e-05,
+      "loss": 0.8226,
+      "step": 2097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058287929799042e-05,
+      "loss": 0.797,
+      "step": 2098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052838905735018e-05,
+      "loss": 0.8584,
+      "step": 2099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404738844142113e-05,
+      "loss": 0.8511,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041936538794297e-05,
+      "loss": 0.7778,
+      "step": 2101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036483199791949e-05,
+      "loss": 0.7877,
+      "step": 2102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031028426352032e-05,
+      "loss": 0.8138,
+      "step": 2103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025572220412998e-05,
+      "loss": 0.8491,
+      "step": 2104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402011458391381e-05,
+      "loss": 0.8273,
+      "step": 2105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014655518793936e-05,
+      "loss": 0.8882,
+      "step": 2106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009195026993357e-05,
+      "loss": 0.8476,
+      "step": 2107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003733110452556e-05,
+      "loss": 0.8182,
+      "step": 2108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998269771112523e-05,
+      "loss": 0.8499,
+      "step": 2109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992805010914754e-05,
+      "loss": 0.785,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987338831801257e-05,
+      "loss": 0.2676,
+      "step": 2111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981871235714532e-05,
+      "loss": 0.8145,
+      "step": 2112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976402224597587e-05,
+      "loss": 0.7881,
+      "step": 2113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970931800393943e-05,
+      "loss": 0.8315,
+      "step": 2114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965459965047605e-05,
+      "loss": 0.7919,
+      "step": 2115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959986720503093e-05,
+      "loss": 0.8663,
+      "step": 2116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954512068705425e-05,
+      "loss": 0.8504,
+      "step": 2117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394903601160012e-05,
+      "loss": 0.8095,
+      "step": 2118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943558551133186e-05,
+      "loss": 0.2528,
+      "step": 2119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3938079689251147e-05,
+      "loss": 0.8438,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932599427901008e-05,
+      "loss": 0.8276,
+      "step": 2121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927117769030287e-05,
+      "loss": 0.8245,
+      "step": 2122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392163471458699e-05,
+      "loss": 0.8077,
+      "step": 2123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391615026651961e-05,
+      "loss": 0.2924,
+      "step": 2124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910664426777154e-05,
+      "loss": 0.8498,
+      "step": 2125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905177197309116e-05,
+      "loss": 0.8221,
+      "step": 2126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899688580065478e-05,
+      "loss": 0.8361,
+      "step": 2127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894198576996722e-05,
+      "loss": 0.8338,
+      "step": 2128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888707190053823e-05,
+      "loss": 0.8324,
+      "step": 2129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3883214421188239e-05,
+      "loss": 0.8168,
+      "step": 2130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877720272351927e-05,
+      "loss": 0.7897,
+      "step": 2131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872224745497334e-05,
+      "loss": 0.8801,
+      "step": 2132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866727842577394e-05,
+      "loss": 0.8188,
+      "step": 2133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861229565545532e-05,
+      "loss": 0.7898,
+      "step": 2134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385572991635566e-05,
+      "loss": 0.8,
+      "step": 2135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850228896962178e-05,
+      "loss": 0.7727,
+      "step": 2136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844726509319979e-05,
+      "loss": 0.8254,
+      "step": 2137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839222755384428e-05,
+      "loss": 0.788,
+      "step": 2138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833717637111384e-05,
+      "loss": 0.8843,
+      "step": 2139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828211156457196e-05,
+      "loss": 0.8311,
+      "step": 2140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822703315378686e-05,
+      "loss": 0.8258,
+      "step": 2141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817194115833174e-05,
+      "loss": 0.8336,
+      "step": 2142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811683559778446e-05,
+      "loss": 0.2805,
+      "step": 2143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806171649172782e-05,
+      "loss": 0.8356,
+      "step": 2144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800658385974935e-05,
+      "loss": 0.8546,
+      "step": 2145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795143772144151e-05,
+      "loss": 0.8118,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378962780964014e-05,
+      "loss": 0.84,
+      "step": 2147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784110500423104e-05,
+      "loss": 0.8651,
+      "step": 2148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377859184645372e-05,
+      "loss": 0.8221,
+      "step": 2149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773071849693137e-05,
+      "loss": 0.2621,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376755051210299e-05,
+      "loss": 0.8072,
+      "step": 2151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762027835645384e-05,
+      "loss": 0.8124,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756503822282903e-05,
+      "loss": 0.8398,
+      "step": 2153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750978473978611e-05,
+      "loss": 0.7874,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745451792696031e-05,
+      "loss": 0.7788,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739923780399176e-05,
+      "loss": 0.8744,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734394439052525e-05,
+      "loss": 0.8169,
+      "step": 2157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728863770621028e-05,
+      "loss": 0.854,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372333177707011e-05,
+      "loss": 0.8221,
+      "step": 2159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717798460365663e-05,
+      "loss": 0.8402,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712263822474052e-05,
+      "loss": 0.8172,
+      "step": 2161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706727865362113e-05,
+      "loss": 0.796,
+      "step": 2162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701190590997149e-05,
+      "loss": 0.7886,
+      "step": 2163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3695652001346928e-05,
+      "loss": 0.837,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690112098379693e-05,
+      "loss": 0.853,
+      "step": 2165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684570884064146e-05,
+      "loss": 0.2635,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679028360369455e-05,
+      "loss": 0.8497,
+      "step": 2167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673484529265262e-05,
+      "loss": 0.8408,
+      "step": 2168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667939392721666e-05,
+      "loss": 0.8187,
+      "step": 2169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366239295270923e-05,
+      "loss": 0.8437,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656845211198984e-05,
+      "loss": 0.7943,
+      "step": 2171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651296170162417e-05,
+      "loss": 0.8118,
+      "step": 2172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645745831571486e-05,
+      "loss": 0.8366,
+      "step": 2173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36401941973986e-05,
+      "loss": 0.8487,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634641269616635e-05,
+      "loss": 0.8451,
+      "step": 2175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3629087050198924e-05,
+      "loss": 0.8035,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623531541119259e-05,
+      "loss": 0.7767,
+      "step": 2177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617974744351891e-05,
+      "loss": 0.8271,
+      "step": 2178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612416661871532e-05,
+      "loss": 0.8159,
+      "step": 2179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606857295653341e-05,
+      "loss": 0.8628,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601296647672947e-05,
+      "loss": 0.7991,
+      "step": 2181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595734719906421e-05,
+      "loss": 0.841,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35901715143303e-05,
+      "loss": 0.2603,
+      "step": 2183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584607032921566e-05,
+      "loss": 0.8049,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579041277657661e-05,
+      "loss": 0.8276,
+      "step": 2185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357347425051648e-05,
+      "loss": 0.8245,
+      "step": 2186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567905953476357e-05,
+      "loss": 0.7723,
+      "step": 2187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562336388516097e-05,
+      "loss": 0.8763,
+      "step": 2188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556765557614942e-05,
+      "loss": 0.836,
+      "step": 2189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551193462752587e-05,
+      "loss": 0.8123,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545620105909182e-05,
+      "loss": 0.8506,
+      "step": 2191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540045489065318e-05,
+      "loss": 0.8005,
+      "step": 2192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534469614202034e-05,
+      "loss": 0.7779,
+      "step": 2193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528892483300821e-05,
+      "loss": 0.8091,
+      "step": 2194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523314098343613e-05,
+      "loss": 0.8019,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517734461312789e-05,
+      "loss": 0.8579,
+      "step": 2196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512153574191176e-05,
+      "loss": 0.8032,
+      "step": 2197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350657143896204e-05,
+      "loss": 0.8325,
+      "step": 2198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500988057609098e-05,
+      "loss": 0.7997,
+      "step": 2199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495403432116507e-05,
+      "loss": 0.8029,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348981756446886e-05,
+      "loss": 0.8225,
+      "step": 2201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484230456651202e-05,
+      "loss": 0.265,
+      "step": 2202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478642110649012e-05,
+      "loss": 0.8282,
+      "step": 2203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473052528448203e-05,
+      "loss": 0.8297,
+      "step": 2204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467461712035146e-05,
+      "loss": 0.73,
+      "step": 2205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461869663396629e-05,
+      "loss": 0.8187,
+      "step": 2206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456276384519893e-05,
+      "loss": 0.8579,
+      "step": 2207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345068187739261e-05,
+      "loss": 0.2881,
+      "step": 2208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445086144002892e-05,
+      "loss": 0.8509,
+      "step": 2209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439489186339283e-05,
+      "loss": 0.8362,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433891006390761e-05,
+      "loss": 0.8248,
+      "step": 2211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428291606146747e-05,
+      "loss": 0.812,
+      "step": 2212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422690987597078e-05,
+      "loss": 0.788,
+      "step": 2213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417089152732049e-05,
+      "loss": 0.8022,
+      "step": 2214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411486103542361e-05,
+      "loss": 0.8045,
+      "step": 2215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340588184201917e-05,
+      "loss": 0.8104,
+      "step": 2216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400276370154049e-05,
+      "loss": 0.8459,
+      "step": 2217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394669689939002e-05,
+      "loss": 0.8412,
+      "step": 2218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389061803366463e-05,
+      "loss": 0.2884,
+      "step": 2219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3383452712429304e-05,
+      "loss": 0.891,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377842419120807e-05,
+      "loss": 0.8439,
+      "step": 2221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372230925434699e-05,
+      "loss": 0.8576,
+      "step": 2222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366618233365123e-05,
+      "loss": 0.8531,
+      "step": 2223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361004344906652e-05,
+      "loss": 0.8237,
+      "step": 2224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355389262054283e-05,
+      "loss": 0.827,
+      "step": 2225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349772986803438e-05,
+      "loss": 0.7913,
+      "step": 2226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334415552114996e-05,
+      "loss": 0.2661,
+      "step": 2227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333853686709012e-05,
+      "loss": 0.8548,
+      "step": 2228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332917026620611e-05,
+      "loss": 0.8383,
+      "step": 2229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327296001738536e-05,
+      "loss": 0.7818,
+      "step": 2230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321673794441442e-05,
+      "loss": 0.7758,
+      "step": 2231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331605040672727e-05,
+      "loss": 0.8595,
+      "step": 2232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310425840594398e-05,
+      "loss": 0.8153,
+      "step": 2233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330480009804162e-05,
+      "loss": 0.8723,
+      "step": 2234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299173181068138e-05,
+      "loss": 0.8331,
+      "step": 2235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293545091673587e-05,
+      "loss": 0.2655,
+      "step": 2236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287915831858011e-05,
+      "loss": 0.7942,
+      "step": 2237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3282285403621864e-05,
+      "loss": 0.8819,
+      "step": 2238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276653808966025e-05,
+      "loss": 0.7978,
+      "step": 2239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271021049891778e-05,
+      "loss": 0.8696,
+      "step": 2240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265387128400833e-05,
+      "loss": 0.7539,
+      "step": 2241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259752046495304e-05,
+      "loss": 0.8082,
+      "step": 2242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254115806177721e-05,
+      "loss": 0.8154,
+      "step": 2243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248478409451017e-05,
+      "loss": 0.7999,
+      "step": 2244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242839858318548e-05,
+      "loss": 0.8282,
+      "step": 2245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237200154784083e-05,
+      "loss": 0.2827,
+      "step": 2246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231559300851778e-05,
+      "loss": 0.7654,
+      "step": 2247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225917298526225e-05,
+      "loss": 0.847,
+      "step": 2248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220274149812408e-05,
+      "loss": 0.7837,
+      "step": 2249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214629856715721e-05,
+      "loss": 0.2677,
+      "step": 2250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208984421241968e-05,
+      "loss": 0.8104,
+      "step": 2251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203337845397358e-05,
+      "loss": 0.8161,
+      "step": 2252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197690131188497e-05,
+      "loss": 0.8272,
+      "step": 2253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192041280622409e-05,
+      "loss": 0.8145,
+      "step": 2254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186391295706517e-05,
+      "loss": 0.8226,
+      "step": 2255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180740178448641e-05,
+      "loss": 0.8162,
+      "step": 2256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175087930857013e-05,
+      "loss": 0.7588,
+      "step": 2257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3169434554940259e-05,
+      "loss": 0.8259,
+      "step": 2258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163780052707407e-05,
+      "loss": 0.882,
+      "step": 2259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158124426167891e-05,
+      "loss": 0.2911,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152467677331537e-05,
+      "loss": 0.7989,
+      "step": 2261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146809808208574e-05,
+      "loss": 0.7807,
+      "step": 2262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314115082080963e-05,
+      "loss": 0.8713,
+      "step": 2263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135490717145726e-05,
+      "loss": 0.8112,
+      "step": 2264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129829499228289e-05,
+      "loss": 0.8026,
+      "step": 2265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312416716906913e-05,
+      "loss": 0.8159,
+      "step": 2266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311850372868046e-05,
+      "loss": 0.8738,
+      "step": 2267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3112839180074892e-05,
+      "loss": 0.7923,
+      "step": 2268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107173525265417e-05,
+      "loss": 0.7682,
+      "step": 2269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101506766265435e-05,
+      "loss": 0.7967,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309583890508873e-05,
+      "loss": 0.7852,
+      "step": 2271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.773,
+      "step": 2272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084499884262245e-05,
+      "loss": 0.8482,
+      "step": 2273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078828728641994e-05,
+      "loss": 0.8777,
+      "step": 2274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073156478904077e-05,
+      "loss": 0.8678,
+      "step": 2275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306748313706422e-05,
+      "loss": 0.8066,
+      "step": 2276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061808705138557e-05,
+      "loss": 0.8303,
+      "step": 2277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056133185143596e-05,
+      "loss": 0.8558,
+      "step": 2278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050456579096236e-05,
+      "loss": 0.8344,
+      "step": 2279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3044778889013764e-05,
+      "loss": 0.8556,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303910011691385e-05,
+      "loss": 0.8714,
+      "step": 2281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3033420264814547e-05,
+      "loss": 0.7649,
+      "step": 2282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3027739334734296e-05,
+      "loss": 0.7819,
+      "step": 2283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022057328691915e-05,
+      "loss": 0.7764,
+      "step": 2284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016374248706612e-05,
+      "loss": 0.8115,
+      "step": 2285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3010690096797971e-05,
+      "loss": 0.8199,
+      "step": 2286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005004874985955e-05,
+      "loss": 0.8176,
+      "step": 2287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999318585290919e-05,
+      "loss": 0.8039,
+      "step": 2288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993631229733584e-05,
+      "loss": 0.8201,
+      "step": 2289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298794281033505e-05,
+      "loss": 0.8267,
+      "step": 2290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982253329116809e-05,
+      "loss": 0.7441,
+      "step": 2291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297656278810072e-05,
+      "loss": 0.7739,
+      "step": 2292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970871189309013e-05,
+      "loss": 0.8422,
+      "step": 2293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965178534764311e-05,
+      "loss": 0.8178,
+      "step": 2294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959484826489596e-05,
+      "loss": 0.788,
+      "step": 2295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295379006650823e-05,
+      "loss": 0.8377,
+      "step": 2296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294809425684395e-05,
+      "loss": 0.8005,
+      "step": 2297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2942397399520867e-05,
+      "loss": 0.7731,
+      "step": 2298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2936699496563463e-05,
+      "loss": 0.7839,
+      "step": 2299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293100054999659e-05,
+      "loss": 0.8366,
+      "step": 2300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925300561845471e-05,
+      "loss": 0.8322,
+      "step": 2301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919599534135703e-05,
+      "loss": 0.7425,
+      "step": 2302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913897468893249e-05,
+      "loss": 0.8715,
+      "step": 2303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908194368144437e-05,
+      "loss": 0.814,
+      "step": 2304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902490233915975e-05,
+      "loss": 0.7936,
+      "step": 2305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2896785068234925e-05,
+      "loss": 0.8304,
+      "step": 2306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891078873128722e-05,
+      "loss": 0.8609,
+      "step": 2307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885371650625164e-05,
+      "loss": 0.8498,
+      "step": 2308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879663402752423e-05,
+      "loss": 0.7944,
+      "step": 2309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873954131539024e-05,
+      "loss": 0.8379,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868243839013859e-05,
+      "loss": 0.2816,
+      "step": 2311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862532527206186e-05,
+      "loss": 0.8031,
+      "step": 2312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856820198145623e-05,
+      "loss": 0.7716,
+      "step": 2313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285110685386215e-05,
+      "loss": 0.8718,
+      "step": 2314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845392496386107e-05,
+      "loss": 0.8334,
+      "step": 2315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839677127748195e-05,
+      "loss": 0.7925,
+      "step": 2316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833960749979473e-05,
+      "loss": 0.8132,
+      "step": 2317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828243365111364e-05,
+      "loss": 0.8197,
+      "step": 2318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822524975175636e-05,
+      "loss": 0.2723,
+      "step": 2319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281680558220443e-05,
+      "loss": 0.8577,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811085188230234e-05,
+      "loss": 0.857,
+      "step": 2321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805363795285895e-05,
+      "loss": 0.8304,
+      "step": 2322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799641405404611e-05,
+      "loss": 0.263,
+      "step": 2323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793918020619937e-05,
+      "loss": 0.8296,
+      "step": 2324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788193642965785e-05,
+      "loss": 0.7999,
+      "step": 2325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782468274476416e-05,
+      "loss": 0.7892,
+      "step": 2326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776741917186437e-05,
+      "loss": 0.8078,
+      "step": 2327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771014573130822e-05,
+      "loss": 0.8401,
+      "step": 2328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765286244344883e-05,
+      "loss": 0.8291,
+      "step": 2329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759556932864285e-05,
+      "loss": 0.848,
+      "step": 2330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753826640725045e-05,
+      "loss": 0.8656,
+      "step": 2331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748095369963524e-05,
+      "loss": 0.8349,
+      "step": 2332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742363122616432e-05,
+      "loss": 0.8123,
+      "step": 2333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 0.2514,
+      "step": 2334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730895706314122e-05,
+      "loss": 0.8143,
+      "step": 2335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725160541434053e-05,
+      "loss": 0.8377,
+      "step": 2336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271942440811873e-05,
+      "loss": 0.8743,
+      "step": 2337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713687308406581e-05,
+      "loss": 0.2425,
+      "step": 2338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2707949244336391e-05,
+      "loss": 0.2873,
+      "step": 2339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702210217947289e-05,
+      "loss": 0.8148,
+      "step": 2340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696470231278735e-05,
+      "loss": 0.8689,
+      "step": 2341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690729286370546e-05,
+      "loss": 0.8812,
+      "step": 2342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684987385262866e-05,
+      "loss": 0.8575,
+      "step": 2343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679244529996182e-05,
+      "loss": 0.7936,
+      "step": 2344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673500722611327e-05,
+      "loss": 0.8439,
+      "step": 2345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667755965149464e-05,
+      "loss": 0.7919,
+      "step": 2346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662010259652096e-05,
+      "loss": 0.7635,
+      "step": 2347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656263608161067e-05,
+      "loss": 0.8442,
+      "step": 2348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2650516012718553e-05,
+      "loss": 0.7553,
+      "step": 2349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644767475367063e-05,
+      "loss": 0.8052,
+      "step": 2350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263901799814945e-05,
+      "loss": 0.8127,
+      "step": 2351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263326758310889e-05,
+      "loss": 0.2852,
+      "step": 2352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627516232288897e-05,
+      "loss": 0.816,
+      "step": 2353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262176394773332e-05,
+      "loss": 0.8123,
+      "step": 2354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616010731486342e-05,
+      "loss": 0.8561,
+      "step": 2355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610256585592464e-05,
+      "loss": 0.8162,
+      "step": 2356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604501512096534e-05,
+      "loss": 0.7985,
+      "step": 2357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598745513043716e-05,
+      "loss": 0.8048,
+      "step": 2358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2592988590479514e-05,
+      "loss": 0.8332,
+      "step": 2359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258723074644975e-05,
+      "loss": 0.8015,
+      "step": 2360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581471983000579e-05,
+      "loss": 0.9027,
+      "step": 2361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575712302178489e-05,
+      "loss": 0.8817,
+      "step": 2362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569951706030277e-05,
+      "loss": 0.7989,
+      "step": 2363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.256419019660308e-05,
+      "loss": 0.8275,
+      "step": 2364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558427775944357e-05,
+      "loss": 0.2851,
+      "step": 2365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552664446101888e-05,
+      "loss": 0.8233,
+      "step": 2366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546900209123773e-05,
+      "loss": 0.8486,
+      "step": 2367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541135067058443e-05,
+      "loss": 0.8127,
+      "step": 2368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535369021954644e-05,
+      "loss": 0.7988,
+      "step": 2369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529602075861443e-05,
+      "loss": 0.8376,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523834230828233e-05,
+      "loss": 0.8801,
+      "step": 2371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518065488904717e-05,
+      "loss": 0.8277,
+      "step": 2372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512295852140927e-05,
+      "loss": 0.2646,
+      "step": 2373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506525322587207e-05,
+      "loss": 0.8262,
+      "step": 2374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2500753902294217e-05,
+      "loss": 0.8343,
+      "step": 2375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249498159331294e-05,
+      "loss": 0.8152,
+      "step": 2376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489208397694668e-05,
+      "loss": 0.2839,
+      "step": 2377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483434317491008e-05,
+      "loss": 0.7683,
+      "step": 2378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477659354753892e-05,
+      "loss": 0.8422,
+      "step": 2379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471883511535552e-05,
+      "loss": 0.7778,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246610678988854e-05,
+      "loss": 0.8027,
+      "step": 2381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460329191865716e-05,
+      "loss": 0.8744,
+      "step": 2382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454550719520263e-05,
+      "loss": 0.2273,
+      "step": 2383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448771374905655e-05,
+      "loss": 0.7599,
+      "step": 2384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2442991160075695e-05,
+      "loss": 0.8078,
+      "step": 2385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437210077084485e-05,
+      "loss": 0.8043,
+      "step": 2386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431428127986435e-05,
+      "loss": 0.2867,
+      "step": 2387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242564531483627e-05,
+      "loss": 0.7948,
+      "step": 2388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419861639689017e-05,
+      "loss": 0.2828,
+      "step": 2389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414077104600004e-05,
+      "loss": 0.8121,
+      "step": 2390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408291711624878e-05,
+      "loss": 0.7329,
+      "step": 2391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240250546281958e-05,
+      "loss": 0.8298,
+      "step": 2392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396718360240358e-05,
+      "loss": 0.8883,
+      "step": 2393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390930405943766e-05,
+      "loss": 0.8127,
+      "step": 2394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238514160198666e-05,
+      "loss": 0.822,
+      "step": 2395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379351950426188e-05,
+      "loss": 0.7828,
+      "step": 2396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237356145331982e-05,
+      "loss": 0.8494,
+      "step": 2397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367770112725303e-05,
+      "loss": 0.8831,
+      "step": 2398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361977930700704e-05,
+      "loss": 0.8719,
+      "step": 2399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356184909304373e-05,
+      "loss": 0.2714,
+      "step": 2400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235039105059497e-05,
+      "loss": 0.8413,
+      "step": 2401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344596356631446e-05,
+      "loss": 0.7991,
+      "step": 2402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338800829473053e-05,
+      "loss": 0.8081,
+      "step": 2403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233300447117933e-05,
+      "loss": 0.8356,
+      "step": 2404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327207283810127e-05,
+      "loss": 0.7876,
+      "step": 2405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321409269425575e-05,
+      "loss": 0.7898,
+      "step": 2406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315610430086102e-05,
+      "loss": 0.2748,
+      "step": 2407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309810767852435e-05,
+      "loss": 0.7683,
+      "step": 2408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304010284785586e-05,
+      "loss": 0.7849,
+      "step": 2409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298208982946862e-05,
+      "loss": 0.8051,
+      "step": 2410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292406864397866e-05,
+      "loss": 0.8325,
+      "step": 2411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228660393120048e-05,
+      "loss": 0.8405,
+      "step": 2412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280800185416884e-05,
+      "loss": 0.8339,
+      "step": 2413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274995629109545e-05,
+      "loss": 0.8105,
+      "step": 2414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269190264341216e-05,
+      "loss": 0.7817,
+      "step": 2415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263384093174939e-05,
+      "loss": 0.7525,
+      "step": 2416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257577117674045e-05,
+      "loss": 0.8025,
+      "step": 2417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251769339902143e-05,
+      "loss": 0.8233,
+      "step": 2418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245960761923134e-05,
+      "loss": 0.8687,
+      "step": 2419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224015138580121e-05,
+      "loss": 0.8258,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2234341213600823e-05,
+      "loss": 0.8027,
+      "step": 2421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228530247386737e-05,
+      "loss": 0.788,
+      "step": 2422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222718489223977e-05,
+      "loss": 0.7839,
+      "step": 2423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216905941177854e-05,
+      "loss": 0.8278,
+      "step": 2424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2211092605313971e-05,
+      "loss": 0.7984,
+      "step": 2425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22052784836982e-05,
+      "loss": 0.7648,
+      "step": 2426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199463578396688e-05,
+      "loss": 0.8247,
+      "step": 2427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193647891475873e-05,
+      "loss": 0.8213,
+      "step": 2428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187831425002466e-05,
+      "loss": 0.8312,
+      "step": 2429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218201418104345e-05,
+      "loss": 0.7947,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217619616166609e-05,
+      "loss": 0.8181,
+      "step": 2431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170377368937926e-05,
+      "loss": 0.8,
+      "step": 2432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216455780492677e-05,
+      "loss": 0.8471,
+      "step": 2433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215873747170071e-05,
+      "loss": 0.8612,
+      "step": 2434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152916371328104e-05,
+      "loss": 0.2581,
+      "step": 2435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147094505877593e-05,
+      "loss": 0.8207,
+      "step": 2436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141271877418075e-05,
+      "loss": 0.7925,
+      "step": 2437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135448488018734e-05,
+      "loss": 0.7972,
+      "step": 2438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212962433974901e-05,
+      "loss": 0.8268,
+      "step": 2439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123799434678624e-05,
+      "loss": 0.8438,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211797377487756e-05,
+      "loss": 0.837,
+      "step": 2441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112147362416076e-05,
+      "loss": 0.8128,
+      "step": 2442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106320199364687e-05,
+      "loss": 0.8454,
+      "step": 2443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100492287794186e-05,
+      "loss": 0.8058,
+      "step": 2444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094663629775625e-05,
+      "loss": 0.8087,
+      "step": 2445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088834227380321e-05,
+      "loss": 0.8057,
+      "step": 2446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208300408267986e-05,
+      "loss": 0.8412,
+      "step": 2447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077173197746097e-05,
+      "loss": 0.8062,
+      "step": 2448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071341574651134e-05,
+      "loss": 0.8068,
+      "step": 2449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065509215467345e-05,
+      "loss": 0.8149,
+      "step": 2450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059676122267365e-05,
+      "loss": 0.843,
+      "step": 2451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053842297124094e-05,
+      "loss": 0.2681,
+      "step": 2452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048007742110681e-05,
+      "loss": 0.8041,
+      "step": 2453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042172459300546e-05,
+      "loss": 0.7887,
+      "step": 2454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036336450767359e-05,
+      "loss": 0.7676,
+      "step": 2455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030499718585054e-05,
+      "loss": 0.7906,
+      "step": 2456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2024662264827822e-05,
+      "loss": 0.7938,
+      "step": 2457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018824091570103e-05,
+      "loss": 0.8248,
+      "step": 2458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012985200886602e-05,
+      "loss": 0.8266,
+      "step": 2459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007145594852274e-05,
+      "loss": 0.285,
+      "step": 2460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001305275542325e-05,
+      "loss": 0.8092,
+      "step": 2461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1995464245032222e-05,
+      "loss": 0.8353,
+      "step": 2462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989622505397685e-05,
+      "loss": 0.8031,
+      "step": 2463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198378005871467e-05,
+      "loss": 0.7879,
+      "step": 2464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977936907059409e-05,
+      "loss": 0.8471,
+      "step": 2465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197209305250837e-05,
+      "loss": 0.7518,
+      "step": 2466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966248497138264e-05,
+      "loss": 0.7893,
+      "step": 2467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960403243026072e-05,
+      "loss": 0.8026,
+      "step": 2468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954557292249003e-05,
+      "loss": 0.8713,
+      "step": 2469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948710646884522e-05,
+      "loss": 0.8058,
+      "step": 2470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942863309010346e-05,
+      "loss": 0.8569,
+      "step": 2471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1937015280704425e-05,
+      "loss": 0.8014,
+      "step": 2472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.193116656404497e-05,
+      "loss": 0.8278,
+      "step": 2473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192531716111042e-05,
+      "loss": 0.8447,
+      "step": 2474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1919467073979476e-05,
+      "loss": 0.8397,
+      "step": 2475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913616304731064e-05,
+      "loss": 0.8179,
+      "step": 2476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907764855444369e-05,
+      "loss": 0.8463,
+      "step": 2477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901912728198802e-05,
+      "loss": 0.8658,
+      "step": 2478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896059925074029e-05,
+      "loss": 0.2686,
+      "step": 2479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890206448149945e-05,
+      "loss": 0.7975,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884352299506693e-05,
+      "loss": 0.8318,
+      "step": 2481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878497481224649e-05,
+      "loss": 0.8612,
+      "step": 2482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872641995384426e-05,
+      "loss": 0.8238,
+      "step": 2483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866785844066884e-05,
+      "loss": 0.878,
+      "step": 2484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860929029353108e-05,
+      "loss": 0.8644,
+      "step": 2485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1855071553324427e-05,
+      "loss": 0.8325,
+      "step": 2486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849213418062395e-05,
+      "loss": 0.7859,
+      "step": 2487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843354625648813e-05,
+      "loss": 0.8422,
+      "step": 2488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837495178165706e-05,
+      "loss": 0.7639,
+      "step": 2489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831635077695337e-05,
+      "loss": 0.8307,
+      "step": 2490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825774326320197e-05,
+      "loss": 0.8281,
+      "step": 2491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181991292612301e-05,
+      "loss": 0.8188,
+      "step": 2492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1814050879186732e-05,
+      "loss": 0.8683,
+      "step": 2493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808188187594549e-05,
+      "loss": 0.7808,
+      "step": 2494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802324853429877e-05,
+      "loss": 0.8394,
+      "step": 2495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179646087877635e-05,
+      "loss": 0.8391,
+      "step": 2496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790596265717847e-05,
+      "loss": 0.7877,
+      "step": 2497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1784731016338458e-05,
+      "loss": 0.737,
+      "step": 2498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177886513272251e-05,
+      "loss": 0.8265,
+      "step": 2499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177299861695455e-05,
+      "loss": 0.8125,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767131471119347e-05,
+      "loss": 0.7698,
+      "step": 2501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761263697301906e-05,
+      "loss": 0.8278,
+      "step": 2502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755395297587439e-05,
+      "loss": 0.8156,
+      "step": 2503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749526274061394e-05,
+      "loss": 0.8188,
+      "step": 2504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174365662880943e-05,
+      "loss": 0.2797,
+      "step": 2505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737786363917438e-05,
+      "loss": 0.8265,
+      "step": 2506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731915481471516e-05,
+      "loss": 0.846,
+      "step": 2507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726043983557996e-05,
+      "loss": 0.2349,
+      "step": 2508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720171872263415e-05,
+      "loss": 0.8494,
+      "step": 2509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714299149674538e-05,
+      "loss": 0.7728,
+      "step": 2510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708425817878342e-05,
+      "loss": 0.7954,
+      "step": 2511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702551878962025e-05,
+      "loss": 0.8306,
+      "step": 2512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696677335012994e-05,
+      "loss": 0.8057,
+      "step": 2513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1690802188118878e-05,
+      "loss": 0.8016,
+      "step": 2514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684926440367511e-05,
+      "loss": 0.785,
+      "step": 2515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679050093846956e-05,
+      "loss": 0.8105,
+      "step": 2516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673173150645472e-05,
+      "loss": 0.8217,
+      "step": 2517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166729561285154e-05,
+      "loss": 0.2731,
+      "step": 2518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661417482553848e-05,
