diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16420 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998856280165611,
+  "eval_steps": 500,
+  "global_step": 2732,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.439024390243903e-07,
+      "loss": 1.7746,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 2.0489,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.317073170731707e-07,
+      "loss": 2.0594,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 2.0631,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 1.9292,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 1.6238,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.707317073170732e-06,
+      "loss": 1.5316,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.3907,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1951219512195125e-06,
+      "loss": 1.465,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 1.4066,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.682926829268293e-06,
+      "loss": 1.3394,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 1.3572,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1707317073170736e-06,
+      "loss": 1.3407,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 1.2902,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 1.326,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 1.2057,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.146341463414634e-06,
+      "loss": 1.2199,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 1.1638,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 1.1217,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 1.2086,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 1.1933,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.1808,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.609756097560977e-06,
+      "loss": 1.1663,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 1.2118,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 1.1982,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 1.0557,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.585365853658538e-06,
+      "loss": 1.1564,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 1.1116,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0731707317073175e-06,
+      "loss": 1.1466,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 1.1846,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.560975609756098e-06,
+      "loss": 1.0992,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 1.1256,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.048780487804879e-06,
+      "loss": 1.1361,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 1.111,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 1.1135,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 1.122,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.02439024390244e-06,
+      "loss": 1.0