+      "loss": 0.8159,
+      "step": 2519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16555387618413e-05,
+      "loss": 0.8077,
+      "step": 2520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649659452803005e-05,
+      "loss": 0.8131,
+      "step": 2521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643779557528278e-05,
+      "loss": 0.8182,
+      "step": 2522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637899078106654e-05,
+      "loss": 0.8307,
+      "step": 2523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632018016627859e-05,
+      "loss": 0.7836,
+      "step": 2524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1626136375181845e-05,
+      "loss": 0.8262,
+      "step": 2525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620254155858752e-05,
+      "loss": 0.7724,
+      "step": 2526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1614371360748934e-05,
+      "loss": 0.8105,
+      "step": 2527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608487991942956e-05,
+      "loss": 0.8173,
+      "step": 2528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602604051531573e-05,
+      "loss": 0.7852,
+      "step": 2529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159671954160575e-05,
+      "loss": 0.8067,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1590834464256657e-05,
+      "loss": 0.7541,
+      "step": 2531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584948821575666e-05,
+      "loss": 0.8381,
+      "step": 2532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1579062615654339e-05,
+      "loss": 0.8031,
+      "step": 2533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573175848584455e-05,
+      "loss": 0.8146,
+      "step": 2534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567288522457976e-05,
+      "loss": 0.8121,
+      "step": 2535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1561400639367075e-05,
+      "loss": 0.8805,
+      "step": 2536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555512201404117e-05,
+      "loss": 0.2692,
+      "step": 2537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549623210661663e-05,
+      "loss": 0.8606,
+      "step": 2538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543733669232473e-05,
+      "loss": 0.7488,
+      "step": 2539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537843579209509e-05,
+      "loss": 0.8703,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153195294268591e-05,
+      "loss": 0.8511,
+      "step": 2541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526061761755033e-05,
+      "loss": 0.7825,
+      "step": 2542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152017003851041e-05,
+      "loss": 0.7876,
+      "step": 2543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514277775045768e-05,
+      "loss": 0.8257,
+      "step": 2544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508384973455036e-05,
+      "loss": 0.7825,
+      "step": 2545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502491635832327e-05,
+      "loss": 0.8552,
+      "step": 2546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496597764271937e-05,
+      "loss": 0.8375,
+      "step": 2547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490703360868373e-05,
+      "loss": 0.8644,
+      "step": 2548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484808427716314e-05,
+      "loss": 0.844,
+      "step": 2549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478912966910627e-05,
+      "loss": 0.7441,
+      "step": 2550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473016980546377e-05,
+      "loss": 0.8162,
+      "step": 2551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467120470718805e-05,
+      "loss": 0.8312,
+      "step": 2552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461223439523344e-05,
+      "loss": 0.7366,
+      "step": 2553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455325889055616e-05,
+      "loss": 0.7654,
+      "step": 2554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449427821411412e-05,
+      "loss": 0.8297,
+      "step": 2555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443529238686726e-05,
+      "loss": 0.8035,
+      "step": 2556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437630142977724e-05,
+      "loss": 0.7959,
+      "step": 2557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431730536380759e-05,
+      "loss": 0.8562,
+      "step": 2558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425830420992358e-05,
+      "loss": 0.7872,
+      "step": 2559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419929798909241e-05,
+      "loss": 0.7799,
+      "step": 2560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1414028672228294e-05,
+      "loss": 0.8058,
+      "step": 2561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408127043046598e-05,
+      "loss": 0.7969,
+      "step": 2562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402224913461396e-05,
+      "loss": 0.823,
+      "step": 2563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396322285570119e-05,
+      "loss": 0.8154,
+      "step": 2564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390419161470372e-05,
+      "loss": 0.7937,
+      "step": 2565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384515543259943e-05,
+      "loss": 0.2651,
+      "step": 2566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378611433036782e-05,
+      "loss": 0.8292,
+      "step": 2567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372706832899027e-05,
+      "loss": 0.7935,
+      "step": 2568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366801744944986e-05,
+      "loss": 0.7786,
+      "step": 2569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360896171273134e-05,
+      "loss": 0.7883,
+      "step": 2570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135499011398213e-05,
+      "loss": 0.7695,
+      "step": 2571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134908357517079e-05,
+      "loss": 0.8112,
+      "step": 2572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343176556938118e-05,
+      "loss": 0.8615,
+      "step": 2573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337269061383278e-05,
+      "loss": 0.8506,
+      "step": 2574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1331361090605601e-05,
+      "loss": 0.2746,
+      "step": 2575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325452646704597e-05,
+      "loss": 0.8309,
+      "step": 2576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131954373177994e-05,
+      "loss": 0.8006,
+      "step": 2577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313634347931466e-05,
+      "loss": 0.8128,
+      "step": 2578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307724497259183e-05,
+      "loss": 0.8767,
+      "step": 2579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301814181863265e-05,
+      "loss": 0.794,
+      "step": 2580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295903403844047e-05,
+      "loss": 0.8302,
+      "step": 2581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289992165302036e-05,
+      "loss": 0.8169,
+      "step": 2582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284080468337892e-05,
+      "loss": 0.799,
+      "step": 2583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278168315052445e-05,
+      "loss": 0.8251,
+      "step": 2584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272255707546691e-05,
+      "loss": 0.8745,
+      "step": 2585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126634264792178e-05,
+      "loss": 0.8151,
+      "step": 2586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126042913827902e-05,
+      "loss": 0.7748,
+      "step": 2587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254515180719893e-05,
+      "loss": 0.8533,
+      "step": 2588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248600777346026e-05,
+      "loss": 0.7798,
+      "step": 2589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242685930259207e-05,
+      "loss": 0.8297,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236770641561389e-05,
+      "loss": 0.8247,
+      "step": 2591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230854913354674e-05,
+      "loss": 0.8403,
+      "step": 2592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224938747741325e-05,
+      "loss": 0.8462,
+      "step": 2593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219022146823762e-05,
+      "loss": 0.8497,
+      "step": 2594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121310511270455e-05,
+      "loss": 0.8088,
+      "step": 2595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207187647486418e-05,
+      "loss": 0.7649,
+      "step": 2596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201269753272245e-05,
+      "loss": 0.8158,
+      "step": 2597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119535143216506e-05,
+      "loss": 0.8409,
+      "step": 2598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118943268626805e-05,
+      "loss": 0.8361,
+      "step": 2599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183513517684546e-05,
+      "loss": 0.8075,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177593928518029e-05,
+      "loss": 0.8058,
+      "step": 2601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171673920872134e-05,
+      "loss": 0.7712,
+      "step": 2602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116575349685065e-05,
+      "loss": 0.8887,
+      "step": 2603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159832658557498e-05,
+      "loss": 0.9076,
+      "step": 2604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153911408096762e-05,
+      "loss": 0.8316,
+      "step": 2605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147989747572662e-05,
+      "loss": 0.8299,
+      "step": 2606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142067679089568e-05,
+      "loss": 0.7756,
+      "step": 2607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136145204751995e-05,
+      "loss": 0.8017,
+      "step": 2608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130222326664601e-05,
+      "loss": 0.8596,
+      "step": 2609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124299046932192e-05,
+      "loss": 0.8237,
+      "step": 2610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1118375367659706e-05,
+      "loss": 0.7914,
+      "step": 2611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1112451290952238e-05,
+      "loss": 0.7531,
+      "step": 2612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106526818915008e-05,
+      "loss": 0.8099,
+      "step": 2613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100601953653393e-05,
+      "loss": 0.3035,
+      "step": 2614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094676697272894e-05,
+      "loss": 0.8675,
+      "step": 2615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088751051879166e-05,
+      "loss": 0.7791,
+      "step": 2616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082825019577986e-05,
+      "loss": 0.7676,
+      "step": 2617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107689860247528e-05,
+      "loss": 0.786,
+      "step": 2618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070971802677115e-05,
+      "loss": 0.7785,
+      "step": 2619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065044622289675e-05,
+      "loss": 0.8642,
+      "step": 2620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10591170634193e-05,
+      "loss": 0.8223,
+      "step": 2621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053189128172454e-05,
+      "loss": 0.8395,
+      "step": 2622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047260818655737e-05,
+      "loss": 0.8265,
+      "step": 2623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1041332136975874e-05,
+      "loss": 0.8442,
+      "step": 2624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035403085239742e-05,
+      "loss": 0.8378,
+      "step": 2625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029473665554327e-05,
+      "loss": 0.8239,
+      "step": 2626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023543880026759e-05,
+      "loss": 0.8246,
+      "step": 2627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017613730764295e-05,
+      "loss": 0.8116,
+      "step": 2628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011683219874324e-05,
+      "loss": 0.8193,
+      "step": 2629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005752349464353e-05,
+      "loss": 0.8011,
+      "step": 2630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999821121642036e-05,
+      "loss": 0.8022,
+      "step": 2631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993889538515135e-05,
+      "loss": 0.8017,
+      "step": 2632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987957602191543e-05,
+      "loss": 0.8222,
+      "step": 2633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982025314779287e-05,
+      "loss": 0.8241,
+      "step": 2634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0976092678386512e-05,
+      "loss": 0.7556,
+      "step": 2635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0970159695121488e-05,
+      "loss": 0.7716,
+      "step": 2636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964226367092606e-05,
+      "loss": 0.762,
+      "step": 2637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958292696408381e-05,
+      "loss": 0.8331,
+      "step": 2638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0952358685177455e-05,
+      "loss": 0.7557,
+      "step": 2639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946424335508585e-05,
+      "loss": 0.2614,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940489649510649e-05,
+      "loss": 0.7992,
+      "step": 2641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934554629292645e-05,
+      "loss": 0.8045,
+      "step": 2642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928619276963693e-05,
+      "loss": 0.847,
+      "step": 2643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 0.8468,
+      "step": 2644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916747584409991e-05,
+      "loss": 0.8607,
+      "step": 2645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910811248404064e-05,
+      "loss": 0.8251,
+      "step": 2646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0904874588724829e-05,
+      "loss": 0.7862,
+      "step": 2647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898937607481985e-05,
+      "loss": 0.7719,
+      "step": 2648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089300030678535e-05,
+      "loss": 0.7838,
+      "step": 2649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887062688744843e-05,
+      "loss": 0.7266,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881124755470509e-05,
+      "loss": 0.7668,
+      "step": 2651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875186509072502e-05,
+      "loss": 0.828,
+      "step": 2652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0869247951661084e-05,
+      "loss": 0.7996,
+      "step": 2653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086330908534663e-05,
+      "loss": 0.8551,
+      "step": 2654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085736991223962e-05,
+      "loss": 0.8705,
+      "step": 2655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851430434450652e-05,
+      "loss": 0.7753,
+      "step": 2656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845490654090423e-05,
+      "loss": 0.8065,
+      "step": 2657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839550573269744e-05,
+      "loss": 0.8291,
+      "step": 2658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833610194099531e-05,
+      "loss": 0.8586,
+      "step": 2659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827669518690806e-05,
+      "loss": 0.7891,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821728549154696e-05,
+      "loss": 0.7817,
+      "step": 2661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815787287602428e-05,
+      "loss": 0.8326,
+      "step": 2662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809845736145346e-05,
+      "loss": 0.8289,
+      "step": 2663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803903896894877e-05,
+      "loss": 0.854,
+      "step": 2664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079796177196257e-05,
+      "loss": 0.851,
+      "step": 2665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079201936346006e-05,
+      "loss": 0.2552,
+      "step": 2666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786076673499096e-05,
+      "loss": 0.7769,
+      "step": 2667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780133704191517e-05,
+      "loss": 0.8323,
+      "step": 2668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774190457649269e-05,
+      "loss": 0.7997,
+      "step": 2669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768246935984387e-05,
+      "loss": 0.8424,
+      "step": 2670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762303141309016e-05,
+      "loss": 0.7404,
+      "step": 2671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756359075735385e-05,
+      "loss": 0.7875,
+      "step": 2672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750414741375829e-05,
+      "loss": 0.7611,
+      "step": 2673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744470140342775e-05,
+      "loss": 0.8222,
+      "step": 2674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073852527474874e-05,
+      "loss": 0.7635,
+      "step": 2675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732580146706347e-05,
+      "loss": 0.8067,
+      "step": 2676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726634758328302e-05,
+      "loss": 0.2718,
+      "step": 2677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720689111727407e-05,
+      "loss": 0.7898,
+      "step": 2678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714743209016554e-05,
+      "loss": 0.7715,
+      "step": 2679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070879705230873e-05,
+      "loss": 0.8527,
+      "step": 2680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0702850643717009e-05,
+      "loss": 0.8423,
+      "step": 2681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696903985354555e-05,
+      "loss": 0.7606,
+      "step": 2682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0690957079334619e-05,
+      "loss": 0.8644,
+      "step": 2683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685009927770542e-05,
+      "loss": 0.827,
+      "step": 2684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679062532775755e-05,
+      "loss": 0.8186,
+      "step": 2685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673114896463772e-05,
+      "loss": 0.7693,
+      "step": 2686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667167020948192e-05,
+      "loss": 0.8414,
+      "step": 2687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0661218908342705e-05,
+      "loss": 0.7971,
+      "step": 2688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655270560761077e-05,
+      "loss": 0.8267,
+      "step": 2689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0649321980317158e-05,
+      "loss": 0.8269,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643373169124892e-05,
+      "loss": 0.8098,
+      "step": 2691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637424129298288e-05,
+      "loss": 0.7759,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631474862951455e-05,
+      "loss": 0.7834,
+      "step": 2693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0625525372198564e-05,
+      "loss": 0.7797,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619575659153878e-05,
+      "loss": 0.8041,
+      "step": 2695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613625725931738e-05,
+      "loss": 0.8415,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060767557464656e-05,
+      "loss": 0.8136,
+      "step": 2697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601725207412835e-05,
+      "loss": 0.808,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595774626345137e-05,
+      "loss": 0.819,
+      "step": 2699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589823833558111e-05,
+      "loss": 0.7888,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583872831166479e-05,
+      "loss": 0.7711,
+      "step": 2701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577921621285041e-05,
+      "loss": 0.7732,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571970206028663e-05,
+      "loss": 0.7751,
+      "step": 2703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056601858751229e-05,
+      "loss": 0.8031,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560066767850943e-05,
+      "loss": 0.7697,
+      "step": 2705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05541147491597e-05,
+      "loss": 0.8392,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548162533553725e-05,
+      "loss": 0.8215,
+      "step": 2707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542210123148246e-05,
+      "loss": 0.8252,
+      "step": 2708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536257520058555e-05,
+      "loss": 0.7698,
+      "step": 2709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530304726400025e-05,
+      "loss": 0.7755,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524351744288084e-05,
+      "loss": 0.8761,
+      "step": 2711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051839857583823e-05,
+      "loss": 0.845,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512445223166035e-05,
+      "loss": 0.7404,
+      "step": 2713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506491688387128e-05,
+      "loss": 0.7567,
+      "step": 2714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500537973617207e-05,
+      "loss": 0.8316,
+      "step": 2715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494584080972032e-05,
+      "loss": 0.8295,
+      "step": 2716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048863001256743e-05,
+      "loss": 0.7669,
+      "step": 2717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482675770519279e-05,
+      "loss": 0.7896,
+      "step": 2718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476721356943536e-05,
+      "loss": 0.7424,
+      "step": 2719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0470766773956205e-05,
+      "loss": 0.863,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464812023673352e-05,
+      "loss": 0.8106,
+      "step": 2721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458857108211113e-05,
+      "loss": 0.808,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452902029685674e-05,
+      "loss": 0.813,
+      "step": 2723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446946790213275e-05,
+      "loss": 0.8206,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440991391910224e-05,
+      "loss": 0.8277,
+      "step": 2725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435035836892879e-05,
+      "loss": 0.853,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429080127277651e-05,
+      "loss": 0.7846,
+      "step": 2727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423124265181012e-05,
+      "loss": 0.8598,
+      "step": 2728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417168252719485e-05,
+      "loss": 0.781,
+      "step": 2729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411212092009647e-05,
+      "loss": 0.8452,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405255785168131e-05,
+      "loss": 0.8276,
+      "step": 2731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0399299334311615e-05,
+      "loss": 0.8566,
+      "step": 2732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393342741556831e-05,
+      "loss": 0.8534,
+      "step": 2733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387386009020569e-05,
+      "loss": 0.7412,
+      "step": 2734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381429138819652e-05,
+      "loss": 0.7592,
+      "step": 2735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0375472133070969e-05,
+      "loss": 0.8098,
+      "step": 2736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369514993891451e-05,
+      "loss": 0.795,
+      "step": 2737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036355772339807e-05,
+      "loss": 0.8168,
+      "step": 2738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357600323707856e-05,
+      "loss": 0.2697,
+      "step": 2739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0351642796937873e-05,
+      "loss": 0.8299,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345685145205243e-05,
+      "loss": 0.8495,
+      "step": 2741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339727370627119e-05,
+      "loss": 0.7904,
+      "step": 2742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333769475320712e-05,
+      "loss": 0.8204,
+      "step": 2743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032781146140326e-05,
+      "loss": 0.8166,
+      "step": 2744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321853330992056e-05,
+      "loss": 0.8465,
+      "step": 2745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031589508620443e-05,
+      "loss": 0.7665,
+      "step": 2746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030993672915775e-05,
+      "loss": 0.7722,
+      "step": 2747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030397826196943e-05,
+      "loss": 0.834,
+      "step": 2748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298019686756915e-05,
+      "loss": 0.8055,
+      "step": 2749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0292061005637695e-05,
+      "loss": 0.8657,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286102220729297e-05,
+      "loss": 0.8344,
+      "step": 2751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028014333414928e-05,
+      "loss": 0.7512,
+      "step": 2752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274184348015245e-05,
+      "loss": 0.7787,
+      "step": 2753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268225264444829e-05,
+      "loss": 0.8374,
+      "step": 2754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262266085555691e-05,
+      "loss": 0.8095,
+      "step": 2755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256306813465545e-05,
+      "loss": 0.7989,
+      "step": 2756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025034745029212e-05,
+      "loss": 0.8309,
+      "step": 2757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244387998153179e-05,
+      "loss": 0.8789,
+      "step": 2758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023842845916653e-05,
+      "loss": 0.8342,
+      "step": 2759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232468835450002e-05,
+      "loss": 0.8274,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022650912912145e-05,
+      "loss": 0.736,
+      "step": 2761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220549342298774e-05,
+      "loss": 0.276,
+      "step": 2762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214589477099883e-05,
+      "loss": 0.8348,
+      "step": 2763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208629535642726e-05,
+      "loss": 0.839,
+      "step": 2764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202669520045278e-05,
+      "loss": 0.853,
+      "step": 2765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196709432425535e-05,
+      "loss": 0.2788,
+      "step": 2766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0190749274901525e-05,
+      "loss": 0.266,
+      "step": 2767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01847890495913e-05,
+      "loss": 0.7916,
+      "step": 2768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178828758612933e-05,
+      "loss": 0.7913,
+      "step": 2769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172868404084518e-05,
+      "loss": 0.8188,
+      "step": 2770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166907988124179e-05,
+      "loss": 0.8472,
+      "step": 2771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160947512850057e-05,
+      "loss": 0.7873,
+      "step": 2772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154986980380314e-05,
+      "loss": 0.7752,
+      "step": 2773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149026392833137e-05,
+      "loss": 0.8087,
+      "step": 2774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143065752326724e-05,
+      "loss": 0.7574,
+      "step": 2775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137105060979301e-05,
+      "loss": 0.7893,
+      "step": 2776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131144320909105e-05,
+      "loss": 0.7723,
+      "step": 2777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0125183534234392e-05,
+      "loss": 0.864,
+      "step": 2778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119222703073441e-05,
+      "loss": 0.7936,
+      "step": 2779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113261829544541e-05,
+      "loss": 0.8564,
+      "step": 2780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107300915765987e-05,
+      "loss": 0.7959,
+      "step": 2781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101339963856112e-05,
+      "loss": 0.8313,
+      "step": 2782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095378975933238e-05,
+      "loss": 0.766,
+      "step": 2783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089417954115715e-05,
+      "loss": 0.8104,
+      "step": 2784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083456900521898e-05,
+      "loss": 0.8325,
+      "step": 2785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077495817270155e-05,
+      "loss": 0.814,
+      "step": 2786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071534706478868e-05,
+      "loss": 0.7654,
+      "step": 2787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065573570266424e-05,
+      "loss": 0.8603,
+      "step": 2788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059612410751225e-05,
+      "loss": 0.8325,
+      "step": 2789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053651230051671e-05,
+      "loss": 0.8649,
+      "step": 2790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004769003028618e-05,
+      "loss": 0.8282,
+      "step": 2791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041728813573168e-05,
+      "loss": 0.7708,
+      "step": 2792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003576758203107e-05,
+      "loss": 0.7161,
+      "step": 2793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002980633777831e-05,
+      "loss": 0.7729,
+      "step": 2794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023845082933323e-05,
+      "loss": 0.8051,
+      "step": 2795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0017883819614558e-05,
+      "loss": 0.2562,
+      "step": 2796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001192254994045e-05,
+      "loss": 0.8808,
+      "step": 2797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000596127602945e-05,
+      "loss": 0.8372,
+      "step": 2798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.8268,
+      "step": 2799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994038723970551e-06,
+      "loss": 0.7789,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988077450059551e-06,
+      "loss": 0.8366,
+      "step": 2801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982116180385444e-06,
+      "loss": 0.7916,
+      "step": 2802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97615491706668e-06,
+      "loss": 0.7569,
+      "step": 2803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970193662221694e-06,
+      "loss": 0.8585,
+      "step": 2804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964232417968936e-06,
+      "loss": 0.8865,
+      "step": 2805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958271186426834e-06,
+      "loss": 0.8056,
+      "step": 2806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.952309969713826e-06,
+      "loss": 0.8436,
+      "step": 2807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946348769948332e-06,
+      "loss": 0.787,
+      "step": 2808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940387589248778e-06,
+      "loss": 0.7582,
+      "step": 2809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934426429733577e-06,
+      "loss": 0.8171,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928465293521134e-06,
+      "loss": 0.293,
+      "step": 2811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922504182729848e-06,
+      "loss": 0.8147,
+      "step": 2812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916543099478106e-06,
+      "loss": 0.8577,
+      "step": 2813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910582045884292e-06,
+      "loss": 0.8102,
+      "step": 2814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904621024066763e-06,
+      "loss": 0.8048,
+      "step": 2815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.898660036143893e-06,
+      "loss": 0.7972,
+      "step": 2816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892699084234015e-06,
+      "loss": 0.8149,
+      "step": 2817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886738170455464e-06,
+      "loss": 0.8105,
+      "step": 2818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880777296926562e-06,
+      "loss": 0.7968,
+      "step": 2819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.874816465765607e-06,
+      "loss": 0.2997,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868855679090898e-06,
+      "loss": 0.8245,
+      "step": 2821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862894939020702e-06,
+      "loss": 0.8261,
+      "step": 2822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85693424767328e-06,
+      "loss": 0.791,
+      "step": 2823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850973607166865e-06,
+      "loss": 0.816,
+      "step": 2824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84501301961969e-06,
+      "loss": 0.7878,
+      "step": 2825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839052487149947e-06,
+      "loss": 0.853,
+      "step": 2826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833092011875821e-06,
+      "loss": 0.7832,
+      "step": 2827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827131595915486e-06,
+      "loss": 0.809,
+      "step": 2828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82117124138707e-06,
+      "loss": 0.8184,
+      "step": 2829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815210950408703e-06,
+      "loss": 0.7298,
+      "step": 2830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809250725098476e-06,
+      "loss": 0.7272,
+      "step": 2831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803290567574468e-06,
+      "loss": 0.7945,
+      "step": 2832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797330479954725e-06,
+      "loss": 0.8282,
+      "step": 2833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791370464357279e-06,
+      "loss": 0.2844,
+      "step": 2834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78541052290012e-06,
+      "loss": 0.2582,
+      "step": 2835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779450657701227e-06,
+      "loss": 0.8176,
+      "step": 2836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773490870878551e-06,
+      "loss": 0.8323,
+      "step": 2837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76753116455e-06,
+      "loss": 0.847,
+      "step": 2838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761571540833471e-06,
+      "loss": 0.8762,
+      "step": 2839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755612001846823e-06,
+      "loss": 0.8094,
+      "step": 2840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.749652549707887e-06,
+      "loss": 0.8197,
+      "step": 2841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743693186534458e-06,
+      "loss": 0.8126,
+      "step": 2842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73773391444431e-06,
+      "loss": 0.8138,
+      "step": 2843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731774735555174e-06,
+      "loss": 0.7713,
+      "step": 2844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725815651984755e-06,
+      "loss": 0.7734,
+      "step": 2845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719856665850724e-06,
+      "loss": 0.7723,
+      "step": 2846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713897779270705e-06,
+      "loss": 0.8107,
+      "step": 2847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707938994362309e-06,
+      "loss": 0.7954,
+      "step": 2848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701980313243089e-06,
+      "loss": 0.7652,
+      "step": 2849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.696021738030575e-06,
+      "loss": 0.7755,
+      "step": 2850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690063270842252e-06,
+      "loss": 0.8823,
+      "step": 2851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684104913795575e-06,
+      "loss": 0.7573,
+      "step": 2852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.678146669007946e-06,
+      "loss": 0.8372,
+      "step": 2853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672188538596746e-06,
+      "loss": 0.8794,
+      "step": 2854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666230524679292e-06,
+      "loss": 0.839,
+      "step": 2855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660272629372881e-06,
+      "loss": 0.7959,
+      "step": 2856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654314854794762e-06,
+      "loss": 0.8212,
+      "step": 2857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648357203062127e-06,
+      "loss": 0.8348,
+      "step": 2858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642399676292149e-06,
+      "loss": 0.7633,
+      "step": 2859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636442276601932e-06,
+      "loss": 0.7601,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630485006108554e-06,
+      "loss": 0.8236,
+      "step": 2861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624527866929033e-06,
+      "loss": 0.7921,
+      "step": 2862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618570861180352e-06,
+      "loss": 0.8463,
+      "step": 2863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612613990979436e-06,
+      "loss": 0.7648,
+      "step": 2864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606657258443169e-06,
+      "loss": 0.7729,
+      "step": 2865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600700665688388e-06,
+      "loss": 0.8161,
+      "step": 2866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59474421483187e-06,
+      "loss": 0.8246,
+      "step": 2867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588787907990356e-06,
+      "loss": 0.8035,
+      "step": 2868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582831747280517e-06,
+      "loss": 0.7881,
+      "step": 2869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576875734818993e-06,
+      "loss": 0.8242,
+      "step": 2870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570919872722352e-06,
+      "loss": 0.8207,
+      "step": 2871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564964163107125e-06,
+      "loss": 0.813,
+      "step": 2872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559008608089777e-06,
+      "loss": 0.2636,
+      "step": 2873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553053209786725e-06,
+      "loss": 0.7848,
+      "step": 2874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547097970314328e-06,
+      "loss": 0.8418,
+      "step": 2875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541142891788887e-06,
+      "loss": 0.8314,
+      "step": 2876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53518797632665e-06,
+      "loss": 0.2568,
+      "step": 2877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529233226043799e-06,
+      "loss": 0.8061,
+      "step": 2878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52327864305647e-06,
+      "loss": 0.788,
+      "step": 2879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517324229480724e-06,
+      "loss": 0.8025,
+      "step": 2880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511369987432575e-06,
+      "loss": 0.7705,
+      "step": 2881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505415919027971e-06,
+      "loss": 0.7778,
+      "step": 2882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499462026382795e-06,
+      "loss": 0.8687,
+      "step": 2883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493508311612874e-06,
+      "loss": 0.7743,
+      "step": 2884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487554776833966e-06,
+      "loss": 0.8368,
+      "step": 2885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481601424161772e-06,
+      "loss": 0.747,
+      "step": 2886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47564825571192e-06,
+      "loss": 0.8173,
+      "step": 2887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469695273599982e-06,
+      "loss": 0.7942,
+      "step": 2888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463742479941447e-06,
+      "loss": 0.7786,
+      "step": 2889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457789876851759e-06,
+      "loss": 0.7463,
+      "step": 2890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451837466446278e-06,
+      "loss": 0.7908,
+      "step": 2891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445885250840301e-06,
+      "loss": 0.8798,
+      "step": 2892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43993323214906e-06,
+      "loss": 0.8226,
+      "step": 2893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433981412487711e-06,
+      "loss": 0.845,
+      "step": 2894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42802979397134e-06,
+      "loss": 0.8227,
+      "step": 2895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422078378714962e-06,
+      "loss": 0.8097,
+      "step": 2896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416127168833526e-06,
+      "loss": 0.8413,
+      "step": 2897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410176166441892e-06,
+      "loss": 0.7926,
+      "step": 2898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40422537365487e-06,
+      "loss": 0.7336,
+      "step": 2899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398274792587169e-06,
+      "loss": 0.7733,
+      "step": 2900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392324425353441e-06,
+      "loss": 0.7927,
+      "step": 2901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386374274068263e-06,
+      "loss": 0.8207,
+      "step": 2902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.380424340846122e-06,
+      "loss": 0.7804,
+      "step": 2903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374474627801439e-06,
+      "loss": 0.8034,
+      "step": 2904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368525137048549e-06,
+      "loss": 0.7955,
+      "step": 2905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362575870701715e-06,
+      "loss": 0.7737,
+      "step": 2906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356626830875112e-06,
+      "loss": 0.8135,
+      "step": 2907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350678019682847e-06,
+      "loss": 0.8211,
+      "step": 2908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.344729439238926e-06,
+      "loss": 0.8363,
+      "step": 2909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338781091657297e-06,
+      "loss": 0.7691,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33283297905181e-06,
+      "loss": 0.7535,
+      "step": 2911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32688510353623e-06,
+      "loss": 0.8247,
+      "step": 2912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320937467224248e-06,
+      "loss": 0.8538,
+      "step": 2913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314990072229461e-06,
+      "loss": 0.8157,
+      "step": 2914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.309042920665386e-06,
+      "loss": 0.7985,
+      "step": 2915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30309601464545e-06,
+      "loss": 0.8442,
+      "step": 2916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297149356282995e-06,
+      "loss": 0.8222,
+      "step": 2917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291202947691272e-06,
+      "loss": 0.8032,
+      "step": 2918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285256790983446e-06,
+      "loss": 0.793,
+      "step": 2919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279310888272596e-06,
+      "loss": 0.8178,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273365241671698e-06,
+      "loss": 0.8212,
+      "step": 2921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267419853293656e-06,
+      "loss": 0.7494,
+      "step": 2922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261474725251261e-06,
+      "loss": 0.8332,
+      "step": 2923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25552985965723e-06,
+      "loss": 0.8191,
+      "step": 2924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249585258624174e-06,
+      "loss": 0.8401,
+      "step": 2925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243640924264619e-06,
+      "loss": 0.8507,
+      "step": 2926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237696858690987e-06,
+      "loss": 0.8333,
+      "step": 2927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231753064015611e-06,
+      "loss": 0.8423,
+      "step": 2928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225809542350733e-06,
+      "loss": 0.7314,
+      "step": 2929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219866295808481e-06,
+      "loss": 0.7935,
+      "step": 2930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213923326500907e-06,
+      "loss": 0.8081,
+      "step": 2931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207980636539943e-06,
+      "loss": 0.766,
+      "step": 2932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202038228037436e-06,
+      "loss": 0.8498,
+      "step": 2933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196096103105127e-06,
+      "loss": 0.791,
+      "step": 2934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19015426385466e-06,
+      "loss": 0.8209,
+      "step": 2935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184212712397574e-06,
+      "loss": 0.8799,
+      "step": 2936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178271450845306e-06,
+      "loss": 0.7698,
+      "step": 2937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172330481309195e-06,
+      "loss": 0.847,
+      "step": 2938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166389805900469e-06,
+      "loss": 0.8261,
+      "step": 2939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160449426730261e-06,
+      "loss": 0.7593,
+      "step": 2940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154509345909578e-06,
+      "loss": 0.7558,
+      "step": 2941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148569565549355e-06,
+      "loss": 0.787,
+      "step": 2942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142630087760384e-06,
+      "loss": 0.7697,
+      "step": 2943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136690914653377e-06,
+      "loss": 0.7506,
+      "step": 2944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13075204833892e-06,
+      "loss": 0.8683,
+      "step": 2945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1248134909275e-06,
+      "loss": 0.7388,
+      "step": 2946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118875244529493e-06,
+      "loss": 0.8169,
+      "step": 2947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112937311255158e-06,
+      "loss": 0.7879,
+      "step": 2948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106999693214654e-06,
+      "loss": 0.8159,
+      "step": 2949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101062392518015e-06,
+      "loss": 0.7695,
+      "step": 2950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.095125411275174e-06,
+      "loss": 0.8271,
+      "step": 2951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089188751595937e-06,
+      "loss": 0.72,
+      "step": 2952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083252415590014e-06,
+      "loss": 0.7814,
+      "step": 2953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 0.8088,
+      "step": 2954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071380723036314e-06,
+      "loss": 0.8674,
+      "step": 2955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065445370707359e-06,
+      "loss": 0.8399,
+      "step": 2956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059510350489353e-06,
+      "loss": 0.8157,
+      "step": 2957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053575664491417e-06,
+      "loss": 0.8454,
+      "step": 2958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047641314822546e-06,
+      "loss": 0.8111,
+      "step": 2959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04170730359162e-06,
+      "loss": 0.2608,
+      "step": 2960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035773632907396e-06,
+      "loss": 0.7976,
+      "step": 2961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029840304878517e-06,
+      "loss": 0.7533,
+      "step": 2962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023907321613491e-06,
+      "loss": 0.7968,
+      "step": 2963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017974685220716e-06,
+      "loss": 0.8176,
+      "step": 2964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012042397808459e-06,
+      "loss": 0.8195,
+      "step": 2965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006110461484869e-06,
+      "loss": 0.7874,
+      "step": 2966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000178878357968e-06,
+      "loss": 0.7603,
+      "step": 2967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994247650535645e-06,
+      "loss": 0.8201,
+      "step": 2968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98831678012568e-06,
+      "loss": 0.8402,
+      "step": 2969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982386269235706e-06,
+      "loss": 0.7833,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976456119973246e-06,
+      "loss": 0.7788,
+      "step": 2971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970526334445675e-06,
+      "loss": 0.816,
+      "step": 2972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964596914760265e-06,
+      "loss": 0.7836,
+      "step": 2973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958667863024127e-06,
+      "loss": 0.7728,
+      "step": 2974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952739181344267e-06,
+      "loss": 0.3223,
+      "step": 2975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946810871827548e-06,
+      "loss": 0.751,
+      "step": 2976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9408829365807e-06,
+      "loss": 0.7482,
+      "step": 2977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934955377710326e-06,
+      "loss": 0.7974,
+      "step": 2978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929028197322888e-06,
+      "loss": 0.7214,
+      "step": 2979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923101397524721e-06,
+      "loss": 0.8069,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917174980422016e-06,
+      "loss": 0.7946,
+      "step": 2981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91124894812084e-06,
+      "loss": 0.8362,
+      "step": 2982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90532330272711e-06,
+      "loss": 0.8174,
+      "step": 2983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.899398046346608e-06,
+      "loss": 0.7797,
+      "step": 2984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893473181084993e-06,
+      "loss": 0.2959,
+      "step": 2985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887548709047765e-06,
+      "loss": 0.8641,
+      "step": 2986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881624632340296e-06,
+      "loss": 0.831,
+      "step": 2987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87570095306781e-06,
+      "loss": 0.7537,
+      "step": 2988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8697776733354e-06,
+      "loss": 0.8358,
+      "step": 2989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863854795248007e-06,
+      "loss": 0.7786,
+      "step": 2990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857932320910437e-06,
+      "loss": 0.789,
+      "step": 2991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85201025242734e-06,
+      "loss": 0.7942,
+      "step": 2992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84608859190324e-06,
+      "loss": 0.7714,
+      "step": 2993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.840167341442505e-06,
+      "loss": 0.2668,
+      "step": 2994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834246503149351e-06,
+      "loss": 0.7636,
+      "step": 2995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828326079127867e-06,
+      "loss": 0.815,
+      "step": 2996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822406071481975e-06,
+      "loss": 0.8395,
+      "step": 2997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816486482315459e-06,
+      "loss": 0.7812,
+      "step": 2998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810567313731952e-06,
+      "loss": 0.7912,
+      "step": 2999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804648567834943e-06,
+      "loss": 0.7971,
+      "step": 3000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798730246727758e-06,
+      "loss": 0.8108,
+      "step": 3001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792812352513584e-06,
+      "loss": 0.8189,
+      "step": 3002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786894887295451e-06,
+      "loss": 0.7793,
+      "step": 3003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78097785317624e-06,
+      "loss": 0.2862,
+      "step": 3004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775061252258677e-06,
+      "loss": 0.7908,
+      "step": 3005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769145086645328e-06,
+      "loss": 0.7634,
+      "step": 3006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763229358438616e-06,
+      "loss": 0.759,
+      "step": 3007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757314069740795e-06,
+      "loss": 0.7709,
+      "step": 3008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751399222653979e-06,
+      "loss": 0.7393,
+      "step": 3009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745484819280108e-06,
+      "loss": 0.7246,
+      "step": 3010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739570861720978e-06,
+      "loss": 0.8376,
+      "step": 3011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733657352078223e-06,
+      "loss": 0.8156,
+      "step": 3012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727744292453309e-06,
+      "loss": 0.7382,
+      "step": 3013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721831684947557e-06,
+      "loss": 0.8168,
+      "step": 3014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715919531662111e-06,
+      "loss": 0.7789,
+      "step": 3015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.71000783469797e-06,
+      "loss": 0.7783,
+      "step": 3016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704096596155955e-06,
+      "loss": 0.7683,
+      "step": 3017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69818581813674e-06,
+      "loss": 0.8276,
+      "step": 3018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692275502740818e-06,
+      "loss": 0.7981,
+      "step": 3019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686365652068536e-06,
+      "loss": 0.8066,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680456268220063e-06,
+      "loss": 0.7563,
+      "step": 3021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674547353295401e-06,
+      "loss": 0.8058,
+      "step": 3022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6686389093944e-06,
+      "loss": 0.8026,
+      "step": 3023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662730938616724e-06,
+      "loss": 0.7957,
+      "step": 3024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656823443061885e-06,
+      "loss": 0.7882,
+      "step": 3025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650916424829213e-06,
+      "loss": 0.7806,
+      "step": 3026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645009886017877e-06,
+      "loss": 0.8299,
+      "step": 3027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63910382872687e-06,
+      "loss": 0.808,
+      "step": 3028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633198255055017e-06,
+      "loss": 0.7416,
+      "step": 3029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627293167100975e-06,
+      "loss": 0.7635,
+      "step": 3030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62138856696322e-06,
+      "loss": 0.7664,
+      "step": 3031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615484456740062e-06,
+      "loss": 0.8154,
+      "step": 3032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60958083852963e-06,
+      "loss": 0.7528,
+      "step": 3033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603677714429888e-06,
+      "loss": 0.7895,
+      "step": 3034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597775086538608e-06,
+      "loss": 0.7761,
+      "step": 3035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591872956953409e-06,
+      "loss": 0.8215,
+      "step": 3036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.585971327771707e-06,
+      "loss": 0.2513,
+      "step": 3037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580070201090759e-06,
+      "loss": 0.8202,
+      "step": 3038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574169579007643e-06,
+      "loss": 0.7416,
+      "step": 3039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568269463619243e-06,
+      "loss": 0.2911,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562369857022278e-06,
+      "loss": 0.7369,
+      "step": 3041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556470761313275e-06,
+      "loss": 0.7687,
+      "step": 3042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55057217858859e-06,
+      "loss": 0.8104,
+      "step": 3043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54467411094439e-06,
+      "loss": 0.833,
+      "step": 3044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538776560476661e-06,
+      "loss": 0.7073,
+      "step": 3045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532879529281199e-06,
+      "loss": 0.2869,
+      "step": 3046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526983019453624e-06,
+      "loss": 0.7744,
+      "step": 3047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521087033089374e-06,
+      "loss": 0.8317,
+      "step": 3048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515191572283689e-06,
+      "loss": 0.779,
+      "step": 3049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509296639131628e-06,
+      "loss": 0.8461,
+      "step": 3050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503402235728062e-06,
+      "loss": 0.7834,
+      "step": 3051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497508364167678e-06,
+      "loss": 0.782,
+      "step": 3052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491615026544965e-06,
+      "loss": 0.771,
+      "step": 3053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485722224954237e-06,
+      "loss": 0.818,
+      "step": 3054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479829961489595e-06,
+      "loss": 0.7902,
+      "step": 3055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473938238244972e-06,
+      "loss": 0.2775,
+      "step": 3056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46804705731409e-06,
+      "loss": 0.8225,
+      "step": 3057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462156420790493e-06,
+      "loss": 0.8186,
+      "step": 3058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456266330767529e-06,
+      "loss": 0.8183,
+      "step": 3059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45037678933834e-06,
+      "loss": 0.7633,
+      "step": 3060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444487798595888e-06,
+      "loss": 0.8009,
+      "step": 3061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438599360632927e-06,
+      "loss": 0.7384,
+      "step": 3062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432711477542027e-06,
+      "loss": 0.7971,
+      "step": 3063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426824151415548e-06,
+      "loss": 0.8003,
+      "step": 3064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.420937384345666e-06,
+      "loss": 0.8766,
+      "step": 3065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415051178424337e-06,
+      "loss": 0.2504,
+      "step": 3066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409165535743342e-06,
+      "loss": 0.8087,
+      "step": 3067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403280458394255e-06,
+      "loss": 0.7984,
+      "step": 3068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397395948468432e-06,
+      "loss": 0.8026,
+      "step": 3069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391512008057049e-06,
+      "loss": 0.8374,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385628639251067e-06,
+      "loss": 0.2932,
+      "step": 3071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379745844141253e-06,
+      "loss": 0.8361,
+      "step": 3072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373863624818158e-06,
+      "loss": 0.7775,
+      "step": 3073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367981983372143e-06,
+      "loss": 0.7563,
+      "step": 3074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362100921893349e-06,
+      "loss": 0.8251,
+      "step": 3075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35622044247172e-06,
+      "loss": 0.7549,
+      "step": 3076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350340547197e-06,
+      "loss": 0.7724,
+      "step": 3077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3444612381587e-06,
+      "loss": 0.7891,
+      "step": 3078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338582517446154e-06,
+      "loss": 0.7884,
+      "step": 3079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332704387148463e-06,
+      "loss": 0.7544,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326826849354531e-06,
+      "loss": 0.8381,
+      "step": 3081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320949906153048e-06,
+      "loss": 0.7889,
+      "step": 3082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315073559632492e-06,
+      "loss": 0.8313,
+      "step": 3083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309197811881128e-06,
+      "loss": 0.8104,
+      "step": 3084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303322664987008e-06,
+      "loss": 0.8117,
+      "step": 3085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297448121037978e-06,
+      "loss": 0.839,
+      "step": 3086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.291574182121658e-06,
+      "loss": 0.8197,
+      "step": 3087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285700850325467e-06,
+      "loss": 0.8454,
+      "step": 3088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279828127736588e-06,
+      "loss": 0.8522,
+      "step": 3089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27395601644201e-06,
+      "loss": 0.7194,
+      "step": 3090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268084518528487e-06,
+      "loss": 0.7283,
+      "step": 3091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262213636082567e-06,
+      "loss": 0.8116,
+      "step": 3092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256343371190571e-06,
+      "loss": 0.8301,
+      "step": 3093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250473725938608e-06,
+      "loss": 0.7975,
+      "step": 3094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244604702412563e-06,
+      "loss": 0.8023,
+      "step": 3095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238736302698096e-06,
+      "loss": 0.7624,
+      "step": 3096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232868528880657e-06,
+      "loss": 0.7449,
+      "step": 3097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227001383045453e-06,
+      "loss": 0.7874,
+      "step": 3098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221134867277495e-06,
+      "loss": 0.768,
+      "step": 3099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215268983661547e-06,
+      "loss": 0.8148,
+      "step": 3100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209403734282158e-06,
+      "loss": 0.7739,
+      "step": 3101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.203539121223653e-06,
+      "loss": 0.7977,
+      "step": 3102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197675146570128e-06,
+      "loss": 0.7804,
+      "step": 3103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191811812405453e-06,
+      "loss": 0.7921,
+      "step": 3104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18594912081327e-06,
+      "loss": 0.7055,
+      "step": 3105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180087073876995e-06,
+      "loss": 0.2788,
+      "step": 3106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174225673679808e-06,
+      "loss": 0.8066,
+      "step": 3107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16836492230467e-06,
+      "loss": 0.7639,
+      "step": 3108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162504821834296e-06,
+      "loss": 0.8357,
+      "step": 3109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156645374351194e-06,
+      "loss": 0.7844,
+      "step": 3110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150786581937608e-06,
+      "loss": 0.7666,
+      "step": 3111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144928446675578e-06,
+      "loss": 0.7798,
+      "step": 3112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139070970646895e-06,
+      "loss": 0.7771,
+      "step": 3113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133214155933118e-06,
+      "loss": 0.8379,
+      "step": 3114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127358004615575e-06,
+      "loss": 0.8557,
+      "step": 3115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121502518775355e-06,
+      "loss": 0.7532,
+      "step": 3116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11564770049331e-06,
+      "loss": 0.7422,
+      "step": 3117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109793551850057e-06,
+      "loss": 0.8449,
+      "step": 3118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103940074925976e-06,
+      "loss": 0.7511,
+      "step": 3119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.0980872718012e-06,
+      "loss": 0.8277,
+      "step": 3120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092235144555631e-06,
+      "loss": 0.7446,
+      "step": 3121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086383695268937e-06,
+      "loss": 0.7708,
+      "step": 3122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080532926020527e-06,
+      "loss": 0.7748,
+      "step": 3123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074682838889581e-06,
+      "loss": 0.8096,
+      "step": 3124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068833435955034e-06,
+      "loss": 0.7899,
+      "step": 3125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062984719295576e-06,
+      "loss": 0.7335,
+      "step": 3126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057136690989657e-06,
+      "loss": 0.8199,
+      "step": 3127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051289353115483e-06,
+      "loss": 0.8265,
+      "step": 3128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045442707750999e-06,
+      "loss": 0.806,
+      "step": 3129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039596756973928e-06,
+      "loss": 0.8404,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033751502861738e-06,
+      "loss": 0.817,
+      "step": 3131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027906947491634e-06,
+      "loss": 0.8104,
+      "step": 3132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022063092940593e-06,
+      "loss": 0.2504,
+      "step": 3133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01621994128533e-06,
+      "loss": 0.7379,
+      "step": 3134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010377494602322e-06,
+      "loss": 0.8229,
+      "step": 3135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00453575496778e-06,
+      "loss": 0.8324,
+      "step": 3136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998694724457678e-06,
+      "loss": 0.7517,
+      "step": 3137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99285440514773e-06,
+      "loss": 0.8397,
+      "step": 3138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987014799113398e-06,
+      "loss": 0.7975,
+      "step": 3139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9811759084299e-06,
+      "loss": 0.2853,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975337735172178e-06,
+      "loss": 0.7857,
+      "step": 3141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.969500281414947e-06,
+      "loss": 0.7806,
+      "step": 3142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963663549232643e-06,
+      "loss": 0.8346,
+      "step": 3143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95782754069946e-06,
+      "loss": 0.7728,
+      "step": 3144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951992257889322e-06,
+      "loss": 0.82,
+      "step": 3145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946157702875911e-06,
+      "loss": 0.8836,
+      "step": 3146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940323877732637e-06,
+      "loss": 0.7874,
+      "step": 3147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.934490784532657e-06,
+      "loss": 0.8454,
+      "step": 3148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92865842534887e-06,
+      "loss": 0.8169,
+      "step": 3149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922826802253904e-06,
+      "loss": 0.8222,
+      "step": 3150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916995917320141e-06,
+      "loss": 0.8246,
+      "step": 3151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91116577261968e-06,
+      "loss": 0.8116,
+      "step": 3152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905336370224382e-06,
+      "loss": 0.8073,
+      "step": 3153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899507712205818e-06,
+      "loss": 0.7576,
+      "step": 3154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893679800635316e-06,
+      "loss": 0.2795,
+      "step": 3155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887852637583927e-06,
+      "loss": 0.8332,
+      "step": 3156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882026225122443e-06,
+      "loss": 0.7825,
+      "step": 3157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876200565321377e-06,
+      "loss": 0.8016,
+      "step": 3158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870375660250991e-06,
+      "loss": 0.7783,
+      "step": 3159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864551511981269e-06,
+      "loss": 0.7875,
+      "step": 3160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858728122581925e-06,
+      "loss": 0.8066,
+      "step": 3161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852905494122412e-06,
+      "loss": 0.8391,
+      "step": 3162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847083628671897e-06,
+      "loss": 0.7871,
+      "step": 3163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841262528299296e-06,
+      "loss": 0.8133,
+      "step": 3164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835442195073234e-06,
+      "loss": 0.7853,
+      "step": 3165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829622631062079e-06,
+      "loss": 0.8472,
+      "step": 3166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823803838333912e-06,
+      "loss": 0.7845,
+      "step": 3167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817985818956552e-06,
+      "loss": 0.7753,
+      "step": 3168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812168574997537e-06,
+      "loss": 0.7277,
+      "step": 3169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806352108524129e-06,
+      "loss": 0.8241,
+      "step": 3170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800536421603317e-06,
+      "loss": 0.7789,
+      "step": 3171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794721516301804e-06,
+      "loss": 0.7831,
+      "step": 3172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.788907394686032e-06,
+      "loss": 0.7847,
+      "step": 3173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783094058822147e-06,
+      "loss": 0.8293,
+      "step": 3174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77728151077603e-06,
+      "loss": 0.8165,
+      "step": 3175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771469752613267e-06,
+      "loss": 0.8003,
+      "step": 3176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765658786399177e-06,
+      "loss": 0.8107,
+      "step": 3177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759848614198795e-06,
+      "loss": 0.7667,
+      "step": 3178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.754039238076864e-06,
+      "loss": 0.2746,
+      "step": 3179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74823066009786e-06,
+      "loss": 0.757,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742422882325956e-06,
+      "loss": 0.8699,
+      "step": 3181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736615906825065e-06,
+      "loss": 0.8796,
+      "step": 3182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730809735658788e-06,
+      "loss": 0.8441,
+      "step": 3183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72500437089046e-06,
+      "loss": 0.8318,
+      "step": 3184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71919981458312e-06,
+      "loss": 0.7974,
+      "step": 3185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713396068799521e-06,
+      "loss": 0.8352,
+      "step": 3186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707593135602138e-06,
+      "loss": 0.7658,
+      "step": 3187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701791017053137e-06,
+      "loss": 0.8028,
+      "step": 3188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695989715214418e-06,
+      "loss": 0.8479,
+      "step": 3189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690189232147566e-06,
+      "loss": 0.7822,
+      "step": 3190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.684389569913903e-06,
+      "loss": 0.8637,
+      "step": 3191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678590730574429e-06,
+      "loss": 0.8351,
+      "step": 3192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67279271618988e-06,
+      "loss": 0.7953,
+      "step": 3193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666995528820673e-06,
+      "loss": 0.8301,
+      "step": 3194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661199170526949e-06,
+      "loss": 0.7506,
+      "step": 3195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655403643368557e-06,
+      "loss": 0.8138,
+      "step": 3196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649608949405031e-06,
+      "loss": 0.8756,
+      "step": 3197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64381509069563e-06,
+      "loss": 0.7808,
+      "step": 3198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6380220692993e-06,
+      "loss": 0.7809,
+      "step": 3199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632229887274699e-06,
+      "loss": 0.7888,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626438546680184e-06,
+      "loss": 0.7853,
+      "step": 3201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620648049573815e-06,
+      "loss": 0.8351,
+      "step": 3202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614858398013344e-06,
+      "loss": 0.8362,
+      "step": 3203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.609069594056234e-06,
+      "loss": 0.8043,
+      "step": 3204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603281639759645e-06,
+      "loss": 0.7868,
+      "step": 3205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597494537180423e-06,
+      "loss": 0.7283,
+      "step": 3206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5917082883751256e-06,
+      "loss": 0.7657,
+      "step": 3207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585922895399997e-06,
+      "loss": 0.7939,
+      "step": 3208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580138360310988e-06,
+      "loss": 0.8528,
+      "step": 3209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574354685163732e-06,
+      "loss": 0.8083,
+      "step": 3210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568571872013568e-06,
+      "loss": 0.8063,
+      "step": 3211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562789922915518e-06,
+      "loss": 0.8226,
+      "step": 3212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5570088399243054e-06,
+      "loss": 0.8036,
+      "step": 3213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551228625094349e-06,
+      "loss": 0.744,
+      "step": 3214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54544928047974e-06,
+      "loss": 0.7829,
+      "step": 3215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539670808134286e-06,
+      "loss": 0.7709,
+      "step": 3216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533893210111463e-06,
+      "loss": 0.777,
+      "step": 3217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528116488464452e-06,
+      "loss": 0.7594,
+      "step": 3218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522340645246111e-06,
+      "loss": 0.7928,
+      "step": 3219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516565682508994e-06,
+      "loss": 0.8432,
+      "step": 3220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510791602305335e-06,
+      "loss": 0.7938,
+      "step": 3221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505018406687062e-06,
+      "loss": 0.8534,
+      "step": 3222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499246097705783e-06,
+      "loss": 0.8318,
+      "step": 3223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493474677412795e-06,
+      "loss": 0.7834,
+      "step": 3224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487704147859076e-06,
+      "loss": 0.8374,
+      "step": 3225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481934511095286e-06,
+      "loss": 0.7816,
+      "step": 3226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476165769171773e-06,
+      "loss": 0.804,
+      "step": 3227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4703979241385595e-06,
+      "loss": 0.8111,
+      "step": 3228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4646309780453595e-06,
+      "loss": 0.7994,
+      "step": 3229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458864932941559e-06,
+      "loss": 0.766,
+      "step": 3230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453099790876228e-06,
+      "loss": 0.8642,
+      "step": 3231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447335553898115e-06,
+      "loss": 0.7842,
+      "step": 3232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441572224055644e-06,
+      "loss": 0.7765,
+      "step": 3233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435809803396923e-06,
+      "loss": 0.7437,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430048293969725e-06,
+      "loss": 0.7936,
+      "step": 3235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424287697821517e-06,
+      "loss": 0.7984,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418528016999422e-06,
+      "loss": 0.784,
+      "step": 3237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.412769253550255e-06,
+      "loss": 0.8004,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4070114095204905e-06,
+      "loss": 0.8015,
+      "step": 3239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401254486956285e-06,
+      "loss": 0.7787,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39549848790347e-06,
+      "loss": 0.8619,
+      "step": 3241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389743414407536e-06,
+      "loss": 0.7853,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3839892685136615e-06,
+      "loss": 0.8058,
+      "step": 3243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37823605226668e-06,
+      "loss": 0.7808,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372483767711107e-06,
+      "loss": 0.7806,
+      "step": 3245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3667324168911125e-06,
+      "loss": 0.8209,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360982001850556e-06,
+      "loss": 0.2691,
+      "step": 3247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35523252463294e-06,
+      "loss": 0.7676,
+      "step": 3248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34948398728145e-06,
+      "loss": 0.7794,
+      "step": 3249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.343736391838936e-06,
+      "loss": 0.7581,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337989740347905e-06,
+      "loss": 0.8046,
+      "step": 3251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33224403485054e-06,
+      "loss": 0.806,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3264992773886745e-06,
+      "loss": 0.7883,
+      "step": 3253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320755470003822e-06,
+      "loss": 0.8464,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315012614737137e-06,
+      "loss": 0.8455,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309270713629459e-06,
+      "loss": 0.2787,
+      "step": 3256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303529768721267e-06,
+      "loss": 0.7913,
+      "step": 3257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297789782052716e-06,
+      "loss": 0.7719,
+      "step": 3258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292050755663611e-06,
+      "loss": 0.7999,
+      "step": 3259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2863126915934215e-06,
+      "loss": 0.8056,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280575591881274e-06,
+      "loss": 0.7476,
+      "step": 3261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274839458565945e-06,
+      "loss": 0.8035,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269104293685881e-06,
+      "loss": 0.7728,
+      "step": 3263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 0.7698,
+      "step": 3264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2576368773835725e-06,
+      "loss": 0.7774,
+      "step": 3265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251904630036479e-06,
+      "loss": 0.2576,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246173359274959e-06,
+      "loss": 0.7553,
+      "step": 3267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240443067135718e-06,
+      "loss": 0.8275,
+      "step": 3268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2347137556551186e-06,
+      "loss": 0.8389,
+      "step": 3269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.22898542686918e-06,
+      "loss": 0.7989,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223258082813564e-06,
+      "loss": 0.8524,
+      "step": 3271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21753172552359e-06,
+      "loss": 0.8266,
+      "step": 3272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211806357034217e-06,
+      "loss": 0.7913,
+      "step": 3273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2060819793800665e-06,
+      "loss": 0.7532,
+      "step": 3274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200358594595391e-06,
+      "loss": 0.7779,
+      "step": 3275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194636204714112e-06,
+      "loss": 0.799,
+      "step": 3276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1889148117697695e-06,
+      "loss": 0.8122,
+      "step": 3277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183194417795569e-06,
+      "loss": 0.8368,
+      "step": 3278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1774750248243675e-06,
+      "loss": 0.7751,
+      "step": 3279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171756634888641e-06,
+      "loss": 0.7563,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166039250020529e-06,
+      "loss": 0.2429,
+      "step": 3281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160322872251807e-06,
+      "loss": 0.8066,
+      "step": 3282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1546075036138955e-06,
+      "loss": 0.8011,
+      "step": 3283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148893146137852e-06,
+      "loss": 0.7555,
+      "step": 3284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1431798018543805e-06,
+      "loss": 0.7375,
+      "step": 3285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137467472793815e-06,
+      "loss": 0.7305,
+      "step": 3286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131756160986141e-06,
+      "loss": 0.7552,
+      "step": 3287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12604586846098e-06,
+      "loss": 0.7862,
+      "step": 3288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120336597247577e-06,
+      "loss": 0.8068,
+      "step": 3289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114628349374838e-06,
+      "loss": 0.8048,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108921126871282e-06,
+      "loss": 0.8166,
+      "step": 3291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10321493176508e-06,
+      "loss": 0.7786,
+      "step": 3292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0975097660840285e-06,
+      "loss": 0.7938,
+      "step": 3293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091805631855566e-06,
+      "loss": 0.7757,
+      "step": 3294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086102531106755e-06,
+      "loss": 0.798,
+      "step": 3295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0804004658642975e-06,
+      "loss": 0.7786,
+      "step": 3296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074699438154533e-06,
+      "loss": 0.8415,
+      "step": 3297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068999450003411e-06,
+      "loss": 0.8131,
+      "step": 3298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063300503436541e-06,
+      "loss": 0.8027,
+      "step": 3299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0576026004791345e-06,
+      "loss": 0.8331,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0519057431560535e-06,
+      "loss": 0.8286,
+      "step": 3301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0462099334917745e-06,
+      "loss": 0.8278,
+      "step": 3302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040515173510409e-06,
+      "loss": 0.8329,
+      "step": 3303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.034821465235693e-06,
+      "loss": 0.8271,
+      "step": 3304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029128810690986e-06,
+      "loss": 0.8059,
+      "step": 3305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023437211899285e-06,
+      "loss": 0.7898,
+      "step": 3306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017746670883192e-06,
+      "loss": 0.7819,
+      "step": 3307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012057189664954e-06,
+      "loss": 0.851,
+      "step": 3308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006368770266421e-06,
+      "loss": 0.77,
+      "step": 3309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000681414709086e-06,
+      "loss": 0.7827,
+      "step": 3310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994995125014047e-06,
+      "loss": 0.8364,
+      "step": 3311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989309903202035e-06,
+      "loss": 0.7956,
+      "step": 3312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983625751293392e-06,
+      "loss": 0.7628,
+      "step": 3313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977942671308087e-06,
+      "loss": 0.7517,
+      "step": 3314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972260665265708e-06,
+      "loss": 0.8225,
+      "step": 3315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966579735185455e-06,
+      "loss": 0.7638,
+      "step": 3316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960899883086154e-06,
+      "loss": 0.778,
+      "step": 3317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955221110986237e-06,
+      "loss": 0.7408,
+      "step": 3318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9495434209037686e-06,
+      "loss": 0.7913,
+      "step": 3319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943866814856408e-06,
+      "loss": 0.7602,
+      "step": 3320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9381912948614485e-06,
+      "loss": 0.8507,
+      "step": 3321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932516862935783e-06,
+      "loss": 0.7464,
+      "step": 3322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9268435210959274e-06,
+      "loss": 0.8008,
+      "step": 3323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921171271358007e-06,
+      "loss": 0.789,
+      "step": 3324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915500115737757e-06,
+      "loss": 0.8095,
+      "step": 3325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.7994,
+      "step": 3326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904161094911273e-06,
+      "loss": 0.762,
+      "step": 3327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898493233734571e-06,
+      "loss": 0.7861,
+      "step": 3328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892826474734584e-06,
+      "loss": 0.7975,
+      "step": 3329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8871608199251135e-06,
+      "loss": 0.8189,
+      "step": 3330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881496271319541e-06,
+      "loss": 0.8317,
+      "step": 3331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.875832830930872e-06,
+      "loss": 0.7969,
+      "step": 3332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870170500771713e-06,
+      "loss": 0.8185,
+      "step": 3333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864509282854272e-06,
+      "loss": 0.8207,
+      "step": 3334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8588491791903724e-06,
+      "loss": 0.8126,
+      "step": 3335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853190191791428e-06,
+      "loss": 0.8365,
+      "step": 3336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847532322668467e-06,
+      "loss": 0.7729,
+      "step": 3337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841875573832111e-06,
+      "loss": 0.7795,
+      "step": 3338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836219947292597e-06,
+      "loss": 0.7448,
+      "step": 3339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.830565445059745e-06,
+      "loss": 0.8153,
+      "step": 3340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824912069142986e-06,
+      "loss": 0.7383,
+      "step": 3341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.819259821551361e-06,
+      "loss": 0.8212,
+      "step": 3342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813608704293485e-06,
+      "loss": 0.7881,
+      "step": 3343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8079587193775935e-06,
+      "loss": 0.7442,
+      "step": 3344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802309868811505e-06,
+      "loss": 0.8056,
+      "step": 3345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796662154602648e-06,
+      "loss": 0.8276,
+      "step": 3346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791015578758035e-06,
+      "loss": 0.8407,
+      "step": 3347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785370143284285e-06,
+      "loss": 0.7925,
+      "step": 3348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779725850187594e-06,
+      "loss": 0.8269,
+      "step": 3349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774082701473774e-06,
+      "loss": 0.8112,
+      "step": 3350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768440699148225e-06,
+      "loss": 0.7653,
+      "step": 3351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76279984521592e-06,
+      "loss": 0.7961,
+      "step": 3352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757160141681454e-06,
+      "loss": 0.7602,
+      "step": 3353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751521590548986e-06,
+      "loss": 0.8006,
+      "step": 3354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745884193822285e-06,
+      "loss": 0.792,
+      "step": 3355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740247953504697e-06,
+      "loss": 0.8228,
+      "step": 3356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734612871599169e-06,
+      "loss": 0.7905,
+      "step": 3357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728978950108222e-06,
+      "loss": 0.783,
+      "step": 3358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723346191033981e-06,
+      "loss": 0.7745,
+      "step": 3359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717714596378138e-06,
+      "loss": 0.8194,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7120841681419905e-06,
+      "loss": 0.858,
+      "step": 3361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706454908326414e-06,
+      "loss": 0.7614,
+      "step": 3362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.700826818931862e-06,
+      "loss": 0.8475,
+      "step": 3363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695199901958386e-06,
+      "loss": 0.2534,
+      "step": 3364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689574159405605e-06,
+      "loss": 0.7654,
+      "step": 3365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683949593272734e-06,
+      "loss": 0.8001,
+      "step": 3366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678326205558563e-06,
+      "loss": 0.7636,
+      "step": 3367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672703998261465e-06,
+      "loss": 0.8167,
+      "step": 3368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667082973379393e-06,
+      "loss": 0.8039,
+      "step": 3369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66146313290988e-06,
+      "loss": 0.8444,
+      "step": 3370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655844478850045e-06,
+      "loss": 0.8005,
+      "step": 3371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650227013196564e-06,
+      "loss": 0.7582,
+      "step": 3372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644610737945722e-06,
+      "loss": 0.781,
+      "step": 3373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638995655093351e-06,
+      "loss": 0.8052,
+      "step": 3374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633381766634881e-06,
+      "loss": 0.7484,
+      "step": 3375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6277690745653044e-06,
+      "loss": 0.7914,
+      "step": 3376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622157580879196e-06,
+      "loss": 0.7886,
+      "step": 3377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6165472875707005e-06,
+      "loss": 0.8021,
+      "step": 3378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610938196633536e-06,
+      "loss": 0.7808,
+      "step": 3379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605330310061e-06,
+      "loss": 0.738,
+      "step": 3380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599723629845951e-06,
+      "loss": 0.7927,
+      "step": 3381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594118157980833e-06,
+      "loss": 0.2451,
+      "step": 3382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58851389645764e-06,
+      "loss": 0.8201,
+      "step": 3383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582910847267957e-06,
+      "loss": 0.8215,
+      "step": 3384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577309012402924e-06,
+      "loss": 0.8028,
+      "step": 3385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.57170839385326e-06,
+      "loss": 0.7783,
+      "step": 3386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566108993609242e-06,
+      "loss": 0.8274,
+      "step": 3387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560510813660719e-06,
+      "loss": 0.7925,
+      "step": 3388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554913855997109e-06,
+      "loss": 0.8564,
+      "step": 3389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549318122607389e-06,
+      "loss": 0.761,
+      "step": 3390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543723615480111e-06,
+      "loss": 0.7681,
+      "step": 3391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538130336603372e-06,
+      "loss": 0.8107,
+      "step": 3392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5325382879648605e-06,
+      "loss": 0.7693,
+      "step": 3393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526947471551799e-06,
+      "loss": 0.755,
+      "step": 3394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521357889350994e-06,
+      "loss": 0.7962,
+      "step": 3395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5157695433488e-06,
+      "loss": 0.8203,
+      "step": 3396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.51018243553114e-06,
+      "loss": 0.8125,
+      "step": 3397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504596567883496e-06,
+      "loss": 0.7569,
+      "step": 3398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499011942390903e-06,
+      "loss": 0.7985,
+      "step": 3399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4934285610379624e-06,
+      "loss": 0.7793,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487846425808828e-06,
+      "loss": 0.7742,
+      "step": 3401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482265538687217e-06,
+      "loss": 0.7951,
+      "step": 3402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47668590165639e-06,
+      "loss": 0.7706,
+      "step": 3403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.471107516699183e-06,
+      "loss": 0.7666,
+      "step": 3404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465530385797969e-06,
+      "loss": 0.7889,
+      "step": 3405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.459954510934684e-06,
+      "loss": 0.7652,
+      "step": 3406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4543798940908186e-06,
+      "loss": 0.7527,
+      "step": 3407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448806537247412e-06,
+      "loss": 0.8137,
+      "step": 3408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44323444238506e-06,
+      "loss": 0.8058,
+      "step": 3409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437663611483905e-06,
+      "loss": 0.7677,
+      "step": 3410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4320940465236455e-06,
+      "loss": 0.7681,
+      "step": 3411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426525749483526e-06,
+      "loss": 0.7458,
+      "step": 3412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420958722342343e-06,
+      "loss": 0.8118,
+      "step": 3413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415392967078438e-06,
+      "loss": 0.8278,
+      "step": 3414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409828485669701e-06,
+      "loss": 0.8085,
+      "step": 3415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404265280093581e-06,
+      "loss": 0.8318,
+      "step": 3416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398703352327056e-06,
+      "loss": 0.7943,
+      "step": 3417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393142704346661e-06,
+      "loss": 0.8003,
+      "step": 3418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387583338128471e-06,
+      "loss": 0.7779,
+      "step": 3419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382025255648112e-06,
+      "loss": 0.8247,
+      "step": 3420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376468458880744e-06,
+      "loss": 0.7991,
+      "step": 3421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370912949801083e-06,
+      "loss": 0.8222,
+      "step": 3422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365358730383366e-06,
+      "loss": 0.7835,
+      "step": 3423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3598058026013995e-06,
+      "loss": 0.7777,
+      "step": 3424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354254168428517e-06,
+      "loss": 0.7626,
+      "step": 3425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3487038298375836e-06,
+      "loss": 0.799,
+      "step": 3426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34315478880102e-06,
+      "loss": 0.7756,
+      "step": 3427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337607047290774e-06,
+      "loss": 0.7757,
+      "step": 3428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332060607278338e-06,
+      "loss": 0.2481,
+      "step": 3429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326515470734741e-06,
+      "loss": 0.8132,
+      "step": 3430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3209716396305485e-06,
+      "loss": 0.7991,
+      "step": 3431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315429115935858e-06,
+      "loss": 0.8353,
+      "step": 3432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309887901620309e-06,
+      "loss": 0.8233,
+      "step": 3433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304347998653074e-06,
+      "loss": 0.7909,
+      "step": 3434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2988094090028506e-06,
+      "loss": 0.7867,
+      "step": 3435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293272134637889e-06,
+      "loss": 0.7976,
+      "step": 3436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287736177525949e-06,
+      "loss": 0.7634,
+      "step": 3437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282201539634341e-06,
+      "loss": 0.8012,
+      "step": 3438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276668222929894e-06,
+      "loss": 0.7652,
+      "step": 3439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271136229378976e-06,
+      "loss": 0.7963,
+      "step": 3440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265605560947476e-06,
+      "loss": 0.789,
+      "step": 3441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.260076219600823e-06,
+      "loss": 0.6976,
+      "step": 3442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25454820730397e-06,
+      "loss": 0.8348,
+      "step": 3443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24902152602139e-06,
+      "loss": 0.7964,
+      "step": 3444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2434961777170985e-06,
+      "loss": 0.7956,
+      "step": 3445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237972164354617e-06,
+      "loss": 0.7884,
+      "step": 3446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232449487897014e-06,
+      "loss": 0.7743,
+      "step": 3447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226928150306866e-06,
+      "loss": 0.7698,
+      "step": 3448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221408153546284e-06,
+      "loss": 0.2722,
+      "step": 3449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215889499576898e-06,
+      "loss": 0.7601,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.210372190359861e-06,
+      "loss": 0.7474,
+      "step": 3451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204856227855853e-06,
+      "loss": 0.785,
+      "step": 3452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199341614025065e-06,
+      "loss": 0.7824,
+      "step": 3453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193828350827222e-06,
+      "loss": 0.7683,
+      "step": 3454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188316440221556e-06,
+      "loss": 0.8245,
+      "step": 3455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18280588416683e-06,
+      "loss": 0.7607,
+      "step": 3456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177296684621314e-06,
+      "loss": 0.7921,
+      "step": 3457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171788843542809e-06,
+      "loss": 0.7997,
+      "step": 3458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166282362888618e-06,
+      "loss": 0.8045,
+      "step": 3459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160777244615578e-06,
+      "loss": 0.7476,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155273490680025e-06,
+      "loss": 0.7499,
+      "step": 3461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149771103037821e-06,
+      "loss": 0.8531,
+      "step": 3462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144270083644341e-06,
+      "loss": 0.7762,
+      "step": 3463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1387704344544684e-06,
+      "loss": 0.7886,
+      "step": 3464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133272157422611e-06,
+      "loss": 0.8099,
+      "step": 3465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127775254502668e-06,
+      "loss": 0.761,
+      "step": 3466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122279727648078e-06,
+      "loss": 0.2585,
+      "step": 3467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1167855788117655e-06,
+      "loss": 0.8231,
+      "step": 3468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111292809946183e-06,
+      "loss": 0.8266,
+      "step": 3469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1058014230032795e-06,
+      "loss": 0.7644,
+      "step": 3470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100311419934523e-06,
+      "loss": 0.814,
+      "step": 3471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094822802690886e-06,
+      "loss": 0.7587,
+      "step": 3472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0893355732228455e-06,
+      "loss": 0.8231,
+      "step": 3473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083849733480394e-06,
+      "loss": 0.8058,
+      "step": 3474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078365285413015e-06,
+      "loss": 0.8194,
+      "step": 3475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072882230969716e-06,
+      "loss": 0.7819,
+      "step": 3476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067400572098994e-06,
+      "loss": 0.7468,
+      "step": 3477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061920310748858e-06,
+      "loss": 0.7468,
+      "step": 3478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056441448866817e-06,
+      "loss": 0.8066,
+      "step": 3479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050963988399883e-06,
+      "loss": 0.8734,
+      "step": 3480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0454879312945755e-06,
+      "loss": 0.7588,
+      "step": 3481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040013279496908e-06,
+      "loss": 0.7667,
+      "step": 3482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034540034952398e-06,
+      "loss": 0.7903,
+      "step": 3483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0290681996060605e-06,
+      "loss": 0.7712,
+      "step": 3484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023597775402416e-06,
+      "loss": 0.7916,
+      "step": 3485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018128764285471e-06,
+      "loss": 0.8124,
+      "step": 3486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012661168198748e-06,
+      "loss": 0.8315,
+      "step": 3487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007194989085247e-06,
+      "loss": 0.777,
+      "step": 3488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001730228887479e-06,
+      "loss": 0.7127,
+      "step": 3489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9962668895474486e-06,
+      "loss": 0.804,
+      "step": 3490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9908049730066454e-06,
+      "loss": 0.8104,
+      "step": 3491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985344481206067e-06,
+      "loss": 0.7613,
+      "step": 3492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979885416086193e-06,
+      "loss": 0.8569,
+      "step": 3493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974427779587004e-06,
+      "loss": 0.7205,
+      "step": 3494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96897157364797e-06,
+      "loss": 0.7518,
+      "step": 3495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963516800208056e-06,
+      "loss": 0.8214,
+      "step": 3496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958063461205706e-06,
+      "loss": 0.2735,
+      "step": 3497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952611558578871e-06,
+      "loss": 0.7565,
+      "step": 3498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947161094264984e-06,
+      "loss": 0.7928,
+      "step": 3499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9417120702009604e-06,
+      "loss": 0.8118,
+      "step": 3500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9362644883232135e-06,
+      "loss": 0.8286,
+      "step": 3501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930818350567639e-06,
+      "loss": 0.7721,
+      "step": 3502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.925373658869622e-06,
+      "loss": 0.7649,
+      "step": 3503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919930415164033e-06,
+      "loss": 0.7646,
+      "step": 3504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.914488621385228e-06,
+      "loss": 0.767,
+      "step": 3505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90904827946704e-06,
+      "loss": 0.8017,
+      "step": 3506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903609391342801e-06,
+      "loss": 0.7518,
+      "step": 3507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898171958945322e-06,
+      "loss": 0.8003,
+      "step": 3508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892735984206881e-06,
+      "loss": 0.7924,
+      "step": 3509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887301469059262e-06,
+      "loss": 0.7879,
+      "step": 3510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881868415433711e-06,
+      "loss": 0.8141,
+      "step": 3511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876436825260967e-06,
+      "loss": 0.8043,
+      "step": 3512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87100670047124e-06,
+      "loss": 0.8252,
+      "step": 3513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865578042994227e-06,
+      "loss": 0.7788,
+      "step": 3514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860150854759097e-06,
+      "loss": 0.8296,
+      "step": 3515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854725137694501e-06,
+      "loss": 0.7476,
+      "step": 3516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849300893728569e-06,
+      "loss": 0.7754,
+      "step": 3517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8438781247889e-06,
+      "loss": 0.8135,
+      "step": 3518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838456832802579e-06,
+      "loss": 0.828,
+      "step": 3519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8330370196961506e-06,
+      "loss": 0.7776,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8276186873956555e-06,
+      "loss": 0.7331,
+      "step": 3521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822201837826589e-06,
+      "loss": 0.8311,
+      "step": 3522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816786472913929e-06,
+      "loss": 0.814,
+      "step": 3523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8113725945821245e-06,
+      "loss": 0.8354,
+      "step": 3524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80596020475509e-06,
+      "loss": 0.7982,
+      "step": 3525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800549305356224e-06,
+      "loss": 0.7528,
+      "step": 3526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795139898308386e-06,
+      "loss": 0.8077,
+      "step": 3527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789731985533906e-06,
+      "loss": 0.7623,
+      "step": 3528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7843255689545784e-06,
+      "loss": 0.7763,
+      "step": 3529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7789206504916815e-06,
+      "loss": 0.8093,
+      "step": 3530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773517232065946e-06,
+      "loss": 0.78,
+      "step": 3531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7681153155975755e-06,
+      "loss": 0.7699,
+      "step": 3532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7627149030062344e-06,
+      "loss": 0.8102,
+      "step": 3533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757315996211066e-06,
+      "loss": 0.7793,
+      "step": 3534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751918597130665e-06,
+      "loss": 0.7468,
+      "step": 3535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746522707683093e-06,
+      "loss": 0.7769,
+      "step": 3536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741128329785885e-06,
+      "loss": 0.8624,
+      "step": 3537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735735465356021e-06,
+      "loss": 0.7849,
+      "step": 3538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73034411630996e-06,
+      "loss": 0.8133,
+      "step": 3539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724954284563615e-06,
+      "loss": 0.2403,
+      "step": 3540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719565972032359e-06,
+      "loss": 0.7491,
+      "step": 3541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714179180631024e-06,
+      "loss": 0.8316,
+      "step": 3542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708793912273911e-06,
+      "loss": 0.2563,
+      "step": 3543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703410168874768e-06,
+      "loss": 0.2267,
+      "step": 3544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.698027952346804e-06,
+      "loss": 0.792,
+      "step": 3545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692647264602698e-06,
+      "loss": 0.7701,
+      "step": 3546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68726810755456e-06,
+      "loss": 0.809,
+      "step": 3547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.681890483113982e-06,
+      "loss": 0.7964,
+      "step": 3548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676514393191997e-06,
+      "loss": 0.7702,
+      "step": 3549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671139839699096e-06,
+      "loss": 0.7749,
+      "step": 3550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66576682454522e-06,
+      "loss": 0.7257,
+      "step": 3551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660395349639776e-06,
+      "loss": 0.7651,
+      "step": 3552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6550254168916105e-06,
+      "loss": 0.7731,
+      "step": 3553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649657028209024e-06,
+      "loss": 0.8241,
+      "step": 3554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64429018549978e-06,
+      "loss": 0.7884,
+      "step": 3555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638924890671069e-06,
+      "loss": 0.7722,
+      "step": 3556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63356114562956e-06,
+      "loss": 0.7892,
+      "step": 3557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62819895228135e-06,
+      "loss": 0.813,
+      "step": 3558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622838312531993e-06,
+      "loss": 0.2631,
+      "step": 3559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6174792282864865e-06,
+      "loss": 0.792,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612121701449285e-06,
+      "loss": 0.8487,
+      "step": 3561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6067657339242785e-06,
+      "loss": 0.8097,
+      "step": 3562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601411327614805e-06,
+      "loss": 0.7601,
+      "step": 3563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5960584844236565e-06,
+      "loss": 0.8118,
+      "step": 3564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.59070720625306e-06,
+      "loss": 0.8156,
+      "step": 3565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585357495004689e-06,
+      "loss": 0.8218,
+      "step": 3566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58000935257966e-06,
+      "loss": 0.7829,
+      "step": 3567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574662780878533e-06,
+      "loss": 0.8002,
+      "step": 3568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569317781801305e-06,
+      "loss": 0.7597,
+      "step": 3569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563974357247428e-06,
+      "loss": 0.7919,
+      "step": 3570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558632509115778e-06,
+      "loss": 0.7987,
+      "step": 3571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553292239304675e-06,
+      "loss": 0.8299,
+      "step": 3572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5479535497118906e-06,
+      "loss": 0.7974,
+      "step": 3573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 0.7733,
+      "step": 3574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537280918769497e-06,
+      "loss": 0.8137,
+      "step": 3575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531946981212599e-06,
+      "loss": 0.8141,
+      "step": 3576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526614631459446e-06,
+      "loss": 0.2634,
+      "step": 3577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521283871404972e-06,
+      "loss": 0.7789,
+      "step": 3578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515954702943566e-06,
+      "loss": 0.8541,
+      "step": 3579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510627127969048e-06,
+      "loss": 0.7463,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50530114837466e-06,
+      "loss": 0.8349,
+      "step": 3581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4999767660530925e-06,
+      "loss": 0.7849,
+      "step": 3582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494653982896461e-06,
+      "loss": 0.7609,
+      "step": 3583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48933280079631e-06,
+      "loss": 0.8284,
+      "step": 3584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484013221643616e-06,
+      "loss": 0.8083,
+      "step": 3585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478695247328798e-06,
+      "loss": 0.8012,
+      "step": 3586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473378879741682e-06,
+      "loss": 0.8262,
+      "step": 3587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468064120771544e-06,
+      "loss": 0.7507,
+      "step": 3588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46275097230708e-06,
+      "loss": 0.7754,
+      "step": 3589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457439436236407e-06,
+      "loss": 0.7607,
+      "step": 3590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4521295144470835e-06,
+      "loss": 0.808,
+      "step": 3591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446821208826083e-06,
+      "loss": 0.7943,
+      "step": 3592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4415145212598095e-06,
+      "loss": 0.7651,
+      "step": 3593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436209453634087e-06,
+      "loss": 0.2707,
+      "step": 3594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430906007834178e-06,
+      "loss": 0.7452,
+      "step": 3595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4256041857447415e-06,
+      "loss": 0.8408,
+      "step": 3596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420303989249891e-06,
+      "loss": 0.7858,
+      "step": 3597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415005420233141e-06,
+      "loss": 0.7997,
+      "step": 3598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409708480577431e-06,
+      "loss": 0.7921,
+      "step": 3599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404413172165133e-06,
+      "loss": 0.76,
+      "step": 3600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399119496878027e-06,
+      "loss": 0.8318,
+      "step": 3601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393827456597317e-06,
+      "loss": 0.7906,
+      "step": 3602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38853705320362e-06,
+      "loss": 0.7566,
+      "step": 3603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3832482885769855e-06,
+      "loss": 0.7587,
+      "step": 3604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3779611645968696e-06,
+      "loss": 0.7761,
+      "step": 3605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372675683142146e-06,
+      "loss": 0.8002,
+      "step": 3606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3673918460911025e-06,
+      "loss": 0.7456,
+      "step": 3607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362109655321455e-06,
+      "loss": 0.8079,
+      "step": 3608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3568291127103225e-06,
+      "loss": 0.781,
+      "step": 3609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35155022013424e-06,
+      "loss": 0.7366,
+      "step": 3610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346272979469159e-06,
+      "loss": 0.7637,
+      "step": 3611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340997392590439e-06,
+      "loss": 0.7986,
+      "step": 3612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335723461372863e-06,
+      "loss": 0.8356,
+      "step": 3613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330451187690614e-06,
+      "loss": 0.8005,
+      "step": 3614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32518057341729e-06,
+      "loss": 0.7753,
+      "step": 3615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319911620425897e-06,
+      "loss": 0.8049,
+      "step": 3616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3146443305888586e-06,
+      "loss": 0.7629,
+      "step": 3617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309378705778e-06,
+      "loss": 0.7603,
+      "step": 3618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304114747864551e-06,
+      "loss": 0.8241,
+      "step": 3619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298852458719168e-06,
+      "loss": 0.7944,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293591840211885e-06,
+      "loss": 0.7559,
+      "step": 3621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288332894212168e-06,
+      "loss": 0.8107,
+      "step": 3622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283075622588878e-06,
+      "loss": 0.8149,
+      "step": 3623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277820027210279e-06,
+      "loss": 0.7747,
+      "step": 3624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272566109944039e-06,
+      "loss": 0.7566,
+      "step": 3625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267313872657242e-06,
+      "loss": 0.2484,
+      "step": 3626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2620633172163625e-06,
+      "loss": 0.8269,
+      "step": 3627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256814445487275e-06,
+      "loss": 0.8153,
+      "step": 3628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251567259335273e-06,
+      "loss": 0.7602,
+      "step": 3629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246321760625025e-06,
+      "loss": 0.8484,
+      "step": 3630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241077951220626e-06,
+      "loss": 0.7572,
+      "step": 3631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235835832985552e-06,
+      "loss": 0.7979,
+      "step": 3632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230595407782688e-06,
+      "loss": 0.7768,
+      "step": 3633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225356677474309e-06,
+      "loss": 0.8136,
+      "step": 3634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220119643922102e-06,
+      "loss": 0.8006,
+      "step": 3635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214884308987136e-06,
+      "loss": 0.8094,
+      "step": 3636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20965067452988e-06,
+      "loss": 0.7668,
+      "step": 3637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204418742410212e-06,
+      "loss": 0.8122,
+      "step": 3638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199188514487379e-06,
+      "loss": 0.7851,
+      "step": 3639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193959992620048e-06,
+      "loss": 0.7812,
+      "step": 3640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1887331786662675e-06,
+      "loss": 0.7938,
+      "step": 3641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183508074483478e-06,
+      "loss": 0.7911,
+      "step": 3642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1782846819285114e-06,
+      "loss": 0.7649,
+      "step": 3643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1730630028576055e-06,
+      "loss": 0.7363,
+      "step": 3644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167843039126372e-06,
+      "loss": 0.7813,
+      "step": 3645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1626247925898175e-06,
+      "loss": 0.7397,
+      "step": 3646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1574082651023484e-06,
+      "loss": 0.7888,
+      "step": 3647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1521934585177465e-06,
+      "loss": 0.7869,
+      "step": 3648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146980374689192e-06,
+      "loss": 0.7833,
+      "step": 3649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14176901546924e-06,
+      "loss": 0.7859,
+      "step": 3650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136559382709858e-06,
+      "loss": 0.293,
+      "step": 3651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131351478262364e-06,
+      "loss": 0.8333,
+      "step": 3652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126145303977497e-06,
+      "loss": 0.783,
+      "step": 3653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120940861705357e-06,
+      "loss": 0.7237,
+      "step": 3654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115738153295436e-06,
+      "loss": 0.7802,
+      "step": 3655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11053718059662e-06,
+      "loss": 0.7295,
+      "step": 3656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105337945457161e-06,
+      "loss": 0.7691,
+      "step": 3657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1001404497247055e-06,
+      "loss": 0.7771,
+      "step": 3658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0949446952462724e-06,
+      "loss": 0.7802,
+      "step": 3659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089750683868279e-06,
+      "loss": 0.8096,
+      "step": 3660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084558417436496e-06,
+      "loss": 0.7671,
+      "step": 3661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079367897796102e-06,
+      "loss": 0.7879,
+      "step": 3662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0741791267916365e-06,
+      "loss": 0.7911,
+      "step": 3663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068992106267021e-06,
+      "loss": 0.7856,
+      "step": 3664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.063806838065566e-06,
+      "loss": 0.7461,
+      "step": 3665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058623324029944e-06,
+      "loss": 0.7251,
+      "step": 3666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053441566002214e-06,
+      "loss": 0.7369,
+      "step": 3667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.048261565823802e-06,
+      "loss": 0.8027,
+      "step": 3668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043083325335527e-06,
+      "loss": 0.8459,
+      "step": 3669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037906846377556e-06,
+      "loss": 0.812,
+      "step": 3670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032732130789456e-06,
+      "loss": 0.755,
+      "step": 3671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027559180410151e-06,
+      "loss": 0.7893,
+      "step": 3672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022387997077942e-06,
+      "loss": 0.7412,
+      "step": 3673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017218582630507e-06,
+      "loss": 0.8136,
+      "step": 3674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012050938904889e-06,
+      "loss": 0.802,
+      "step": 3675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0068850677375036e-06,
+      "loss": 0.8303,
+      "step": 3676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001720970964135e-06,
+      "loss": 0.7961,
+      "step": 3677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996558650419942e-06,
+      "loss": 0.2839,
+      "step": 3678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99139810793945e-06,
+      "loss": 0.7922,
+      "step": 3679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98623934535655e-06,
+      "loss": 0.7868,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981082364504499e-06,
+      "loss": 0.7576,
+      "step": 3681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975927167215924e-06,
+      "loss": 0.8247,
+      "step": 3682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970773755322823e-06,
+      "loss": 0.7543,
+      "step": 3683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965622130656551e-06,
+      "loss": 0.79,
+      "step": 3684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960472295047832e-06,
+      "loss": 0.765,
+      "step": 3685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955324250326749e-06,
+      "loss": 0.7438,
+      "step": 3686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9501779983227616e-06,
+      "loss": 0.7811,
+      "step": 3687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94503354086468e-06,
+      "loss": 0.7981,
+      "step": 3688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9398908797806785e-06,
+      "loss": 0.7024,
+      "step": 3689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934750016898295e-06,
+      "loss": 0.7831,
+      "step": 3690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929610954044433e-06,
+      "loss": 0.7573,
+      "step": 3691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924473693045349e-06,
+      "loss": 0.7598,
+      "step": 3692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919338235726657e-06,
+      "loss": 0.8379,
+      "step": 3693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914204583913349e-06,
+      "loss": 0.7974,
+      "step": 3694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909072739429744e-06,
+      "loss": 0.8339,
+      "step": 3695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903942704099548e-06,
+      "loss": 0.7903,
+      "step": 3696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898814479745808e-06,
+      "loss": 0.8202,
+      "step": 3697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893688068190933e-06,
+      "loss": 0.8503,
+      "step": 3698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888563471256681e-06,
+      "loss": 0.7968,
+      "step": 3699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8834406907641784e-06,
+      "loss": 0.7805,
+      "step": 3700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878319728533896e-06,
+      "loss": 0.2846,
+      "step": 3701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8732005863856545e-06,
+      "loss": 0.7105,
+      "step": 3702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868083266138647e-06,
+      "loss": 0.7623,
+      "step": 3703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862967769611389e-06,
+      "loss": 0.781,
+      "step": 3704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8578540986217805e-06,
+      "loss": 0.819,
+      "step": 3705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8527422549870495e-06,
+      "loss": 0.8562,
+      "step": 3706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847632240523784e-06,
+      "loss": 0.7841,
+      "step": 3707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842524057047916e-06,
+      "loss": 0.8296,
+      "step": 3708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837417706374739e-06,
+      "loss": 0.8247,
+      "step": 3709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832313190318886e-06,
+      "loss": 0.8263,
+      "step": 3710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827210510694331e-06,
+      "loss": 0.7649,
+      "step": 3711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822109669314419e-06,
+      "loss": 0.787,
+      "step": 3712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817010667991809e-06,
+      "loss": 0.7801,
+      "step": 3713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8119135085385375e-06,
+      "loss": 0.7797,
+      "step": 3714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806818192765966e-06,
+      "loss": 0.8221,
+      "step": 3715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801724722484809e-06,
+      "loss": 0.8272,
+      "step": 3716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79663309950512e-06,
+      "loss": 0.8206,
+      "step": 3717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791543325636307e-06,
+      "loss": 0.7748,
+      "step": 3718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78645540268711e-06,
+      "loss": 0.7927,
+      "step": 3719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78136933246561e-06,
+      "loss": 0.8627,
+      "step": 3720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776285116779244e-06,
+      "loss": 0.8049,
+      "step": 3721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771202757434774e-06,
+      "loss": 0.8131,
+      "step": 3722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7661222562383104e-06,
+      "loss": 0.8185,
+      "step": 3723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7610436149953e-06,
+      "loss": 0.7756,
+      "step": 3724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755966835510531e-06,
+      "loss": 0.7946,
+      "step": 3725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750891919588127e-06,
+      "loss": 0.7892,
+      "step": 3726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7458188690315555e-06,
+      "loss": 0.8171,
+      "step": 3727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7407476856436166e-06,
+      "loss": 0.7972,
+      "step": 3728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7356783712264405e-06,
+      "loss": 0.7865,
+      "step": 3729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730610927581511e-06,
+      "loss": 0.7709,
+      "step": 3730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.725545356509631e-06,
+      "loss": 0.8012,
+      "step": 3731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720481659810941e-06,
+      "loss": 0.7809,
+      "step": 3732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715419839284917e-06,
+      "loss": 0.776,
+      "step": 3733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710359896730379e-06,
+      "loss": 0.8252,
+      "step": 3734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705301833945454e-06,
+      "loss": 0.7775,
+      "step": 3735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700245652727623e-06,
+      "loss": 0.7854,
+      "step": 3736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695191354873702e-06,
+      "loss": 0.761,
+      "step": 3737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690138942179809e-06,
+      "loss": 0.7687,
+      "step": 3738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.685088416441424e-06,
+      "loss": 0.7859,
+      "step": 3739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68003977945334e-06,
+      "loss": 0.2567,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6749930330096805e-06,
+      "loss": 0.7868,
+      "step": 3741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669948178903895e-06,
+      "loss": 0.7968,
+      "step": 3742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664905218928773e-06,
+      "loss": 0.8134,
+      "step": 3743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659864154876411e-06,
+      "loss": 0.7831,
+      "step": 3744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654824988538248e-06,
+      "loss": 0.7735,
+      "step": 3745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6497877217050505e-06,
+      "loss": 0.7369,
+      "step": 3746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644752356166887e-06,
+      "loss": 0.8068,
+      "step": 3747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6397188937131785e-06,
+      "loss": 0.7123,
+      "step": 3748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6346873361326535e-06,
+      "loss": 0.2486,
+      "step": 3749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629657685213368e-06,
+      "loss": 0.8255,
+      "step": 3750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624629942742695e-06,
+      "loss": 0.7632,
+      "step": 3751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6196041105073444e-06,
+      "loss": 0.7658,
+      "step": 3752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614580190293323e-06,
+      "loss": 0.2505,
+      "step": 3753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609558183885979e-06,
+      "loss": 0.7331,
+      "step": 3754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6045380930699824e-06,
+      "loss": 0.746,
+      "step": 3755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599519919629297e-06,
+      "loss": 0.8039,
+      "step": 3756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594503665347233e-06,
+      "loss": 0.8084,
+      "step": 3757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589489332006406e-06,
+      "loss": 0.7671,
+      "step": 3758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584476921388748e-06,
+      "loss": 0.7499,
+      "step": 3759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579466435275506e-06,
+      "loss": 0.2873,
+      "step": 3760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574457875447257e-06,
+      "loss": 0.7543,
+      "step": 3761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569451243683878e-06,
+      "loss": 0.8273,
+      "step": 3762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564446541764567e-06,
+      "loss": 0.7836,
+      "step": 3763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559443771467833e-06,
+      "loss": 0.2425,
+      "step": 3764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554442934571507e-06,
+      "loss": 0.7562,
+      "step": 3765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549444032852724e-06,
+      "loss": 0.827,
+      "step": 3766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544447068087934e-06,
+      "loss": 0.792,
+      "step": 3767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539452042052901e-06,
+      "loss": 0.7763,
+      "step": 3768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534458956522692e-06,
+      "loss": 0.8493,
+      "step": 3769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5294678132717e-06,
+      "loss": 0.7405,
+      "step": 3770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524478614073614e-06,
+      "loss": 0.2657,
+      "step": 3771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519491360701435e-06,
+      "loss": 0.8083,
+      "step": 3772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514506054927473e-06,
+      "loss": 0.7206,
+      "step": 3773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509522698523352e-06,
+      "loss": 0.7347,
+      "step": 3774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504541293259996e-06,
+      "loss": 0.7757,
+      "step": 3775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499561840907633e-06,
+      "loss": 0.7556,
+      "step": 3776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494584343235814e-06,
+      "loss": 0.7484,
+      "step": 3777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489608802013367e-06,
+      "loss": 0.7964,
+      "step": 3778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484635219008452e-06,
+      "loss": 0.8524,
+      "step": 3779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4796635959885195e-06,
+      "loss": 0.7676,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474693934720325e-06,
+      "loss": 0.892,
+      "step": 3781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469726236969926e-06,
+      "loss": 0.2866,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4647605045026885e-06,
+      "loss": 0.7939,
+      "step": 3783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4597967390832745e-06,
+      "loss": 0.7868,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454834942475642e-06,
+      "loss": 0.7993,
+      "step": 3785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449875116443069e-06,
+      "loss": 0.7785,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444917262748103e-06,
+      "loss": 0.8047,
+      "step": 3787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43996138315262e-06,
+      "loss": 0.7671,
+      "step": 3788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435007479417777e-06,
+      "loss": 0.7245,
+      "step": 3789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430055553304034e-06,
+      "loss": 0.254,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425105606571145e-06,
+      "loss": 0.8433,
+      "step": 3791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420157640978169e-06,
+      "loss": 0.78,
+      "step": 3792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415211658283452e-06,
+      "loss": 0.8476,
+      "step": 3793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4102676602446375e-06,
+      "loss": 0.8279,
+      "step": 3794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405325648618673e-06,
+      "loss": 0.7407,
+      "step": 3795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4003856251617775e-06,
+      "loss": 0.2798,
+      "step": 3796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39544759162949e-06,
+      "loss": 0.8234,
+      "step": 3797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390511549776628e-06,
+      "loss": 0.754,
+      "step": 3798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385577501357301e-06,
+      "loss": 0.8348,
+      "step": 3799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380645448124911e-06,
+      "loss": 0.8111,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375715391832159e-06,
+      "loss": 0.7641,
+      "step": 3801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370787334231026e-06,
+      "loss": 0.7908,
+      "step": 3802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365861277072784e-06,
+      "loss": 0.8116,
+      "step": 3803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360937222108002e-06,
+      "loss": 0.7373,
+      "step": 3804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3560151710865324e-06,
+      "loss": 0.8254,
+      "step": 3805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351095125757513e-06,
+      "loss": 0.8399,
+      "step": 3806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346177087869368e-06,
+      "loss": 0.7399,
+      "step": 3807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341261059169821e-06,
+      "loss": 0.7328,
+      "step": 3808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336347041405859e-06,
+      "loss": 0.7707,
+      "step": 3809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331435036323778e-06,
+      "loss": 0.7865,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326525045669145e-06,
+      "loss": 0.7132,
+      "step": 3811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32161707118681e-06,
+      "loss": 0.7991,
+      "step": 3812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3167111146209175e-06,
+      "loss": 0.7126,
+      "step": 3813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3118071777148865e-06,
+      "loss": 0.2807,
+      "step": 3814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306905262211419e-06,
+      "loss": 0.747,
+      "step": 3815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3020053698524946e-06,
+      "loss": 0.2505,
+      "step": 3816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2971075023793905e-06,
+      "loss": 0.8083,
+      "step": 3817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292211661532641e-06,
+      "loss": 0.7791,
+      "step": 3818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287317849052075e-06,
+      "loss": 0.7589,
+      "step": 3819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282426066676808e-06,
+      "loss": 0.818,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277536316145209e-06,
+      "loss": 0.7178,
+      "step": 3821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272648599194948e-06,
+      "loss": 0.2597,
+      "step": 3822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267762917562963e-06,
+      "loss": 0.7719,
+      "step": 3823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262879272985468e-06,
+      "loss": 0.7674,
+      "step": 3824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257997667197953e-06,
+      "loss": 0.7978,
+      "step": 3825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253118101935193e-06,
+      "loss": 0.292,
+      "step": 3826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2482405789312186e-06,
+      "loss": 0.7657,
+      "step": 3827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243365099919352e-06,
+      "loss": 0.7686,
+      "step": 3828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238491666632191e-06,
+      "loss": 0.7736,
+      "step": 3829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233620280801582e-06,
+      "loss": 0.8085,
+      "step": 3830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.228750944158674e-06,
+      "loss": 0.7778,
+      "step": 3831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223883658433869e-06,
+      "loss": 0.7929,
+      "step": 3832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2190184253568465e-06,
+      "loss": 0.8228,
+      "step": 3833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21415524665655e-06,
+      "loss": 0.7594,
+      "step": 3834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209294124061207e-06,
+      "loss": 0.8517,
+      "step": 3835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204435059298303e-06,
+      "loss": 0.8497,
+      "step": 3836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199578054094589e-06,
+      "loss": 0.8347,
+      "step": 3837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1947231101761e-06,
+      "loss": 0.7539,
+      "step": 3838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189870229268117e-06,
+      "loss": 0.7749,
+      "step": 3839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185019413095208e-06,
+      "loss": 0.8218,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180170663381196e-06,
+      "loss": 0.7745,
+      "step": 3841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175323981849173e-06,
+      "loss": 0.8204,
+      "step": 3842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170479370221491e-06,
+      "loss": 0.7543,
+      "step": 3843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165636830219776e-06,
+      "loss": 0.7352,
+      "step": 3844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1607963635649105e-06,
+      "loss": 0.8039,
+      "step": 3845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155957971977038e-06,
+      "loss": 0.7179,
+      "step": 3846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151121657175577e-06,
+      "loss": 0.7771,
+      "step": 3847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146287420879196e-06,
+      "loss": 0.8103,
+      "step": 3848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141455264805828e-06,
+      "loss": 0.7473,
+      "step": 3849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136625190672664e-06,
+      "loss": 0.8409,
+      "step": 3850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1317972001961694e-06,
+      "loss": 0.8273,
+      "step": 3851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126971295092043e-06,
+      "loss": 0.7608,
+      "step": 3852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.7205,
+      "step": 3853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117325747860077e-06,
+      "loss": 0.7963,
+      "step": 3854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1125061091599495e-06,
+      "loss": 0.8017,
+      "step": 3855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107688562687642e-06,
+      "loss": 0.7844,
+      "step": 3856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102873110155151e-06,
+      "loss": 0.8164,
+      "step": 3857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098059753273738e-06,
+      "loss": 0.7,
+      "step": 3858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093248493753911e-06,
+      "loss": 0.7176,
+      "step": 3859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088439333305449e-06,
+      "loss": 0.8117,
+      "step": 3860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083632273637362e-06,
+      "loss": 0.7864,
+      "step": 3861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078827316457935e-06,
+      "loss": 0.756,
+      "step": 3862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074024463474694e-06,
+      "loss": 0.8107,
+      "step": 3863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069223716394419e-06,
+      "loss": 0.7426,
+      "step": 3864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064425076923141e-06,
+      "loss": 0.7402,
+      "step": 3865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059628546766149e-06,
+      "loss": 0.7655,
+      "step": 3866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054834127627974e-06,
+      "loss": 0.8137,
+      "step": 3867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050041821212396e-06,
+      "loss": 0.2577,
+      "step": 3868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0452516292224595e-06,
+      "loss": 0.8042,
+      "step": 3869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040463553360431e-06,
+      "loss": 0.8035,
+      "step": 3870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035677595327851e-06,
+      "loss": 0.7652,
+      "step": 3871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030893756825495e-06,
+      "loss": 0.8044,
+      "step": 3872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026112039553384e-06,
+      "loss": 0.7514,
+      "step": 3873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021332445210785e-06,
+      "loss": 0.8,
+      "step": 3874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016554975496222e-06,
+      "loss": 0.7858,
+      "step": 3875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011779632107451e-06,
+      "loss": 0.7547,
+      "step": 3876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007006416741474e-06,
+      "loss": 0.7456,
+      "step": 3877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0022353310945474e-06,
+      "loss": 0.7347,
+      "step": 3878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997466376862158e-06,
+      "loss": 0.7909,
+      "step": 3879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992699555739041e-06,
+      "loss": 0.8107,
+      "step": 3880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987934869419172e-06,
+      "loss": 0.8099,
+      "step": 3881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98317231959577e-06,
+      "loss": 0.2714,
+      "step": 3882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97841190796129e-06,
+      "loss": 0.7427,
+      "step": 3883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973653636207437e-06,
+      "loss": 0.7582,
+      "step": 3884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968897506025147e-06,
+      "loss": 0.7741,
+      "step": 3885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964143519104591e-06,
+      "loss": 0.7774,
+      "step": 3886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959391677135194e-06,
+      "loss": 0.7732,
+      "step": 3887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9546419818056045e-06,
+      "loss": 0.2777,
+      "step": 3888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949894434803714e-06,
+      "loss": 0.753,
+      "step": 3889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945149037816644e-06,
+      "loss": 0.7394,
+      "step": 3890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9404057925307714e-06,
+      "loss": 0.8248,
+      "step": 3891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935664700631679e-06,
+      "loss": 0.7853,
+      "step": 3892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930925763804204e-06,
+      "loss": 0.7834,
+      "step": 3893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9261889837324245e-06,
+      "loss": 0.7205,
+      "step": 3894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9214543620996245e-06,
+      "loss": 0.7867,
+      "step": 3895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9167219005883495e-06,
+      "loss": 0.7745,
+      "step": 3896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911991600880363e-06,
+      "loss": 0.243,
+      "step": 3897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907263464656662e-06,
+      "loss": 0.2605,
+      "step": 3898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9025374935974715e-06,
+      "loss": 0.7348,
+      "step": 3899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897813689382262e-06,
+      "loss": 0.2594,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893092053689709e-06,
+      "loss": 0.7551,
+      "step": 3901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888372588197738e-06,
+      "loss": 0.7675,
+      "step": 3902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8836552945835045e-06,
+      "loss": 0.8031,
+      "step": 3903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878940174523371e-06,
+      "loss": 0.8042,
+      "step": 3904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8742272296929506e-06,
+      "loss": 0.8492,
+      "step": 3905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86951646176707e-06,
+      "loss": 0.7632,
+      "step": 3906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8648078724197866e-06,
+      "loss": 0.6997,
+      "step": 3907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860101463324379e-06,
+      "loss": 0.809,
+      "step": 3908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855397236153363e-06,
+      "loss": 0.8205,
+      "step": 3909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850695192578467e-06,
+      "loss": 0.7849,
+      "step": 3910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845995334270644e-06,
+      "loss": 0.7734,
+      "step": 3911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841297662900085e-06,
+      "loss": 0.7545,
+      "step": 3912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836602180136179e-06,
+      "loss": 0.7312,
+      "step": 3913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8319088876475595e-06,
+      "loss": 0.796,
+      "step": 3914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827217787102072e-06,
+      "loss": 0.7224,
+      "step": 3915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822528880166783e-06,
+      "loss": 0.8326,
+      "step": 3916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817842168507978e-06,
+      "loss": 0.7737,
+      "step": 3917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813157653791171e-06,
+      "loss": 0.7433,
+      "step": 3918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8084753376810867e-06,
+      "loss": 0.7684,
+      "step": 3919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8037952218416672e-06,
+      "loss": 0.7406,
+      "step": 3920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7991173079360845e-06,
+      "loss": 0.7861,
+      "step": 3921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7944415976267147e-06,
+      "loss": 0.7178,
+      "step": 3922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789768092575158e-06,
+      "loss": 0.7492,
+      "step": 3923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.785096794442229e-06,
+      "loss": 0.7956,
+      "step": 3924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7804277048879577e-06,
+      "loss": 0.825,
+      "step": 3925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775760825571587e-06,
+      "loss": 0.7896,
+      "step": 3926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7710961581515838e-06,
+      "loss": 0.7681,
+      "step": 3927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.766433704285619e-06,
+      "loss": 0.8002,
+      "step": 3928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761773465630577e-06,
+      "loss": 0.7836,
+      "step": 3929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.757115443842565e-06,
+      "loss": 0.7545,
+      "step": 3930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.752459640576891e-06,
+      "loss": 0.7433,
+      "step": 3931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7478060574880805e-06,
+      "loss": 0.8078,
+      "step": 3932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743154696229865e-06,
+      "loss": 0.8134,
+      "step": 3933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7385055584552e-06,
+      "loss": 0.7629,
+      "step": 3934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7338586458162274e-06,
+      "loss": 0.7531,
+      "step": 3935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.729213959964323e-06,
+      "loss": 0.7817,
+      "step": 3936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724571502550056e-06,
+      "loss": 0.7944,
+      "step": 3937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7199312752232053e-06,
+      "loss": 0.7579,
+      "step": 3938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7152932796327655e-06,
+      "loss": 0.8543,
+      "step": 3939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.71065751742693e-06,
+      "loss": 0.7912,
+      "step": 3940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7060239902531004e-06,
+      "loss": 0.725,
+      "step": 3941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701392699757882e-06,
+      "loss": 0.738,
+      "step": 3942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6967636475870983e-06,
+      "loss": 0.7331,
+      "step": 3943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6921368353857524e-06,
+      "loss": 0.7697,
+      "step": 3944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6875122647980764e-06,
+      "loss": 0.748,
+      "step": 3945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6828899374674933e-06,
+      "loss": 0.2426,
+      "step": 3946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6782698550366257e-06,
+      "loss": 0.7835,
+      "step": 3947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673652019147311e-06,
+      "loss": 0.7872,
+      "step": 3948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.669036431440579e-06,
+      "loss": 0.769,
+      "step": 3949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6644230935566614e-06,
+      "loss": 0.7377,
+      "step": 3950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6598120071349884e-06,
+      "loss": 0.8009,
+      "step": 3951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6552031738142004e-06,
+      "loss": 0.7477,
+      "step": 3952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6505965952321267e-06,
+      "loss": 0.8117,
+      "step": 3953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645992273025797e-06,
+      "loss": 0.7959,
+      "step": 3954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6413902088314425e-06,
+      "loss": 0.7379,
+      "step": 3955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6367904042844857e-06,
+      "loss": 0.7773,
+      "step": 3956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.632192861019559e-06,
+      "loss": 0.7689,
+      "step": 3957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6275975806704777e-06,
+      "loss": 0.8018,
+      "step": 3958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6230045648702585e-06,
+      "loss": 0.7719,
+      "step": 3959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6184138152511107e-06,
+      "loss": 0.752,
+      "step": 3960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613825333444446e-06,
+      "loss": 0.7673,
+      "step": 3961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609239121080862e-06,
+      "loss": 0.72,
+      "step": 3962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.604655179790152e-06,
+      "loss": 0.8088,
+      "step": 3963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6000735112012984e-06,
+      "loss": 0.7385,
+      "step": 3964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595494116942494e-06,
+      "loss": 0.7269,
+      "step": 3965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590916998641092e-06,
+      "loss": 0.7334,
+      "step": 3966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5863421579236668e-06,
+      "loss": 0.8384,
+      "step": 3967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5817695964159673e-06,
+      "loss": 0.7815,
+      "step": 3968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.577199315742933e-06,
+      "loss": 0.7721,
+      "step": 3969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.572631317528703e-06,
+      "loss": 0.7337,
+      "step": 3970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5680656033965944e-06,
+      "loss": 0.8145,
+      "step": 3971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563502174969117e-06,
+      "loss": 0.7386,
+      "step": 3972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558941033867964e-06,
+      "loss": 0.7852,
+      "step": 3973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5543821817140313e-06,
+      "loss": 0.7915,
+      "step": 3974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549825620127374e-06,
+      "loss": 0.7591,
+      "step": 3975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.545271350727257e-06,
+      "loss": 0.7649,
+      "step": 3976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540719375132129e-06,
+      "loss": 0.794,
+      "step": 3977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5361696949596046e-06,
+      "loss": 0.7799,
+      "step": 3978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5316223118265046e-06,
+      "loss": 0.8669,
+      "step": 3979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5270772273488206e-06,
+      "loss": 0.7461,
+      "step": 3980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522534443141731e-06,
+      "loss": 0.267,
+      "step": 3981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5179939608195935e-06,
+      "loss": 0.7767,
+      "step": 3982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5134557819959603e-06,
+      "loss": 0.7685,
+      "step": 3983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5089199082835436e-06,
+      "loss": 0.7911,
+      "step": 3984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5043863412942535e-06,
+      "loss": 0.8317,
+      "step": 3985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.499855082639183e-06,
+      "loss": 0.7833,
+      "step": 3986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4953261339285836e-06,
+      "loss": 0.7507,
+      "step": 3987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4907994967719096e-06,
+      "loss": 0.832,
+      "step": 3988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48627517277778e-06,
+      "loss": 0.7662,
+      "step": 3989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4817531635539946e-06,
+      "loss": 0.7863,
+      "step": 3990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4772334707075304e-06,
+      "loss": 0.7503,
+      "step": 3991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472716095844547e-06,
+      "loss": 0.7915,
+      "step": 3992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4682010405703726e-06,
+      "loss": 0.7473,
+      "step": 3993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.463688306489511e-06,
+      "loss": 0.7713,
+      "step": 3994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4591778952056543e-06,
+      "loss": 0.7854,
+      "step": 3995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.454669808321646e-06,
+      "loss": 0.7324,
+      "step": 3996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4501640474395247e-06,
+      "loss": 0.7482,
+      "step": 3997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4456606141604932e-06,
+      "loss": 0.7606,
+      "step": 3998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.441159510084927e-06,
+      "loss": 0.7885,
+      "step": 3999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4366607368123727e-06,
+      "loss": 0.7927,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432164295941558e-06,
+      "loss": 0.2647,
+      "step": 4001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427670189070369e-06,
+      "loss": 0.7995,
+      "step": 4002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423178417795867e-06,
+      "loss": 0.7849,
+      "step": 4003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418688983714291e-06,
+      "loss": 0.7462,
+      "step": 4004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4142018884210393e-06,
+      "loss": 0.7762,
+      "step": 4005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409717133510683e-06,
+      "loss": 0.8027,
+      "step": 4006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4052347205769567e-06,
+      "loss": 0.8106,
+      "step": 4007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4007546512127764e-06,
+      "loss": 0.7668,
+      "step": 4008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3962769270102057e-06,
+      "loss": 0.763,
+      "step": 4009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.391801549560494e-06,
+      "loss": 0.7694,
+      "step": 4010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387328520454043e-06,
+      "loss": 0.7654,
+      "step": 4011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3828578412804235e-06,
+      "loss": 0.7529,
+      "step": 4012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3783895136283774e-06,
+      "loss": 0.7564,
+      "step": 4013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373923539085805e-06,
+      "loss": 0.7483,
+      "step": 4014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369459919239768e-06,
+      "loss": 0.7871,
+      "step": 4015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.364998655676496e-06,
+      "loss": 0.77,
+      "step": 4016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360539749981385e-06,
+      "loss": 0.839,
+      "step": 4017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356083203738978e-06,
+      "loss": 0.8047,
+      "step": 4018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.351629018532999e-06,
+      "loss": 0.7726,
+      "step": 4019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3471771959463195e-06,
+      "loss": 0.7774,
+      "step": 4020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3427277375609733e-06,
+      "loss": 0.7474,
+      "step": 4021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.338280644958162e-06,
+      "loss": 0.804,
+      "step": 4022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3338359197182366e-06,
+      "loss": 0.7595,
+      "step": 4023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329393563420713e-06,
+      "loss": 0.7261,
+      "step": 4024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3249535776442575e-06,
+      "loss": 0.7814,
+      "step": 4025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3205159639667117e-06,
+      "loss": 0.7525,
+      "step": 4026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316080723965047e-06,
+      "loss": 0.7268,
+      "step": 4027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3116478592154177e-06,
+      "loss": 0.7464,
+      "step": 4028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.307217371293119e-06,
+      "loss": 0.8518,
+      "step": 4029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302789261772601e-06,
+      "loss": 0.769,
+      "step": 4030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2983635322274808e-06,
+      "loss": 0.7542,
+      "step": 4031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2939401842305187e-06,
+      "loss": 0.7197,
+      "step": 4032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2895192193536307e-06,
+      "loss": 0.8138,
+      "step": 4033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285100639167883e-06,
+      "loss": 0.7468,
+      "step": 4034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280684445243506e-06,
+      "loss": 0.7673,
+      "step": 4035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2762706391498712e-06,
+      "loss": 0.7917,
+      "step": 4036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2718592224555035e-06,
+      "loss": 0.8488,
+      "step": 4037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2674501967280814e-06,
+      "loss": 0.8283,
+      "step": 4038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2630435635344283e-06,
+      "loss": 0.7897,
+      "step": 4039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.258639324440527e-06,
+      "loss": 0.7604,
+      "step": 4040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254237481011502e-06,
+      "loss": 0.691,
+      "step": 4041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2498380348116264e-06,
+      "loss": 0.7947,
+      "step": 4042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2454409874043223e-06,
+      "loss": 0.7706,
+      "step": 4043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2410463403521653e-06,
+      "loss": 0.7782,
+      "step": 4044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236654095216869e-06,
+      "loss": 0.7119,
+      "step": 4045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2322642535592994e-06,
+      "loss": 0.8311,
+      "step": 4046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2278768169394627e-06,
+      "loss": 0.7381,
+      "step": 4047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2234917869165203e-06,
+      "loss": 0.2707,
+      "step": 4048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2191091650487683e-06,
+      "loss": 0.7933,
+      "step": 4049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214728952893649e-06,
+      "loss": 0.7377,
+      "step": 4050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210351152007758e-06,
+      "loss": 0.8027,
+      "step": 4051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205975763946815e-06,
+      "loss": 0.7996,
+      "step": 4052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2016027902657034e-06,
+      "loss": 0.8421,
+      "step": 4053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1972322325184347e-06,
+      "loss": 0.7924,
+      "step": 4054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1928640922581665e-06,
+      "loss": 0.7751,
+      "step": 4055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.188498371037193e-06,
+      "loss": 0.7615,
+      "step": 4056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1841350704069583e-06,
+      "loss": 0.7243,
+      "step": 4057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1797741919180403e-06,
+      "loss": 0.7927,
+      "step": 4058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.175415737120149e-06,
+      "loss": 0.784,
+      "step": 4059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171059707562153e-06,
+      "loss": 0.7792,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1667061047920324e-06,
+      "loss": 0.7467,
+      "step": 4061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162354930356929e-06,
+      "loss": 0.8085,
+      "step": 4062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1580061858031075e-06,
+      "loss": 0.7771,
+      "step": 4063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1536598726759747e-06,
+      "loss": 0.7679,
+      "step": 4064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149315992520067e-06,
+      "loss": 0.7292,
+      "step": 4065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144974546879069e-06,
+      "loss": 0.8276,
+      "step": 4066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1406355372957887e-06,
+      "loss": 0.7919,
+      "step": 4067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136298965312168e-06,
+      "loss": 0.8149,
+      "step": 4068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1319648324692954e-06,
+      "loss": 0.8174,
+      "step": 4069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1276331403073733e-06,
+      "loss": 0.7946,
+      "step": 4070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123303890365754e-06,
+      "loss": 0.7626,
+      "step": 4071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1189770841829147e-06,
+      "loss": 0.7311,
+      "step": 4072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1146527232964617e-06,
+      "loss": 0.7633,
+      "step": 4073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.110330809243134e-06,
+      "loss": 0.774,
+      "step": 4074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.106011343558808e-06,
+      "loss": 0.789,
+      "step": 4075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1016943277784806e-06,
+      "loss": 0.7394,
+      "step": 4076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097379763436279e-06,
+      "loss": 0.8128,
+      "step": 4077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093067652065468e-06,
+      "loss": 0.7674,
+      "step": 4078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0887579951984303e-06,
+      "loss": 0.7935,
+      "step": 4079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0844507943666834e-06,
+      "loss": 0.7851,
+      "step": 4080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0801460511008675e-06,
+      "loss": 0.7082,
+      "step": 4081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0758437669307516e-06,
+      "loss": 0.7873,
+      "step": 4082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0715439433852268e-06,
+      "loss": 0.8144,
+      "step": 4083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0672465819923215e-06,
+      "loss": 0.7453,
+      "step": 4084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062951684279175e-06,
+      "loss": 0.7915,
+      "step": 4085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058659251772057e-06,
+      "loss": 0.8277,
+      "step": 4086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.054369285996366e-06,
+      "loss": 0.7526,
+      "step": 4087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0500817884766155e-06,
+      "loss": 0.7557,
+      "step": 4088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045796760736447e-06,
+      "loss": 0.7256,
+      "step": 4089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041514204298619e-06,
+      "loss": 0.2506,
+      "step": 4090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0372341206850265e-06,
+      "loss": 0.7727,
+      "step": 4091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0329565114166592e-06,
+      "loss": 0.7452,
+      "step": 4092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0286813780136558e-06,
+      "loss": 0.7863,
+      "step": 4093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0244087219952565e-06,
+      "loss": 0.7136,
+      "step": 4094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0201385448798273e-06,
+      "loss": 0.7283,
+      "step": 4095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0158708481848577e-06,
+      "loss": 0.7943,
+      "step": 4096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011605633426947e-06,
+      "loss": 0.7547,
+      "step": 4097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007342902121818e-06,
+      "loss": 0.7854,
+      "step": 4098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003082655784307e-06,
+      "loss": 0.7565,
+      "step": 4099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9988248959283784e-06,
+      "loss": 0.756,
+      "step": 4100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9945696240670905e-06,
+      "loss": 0.8273,
+      "step": 4101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.990316841712644e-06,
+      "loss": 0.784,
+      "step": 4102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9860665503763377e-06,
+      "loss": 0.7554,
+      "step": 4103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981818751568586e-06,
+      "loss": 0.7677,
+      "step": 4104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977573446798927e-06,
+      "loss": 0.2826,
+      "step": 4105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.973330637576006e-06,
+      "loss": 0.7472,
+      "step": 4106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.96909032540758e-06,
+      "loss": 0.7988,
+      "step": 4107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.964852511800519e-06,
+      "loss": 0.7933,
+      "step": 4108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9606171982608133e-06,
+      "loss": 0.2414,
+      "step": 4109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.956384386293555e-06,
+      "loss": 0.8106,
+      "step": 4110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.95215407740295e-06,
+      "loss": 0.7704,
+      "step": 4111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9479262730923165e-06,
+      "loss": 0.7496,
+      "step": 4112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9437009748640776e-06,
+      "loss": 0.7543,
+      "step": 4113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939478184219777e-06,
+      "loss": 0.7961,
+      "step": 4114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935257902660056e-06,
+      "loss": 0.8217,
+      "step": 4115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.931040131684667e-06,
+      "loss": 0.7528,
+      "step": 4116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926824872792471e-06,
+      "loss": 0.7602,
+      "step": 4117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.922612127481441e-06,
+      "loss": 0.7898,
+      "step": 4118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9184018972486506e-06,
+      "loss": 0.7959,
+      "step": 4119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9141941835902796e-06,
+      "loss": 0.7507,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9099889880016152e-06,
+      "loss": 0.7408,
+      "step": 4121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.905786311977055e-06,
+      "loss": 0.7908,
+      "step": 4122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9015861570100924e-06,
+      "loss": 0.7626,
+      "step": 4123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8973885245933287e-06,
+      "loss": 0.7439,
+      "step": 4124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893193416218469e-06,
+      "loss": 0.765,
+      "step": 4125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8890008333763187e-06,
+      "loss": 0.766,
+      "step": 4126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8848107775567936e-06,
+      "loss": 0.7822,
+      "step": 4127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.880623250248904e-06,
+      "loss": 0.7847,
+      "step": 4128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876438252940763e-06,
+      "loss": 0.816,
+      "step": 4129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8722557871195822e-06,
+      "loss": 0.7859,
+      "step": 4130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8680758542716837e-06,
+      "loss": 0.7265,
+      "step": 4131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8638984558824777e-06,
+      "loss": 0.7723,
+      "step": 4132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8597235934364776e-06,
+      "loss": 0.7263,
+      "step": 4133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855551268417305e-06,
+      "loss": 0.7891,
+      "step": 4134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851381482307658e-06,
+      "loss": 0.7598,
+      "step": 4135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847214236589356e-06,
+      "loss": 0.8003,
+      "step": 4136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8430495327433017e-06,
+      "loss": 0.8102,
+      "step": 4137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838887372249499e-06,
+      "loss": 0.7551,
+      "step": 4138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8347277565870447e-06,
+      "loss": 0.8171,
+      "step": 4139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830570687234138e-06,
+      "loss": 0.7143,
+      "step": 4140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8264161656680666e-06,
+      "loss": 0.7178,
+      "step": 4141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822264193365212e-06,
+      "loss": 0.7769,
+      "step": 4142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8181147718010624e-06,
+      "loss": 0.7662,
+      "step": 4143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813967902450179e-06,
+      "loss": 0.815,
+      "step": 4144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8098235867862346e-06,
+      "loss": 0.7735,
+      "step": 4145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805681826281983e-06,
+      "loss": 0.7667,
+      "step": 4146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8015426224092766e-06,
+      "loss": 0.7946,
+      "step": 4147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7974059766390527e-06,
+      "loss": 0.77,
+      "step": 4148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793271890441348e-06,
+      "loss": 0.8081,
+      "step": 4149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7891403652852844e-06,
+      "loss": 0.826,
+      "step": 4150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7850114026390695e-06,
+      "loss": 0.8057,
+      "step": 4151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.780885003970012e-06,
+      "loss": 0.8262,
+      "step": 4152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7767611707445e-06,
+      "loss": 0.7654,
+      "step": 4153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7726399044280107e-06,
+      "loss": 0.7817,
+      "step": 4154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7685212064851118e-06,
+      "loss": 0.7613,
+      "step": 4155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7644050783794586e-06,
+      "loss": 0.7734,
+      "step": 4156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760291521573788e-06,
+      "loss": 0.7708,
+      "step": 4157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756180537529932e-06,
+      "loss": 0.7608,
+      "step": 4158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7520721277088023e-06,
+      "loss": 0.7583,
+      "step": 4159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747966293570392e-06,
+      "loss": 0.7725,
+      "step": 4160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743863036573792e-06,
+      "loss": 0.8327,
+      "step": 4161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7397623581771638e-06,
+      "loss": 0.8339,
+      "step": 4162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7356642598377604e-06,
+      "loss": 0.7886,
+      "step": 4163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7315687430119097e-06,
+      "loss": 0.8502,
+      "step": 4164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.727475809155039e-06,
+      "loss": 0.7619,
+      "step": 4165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7233854597216335e-06,
+      "loss": 0.7768,
+      "step": 4166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719297696165284e-06,
+      "loss": 0.7827,
+      "step": 4167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715212519938646e-06,
+      "loss": 0.7531,
+      "step": 4168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.711129932493458e-06,
+      "loss": 0.7551,
+      "step": 4169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.70704993528055e-06,
+      "loss": 0.7765,
+      "step": 4170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7029725297498177e-06,
+      "loss": 0.7258,
+      "step": 4171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.698897717350243e-06,
+      "loss": 0.7493,
+      "step": 4172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6948254995298804e-06,
+      "loss": 0.7278,
+      "step": 4173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6907558777358756e-06,
+      "loss": 0.7896,
+      "step": 4174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6866888534144297e-06,
+      "loss": 0.7394,
+      "step": 4175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6826244280108438e-06,
+      "loss": 0.2304,
+      "step": 4176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6785626029694823e-06,
+      "loss": 0.8028,
+      "step": 4177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674503379733785e-06,
+      "loss": 0.7946,
+      "step": 4178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.670446759746277e-06,
+      "loss": 0.2396,
+      "step": 4179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6663927444485484e-06,
+      "loss": 0.7803,
+      "step": 4180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6623413352812666e-06,
+      "loss": 0.7426,
+      "step": 4181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6582925336841705e-06,
+      "loss": 0.7626,
+      "step": 4182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654246341096085e-06,
+      "loss": 0.7405,
+      "step": 4183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.650202758954886e-06,
+      "loss": 0.7641,
+      "step": 4184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646161788697543e-06,
+      "loss": 0.7762,
+      "step": 4185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6421234317600842e-06,
+      "loss": 0.8053,
+      "step": 4186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6380876895776107e-06,
+      "loss": 0.775,
+      "step": 4187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6340545635843027e-06,
+      "loss": 0.7514,
+      "step": 4188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6300240552134014e-06,
+      "loss": 0.7958,
+      "step": 4189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625996165897222e-06,
+      "loss": 0.7858,
+      "step": 4190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6219708970671433e-06,
+      "loss": 0.8251,
+      "step": 4191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617948250153626e-06,
+      "loss": 0.7313,
+      "step": 4192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6139282265861855e-06,
+      "loss": 0.8178,
+      "step": 4193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6099108277934105e-06,
+      "loss": 0.7536,
+      "step": 4194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6058960552029543e-06,
+      "loss": 0.7869,
+      "step": 4195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.601883910241546e-06,
+      "loss": 0.7194,
+      "step": 4196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597874394334969e-06,
+      "loss": 0.8447,
+      "step": 4197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59386750890808e-06,
+      "loss": 0.7766,
+      "step": 4198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5898632553847967e-06,
+      "loss": 0.795,
+      "step": 4199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5858616351881006e-06,
+      "loss": 0.765,
+      "step": 4200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581862649740047e-06,
+      "loss": 0.7129,
+      "step": 4201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577866300461743e-06,
+      "loss": 0.8117,
+      "step": 4202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573872588773365e-06,
+      "loss": 0.8527,
+      "step": 4203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5698815160941494e-06,
+      "loss": 0.7841,
+      "step": 4204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5658930838424e-06,
+      "loss": 0.7969,
+      "step": 4205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5619072934354762e-06,
+      "loss": 0.7285,
+      "step": 4206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5579241462897986e-06,
+      "loss": 0.8673,
+      "step": 4207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55394364382086e-06,
+      "loss": 0.7737,
+      "step": 4208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5499657874431903e-06,
+      "loss": 0.2594,
+      "step": 4209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545990578570404e-06,
+      "loss": 0.8055,
+      "step": 4210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.542018018615161e-06,
+      "loss": 0.7801,
+      "step": 4211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5380481089891804e-06,
+      "loss": 0.7795,
+      "step": 4212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534080851103241e-06,
+      "loss": 0.7758,
+      "step": 4213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5301162463671845e-06,
+      "loss": 0.2626,
+      "step": 4214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526154296189902e-06,
+      "loss": 0.7461,
+      "step": 4215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.522195001979343e-06,
+      "loss": 0.8334,
+      "step": 4216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5182383651425214e-06,
+      "loss": 0.2402,
+      "step": 4217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514284387085488e-06,
+      "loss": 0.7513,
+      "step": 4218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.510333069213372e-06,
+      "loss": 0.7745,
+      "step": 4219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5063844129303416e-06,
+      "loss": 0.769,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5024384196396235e-06,
+      "loss": 0.7379,
+      "step": 4221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4984950907434934e-06,
+      "loss": 0.7599,
+      "step": 4222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494554427643294e-06,
+      "loss": 0.7919,
+      "step": 4223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4906164317394067e-06,
+      "loss": 0.74,
+      "step": 4224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4866811044312667e-06,
+      "loss": 0.7481,
+      "step": 4225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4827484471173726e-06,
+      "loss": 0.7761,
+      "step": 4226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478818461195255e-06,
+      "loss": 0.7582,
+      "step": 4227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4748911480615135e-06,
+      "loss": 0.7256,
+      "step": 4228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4709665091117873e-06,
+      "loss": 0.7233,
+      "step": 4229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.467044545740769e-06,
+      "loss": 0.7661,
+      "step": 4230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.463125259342195e-06,
+      "loss": 0.7547,
+      "step": 4231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459208651308862e-06,
+      "loss": 0.7666,
+      "step": 4232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455294723032604e-06,
+      "loss": 0.8423,
+      "step": 4233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451383475904304e-06,
+      "loss": 0.7213,
+      "step": 4234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4474749113139007e-06,
+      "loss": 0.8129,
+      "step": 4235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44356903065037e-06,
+      "loss": 0.7755,
+      "step": 4236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4396658353017377e-06,
+      "loss": 0.7745,
+      "step": 4237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.435765326655073e-06,
+      "loss": 0.7636,
+      "step": 4238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.431867506096499e-06,
+      "loss": 0.7592,
+      "step": 4239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4279723750111683e-06,
+      "loss": 0.7738,
+      "step": 4240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4240799347832934e-06,
+      "loss": 0.8263,
+      "step": 4241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42019018679612e-06,
+      "loss": 0.2492,
+      "step": 4242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4163031324319385e-06,
+      "loss": 0.8046,
+      "step": 4243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4124187730720916e-06,
+      "loss": 0.757,
+      "step": 4244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4085371100969514e-06,
+      "loss": 0.2875,
+      "step": 4245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4046581448859387e-06,
+      "loss": 0.7489,
+      "step": 4246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400781878817511e-06,
+      "loss": 0.758,
+      "step": 4247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3969083132691785e-06,
+      "loss": 0.7182,
+      "step": 4248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3930374496174726e-06,
+      "loss": 0.7429,
+      "step": 4249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389169289237978e-06,
+      "loss": 0.7709,
+      "step": 4250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3853038335053234e-06,
+      "loss": 0.7534,
+      "step": 4251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3814410837931577e-06,
+      "loss": 0.7984,
+      "step": 4252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377581041474186e-06,
+      "loss": 0.8221,
+      "step": 4253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3737237079201437e-06,
+      "loss": 0.7792,
+      "step": 4254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3698690845018023e-06,
+      "loss": 0.774,
+      "step": 4255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3660171725889703e-06,
+      "loss": 0.7958,
+      "step": 4256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.362167973550502e-06,
+      "loss": 0.7638,
+      "step": 4257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3583214887542705e-06,
+      "loss": 0.8023,
+      "step": 4258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3544777195671976e-06,
+      "loss": 0.2694,
+      "step": 4259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350636667355244e-06,
+      "loss": 0.7672,
+      "step": 4260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3467983334833855e-06,
+      "loss": 0.8007,
+      "step": 4261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3429627193156513e-06,
+      "loss": 0.8047,
+      "step": 4262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3391298262150953e-06,
+      "loss": 0.7494,
+      "step": 4263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3352996555438036e-06,
+      "loss": 0.7658,
+      "step": 4264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331472208662896e-06,
+      "loss": 0.7317,
+      "step": 4265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3276474869325295e-06,
+      "loss": 0.8455,
+      "step": 4266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3238254917118862e-06,
+      "loss": 0.8182,
+      "step": 4267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320006224359178e-06,
+      "loss": 0.7569,
+      "step": 4268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31618968623166e-06,
+      "loss": 0.7628,
+      "step": 4269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312375878685598e-06,
+      "loss": 0.7662,
+      "step": 4270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3085648030763042e-06,
+      "loss": 0.7606,
+      "step": 4271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.304756460758111e-06,
+      "loss": 0.741,
+      "step": 4272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3009508530843827e-06,
+      "loss": 0.7735,
+      "step": 4273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297147981407509e-06,
+      "loss": 0.7312,
+      "step": 4274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2933478470789127e-06,
+      "loss": 0.7726,
+      "step": 4275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2895504514490397e-06,
+      "loss": 0.7909,
+      "step": 4276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2857557958673626e-06,
+      "loss": 0.8255,
+      "step": 4277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2819638816823796e-06,
+      "loss": 0.7697,
+      "step": 4278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2781747102416195e-06,
+      "loss": 0.2325,
+      "step": 4279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274388282891632e-06,
+      "loss": 0.7089,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2706046009779924e-06,
+      "loss": 0.84,
+      "step": 4281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2668236658453e-06,
+      "loss": 0.782,
+      "step": 4282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263045478837178e-06,
+      "loss": 0.7835,
+      "step": 4283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2592700412962775e-06,
+      "loss": 0.7109,
+      "step": 4284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2554973545642656e-06,
+      "loss": 0.7917,
+      "step": 4285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251727419981836e-06,
+      "loss": 0.7141,
+      "step": 4286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2479602388887013e-06,
+      "loss": 0.7627,
+      "step": 4287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.244195812623602e-06,
+      "loss": 0.8159,
+      "step": 4288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2404341425242936e-06,
+      "loss": 0.7271,
+      "step": 4289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2366752299275497e-06,
+      "loss": 0.6977,
+      "step": 4290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2329190761691766e-06,
+      "loss": 0.7639,
+      "step": 4291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2291656825839814e-06,
+      "loss": 0.8301,
+      "step": 4292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225415050505809e-06,
+      "loss": 0.7765,
+      "step": 4293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2216671812675118e-06,
+      "loss": 0.7395,
+      "step": 4294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2179220762009624e-06,
+      "loss": 0.7709,
+      "step": 4295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2141797366370498e-06,
+      "loss": 0.782,
+      "step": 4296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.210440163905687e-06,
+      "loss": 0.772,
+      "step": 4297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2067033593357977e-06,
+      "loss": 0.7756,
+      "step": 4298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2029693242553195e-06,
+      "loss": 0.764,
+      "step": 4299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1992380599912176e-06,
+      "loss": 0.7319,
+      "step": 4300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195509567869454e-06,
+      "loss": 0.7578,
+      "step": 4301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1917838492150245e-06,
+      "loss": 0.7241,
+      "step": 4302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1880609053519263e-06,
+      "loss": 0.7147,
+      "step": 4303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184340737603178e-06,
+      "loss": 0.7475,
+      "step": 4304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1806233472908046e-06,
+      "loss": 0.804,
+      "step": 4305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176908735735852e-06,
+      "loss": 0.7667,
+      "step": 4306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1731969042583747e-06,
+      "loss": 0.2501,
+      "step": 4307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1694878541774355e-06,
+      "loss": 0.7192,
+      "step": 4308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1657815868111187e-06,
+      "loss": 0.7481,
+      "step": 4309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1620781034765106e-06,
+      "loss": 0.8159,
+      "step": 4310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.15837740548971e-06,
+      "loss": 0.2474,
+      "step": 4311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154679494165829e-06,
+      "loss": 0.8061,
+      "step": 4312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150984370818986e-06,
+      "loss": 0.7377,
+      "step": 4313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1472920367623094e-06,
+      "loss": 0.7954,
+      "step": 4314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1436024933079403e-06,
+      "loss": 0.7724,
+      "step": 4315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1399157417670233e-06,
+      "loss": 0.7653,
+      "step": 4316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1362317834497103e-06,
+      "loss": 0.8168,
+      "step": 4317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132550619665168e-06,
+      "loss": 0.7689,
+      "step": 4318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128872251721561e-06,
+      "loss": 0.7641,
+      "step": 4319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1251966809260647e-06,
+      "loss": 0.2468,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.121523908584857e-06,
+      "loss": 0.811,
+      "step": 4321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1178539360031335e-06,
+      "loss": 0.7196,
+      "step": 4322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114186764485073e-06,
+      "loss": 0.7336,
+      "step": 4323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1105223953338805e-06,
+      "loss": 0.7968,
+      "step": 4324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1068608298517532e-06,
+      "loss": 0.8307,
+      "step": 4325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1032020693398925e-06,
+      "loss": 0.8229,
+      "step": 4326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099546115098511e-06,
+      "loss": 0.8067,
+      "step": 4327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.095892968426815e-06,
+      "loss": 0.7304,
+      "step": 4328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092242630623016e-06,
+      "loss": 0.7895,
+      "step": 4329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0885951029843256e-06,
+      "loss": 0.7724,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0849503868069676e-06,
+      "loss": 0.8017,
+      "step": 4331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0813084833861473e-06,
+      "loss": 0.2502,
+      "step": 4332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077669394016086e-06,
+      "loss": 0.789,
+      "step": 4333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0740331199900053e-06,
+      "loss": 0.7679,
+      "step": 4334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070399662600111e-06,
+      "loss": 0.2569,
+      "step": 4335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0667690231376257e-06,
+      "loss": 0.7985,
+      "step": 4336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063141202892762e-06,
+      "loss": 0.2554,
+      "step": 4337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0595162031547287e-06,
+      "loss": 0.724,
+      "step": 4338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558940252117353e-06,
+      "loss": 0.7537,
+      "step": 4339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0522746703509944e-06,
+      "loss": 0.771,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0486581398587002e-06,
+      "loss": 0.7525,
+      "step": 4341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0450444350200584e-06,
+      "loss": 0.8017,
+      "step": 4342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041433557119268e-06,
+      "loss": 0.7538,
+      "step": 4343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0378255074395094e-06,
+      "loss": 0.792,
+      "step": 4344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034220287262978e-06,
+      "loss": 0.7922,
+      "step": 4345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030617897870851e-06,
+      "loss": 0.7841,
+      "step": 4346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027018340543303e-06,
+      "loss": 0.819,
+      "step": 4347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0234216165594988e-06,
+      "loss": 0.753,
+      "step": 4348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.019827727197605e-06,
+      "loss": 0.7734,
+      "step": 4349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0162366737347738e-06,
+      "loss": 0.7763,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0126484574471473e-06,
+      "loss": 0.7623,
+      "step": 4351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0090630796098688e-06,
+      "loss": 0.8187,
+      "step": 4352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0054805414970658e-06,
+      "loss": 0.7833,
+      "step": 4353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.001900844381857e-06,
+      "loss": 0.7315,
+      "step": 4354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9983239895363528e-06,
+      "loss": 0.773,
+      "step": 4355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9947499782316537e-06,
+      "loss": 0.8152,
+      "step": 4356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991178811737847e-06,
+      "loss": 0.7868,
+      "step": 4357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9876104913240167e-06,
+      "loss": 0.7878,
+      "step": 4358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984045018258228e-06,
+      "loss": 0.7575,
+      "step": 4359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9804823938075344e-06,
+      "loss": 0.7678,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976922619237982e-06,
+      "loss": 0.7759,
+      "step": 4361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973365695814602e-06,
+      "loss": 0.7722,
+      "step": 4362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9698116248014088e-06,
+      "loss": 0.7463,
+      "step": 4363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9662604074614044e-06,
+      "loss": 0.8158,
+      "step": 4364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962712045056587e-06,
+      "loss": 0.7239,
+      "step": 4365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9591665388479196e-06,
+      "loss": 0.7425,
+      "step": 4366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9556238900953717e-06,
+      "loss": 0.7826,
+      "step": 4367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952084100057884e-06,
+      "loss": 0.8073,
+      "step": 4368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9485471699933833e-06,
+      "loss": 0.7849,
+      "step": 4369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945013101158787e-06,
+      "loss": 0.796,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941481894809987e-06,
+      "loss": 0.7864,
+      "step": 4371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9379535522018623e-06,
+      "loss": 0.7438,
+      "step": 4372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9344280745882716e-06,
+      "loss": 0.7885,
+      "step": 4373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9309054632220645e-06,
+      "loss": 0.7787,
+      "step": 4374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9273857193550537e-06,
+      "loss": 0.7482,
+      "step": 4375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923868844238054e-06,
+      "loss": 0.7448,
+      "step": 4376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920354839120845e-06,
+      "loss": 0.7578,
+      "step": 4377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916843705252195e-06,
+      "loss": 0.7797,
+      "step": 4378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9133354438798456e-06,
+      "loss": 0.2451,
+      "step": 4379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.7469,
+      "step": 4380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9063275436099382e-06,
+      "loss": 0.7697,
+      "step": 4381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9028279072027589e-06,
+      "loss": 0.7576,
+      "step": 4382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8993311482726562e-06,
+      "loss": 0.7553,
+      "step": 4383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.895837268062256e-06,
+      "loss": 0.7609,
+      "step": 4384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.892346267813182e-06,
+      "loss": 0.7666,
+      "step": 4385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8888581487660185e-06,
+      "loss": 0.734,
+      "step": 4386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8853729121603348e-06,
+      "loss": 0.8189,
+      "step": 4387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.881890559234668e-06,
+      "loss": 0.8083,
+      "step": 4388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.878411091226543e-06,
+      "loss": 0.7504,
+      "step": 4389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874934509372448e-06,
+      "loss": 0.7595,
+      "step": 4390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8714608149078473e-06,
+      "loss": 0.7774,
+      "step": 4391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8679900090671854e-06,
+      "loss": 0.7355,
+      "step": 4392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8645220930838748e-06,
+      "loss": 0.8059,
+      "step": 4393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8610570681903018e-06,
+      "loss": 0.783,
+      "step": 4394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8575949356178224e-06,
+      "loss": 0.785,
+      "step": 4395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8541356965967782e-06,
+      "loss": 0.7702,
+      "step": 4396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8506793523564615e-06,
+      "loss": 0.8228,
+      "step": 4397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8472259041251538e-06,
+      "loss": 0.7412,
+      "step": 4398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8437753531300984e-06,
+      "loss": 0.8211,
+      "step": 4399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8403277005975084e-06,
+      "loss": 0.781,
+      "step": 4400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8368829477525763e-06,
+      "loss": 0.7401,
+      "step": 4401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8334410958194527e-06,
+      "loss": 0.2518,
+      "step": 4402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8300021460212635e-06,
+      "loss": 0.8134,
+      "step": 4403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8265660995801004e-06,
+      "loss": 0.8215,
+      "step": 4404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8231329577170299e-06,
+      "loss": 0.7513,
+      "step": 4405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8197027216520734e-06,
+      "loss": 0.761,
+      "step": 4406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.816275392604232e-06,
+      "loss": 0.7615,
+      "step": 4407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8128509717914733e-06,
+      "loss": 0.7935,
+      "step": 4408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8094294604307183e-06,
+      "loss": 0.7711,
+      "step": 4409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8060108597378712e-06,
+      "loss": 0.7549,
+      "step": 4410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80259517092779e-06,
+      "loss": 0.6877,
+      "step": 4411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7991823952143028e-06,
+      "loss": 0.7563,
+      "step": 4412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7957725338101972e-06,
+      "loss": 0.804,
+      "step": 4413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7923655879272395e-06,
+      "loss": 0.8068,
+      "step": 4414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7889615587761378e-06,
+      "loss": 0.7998,
+      "step": 4415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.785560447566581e-06,
+      "loss": 0.7419,
+      "step": 4416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7821622555072216e-06,
+      "loss": 0.8366,
+      "step": 4417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7787669838056575e-06,
+      "loss": 0.7877,
+      "step": 4418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7753746336684674e-06,
+      "loss": 0.7952,
+      "step": 4419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7719852063011844e-06,
+      "loss": 0.7748,
+      "step": 4420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7685987029083008e-06,
+      "loss": 0.8004,
+      "step": 4421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7652151246932703e-06,
+      "loss": 0.7278,
+      "step": 4422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7618344728585135e-06,
+      "loss": 0.8212,
+      "step": 4423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7584567486054039e-06,
+      "loss": 0.2742,
+      "step": 4424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7550819531342755e-06,
+      "loss": 0.7676,
+      "step": 4425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7517100876444294e-06,
+      "loss": 0.757,
+      "step": 4426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74834115333411e-06,
+      "loss": 0.8108,
+      "step": 4427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7449751514005365e-06,
+      "loss": 0.7552,
+      "step": 4428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7416120830398765e-06,
+      "loss": 0.7939,
+      "step": 4429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7382519494472571e-06,
+      "loss": 0.799,
+      "step": 4430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7348947518167602e-06,
+      "loss": 0.7783,
+      "step": 4431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7315404913414325e-06,
+      "loss": 0.7283,
+      "step": 4432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7281891692132679e-06,
+      "loss": 0.847,
+      "step": 4433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7248407866232175e-06,
+      "loss": 0.8299,
+      "step": 4434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7214953447611948e-06,
+      "loss": 0.7233,
+      "step": 4435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7181528448160611e-06,
+      "loss": 0.8062,
+      "step": 4436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714813287975634e-06,
+      "loss": 0.7547,
+      "step": 4437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7114766754266842e-06,
+      "loss": 0.8109,
+      "step": 4438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081430083549434e-06,
+      "loss": 0.7817,
+      "step": 4439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048122879450812e-06,
+      "loss": 0.7951,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701484515380738e-06,
+      "loss": 0.8102,
+      "step": 4441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6981596918444953e-06,
+      "loss": 0.7481,
+      "step": 4442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694837818517886e-06,
+      "loss": 0.7964,
+      "step": 4443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6915188965814034e-06,
+      "loss": 0.7741,
+      "step": 4444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6882029272144852e-06,
+      "loss": 0.8354,
+      "step": 4445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6848899115955208e-06,
+      "loss": 0.7098,
+      "step": 4446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6815798509018467e-06,
+      "loss": 0.7721,
+      "step": 4447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6782727463097626e-06,
+      "loss": 0.7585,
+      "step": 4448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6749685989944974e-06,
+      "loss": 0.7551,
+      "step": 4449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6716674101302465e-06,
+      "loss": 0.7867,
+      "step": 4450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6683691808901448e-06,
+      "loss": 0.7529,
+      "step": 4451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6650739124462766e-06,
+      "loss": 0.8022,
+      "step": 4452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6617816059696801e-06,
+      "loss": 0.778,
+      "step": 4453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6584922626303325e-06,
+      "loss": 0.8233,
+      "step": 4454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655205883597163e-06,
+      "loss": 0.7226,
+      "step": 4455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6519224700380422e-06,
+      "loss": 0.749,
+      "step": 4456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6486420231197997e-06,
+      "loss": 0.7942,
+      "step": 4457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645364544008191e-06,
+      "loss": 0.7295,
+      "step": 4458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6420900338679357e-06,
+      "loss": 0.2392,
+      "step": 4459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6388184938626871e-06,
+      "loss": 0.8292,
+      "step": 4460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6355499251550456e-06,
+      "loss": 0.7234,
+      "step": 4461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6322843289065581e-06,
+      "loss": 0.7517,
+      "step": 4462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6290217062777147e-06,
+      "loss": 0.8265,
+      "step": 4463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6257620584279454e-06,
+      "loss": 0.2738,
+      "step": 4464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6225053865156237e-06,
+      "loss": 0.7637,
+      "step": 4465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6192516916980717e-06,
+      "loss": 0.8268,
+      "step": 4466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6160009751315454e-06,
+      "loss": 0.7534,
+      "step": 4467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6127532379712473e-06,
+      "loss": 0.7595,
+      "step": 4468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6095084813713201e-06,
+      "loss": 0.7795,
+      "step": 4469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6062667064848424e-06,
+      "loss": 0.8317,
+      "step": 4470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603027914463845e-06,
+      "loss": 0.723,
+      "step": 4471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.599792106459288e-06,
+      "loss": 0.7688,
+      "step": 4472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596559283621074e-06,
+      "loss": 0.799,
+      "step": 4473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5933294470980443e-06,
+      "loss": 0.7998,
+      "step": 4474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5901025980379826e-06,
+      "loss": 0.7332,
+      "step": 4475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5868787375876082e-06,
+      "loss": 0.7618,
+      "step": 4476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5836578668925762e-06,
+      "loss": 0.7759,
+      "step": 4477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5804399870974806e-06,
+      "loss": 0.7658,
+      "step": 4478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5772250993458605e-06,
+      "loss": 0.7555,
+      "step": 4479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5740132047801738e-06,
+      "loss": 0.7697,
+      "step": 4480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.570804304541832e-06,
+      "loss": 0.771,
+      "step": 4481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5675983997711797e-06,
+      "loss": 0.7863,
+      "step": 4482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5643954916074845e-06,
+      "loss": 0.7427,
+      "step": 4483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5611955811889645e-06,
+      "loss": 0.7096,
+      "step": 4484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5579986696527637e-06,
+      "loss": 0.807,
+      "step": 4485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5548047581349624e-06,
+      "loss": 0.2382,
+      "step": 4486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5516138477705732e-06,
+      "loss": 0.8208,
+      "step": 4487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5484259396935508e-06,
+      "loss": 0.765,
+      "step": 4488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5452410350367675e-06,
+      "loss": 0.7291,
+      "step": 4489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5420591349320401e-06,
+      "loss": 0.7258,
+      "step": 4490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538880240510121e-06,
+      "loss": 0.7583,
+      "step": 4491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5357043529006777e-06,
+      "loss": 0.8223,
+      "step": 4492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.532531473232326e-06,
+      "loss": 0.7964,
+      "step": 4493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5293616026326053e-06,
+      "loss": 0.7313,
+      "step": 4494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5261947422279866e-06,
+      "loss": 0.7971,
+      "step": 4495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5230308931438675e-06,
+      "loss": 0.7283,
+      "step": 4496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5198700565045865e-06,
+      "loss": 0.7759,
+      "step": 4497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.516712233433396e-06,
+      "loss": 0.8097,
+      "step": 4498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5135574250524898e-06,
+      "loss": 0.731,
+      "step": 4499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5104056324829908e-06,
+      "loss": 0.7675,
+      "step": 4500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5072568568449365e-06,
+      "loss": 0.7629,
+      "step": 4501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5041110992573083e-06,
+      "loss": 0.7807,
+      "step": 4502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5009683608380067e-06,
+      "loss": 0.8008,
+      "step": 4503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4978286427038602e-06,
+      "loss": 0.7501,
+      "step": 4504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4946919459706222e-06,
+      "loss": 0.8278,
+      "step": 4505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4915582717529797e-06,
+      "loss": 0.7653,
+      "step": 4506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4884276211645388e-06,
+      "loss": 0.7951,
+      "step": 4507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4852999953178293e-06,
+      "loss": 0.2384,
+      "step": 4508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4821753953243168e-06,
+      "loss": 0.7718,
+      "step": 4509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4790538222943795e-06,
+      "loss": 0.7924,
+      "step": 4510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.475935277337327e-06,
+      "loss": 0.7989,
+      "step": 4511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4728197615613905e-06,
+      "loss": 0.8141,
+      "step": 4512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469707276073724e-06,
+      "loss": 0.861,
+      "step": 4513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4665978219804056e-06,
+      "loss": 0.7592,
+      "step": 4514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4634914003864386e-06,
+      "loss": 0.8325,
+      "step": 4515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4603880123957448e-06,
+      "loss": 0.7825,
+      "step": 4516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4572876591111684e-06,
+      "loss": 0.8022,
+      "step": 4517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4541903416344783e-06,
+      "loss": 0.8062,
+      "step": 4518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4510960610663627e-06,
+      "loss": 0.724,
+      "step": 4519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4480048185064288e-06,
+      "loss": 0.2661,
+      "step": 4520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4449166150532036e-06,
+      "loss": 0.805,
+      "step": 4521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4418314518041432e-06,
+      "loss": 0.8291,
+      "step": 4522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4387493298556077e-06,
+      "loss": 0.7886,
+      "step": 4523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.435670250302892e-06,
+      "loss": 0.7967,
+      "step": 4524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4325942142402006e-06,
+      "loss": 0.802,
+      "step": 4525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4295212227606548e-06,
+      "loss": 0.7526,
+      "step": 4526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4264512769563054e-06,
+      "loss": 0.7692,
+      "step": 4527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4233843779181079e-06,
+      "loss": 0.7809,
+      "step": 4528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4203205267359432e-06,
+      "loss": 0.7685,
+      "step": 4529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.417259724498603e-06,
+      "loss": 0.7658,
+      "step": 4530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.414201972293805e-06,
+      "loss": 0.7727,
+      "step": 4531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4111472712081698e-06,
+      "loss": 0.772,
+      "step": 4532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4080956223272458e-06,
+      "loss": 0.7425,
+      "step": 4533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405047026735491e-06,
+      "loss": 0.6981,
+      "step": 4534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4020014855162755e-06,
+      "loss": 0.7916,
+      "step": 4535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3989589997518937e-06,
+      "loss": 0.7988,
+      "step": 4536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395919570523544e-06,
+      "loss": 0.7458,
+      "step": 4537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3928831989113444e-06,
+      "loss": 0.2777,
+      "step": 4538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.389849885994322e-06,
+      "loss": 0.8181,
+      "step": 4539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3868196328504258e-06,
+      "loss": 0.7488,
+      "step": 4540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3837924405565029e-06,
+      "loss": 0.6901,
+      "step": 4541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3807683101883285e-06,
+      "loss": 0.7317,
+      "step": 4542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3777472428205774e-06,
+      "loss": 0.824,
+      "step": 4543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3747292395268407e-06,
+      "loss": 0.8275,
+      "step": 4544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.371714301379624e-06,
+      "loss": 0.7465,
+      "step": 4545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3687024294503392e-06,
+      "loss": 0.2493,
+      "step": 4546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.365693624809309e-06,
+      "loss": 0.7725,
+      "step": 4547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3626878885257644e-06,
+      "loss": 0.81,
+      "step": 4548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3596852216678534e-06,
+      "loss": 0.7864,
+      "step": 4549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.356685625302625e-06,
+      "loss": 0.7629,
+      "step": 4550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3536891004960406e-06,
+      "loss": 0.7797,
+      "step": 4551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3506956483129674e-06,
+      "loss": 0.8207,
+      "step": 4552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3477052698171877e-06,
+      "loss": 0.76,
+      "step": 4553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344717966071385e-06,
+      "loss": 0.801,
+      "step": 4554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.341733738137151e-06,
+      "loss": 0.7946,
+      "step": 4555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3387525870749851e-06,
+      "loss": 0.769,
+      "step": 4556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3357745139442924e-06,
+      "loss": 0.7748,
+      "step": 4557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3327995198033883e-06,
+      "loss": 0.781,
+      "step": 4558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3298276057094894e-06,
+      "loss": 0.777,
+      "step": 4559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3268587727187198e-06,
+      "loss": 0.7337,
+      "step": 4560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3238930218861057e-06,
+      "loss": 0.7532,
+      "step": 4561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3209303542655837e-06,
+      "loss": 0.8212,
+      "step": 4562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.317970770909991e-06,
+      "loss": 0.7788,
+      "step": 4563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3150142728710669e-06,
+      "loss": 0.8055,
+      "step": 4564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3120608611994624e-06,
+      "loss": 0.7679,
+      "step": 4565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3091105369447166e-06,
+      "loss": 0.7231,
+      "step": 4566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3061633011552887e-06,
+      "loss": 0.7817,
+      "step": 4567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30321915487853e-06,
+      "loss": 0.787,
+      "step": 4568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3002780991606967e-06,
+      "loss": 0.7532,
+      "step": 4569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2973401350469428e-06,
+      "loss": 0.7575,
+      "step": 4570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.294405263581332e-06,
+      "loss": 0.7632,
+      "step": 4571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2914734858068233e-06,
+      "loss": 0.746,
+      "step": 4572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2885448027652726e-06,
+      "loss": 0.2485,
+      "step": 4573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2856192154974488e-06,
+      "loss": 0.7709,
+      "step": 4574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2826967250430044e-06,
+      "loss": 0.7779,
+      "step": 4575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2797773324405039e-06,
+      "loss": 0.797,
+      "step": 4576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276861038727405e-06,
+      "loss": 0.764,
+      "step": 4577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2739478449400667e-06,
+      "loss": 0.7436,
+      "step": 4578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710377521137418e-06,
+      "loss": 0.8348,
+      "step": 4579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.26813076128259e-06,
+      "loss": 0.8071,
+      "step": 4580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2652268734796603e-06,
+      "loss": 0.7988,
+      "step": 4581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2623260897368994e-06,
+      "loss": 0.7411,
+      "step": 4582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2594284110851618e-06,
+      "loss": 0.8098,
+      "step": 4583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2565338385541792e-06,
+      "loss": 0.7494,
+      "step": 4584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2536423731725978e-06,
+      "loss": 0.7203,
+      "step": 4585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250754015967951e-06,
+      "loss": 0.7233,
+      "step": 4586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247868767966669e-06,
+      "loss": 0.7429,
+      "step": 4587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2449866301940738e-06,
+      "loss": 0.7683,
+      "step": 4588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2421076036743895e-06,
+      "loss": 0.8026,
+      "step": 4589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.23923168943073e-06,
+      "loss": 0.7334,
+      "step": 4590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2363588884850997e-06,
+      "loss": 0.7505,
+      "step": 4591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2334892018584055e-06,
+      "loss": 0.7552,
+      "step": 4592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2306226305704405e-06,
+      "loss": 0.7239,
+      "step": 4593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2277591756398933e-06,
+      "loss": 0.7489,
+      "step": 4594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2248988380843418e-06,
+      "loss": 0.7756,
+      "step": 4595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2220416189202656e-06,
+      "loss": 0.7592,
+      "step": 4596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2191875191630209e-06,
+      "loss": 0.7558,
+      "step": 4597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.21633653982687e-06,
+      "loss": 0.7724,
+      "step": 4598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213488681924958e-06,
+      "loss": 0.7807,
+      "step": 4599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.210643946469321e-06,
+      "loss": 0.7326,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2078023344708911e-06,
+      "loss": 0.7681,
+      "step": 4601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2049638469394854e-06,
+      "loss": 0.2649,
+      "step": 4602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.202128484883811e-06,
+      "loss": 0.7925,
+      "step": 4603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1992962493114645e-06,
+      "loss": 0.7897,
+      "step": 4604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1964671412289376e-06,
+      "loss": 0.7879,
+      "step": 4605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1936411616415966e-06,
+      "loss": 0.7407,
+      "step": 4606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190818311553712e-06,
+      "loss": 0.731,
+      "step": 4607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1879985919684312e-06,
+      "loss": 0.7785,
+      "step": 4608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1851820038877915e-06,
+      "loss": 0.8184,
+      "step": 4609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1823685483127234e-06,
+      "loss": 0.7782,
+      "step": 4610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.179558226243036e-06,
+      "loss": 0.7658,
+      "step": 4611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1767510386774294e-06,
+      "loss": 0.8173,
+      "step": 4612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1739469866134855e-06,
+      "loss": 0.7652,
+      "step": 4613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.171146071047683e-06,
+      "loss": 0.7528,
+      "step": 4614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1683482929753675e-06,
+      "loss": 0.7962,
+      "step": 4615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1655536533907874e-06,
+      "loss": 0.8287,
+      "step": 4616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.162762153287068e-06,
+      "loss": 0.7444,
+      "step": 4617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159973793656215e-06,
+      "loss": 0.8316,
+      "step": 4618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1571885754891288e-06,
+      "loss": 0.7833,
+      "step": 4619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1544064997755843e-06,
+      "loss": 0.2767,
+      "step": 4620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1516275675042422e-06,
+      "loss": 0.7654,
+      "step": 4621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1488517796626453e-06,
+      "loss": 0.7798,
+      "step": 4622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1460791372372238e-06,
+      "loss": 0.7982,
+      "step": 4623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1433096412132838e-06,
+      "loss": 0.7894,
+      "step": 4624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1405432925750181e-06,
+      "loss": 0.7758,
+      "step": 4625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1377800923054961e-06,
+      "loss": 0.7758,
+      "step": 4626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.135020041386672e-06,
+      "loss": 0.7583,
+      "step": 4627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132263140799381e-06,
+      "loss": 0.8224,
+      "step": 4628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1295093915233391e-06,
+      "loss": 0.7283,
+      "step": 4629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1267587945371383e-06,
+      "loss": 0.7906,
+      "step": 4630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1240113508182514e-06,
+      "loss": 0.7787,
+      "step": 4631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1212670613430388e-06,
+      "loss": 0.7834,
+      "step": 4632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1185259270867289e-06,
+      "loss": 0.8068,
+      "step": 4633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1157879490234346e-06,
+      "loss": 0.7538,
+      "step": 4634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1130531281261448e-06,
+      "loss": 0.7384,
+      "step": 4635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1103214653667306e-06,
+      "loss": 0.6963,
+      "step": 4636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1075929617159364e-06,
+      "loss": 0.7628,
+      "step": 4637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1048676181433837e-06,
+      "loss": 0.7602,
+      "step": 4638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1021454356175776e-06,
+      "loss": 0.7911,
+      "step": 4639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0994264151058897e-06,
+      "loss": 0.2678,
+      "step": 4640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0967105575745762e-06,
+      "loss": 0.7662,
+      "step": 4641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0939978639887661e-06,
+      "loss": 0.7687,
+      "step": 4642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0912883353124647e-06,
+      "loss": 0.7463,
+      "step": 4643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088581972508549e-06,
+      "loss": 0.7589,
+      "step": 4644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085878776538778e-06,
+      "loss": 0.7513,
+      "step": 4645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0831787483637812e-06,
+      "loss": 0.7792,
+      "step": 4646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.080481888943059e-06,
+      "loss": 0.7686,
+      "step": 4647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0777881992349959e-06,
+      "loss": 0.7576,
+      "step": 4648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075097680196835e-06,
+      "loss": 0.7882,
+      "step": 4649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072410332784708e-06,
+      "loss": 0.7312,
+      "step": 4650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06972615795361e-06,
+      "loss": 0.7844,
+      "step": 4651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0670451566574102e-06,
+      "loss": 0.7769,
+      "step": 4652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0643673298488523e-06,
+      "loss": 0.7312,
+      "step": 4653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0616926784795511e-06,
+      "loss": 0.8096,
+      "step": 4654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0590212034999914e-06,
+      "loss": 0.7956,
+      "step": 4655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0563529058595302e-06,
+      "loss": 0.7484,
+      "step": 4656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0536877865063988e-06,
+      "loss": 0.7174,
+      "step": 4657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0510258463876899e-06,
+      "loss": 0.8006,
+      "step": 4658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0483670864493777e-06,
+      "loss": 0.7193,
+      "step": 4659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0457115076362978e-06,
+      "loss": 0.7872,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0430591108921594e-06,
+      "loss": 0.766,
+      "step": 4661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0404098971595378e-06,
+      "loss": 0.8213,
+      "step": 4662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0377638673798817e-06,
+      "loss": 0.7336,
+      "step": 4663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035121022493506e-06,
+      "loss": 0.7388,
+      "step": 4664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0324813634395903e-06,
+      "loss": 0.7791,
+      "step": 4665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0298448911561899e-06,
+      "loss": 0.8133,
+      "step": 4666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0272116065802206e-06,
+      "loss": 0.7954,
+      "step": 4667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0245815106474677e-06,
+      "loss": 0.7819,
+      "step": 4668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0219546042925842e-06,
+      "loss": 0.8045,
+      "step": 4669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0193308884490894e-06,
+      "loss": 0.7485,
+      "step": 4670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167103640493647e-06,
+      "loss": 0.2655,
+      "step": 4671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0140930320246667e-06,
+      "loss": 0.7134,
+      "step": 4672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0114788933051078e-06,
+      "loss": 0.793,
+      "step": 4673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0088679488196695e-06,
+      "loss": 0.2624,
+      "step": 4674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0062601994962018e-06,
+      "loss": 0.7697,
+      "step": 4675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0036556462614133e-06,
+      "loss": 0.8097,
+      "step": 4676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0010542900408804e-06,
+      "loss": 0.7544,
+      "step": 4677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.984561317590392e-07,
+      "loss": 0.7724,
+      "step": 4678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.958611723391986e-07,
+      "loss": 0.7178,
+      "step": 4679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.932694127035169e-07,
+      "loss": 0.7599,
+      "step": 4680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906808537730283e-07,
+      "loss": 0.8182,
+      "step": 4681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.880954964676226e-07,
+      "loss": 0.8169,
+      "step": 4682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.855133417060504e-07,
+      "loss": 0.7917,
+      "step": 4683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.829343904059342e-07,
+      "loss": 0.7381,
+      "step": 4684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.803586434837454e-07,
+      "loss": 0.6843,
+      "step": 4685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77786101854825e-07,
+      "loss": 0.7985,
+      "step": 4686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.7521676643337e-07,
+      "loss": 0.7731,
+      "step": 4687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.726506381324451e-07,
+      "loss": 0.7604,
+      "step": 4688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700877178639633e-07,
+      "loss": 0.7243,
+      "step": 4689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.675280065387117e-07,
+      "loss": 0.7586,
+      "step": 4690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.649715050663277e-07,
+      "loss": 0.77,
+      "step": 4691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.624182143553073e-07,
+      "loss": 0.7568,
+      "step": 4692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.598681353130145e-07,
+      "loss": 0.7528,
+      "step": 4693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.573212688456635e-07,
+      "loss": 0.7846,
+      "step": 4694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.547776158583277e-07,
+      "loss": 0.7505,
+      "step": 4695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.522371772549421e-07,
+      "loss": 0.7485,
+      "step": 4696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.496999539383e-07,
+      "loss": 0.7785,
+      "step": 4697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471659468100436e-07,
+      "loss": 0.7723,
+      "step": 4698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.446351567706847e-07,
+      "loss": 0.7824,
+      "step": 4699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.421075847195827e-07,
+      "loss": 0.7512,
+      "step": 4700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.395832315549536e-07,
+      "loss": 0.7008,
+      "step": 4701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.370620981738776e-07,
+      "loss": 0.7327,
+      "step": 4702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.345441854722826e-07,
+      "loss": 0.7673,
+      "step": 4703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.320294943449537e-07,
+      "loss": 0.7692,
+      "step": 4704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.295180256855308e-07,
+      "loss": 0.7307,
+      "step": 4705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.270097803865141e-07,
+      "loss": 0.6926,
+      "step": 4706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.245047593392519e-07,
+      "loss": 0.7932,
+      "step": 4707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.220029634339489e-07,
+      "loss": 0.7878,
+      "step": 4708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.195043935596615e-07,
+      "loss": 0.7877,
+      "step": 4709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.170090506043061e-07,
+      "loss": 0.8034,
+      "step": 4710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.145169354546457e-07,
+      "loss": 0.7732,
+      "step": 4711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.120280489962985e-07,
+      "loss": 0.7771,
+      "step": 4712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.095423921137369e-07,
+      "loss": 0.7026,
+      "step": 4713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.070599656902801e-07,
+      "loss": 0.7759,
+      "step": 4714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.0458077060811e-07,
+      "loss": 0.2597,
+      "step": 4715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.021048077482486e-07,
+      "loss": 0.7032,
+      "step": 4716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.996320779905765e-07,
+      "loss": 0.7686,
+      "step": 4717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.971625822138197e-07,
+      "loss": 0.7587,
+      "step": 4718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.94696321295564e-07,
+      "loss": 0.7111,
+      "step": 4719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.92233296112236e-07,
+      "loss": 0.7396,
+      "step": 4720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.897735075391156e-07,
+      "loss": 0.7942,
+      "step": 4721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.873169564503392e-07,
+      "loss": 0.7752,
+      "step": 4722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.848636437188785e-07,
+      "loss": 0.7934,
+      "step": 4723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824135702165693e-07,
+      "loss": 0.7996,
+      "step": 4724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.799667368140863e-07,
+      "loss": 0.7767,
+      "step": 4725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.775231443809574e-07,
+      "loss": 0.7322,
+      "step": 4726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750827937855566e-07,
+      "loss": 0.6985,
+      "step": 4727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.726456858951082e-07,
+      "loss": 0.7624,
+      "step": 4728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.702118215756817e-07,
+      "loss": 0.7659,
+      "step": 4729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.677812016921938e-07,
+      "loss": 0.7416,
+      "step": 4730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.653538271084139e-07,
+      "loss": 0.7852,
+      "step": 4731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.629296986869473e-07,
+      "loss": 0.7805,
+      "step": 4732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.605088172892573e-07,
+      "loss": 0.8053,
+      "step": 4733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.580911837756467e-07,
+      "loss": 0.7997,
+      "step": 4734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.556767990052639e-07,
+      "loss": 0.7802,
+      "step": 4735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.532656638361025e-07,
+      "loss": 0.7932,
+      "step": 4736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.508577791250095e-07,
+      "loss": 0.7645,
+      "step": 4737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.484531457276657e-07,
+      "loss": 0.7935,
+      "step": 4738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.460517644985999e-07,
+      "loss": 0.7771,
+      "step": 4739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.43653636291194e-07,
+      "loss": 0.7405,
+      "step": 4740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.412587619576562e-07,
+      "loss": 0.8349,
+      "step": 4741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.388671423490568e-07,
+      "loss": 0.8285,
+      "step": 4742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.364787783152973e-07,
+      "loss": 0.7891,
+      "step": 4743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.340936707051273e-07,
+      "loss": 0.7567,
+      "step": 4744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.317118203661356e-07,
+      "loss": 0.7304,
+      "step": 4745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.293332281447608e-07,
+      "loss": 0.7353,
+      "step": 4746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.269578948862756e-07,
+      "loss": 0.8463,
+      "step": 4747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.245858214347969e-07,
+      "loss": 0.8077,
+      "step": 4748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.222170086332871e-07,
+      "loss": 0.8106,
+      "step": 4749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.19851457323545e-07,
+      "loss": 0.7811,
+      "step": 4750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.174891683462116e-07,
+      "loss": 0.782,
+      "step": 4751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151301425407699e-07,
+      "loss": 0.7154,
+      "step": 4752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127743807455446e-07,
+      "loss": 0.7493,
+      "step": 4753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.10421883797694e-07,
+      "loss": 0.7798,
+      "step": 4754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.080726525332239e-07,
+      "loss": 0.764,
+      "step": 4755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.057266877869763e-07,
+      "loss": 0.8126,
+      "step": 4756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.033839903926299e-07,
+      "loss": 0.7549,
+      "step": 4757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.010445611827067e-07,
+      "loss": 0.8312,
+      "step": 4758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.98708400988566e-07,
+      "loss": 0.2721,
+      "step": 4759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.963755106404036e-07,
+      "loss": 0.7791,
+      "step": 4760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.940458909672522e-07,
+      "loss": 0.8206,
+      "step": 4761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.917195427969904e-07,
+      "loss": 0.8268,
+      "step": 4762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.893964669563215e-07,
+      "loss": 0.8572,
+      "step": 4763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87076664270795e-07,
+      "loss": 0.7512,
+      "step": 4764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.847601355647994e-07,
+      "loss": 0.7375,
+      "step": 4765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.824468816615482e-07,
+      "loss": 0.7867,
+      "step": 4766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.80136903383103e-07,
+      "loss": 0.8034,
+      "step": 4767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.778302015503547e-07,
+      "loss": 0.7952,
+      "step": 4768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.755267769830321e-07,
+      "loss": 0.7317,
+      "step": 4769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.732266304996971e-07,
+      "loss": 0.7791,
+      "step": 4770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.709297629177525e-07,
+      "loss": 0.799,
+      "step": 4771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.686361750534266e-07,
+      "loss": 0.7937,
+      "step": 4772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.663458677217916e-07,
+      "loss": 0.7364,
+      "step": 4773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.6405884173675e-07,
+      "loss": 0.6978,
+      "step": 4774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.61775097911035e-07,
+      "loss": 0.7574,
+      "step": 4775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594946370562184e-07,
+      "loss": 0.7598,
+      "step": 4776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.572174599827031e-07,
+      "loss": 0.7902,
+      "step": 4777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.549435674997252e-07,
+      "loss": 0.7601,
+      "step": 4778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.526729604153516e-07,
+      "loss": 0.7514,
+      "step": 4779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.504056395364879e-07,
+      "loss": 0.2366,
+      "step": 4780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.481416056688662e-07,
+      "loss": 0.7766,
+      "step": 4781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.458808596170497e-07,
+      "loss": 0.7256,
+      "step": 4782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.43623402184438e-07,
+      "loss": 0.7874,
+      "step": 4783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.413692341732582e-07,
+      "loss": 0.7945,
+      "step": 4784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.391183563845706e-07,
+      "loss": 0.7684,
+      "step": 4785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.368707696182653e-07,
+      "loss": 0.7774,
+      "step": 4786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.346264746730636e-07,
+      "loss": 0.8038,
+      "step": 4787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.32385472346514e-07,
+      "loss": 0.7713,
+      "step": 4788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.301477634350019e-07,
+      "loss": 0.7515,
+      "step": 4789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.279133487337342e-07,
+      "loss": 0.7991,
+      "step": 4790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.256822290367516e-07,
+      "loss": 0.8056,
+      "step": 4791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.234544051369218e-07,
+      "loss": 0.8383,
+      "step": 4792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.212298778259441e-07,
+      "loss": 0.7483,
+      "step": 4793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.190086478943459e-07,
+      "loss": 0.7233,
+      "step": 4794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.167907161314769e-07,
+      "loss": 0.6892,
+      "step": 4795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.145760833255267e-07,
+      "loss": 0.8126,
+      "step": 4796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123647502634967e-07,
+      "loss": 0.7906,
+      "step": 4797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.101567177312307e-07,
+      "loss": 0.8008,
+      "step": 4798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.07951986513391e-07,
+      "loss": 0.7693,
+      "step": 4799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.057505573934686e-07,
+      "loss": 0.2722,
+      "step": 4800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0355243115378e-07,
+      "loss": 0.2653,
+      "step": 4801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.013576085754725e-07,
+      "loss": 0.8236,
+      "step": 4802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.991660904385145e-07,
+      "loss": 0.8054,
+      "step": 4803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.969778775217007e-07,
+      "loss": 0.7874,
+      "step": 4804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.947929706026546e-07,
+      "loss": 0.8135,
+      "step": 4805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.926113704578197e-07,
+      "loss": 0.7763,
+      "step": 4806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.904330778624691e-07,
+      "loss": 0.7664,
+      "step": 4807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.88258093590699e-07,
+      "loss": 0.7467,
+      "step": 4808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860864184154281e-07,
+      "loss": 0.7906,
+      "step": 4809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.839180531084e-07,
+      "loss": 0.739,
+      "step": 4810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.81752998440185e-07,
+      "loss": 0.7705,
+      "step": 4811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.795912551801742e-07,
+      "loss": 0.8367,
+      "step": 4812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.774328240965788e-07,
+      "loss": 0.7642,
+      "step": 4813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 0.8095,
+      "step": 4814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.731259015256209e-07,
+      "loss": 0.7499,
+      "step": 4815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709774115687983e-07,
+      "loss": 0.796,
+      "step": 4816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.688322368494804e-07,
+      "loss": 0.7873,
+      "step": 4817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.666903781299927e-07,
+      "loss": 0.7435,
+      "step": 4818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.64551836171482e-07,
+      "loss": 0.733,
+      "step": 4819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.624166117339215e-07,
+      "loss": 0.7714,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.602847055761008e-07,
+      "loss": 0.8021,
+      "step": 4821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.581561184556296e-07,
+      "loss": 0.8209,
+      "step": 4822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.560308511289426e-07,
+      "loss": 0.7742,
+      "step": 4823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.539089043512914e-07,
+      "loss": 0.7996,
+      "step": 4824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.517902788767483e-07,
+      "loss": 0.795,
+      "step": 4825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.496749754582043e-07,
+      "loss": 0.7428,
+      "step": 4826
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.475629948473716e-07,
+      "loss": 0.786,
+      "step": 4827
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.454543377947786e-07,
+      "loss": 0.7958,
+      "step": 4828
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.433490050497792e-07,
+      "loss": 0.2828,
+      "step": 4829
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.412469973605384e-07,
+      "loss": 0.7536,
+      "step": 4830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.39148315474043e-07,
+      "loss": 0.7431,
+      "step": 4831
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.370529601360997e-07,
+      "loss": 0.749,
+      "step": 4832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.349609320913297e-07,
+      "loss": 0.753,
+      "step": 4833
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.328722320831737e-07,
+      "loss": 0.7749,
+      "step": 4834
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.30786860853887e-07,
+      "loss": 0.7536,
+      "step": 4835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.287048191445488e-07,
+      "loss": 0.7361,
+      "step": 4836
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.266261076950442e-07,
+      "loss": 0.7982,
+      "step": 4837
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.245507272440843e-07,
+      "loss": 0.7663,
+      "step": 4838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.22478678529197e-07,
+      "loss": 0.7772,
+      "step": 4839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204099622867144e-07,
+      "loss": 0.24,
+      "step": 4840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.183445792517983e-07,
+      "loss": 0.7887,
+      "step": 4841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.162825301584185e-07,
+      "loss": 0.7877,
+      "step": 4842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.14223815739361e-07,
+      "loss": 0.7585,
+      "step": 4843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.121684367262271e-07,
+      "loss": 0.765,
+      "step": 4844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101163938494359e-07,
+      "loss": 0.7546,
+      "step": 4845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08067687838213e-07,
+      "loss": 0.827,
+      "step": 4846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.06022319420606e-07,
+      "loss": 0.7632,
+      "step": 4847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.039802893234769e-07,
+      "loss": 0.2391,
+      "step": 4848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.01941598272493e-07,
+      "loss": 0.7901,
+      "step": 4849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.999062469921424e-07,
+      "loss": 0.7968,
+      "step": 4850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.978742362057244e-07,
+      "loss": 0.7516,
+      "step": 4851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.958455666353502e-07,
+      "loss": 0.7221,
+      "step": 4852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.938202390019432e-07,
+      "loss": 0.8072,
+      "step": 4853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.917982540252442e-07,
+      "loss": 0.7923,
+      "step": 4854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.897796124237986e-07,
+      "loss": 0.7578,
+      "step": 4855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877643149149669e-07,
+      "loss": 0.7756,
+      "step": 4856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.857523622149275e-07,
+      "loss": 0.737,
+      "step": 4857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83743755038656e-07,
+      "loss": 0.7644,
+      "step": 4858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.817384940999538e-07,
+      "loss": 0.793,
+      "step": 4859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.797365801114241e-07,
+      "loss": 0.7452,
+      "step": 4860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.777380137844835e-07,
+      "loss": 0.7572,
+      "step": 4861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.757427958293582e-07,
+      "loss": 0.7747,
+      "step": 4862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.737509269550867e-07,
+      "loss": 0.8199,
+      "step": 4863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.71762407869515e-07,
+      "loss": 0.7777,
+      "step": 4864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.697772392792988e-07,
+      "loss": 0.7012,
+      "step": 4865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.677954218899063e-07,
+      "loss": 0.7758,
+      "step": 4866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.658169564056093e-07,
+      "loss": 0.274,
+      "step": 4867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.638418435294935e-07,
+      "loss": 0.7528,
+      "step": 4868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.618700839634494e-07,
+      "loss": 0.7469,
+      "step": 4869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.59901678408179e-07,
+      "loss": 0.8014,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.579366275631882e-07,
+      "loss": 0.759,
+      "step": 4871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.559749321267971e-07,
+      "loss": 0.7286,
+      "step": 4872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.540165927961294e-07,
+      "loss": 0.2334,
+      "step": 4873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.520616102671128e-07,
+      "loss": 0.7531,
+      "step": 4874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.501099852344904e-07,
+      "loss": 0.7807,
+      "step": 4875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.481617183918053e-07,
+      "loss": 0.7523,
+      "step": 4876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.462168104314092e-07,
+      "loss": 0.8198,
+      "step": 4877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.442752620444602e-07,
+      "loss": 0.7547,
+      "step": 4878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.423370739209266e-07,
+      "loss": 0.8274,
+      "step": 4879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.404022467495728e-07,
+      "loss": 0.6893,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.384707812179791e-07,
+      "loss": 0.7773,
+      "step": 4881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36542678012525e-07,
+      "loss": 0.7823,
+      "step": 4882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.34617937818398e-07,
+      "loss": 0.7702,
+      "step": 4883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.326965613195867e-07,
+      "loss": 0.2469,
+      "step": 4884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.307785491988904e-07,
+      "loss": 0.2337,
+      "step": 4885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288639021379094e-07,
+      "loss": 0.7499,
+      "step": 4886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.26952620817045e-07,
+      "loss": 0.7311,
+      "step": 4887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.250447059155117e-07,
+      "loss": 0.8293,
+      "step": 4888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.231401581113138e-07,
+      "loss": 0.7876,
+      "step": 4889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.212389780812733e-07,
+      "loss": 0.7691,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193411665010073e-07,
+      "loss": 0.245,
+      "step": 4891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.174467240449366e-07,
+      "loss": 0.246,
+      "step": 4892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.155556513862848e-07,
+      "loss": 0.7807,
+      "step": 4893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.136679491970809e-07,
+      "loss": 0.7575,
+      "step": 4894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.117836181481539e-07,
+      "loss": 0.8003,
+      "step": 4895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.099026589091338e-07,
+      "loss": 0.7462,
+      "step": 4896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.080250721484559e-07,
+      "loss": 0.7913,
+      "step": 4897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.06150858533353e-07,
+      "loss": 0.7354,
+      "step": 4898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042800187298613e-07,
+      "loss": 0.2607,
+      "step": 4899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.024125534028179e-07,
+      "loss": 0.7595,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.005484632158597e-07,
+      "loss": 0.7955,
+      "step": 4901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.986877488314246e-07,
+      "loss": 0.7882,
+      "step": 4902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.968304109107536e-07,
+      "loss": 0.8186,
+      "step": 4903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.949764501138832e-07,
+      "loss": 0.7444,
+      "step": 4904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.93125867099652e-07,
+      "loss": 0.7699,
+      "step": 4905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.912786625257005e-07,
+      "loss": 0.8085,
+      "step": 4906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.7698,
+      "step": 4907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875943913231818e-07,
+      "loss": 0.7493,
+      "step": 4908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.857573260038862e-07,
+      "loss": 0.7856,
+      "step": 4909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.839236417434157e-07,
+      "loss": 0.7714,
+      "step": 4910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.820933391933991e-07,
+      "loss": 0.8079,
+      "step": 4911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.802664190042716e-07,
+      "loss": 0.7884,
+      "step": 4912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.784428818252596e-07,
+      "loss": 0.7592,
+      "step": 4913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.766227283043912e-07,
+      "loss": 0.7822,
+      "step": 4914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7480595908849123e-07,
+      "loss": 0.7501,
+      "step": 4915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7299257482318116e-07,
+      "loss": 0.8341,
+      "step": 4916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7118257615288097e-07,
+      "loss": 0.7455,
+      "step": 4917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6937596372080397e-07,
+      "loss": 0.7005,
+      "step": 4918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6757273816896764e-07,
+      "loss": 0.8302,
+      "step": 4919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.65772900138175e-07,
+      "loss": 0.7749,
+      "step": 4920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.639764502680333e-07,
+      "loss": 0.7455,
+      "step": 4921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.621833891969474e-07,
+      "loss": 0.7181,
+      "step": 4922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.603937175621076e-07,
+      "loss": 0.8025,
+      "step": 4923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5860743599951186e-07,
+      "loss": 0.2441,
+      "step": 4924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.568245451439446e-07,
+      "loss": 0.7516,
+      "step": 4925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550450456289901e-07,
+      "loss": 0.7896,
+      "step": 4926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5326893808702257e-07,
+      "loss": 0.8058,
+      "step": 4927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5149622314921925e-07,
+      "loss": 0.7615,
+      "step": 4928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.497269014455419e-07,
+      "loss": 0.8016,
+      "step": 4929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4796097360475186e-07,
+      "loss": 0.7421,
+      "step": 4930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.461984402544084e-07,
+      "loss": 0.7613,
+      "step": 4931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.444393020208526e-07,
+      "loss": 0.8025,
+      "step": 4932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4268355952923115e-07,
+      "loss": 0.2532,
+      "step": 4933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4093121340347824e-07,
+      "loss": 0.7768,
+      "step": 4934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.391822642663213e-07,
+      "loss": 0.7595,
+      "step": 4935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3743671273927867e-07,
+      "loss": 0.798,
+      "step": 4936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.356945594426676e-07,
+      "loss": 0.7191,
+      "step": 4937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3395580499559276e-07,
+      "loss": 0.8186,
+      "step": 4938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3222045001594993e-07,
+      "loss": 0.8095,
+      "step": 4939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3048849512043245e-07,
+      "loss": 0.7774,
+      "step": 4940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2875994092452024e-07,
+      "loss": 0.7814,
+      "step": 4941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.270347880424863e-07,
+      "loss": 0.7469,
+      "step": 4942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.253130370873948e-07,
+      "loss": 0.8041,
+      "step": 4943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.235946886711018e-07,
+      "loss": 0.7063,
+      "step": 4944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.218797434042532e-07,
+      "loss": 0.7989,
+      "step": 4945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2016820189628603e-07,
+      "loss": 0.2542,
+      "step": 4946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.184600647554282e-07,
+      "loss": 0.7921,
+      "step": 4947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1675533258869525e-07,
+      "loss": 0.7507,
+      "step": 4948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1505400600189706e-07,
+      "loss": 0.7673,
+      "step": 4949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.133560855996299e-07,
+      "loss": 0.7512,
+      "step": 4950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.116615719852812e-07,
+      "loss": 0.2468,
+      "step": 4951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0997046576102597e-07,
+      "loss": 0.807,
+      "step": 4952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0828276752783227e-07,
+      "loss": 0.7961,
+      "step": 4953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0659847788544926e-07,
+      "loss": 0.7644,
+      "step": 4954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0491759743242485e-07,
+      "loss": 0.7821,
+      "step": 4955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.032401267660879e-07,
+      "loss": 0.7121,
+      "step": 4956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0156606648255714e-07,
+      "loss": 0.7486,
+      "step": 4957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.998954171767422e-07,
+      "loss": 0.7451,
+      "step": 4958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.982281794423393e-07,
+      "loss": 0.758,
+      "step": 4959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9656435387183e-07,
+      "loss": 0.7668,
+      "step": 4960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949039410564837e-07,
+      "loss": 0.7964,
+      "step": 4961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9324694158636266e-07,
+      "loss": 0.7613,
+      "step": 4962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.915933560503071e-07,
+      "loss": 0.7864,
+      "step": 4963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.899431850359503e-07,
+      "loss": 0.7673,
+      "step": 4964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88296429129712e-07,
+      "loss": 0.8151,
+      "step": 4965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.86653088916793e-07,
+      "loss": 0.255,
+      "step": 4966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.850131649811883e-07,
+      "loss": 0.7351,
+      "step": 4967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.83376657905673e-07,
+      "loss": 0.7362,
+      "step": 4968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.817435682718096e-07,
+      "loss": 0.7793,
+      "step": 4969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.801138966599449e-07,
+      "loss": 0.7232,
+      "step": 4970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7848764364921573e-07,
+      "loss": 0.7898,
+      "step": 4971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7686480981753537e-07,
+      "loss": 0.7499,
+      "step": 4972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7524539574161133e-07,
+      "loss": 0.7542,
+      "step": 4973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.736294019969311e-07,
+      "loss": 0.7276,
+      "step": 4974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.720168291577653e-07,
+      "loss": 0.7681,
+      "step": 4975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7040767779717435e-07,
+      "loss": 0.7548,
+      "step": 4976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6880194848699756e-07,
+      "loss": 0.7591,
+      "step": 4977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.671996417978596e-07,
+      "loss": 0.7224,
+      "step": 4978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6560075829916943e-07,
+      "loss": 0.7052,
+      "step": 4979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.640052985591191e-07,
+      "loss": 0.7724,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.624132631446864e-07,
+      "loss": 0.7546,
+      "step": 4981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.608246526216275e-07,
+      "loss": 0.7906,
+      "step": 4982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5923946755448413e-07,
+      "loss": 0.8209,
+      "step": 4983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5765770850658244e-07,
+      "loss": 0.755,
+      "step": 4984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.560793760400261e-07,
+      "loss": 0.7661,
+      "step": 4985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.545044707157075e-07,
+      "loss": 0.7966,
+      "step": 4986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5293299309329674e-07,
+      "loss": 0.7578,
+      "step": 4987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5136494373124495e-07,
+      "loss": 0.8282,
+      "step": 4988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.498003231867897e-07,
+      "loss": 0.7685,
+      "step": 4989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4823913201594616e-07,
+      "loss": 0.7575,
+      "step": 4990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4668137077351173e-07,
+      "loss": 0.8133,
+      "step": 4991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451270400130646e-07,
+      "loss": 0.7903,
+      "step": 4992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.435761402869675e-07,
+      "loss": 0.7706,
+      "step": 4993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.420286721463562e-07,
+      "loss": 0.8024,
+      "step": 4994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40484636141154e-07,
+      "loss": 0.7796,
+      "step": 4995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.389440328200644e-07,
+      "loss": 0.7699,
+      "step": 4996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.374068627305638e-07,
+      "loss": 0.7728,
+      "step": 4997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3587312641891747e-07,
+      "loss": 0.7595,
+      "step": 4998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3434282443016387e-07,
+      "loss": 0.7766,
+      "step": 4999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328159573081258e-07,
+      "loss": 0.812,
+      "step": 5000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.312925255954014e-07,
+      "loss": 0.7802,
+      "step": 5001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2977252983337204e-07,
+      "loss": 0.7945,
+      "step": 5002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.282559705621924e-07,
+      "loss": 0.7906,
+      "step": 5003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2674284832080127e-07,
+      "loss": 0.766,
+      "step": 5004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2523316364691527e-07,
+      "loss": 0.7348,
+      "step": 5005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.237269170770263e-07,
+      "loss": 0.8101,
+      "step": 5006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2222410914640733e-07,
+      "loss": 0.7406,
+      "step": 5007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2072474038910895e-07,
+      "loss": 0.7546,
+      "step": 5008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1922881133795827e-07,
+      "loss": 0.7309,
+      "step": 5009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1773632252455886e-07,
+      "loss": 0.7686,
+      "step": 5010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.162472744792977e-07,
+      "loss": 0.7473,
+      "step": 5011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1476166773133254e-07,
+      "loss": 0.7876,
+      "step": 5012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.132795028086011e-07,
+      "loss": 0.7495,
+      "step": 5013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.118007802378198e-07,
+      "loss": 0.7733,
+      "step": 5014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.10325500544475e-07,
+      "loss": 0.7788,
+      "step": 5015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0885366425283835e-07,
+      "loss": 0.7283,
+      "step": 5016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0738527188595046e-07,
+      "loss": 0.2667,
+      "step": 5017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0592032396563385e-07,
+      "loss": 0.2702,
+      "step": 5018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0445882101248217e-07,
+      "loss": 0.7713,
+      "step": 5019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.030007635458687e-07,
+      "loss": 0.7466,
+      "step": 5020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0154615208393913e-07,
+      "loss": 0.7557,
+      "step": 5021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.000949871436165e-07,
+      "loss": 0.7629,
+      "step": 5022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.986472692406006e-07,
+      "loss": 0.7431,
+      "step": 5023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.972029988893621e-07,
+      "loss": 0.7919,
+      "step": 5024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9576217660314823e-07,
+      "loss": 0.7824,
+      "step": 5025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.943248028939838e-07,
+      "loss": 0.7568,
+      "step": 5026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.928908782726636e-07,
+      "loss": 0.788,
+      "step": 5027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9146040324875777e-07,
+      "loss": 0.7954,
+      "step": 5028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.900333783306142e-07,
+      "loss": 0.7545,
+      "step": 5029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8860980402535156e-07,
+      "loss": 0.7946,
+      "step": 5030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.871896808388608e-07,
+      "loss": 0.7373,
+      "step": 5031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.857730092758115e-07,
+      "loss": 0.7811,
+      "step": 5032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.843597898396422e-07,
+      "loss": 0.7988,
+      "step": 5033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8295002303256546e-07,
+      "loss": 0.7741,
+      "step": 5034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.815437093555673e-07,
+      "loss": 0.704,
+      "step": 5035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.80140849308409e-07,
+      "loss": 0.8166,
+      "step": 5036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.787414433896185e-07,
+      "loss": 0.7032,
+      "step": 5037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.773454920965046e-07,
+      "loss": 0.77,
+      "step": 5038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.759529959251406e-07,
+      "loss": 0.767,
+      "step": 5039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7456395537037605e-07,
+      "loss": 0.7679,
+      "step": 5040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.731783709258329e-07,
+      "loss": 0.2386,
+      "step": 5041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7179624308390387e-07,
+      "loss": 0.7212,
+      "step": 5042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.704175723357505e-07,
+      "loss": 0.7996,
+      "step": 5043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6904235917131094e-07,
+      "loss": 0.7493,
+      "step": 5044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.67670604079292e-07,
+      "loss": 0.7843,
+      "step": 5045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6630230754717047e-07,
+      "loss": 0.7042,
+      "step": 5046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.649374700611962e-07,
+      "loss": 0.7701,
+      "step": 5047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.63576092106389e-07,
+      "loss": 0.7768,
+      "step": 5048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.622181741665375e-07,
+      "loss": 0.8111,
+      "step": 5049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6086371672420565e-07,
+      "loss": 0.7388,
+      "step": 5050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.595127202607228e-07,
+      "loss": 0.786,
+      "step": 5051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.581651852561906e-07,
+      "loss": 0.736,
+      "step": 5052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5682111218947815e-07,
+      "loss": 0.8246,
+      "step": 5053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.554805015382289e-07,
+      "loss": 0.8062,
+      "step": 5054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.541433537788518e-07,
+      "loss": 0.7135,
+      "step": 5055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.528096693865267e-07,
+      "loss": 0.7365,
+      "step": 5056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5147944883520237e-07,
+      "loss": 0.7246,
+      "step": 5057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.501526925975961e-07,
+      "loss": 0.7604,
+      "step": 5058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4882940114519747e-07,
+      "loss": 0.2364,
+      "step": 5059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4750957494826033e-07,
+      "loss": 0.7331,
+      "step": 5060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4619321447580834e-07,
+      "loss": 0.8318,
+      "step": 5061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.44880320195634e-07,
+      "loss": 0.7576,
+      "step": 5062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.435708925742997e-07,
+      "loss": 0.7934,
+      "step": 5063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.422649320771331e-07,
+      "loss": 0.8251,
+      "step": 5064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.409624391682308e-07,
+      "loss": 0.7485,
+      "step": 5065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3966341431045813e-07,
+      "loss": 0.7943,
+      "step": 5066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3836785796544693e-07,
+      "loss": 0.7238,
+      "step": 5067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3707577059359665e-07,
+      "loss": 0.2518,
+      "step": 5068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.357871526540745e-07,
+      "loss": 0.7621,
+      "step": 5069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3450200460481298e-07,
+      "loss": 0.7683,
+      "step": 5070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.332203269025124e-07,
+      "loss": 0.7735,
+      "step": 5071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3194212000264283e-07,
+      "loss": 0.7996,
+      "step": 5072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3066738435943647e-07,
+      "loss": 0.7411,
+      "step": 5073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.293961204258932e-07,
+      "loss": 0.7842,
+      "step": 5074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2812832865378054e-07,
+      "loss": 0.2746,
+      "step": 5075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2686400949363142e-07,
+      "loss": 0.7486,
+      "step": 5076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.256031633947453e-07,
+      "loss": 0.7322,
+      "step": 5077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2434579080518382e-07,
+      "loss": 0.8134,
+      "step": 5078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2309189217178172e-07,
+      "loss": 0.8363,
+      "step": 5079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.218414679401304e-07,
+      "loss": 0.7451,
+      "step": 5080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2059451855459323e-07,
+      "loss": 0.7984,
+      "step": 5081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.193510444582958e-07,
+      "loss": 0.7958,
+      "step": 5082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1811104609312904e-07,
+      "loss": 0.2695,
+      "step": 5083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1687452389974829e-07,
+      "loss": 0.8132,
+      "step": 5084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1564147831757643e-07,
+      "loss": 0.7754,
+      "step": 5085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.144119097847941e-07,
+      "loss": 0.7676,
+      "step": 5086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.131858187383551e-07,
+      "loss": 0.7981,
+      "step": 5087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.11963205613972e-07,
+      "loss": 0.7581,
+      "step": 5088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1074407084612169e-07,
+      "loss": 0.7883,
+      "step": 5089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0952841486804653e-07,
+      "loss": 0.7832,
+      "step": 5090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0831623811175095e-07,
+      "loss": 0.7651,
+      "step": 5091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.071075410080059e-07,
+      "loss": 0.7434,
+      "step": 5092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0590232398634114e-07,
+      "loss": 0.7789,
+      "step": 5093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0470058747505516e-07,
+      "loss": 0.8068,
+      "step": 5094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0350233190120528e-07,
+      "loss": 0.7492,
+      "step": 5095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0230755769061306e-07,
+      "loss": 0.7764,
+      "step": 5096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0111626526786553e-07,
+      "loss": 0.7267,
+      "step": 5097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9992845505630741e-07,
+      "loss": 0.775,
+      "step": 5098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9874412747804994e-07,
+      "loss": 0.7494,
+      "step": 5099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9756328295396532e-07,
+      "loss": 0.7469,
+      "step": 5100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9638592190368677e-07,
+      "loss": 0.7141,
+      "step": 5101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9521204474561294e-07,
+      "loss": 0.7893,
+      "step": 5102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9404165189690126e-07,
+      "loss": 0.719,
+      "step": 5103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9287474377347238e-07,
+      "loss": 0.7432,
+      "step": 5104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9171132079000676e-07,
+      "loss": 0.7868,
+      "step": 5105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.905513833599504e-07,
+      "loss": 0.7951,
+      "step": 5106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8939493189550684e-07,
+      "loss": 0.7826,
+      "step": 5107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8824196680764296e-07,
+      "loss": 0.7532,
+      "step": 5108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870924885060843e-07,
+      "loss": 0.7893,
+      "step": 5109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.859464973993208e-07,
+      "loss": 0.8044,
+      "step": 5110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8480399389459892e-07,
+      "loss": 0.8027,
+      "step": 5111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.836649783979294e-07,
+      "loss": 0.7559,
+      "step": 5112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8252945131408183e-07,
+      "loss": 0.7812,
+      "step": 5113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8139741304658566e-07,
+      "loss": 0.795,
+      "step": 5114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8026886399773347e-07,
+      "loss": 0.2642,
+      "step": 5115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.791438045685734e-07,
+      "loss": 0.8197,
+      "step": 5116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.780222351589167e-07,
+      "loss": 0.7611,
+      "step": 5117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7690415616733124e-07,
+      "loss": 0.7728,
+      "step": 5118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7578956799115143e-07,
+      "loss": 0.7638,
+      "step": 5119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7467847102646262e-07,
+      "loss": 0.2868,
+      "step": 5120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7357086566811344e-07,
+      "loss": 0.7462,
+      "step": 5121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7246675230971345e-07,
+      "loss": 0.7899,
+      "step": 5122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7136613134362766e-07,
+      "loss": 0.7838,
+      "step": 5123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7026900316098217e-07,
+      "loss": 0.8072,
+      "step": 5124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.691753681516628e-07,
+      "loss": 0.6967,
+      "step": 5125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6808522670430981e-07,
+      "loss": 0.7052,
+      "step": 5126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6699857920632667e-07,
+      "loss": 0.7803,
+      "step": 5127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6591542604387445e-07,
+      "loss": 0.782,
+      "step": 5128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6483576760186748e-07,
+      "loss": 0.7429,
+      "step": 5129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6375960426398552e-07,
+      "loss": 0.7539,
+      "step": 5130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6268693641266152e-07,
+      "loss": 0.7482,
+      "step": 5131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6161776442908728e-07,
+      "loss": 0.7657,
+      "step": 5132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6055208869321327e-07,
+      "loss": 0.7297,
+      "step": 5133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5948990958374543e-07,
+      "loss": 0.7956,
+      "step": 5134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5843122747814854e-07,
+      "loss": 0.8433,
+      "step": 5135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5737604275264496e-07,
+      "loss": 0.8534,
+      "step": 5136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.563243557822147e-07,
+      "loss": 0.8071,
+      "step": 5137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5527616694059333e-07,
+      "loss": 0.7921,
+      "step": 5138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5423147660027172e-07,
+      "loss": 0.7141,
+      "step": 5139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5319028513250177e-07,
+      "loss": 0.7739,
+      "step": 5140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.521525929072898e-07,
+      "loss": 0.8649,
+      "step": 5141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5111840029339743e-07,
+      "loss": 0.776,
+      "step": 5142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.50087707658344e-07,
+      "loss": 0.7843,
+      "step": 5143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.490605153684066e-07,
+      "loss": 0.7536,
+      "step": 5144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4803682378861317e-07,
+      "loss": 0.2585,
+      "step": 5145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4701663328275383e-07,
+      "loss": 0.808,
+      "step": 5146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4599994421337193e-07,
+      "loss": 0.2302,
+      "step": 5147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4498675694176402e-07,
+      "loss": 0.7715,
+      "step": 5148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4397707182798647e-07,
+      "loss": 0.8145,
+      "step": 5149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4297088923084902e-07,
+      "loss": 0.7276,
+      "step": 5150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4196820950791778e-07,
+      "loss": 0.7715,
+      "step": 5151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4096903301551e-07,
+      "loss": 0.785,
+      "step": 5152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.399733601087061e-07,
+      "loss": 0.7744,
+      "step": 5153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3898119114133192e-07,
+      "loss": 0.7899,
+      "step": 5154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3799252646597428e-07,
+      "loss": 0.2809,
+      "step": 5155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3700736643397437e-07,
+      "loss": 0.7968,
+      "step": 5156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3602571139542554e-07,
+      "loss": 0.8031,
+      "step": 5157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3504756169917533e-07,
+      "loss": 0.7396,
+      "step": 5158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3407291769283015e-07,
+      "loss": 0.8172,
+      "step": 5159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3310177972274517e-07,
+      "loss": 0.245,
+      "step": 5160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3213414813403213e-07,
+      "loss": 0.804,
+      "step": 5161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3117002327055927e-07,
+      "loss": 0.7676,
+      "step": 5162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3020940547494033e-07,
+      "loss": 0.7759,
+      "step": 5163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.292522950885533e-07,
+      "loss": 0.8208,
+      "step": 5164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2829869245152282e-07,
+      "loss": 0.8011,
+      "step": 5165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.273485979027278e-07,
+      "loss": 0.8015,
+      "step": 5166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2640201177980371e-07,
+      "loss": 0.7665,
+      "step": 5167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2545893441913592e-07,
+      "loss": 0.2323,
+      "step": 5168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2451936615586413e-07,
+      "loss": 0.7525,
+      "step": 5169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2358330732388012e-07,
+      "loss": 0.7457,
+      "step": 5170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2265075825583227e-07,
+      "loss": 0.7526,
+      "step": 5171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.217217192831155e-07,
+      "loss": 0.8098,
+      "step": 5172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2079619073588345e-07,
+      "loss": 0.7427,
+      "step": 5173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1987417294303748e-07,
+      "loss": 0.7513,
+      "step": 5174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1895566623223332e-07,
+      "loss": 0.8002,
+      "step": 5175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1804067092987981e-07,
+      "loss": 0.7758,
+      "step": 5176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1712918736113798e-07,
+      "loss": 0.7678,
+      "step": 5177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1622121584991874e-07,
+      "loss": 0.2416,
+      "step": 5178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1531675671888621e-07,
+      "loss": 0.2754,
+      "step": 5179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1441581028945659e-07,
+      "loss": 0.8152,
+      "step": 5180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1351837688179824e-07,
+      "loss": 0.7652,
+      "step": 5181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1262445681483048e-07,
+      "loss": 0.7521,
+      "step": 5182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1173405040622254e-07,
+      "loss": 0.8165,
+      "step": 5183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1084715797239798e-07,
+      "loss": 0.783,
+      "step": 5184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0996377982852801e-07,
+      "loss": 0.7553,
+      "step": 5185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0908391628854042e-07,
+      "loss": 0.8172,
+      "step": 5186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0820756766510843e-07,
+      "loss": 0.8045,
+      "step": 5187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0733473426965846e-07,
+      "loss": 0.7533,
+      "step": 5188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0646541641236797e-07,
+      "loss": 0.7384,
+      "step": 5189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0559961440216538e-07,
+      "loss": 0.7591,
+      "step": 5190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0473732854672903e-07,
+      "loss": 0.7883,
+      "step": 5191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0387855915248712e-07,
+      "loss": 0.771,
+      "step": 5192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0302330652462e-07,
+      "loss": 0.7565,
+      "step": 5193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0217157096705676e-07,
+      "loss": 0.7657,
+      "step": 5194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.013233527824764e-07,
+      "loss": 0.282,
+      "step": 5195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0047865227230891e-07,
+      "loss": 0.7673,
+      "step": 5196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.963746973673416e-08,
+      "loss": 0.7832,
+      "step": 5197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.879980547468193e-08,
+      "loss": 0.738,
+      "step": 5198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.796565978383187e-08,
+      "loss": 0.7719,
+      "step": 5199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.713503296061133e-08,
+      "loss": 0.7135,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.630792530019972e-08,
+      "loss": 0.809,
+      "step": 5201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.548433709652527e-08,
+      "loss": 0.7379,
+      "step": 5202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.466426864226274e-08,
+      "loss": 0.7278,
+      "step": 5203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.384772022884015e-08,
+      "loss": 0.8372,
+      "step": 5204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.303469214643313e-08,
+      "loss": 0.8133,
+      "step": 5205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.222518468396391e-08,
+      "loss": 0.8002,
+      "step": 5206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.141919812910793e-08,
+      "loss": 0.675,
+      "step": 5207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.06167327682872e-08,
+      "loss": 0.7937,
+      "step": 5208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.981778888667025e-08,
+      "loss": 0.6951,
+      "step": 5209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.902236676817666e-08,
+      "loss": 0.7452,
+      "step": 5210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.823046669547586e-08,
+      "loss": 0.6975,
+      "step": 5211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.744208894998384e-08,
+      "loss": 0.7887,
+      "step": 5212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.665723381186319e-08,
+      "loss": 0.8415,
+      "step": 5213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.587590156002635e-08,
+      "loss": 0.7459,
+      "step": 5214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.509809247213562e-08,
+      "loss": 0.7715,
+      "step": 5215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.432380682459884e-08,
+      "loss": 0.7866,
+      "step": 5216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.355304489257254e-08,
+      "loss": 0.7645,
+      "step": 5217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.278580694996097e-08,
+      "loss": 0.7132,
+      "step": 5218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.202209326941491e-08,
+      "loss": 0.7962,
+      "step": 5219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.126190412233615e-08,
+      "loss": 0.2626,
+      "step": 5220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.05052397788697e-08,
+      "loss": 0.7242,
+      "step": 5221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.97521005079116e-08,
+      "loss": 0.7097,
+      "step": 5222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.900248657710108e-08,
+      "loss": 0.2411,
+      "step": 5223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.825639825282949e-08,
+      "loss": 0.745,
+      "step": 5224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.751383580023252e-08,
+      "loss": 0.8063,
+      "step": 5225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.677479948319244e-08,
+      "loss": 0.8054,
+      "step": 5226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.603928956434025e-08,
+      "loss": 0.7437,
+      "step": 5227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.530730630505023e-08,
+      "loss": 0.811,
+      "step": 5228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.457884996544873e-08,
+      "loss": 0.8139,
+      "step": 5229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.385392080440535e-08,
+      "loss": 0.7685,
+      "step": 5230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.313251907953622e-08,
+      "loss": 0.2653,
+      "step": 5231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.241464504720298e-08,
+      "loss": 0.7069,
+      "step": 5232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.17002989625193e-08,
+      "loss": 0.7213,
+      "step": 5233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.098948107933656e-08,
+      "loss": 0.7559,
+      "step": 5234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.02821916502594e-08,
+      "loss": 0.7759,
+      "step": 5235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.957843092663675e-08,
+      "loss": 0.7665,
+      "step": 5236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.887819915856075e-08,
+      "loss": 0.7689,
+      "step": 5237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.818149659487127e-08,
+      "loss": 0.7704,
+      "step": 5238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.748832348315582e-08,
+      "loss": 0.8225,
+      "step": 5239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.679868006974621e-08,
+      "loss": 0.7681,
+      "step": 5240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.611256659971754e-08,
+      "loss": 0.7792,
+      "step": 5241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.542998331689365e-08,
+      "loss": 0.7156,
+      "step": 5242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.475093046384384e-08,
+      "loss": 0.7535,
+      "step": 5243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.407540828188175e-08,
+      "loss": 0.7829,
+      "step": 5244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.340341701106645e-08,
+      "loss": 0.7517,
+      "step": 5245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.273495689020026e-08,
+      "loss": 0.7482,
+      "step": 5246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.207002815683427e-08,
+      "loss": 0.8138,
+      "step": 5247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.140863104726391e-08,
+      "loss": 0.7496,
+      "step": 5248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.075076579652672e-08,
+      "loss": 0.8174,
+      "step": 5249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.009643263840791e-08,
+      "loss": 0.7241,
+      "step": 5250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.944563180543705e-08,
+      "loss": 0.7289,
+      "step": 5251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8798363528886904e-08,
+      "loss": 0.7846,
+      "step": 5252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8154628038776805e-08,
+      "loss": 0.79,
+      "step": 5253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7514425563870436e-08,
+      "loss": 0.8258,
+      "step": 5254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.687775633167469e-08,
+      "loss": 0.7581,
+      "step": 5255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.624462056844082e-08,
+      "loss": 0.7683,
+      "step": 5256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.561501849916662e-08,
+      "loss": 0.7525,
+      "step": 5257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.498895034759088e-08,
+      "loss": 0.736,
+      "step": 5258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.436641633619899e-08,
+      "loss": 0.8093,
+      "step": 5259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.374741668622063e-08,
+      "loss": 0.7564,
+      "step": 5260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.313195161762763e-08,
+      "loss": 0.2712,
+      "step": 5261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.252002134913725e-08,
+      "loss": 0.7602,
+      "step": 5262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.191162609820888e-08,
+      "loss": 0.7675,
+      "step": 5263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1306766081048456e-08,
+      "loss": 0.787,
+      "step": 5264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.070544151260293e-08,
+      "loss": 0.7894,
+      "step": 5265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0107652606563585e-08,
+      "loss": 0.8139,
+      "step": 5266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.951339957536716e-08,
+      "loss": 0.8051,
+      "step": 5267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.892268263019029e-08,
+      "loss": 0.8497,
+      "step": 5268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8335501980955043e-08,
+      "loss": 0.7887,
+      "step": 5269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7751857836328966e-08,
+      "loss": 0.7628,
+      "step": 5270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7171750403718356e-08,
+      "loss": 0.8057,
+      "step": 5271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659517988927609e-08,
+      "loss": 0.724,
+      "step": 5272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6022146497896045e-08,
+      "loss": 0.8189,
+      "step": 5273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.545265043321645e-08,
+      "loss": 0.7477,
+      "step": 5274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.488669189761874e-08,
+      "loss": 0.689,
+      "step": 5275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.432427109222648e-08,
+      "loss": 0.7627,
+      "step": 5276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.376538821690424e-08,
+      "loss": 0.7601,
+      "step": 5277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.321004347026314e-08,
+      "loss": 0.7585,
+      "step": 5278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2658237049655325e-08,
+      "loss": 0.7514,
+      "step": 5279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.210996915117283e-08,
+      "loss": 0.7981,
+      "step": 5280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1565239969655335e-08,
+      "loss": 0.7555,
+      "step": 5281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.102404969868024e-08,
+      "loss": 0.8011,
+      "step": 5282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0486398530570345e-08,
+      "loss": 0.7769,
+      "step": 5283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9952286656389506e-08,
+      "loss": 0.7787,
+      "step": 5284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.942171426594476e-08,
+      "loss": 0.8163,
+      "step": 5285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.889468154778309e-08,
+      "loss": 0.7544,
+      "step": 5286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.837118868919687e-08,
+      "loss": 0.7583,
+      "step": 5287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.785123587621731e-08,
+      "loss": 0.8083,
+      "step": 5288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7334823293619924e-08,
+      "loss": 0.7968,
+      "step": 5289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6821951124921264e-08,
+      "loss": 0.8338,
+      "step": 5290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.631261955237886e-08,
+      "loss": 0.7621,
+      "step": 5291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.580682875699459e-08,
+      "loss": 0.7968,
+      "step": 5292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5304578918508025e-08,
+      "loss": 0.6973,
+      "step": 5293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.480587021540527e-08,
+      "loss": 0.7787,
+      "step": 5294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4310702824910114e-08,
+      "loss": 0.74,
+      "step": 5295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381907692299069e-08,
+      "loss": 0.8042,
+      "step": 5296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3330992684352805e-08,
+      "loss": 0.7877,
+      "step": 5297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.284645028244771e-08,
+      "loss": 0.8086,
+      "step": 5298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.236544988946544e-08,
+      "loss": 0.7864,
+      "step": 5299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.188799167633927e-08,
+      "loss": 0.7523,
+      "step": 5300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.141407581274236e-08,
+      "loss": 0.7558,
+      "step": 5301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.094370246708889e-08,
+      "loss": 0.8422,
+      "step": 5302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.047687180653402e-08,
+      "loss": 0.7645,
+      "step": 5303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.001358399697618e-08,
+      "loss": 0.7944,
+      "step": 5304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9553839203052547e-08,
+      "loss": 0.7571,
+      "step": 5305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9097637588140216e-08,
+      "loss": 0.7565,
+      "step": 5306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.864497931436061e-08,
+      "loss": 0.7006,
+      "step": 5307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8195864542572837e-08,
+      "loss": 0.7712,
+      "step": 5308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7750293432378117e-08,
+      "loss": 0.7448,
+      "step": 5309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7308266142119788e-08,
+      "loss": 0.7655,
+      "step": 5310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6869782828877754e-08,
+      "loss": 0.8175,
+      "step": 5311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.643484364847737e-08,
+      "loss": 0.7743,
+      "step": 5312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6003448755480553e-08,
+      "loss": 0.8357,
+      "step": 5313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.557559830319245e-08,
+      "loss": 0.7883,
+      "step": 5314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5151292443656993e-08,
+      "loss": 0.7671,
+      "step": 5315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4730531327658014e-08,
+      "loss": 0.7717,
+      "step": 5316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.431331510472257e-08,
+      "loss": 0.768,
+      "step": 5317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.389964392311317e-08,
+      "loss": 0.7421,
+      "step": 5318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.348951792983889e-08,
+      "loss": 0.8342,
+      "step": 5319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3082937270643145e-08,
+      "loss": 0.7703,
+      "step": 5320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2679902090012584e-08,
+      "loss": 0.7229,
+      "step": 5321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2280412531172635e-08,
+      "loss": 0.7539,
+      "step": 5322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1884468736088628e-08,
+      "loss": 0.7814,
+      "step": 5323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1492070845468005e-08,
+      "loss": 0.7476,
+      "step": 5324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1103218998754783e-08,
+      "loss": 0.7714,
+      "step": 5325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0717913334136197e-08,
+      "loss": 0.7849,
+      "step": 5326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0336153988536055e-08,
+      "loss": 0.754,
+      "step": 5327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9957941097620283e-08,
+      "loss": 0.7545,
+      "step": 5328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9583274795793584e-08,
+      "loss": 0.7875,
+      "step": 5329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9212155216200567e-08,
+      "loss": 0.8144,
+      "step": 5330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.884458249072574e-08,
+      "loss": 0.7463,
+      "step": 5331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8480556749991274e-08,
+      "loss": 0.7234,
+      "step": 5332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8120078123360364e-08,
+      "loss": 0.7701,
+      "step": 5333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7763146738938307e-08,
+      "loss": 0.7479,
+      "step": 5334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.740976272356365e-08,
+      "loss": 0.2648,
+      "step": 5335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7059926202820376e-08,
+      "loss": 0.3075,
+      "step": 5336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.671363730102793e-08,
+      "loss": 0.7698,
+      "step": 5337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6370896141246762e-08,
+      "loss": 0.2562,
+      "step": 5338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.603170284527611e-08,
+      "loss": 0.7915,
+      "step": 5339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696057533655108e-08,
+      "loss": 0.804,
+      "step": 5340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5363960325660565e-08,
+      "loss": 0.2622,
+      "step": 5341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.503541133930919e-08,
+      "loss": 0.7184,
+      "step": 5342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4710410691357591e-08,
+      "loss": 0.7421,
+      "step": 5343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4388958497300043e-08,
+      "loss": 0.829,
+      "step": 5344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4071054871369616e-08,
+      "loss": 0.2724,
+      "step": 5345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3756699926541495e-08,
+      "loss": 0.7915,
+      "step": 5346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3445893774525209e-08,
+      "loss": 0.7506,
+      "step": 5347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3138636525772408e-08,
+      "loss": 0.7173,
+      "step": 5348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2834928289472415e-08,
+      "loss": 0.2611,
+      "step": 5349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.253476917355334e-08,
+      "loss": 0.7836,
+      "step": 5350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2238159284680972e-08,
+      "loss": 0.7923,
+      "step": 5351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1945098728263215e-08,
+      "loss": 0.2808,
+      "step": 5352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.165558760844343e-08,
+      "loss": 0.8032,
+      "step": 5353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1369626028104874e-08,
+      "loss": 0.7921,
+      "step": 5354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.108721408886959e-08,
+      "loss": 0.2322,
+      "step": 5355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0808351891096191e-08,
+      "loss": 0.7792,
+      "step": 5356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0533039533884293e-08,
+      "loss": 0.2523,
+      "step": 5357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0261277115071188e-08,
+      "loss": 0.7795,
+      "step": 5358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99306473123407e-09,
+      "loss": 0.7399,
+      "step": 5359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.728402477684472e-09,
+      "loss": 0.747,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.467290448476053e-09,
+      "loss": 0.779,
+      "step": 5361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.209728736400136e-09,
+      "loss": 0.8066,
+      "step": 5362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.955717432985733e-09,
+      "loss": 0.7909,
+      "step": 5363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.705256628499525e-09,
+      "loss": 0.7716,
+      "step": 5364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.458346411949203e-09,
+      "loss": 0.736,
+      "step": 5365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.214986871076803e-09,
+      "loss": 0.2286,
+      "step": 5366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.975178092365365e-09,
+      "loss": 0.7957,
+      "step": 5367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.738920161035613e-09,
+      "loss": 0.7384,
+      "step": 5368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.506213161044829e-09,
+      "loss": 0.7854,
+      "step": 5369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.277057175091307e-09,
+      "loss": 0.7461,
+      "step": 5370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.051452284609905e-09,
+      "loss": 0.7789,
+      "step": 5371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.82939856977094e-09,
+      "loss": 0.7501,
+      "step": 5372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.610896109487952e-09,
+      "loss": 0.7403,
+      "step": 5373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.39594498140883e-09,
+      "loss": 0.8028,
+      "step": 5374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.184545261919139e-09,
+      "loss": 0.8134,
+      "step": 5375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.97669702614545e-09,
+      "loss": 0.6864,
+      "step": 5376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.772400347949791e-09,
+      "loss": 0.7822,
+      "step": 5377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.571655299931866e-09,
+      "loss": 0.7716,
+      "step": 5378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.374461953431276e-09,
+      "loss": 0.7707,
+      "step": 5379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.180820378524187e-09,
+      "loss": 0.8078,
+      "step": 5380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.990730644023334e-09,
+      "loss": 0.2269,
+      "step": 5381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.804192817481345e-09,
+      "loss": 0.7858,
+      "step": 5382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.621206965188529e-09,
+      "loss": 0.7939,
+      "step": 5383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4417731521717576e-09,
+      "loss": 0.7675,
+      "step": 5384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.26589144219447e-09,
+      "loss": 0.7441,
+      "step": 5385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.093561897762222e-09,
+      "loss": 0.2525,
+      "step": 5386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.924784580113805e-09,
+      "loss": 0.8006,
+      "step": 5387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.759559549227909e-09,
+      "loss": 0.755,
+      "step": 5388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.597886863818678e-09,
+      "loss": 0.8324,
+      "step": 5389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.439766581342374e-09,
+      "loss": 0.8505,
+      "step": 5390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2851987579873844e-09,
+      "loss": 0.8084,
+      "step": 5391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1341834486831035e-09,
+      "loss": 0.7652,
+      "step": 5392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9867207070954916e-09,
+      "loss": 0.756,
+      "step": 5393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.842810585627076e-09,
+      "loss": 0.7925,
+      "step": 5394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.70245313542139e-09,
+      "loss": 0.7918,
+      "step": 5395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5656484063552035e-09,
+      "loss": 0.7952,
+      "step": 5396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4323964470451823e-09,
+      "loss": 0.7786,
+      "step": 5397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.302697304843449e-09,
+      "loss": 0.7473,
+      "step": 5398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.176551025843132e-09,
+      "loss": 0.8059,
+      "step": 5399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.053957654871708e-09,
+      "loss": 0.8198,
+      "step": 5400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.934917235494327e-09,
+      "loss": 0.2346,
+      "step": 5401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8194298100149277e-09,
+      "loss": 0.7352,
+      "step": 5402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7074954194729044e-09,
+      "loss": 0.8059,
+      "step": 5403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5991141036475478e-09,
+      "loss": 0.7739,
+      "step": 5404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.494285901053605e-09,
+      "loss": 0.8128,
+      "step": 5405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3930108489446093e-09,
+      "loss": 0.7523,
+      "step": 5406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2952889833084404e-09,
+      "loss": 0.7968,
+      "step": 5407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2011203388739844e-09,
+      "loss": 0.8423,
+      "step": 5408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.110504949104474e-09,
+      "loss": 0.7432,
+      "step": 5409
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0234428462030377e-09,
+      "loss": 0.2331,
+      "step": 5410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.39934061109371e-10,
+      "loss": 0.8073,
+      "step": 5411
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.59978623497515e-10,
+      "loss": 0.7683,
+      "step": 5412
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.835765617825175e-10,
+      "loss": 0.7645,
+      "step": 5413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.107279031148828e-10,
+      "loss": 0.7679,
+      "step": 5414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.414326733827913e-10,
+      "loss": 0.794,
+      "step": 5415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.756908972109898e-10,
+      "loss": 0.7172,
+      "step": 5416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.135025979641217e-10,
+      "loss": 0.2286,
+      "step": 5417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5486779773895595e-10,
+      "loss": 0.7969,
+      "step": 5418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.997865173754889e-10,
+      "loss": 0.6885,
+      "step": 5419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4825877644473207e-10,
+      "loss": 0.7484,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.002845932598142e-10,
+      "loss": 0.7108,
+      "step": 5421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5586398486932004e-10,
+      "loss": 0.7721,
+      "step": 5422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.149969670595109e-10,
+      "loss": 0.7376,
+      "step": 5423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.776835543509936e-10,
+      "loss": 0.7469,
+      "step": 5424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.439237600053822e-10,
+      "loss": 0.7911,
+      "step": 5425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1371759601974674e-10,
+      "loss": 0.7818,
+      "step": 5426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.706507312772339e-11,
+      "loss": 0.7675,
+      "step": 5427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.396620080173499e-11,
+      "loss": 0.7283,
+      "step": 5428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.442098724855015e-11,
+      "loss": 0.7831,
+      "step": 5429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8429439417054783e-11,
+      "loss": 0.7472,
+      "step": 5430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5991562987149877e-11,
+      "loss": 0.7642,
+      "step": 5431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.107362379743521e-12,
+      "loss": 0.7387,
+      "step": 5432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.776840752309994e-12,
+      "loss": 0.3795,
+      "step": 5433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6521,
+      "step": 5434
+    },
+    {
+      "epoch": 1.0,
+      "step": 5434,
+      "total_flos": 2229703926349824.0,
+      "train_loss": 0.7875823799409671,
+      "train_runtime": 68017.7385,
+      "train_samples_per_second": 10.225,
+      "train_steps_per_second": 0.08
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5434,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 2229703926349824.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}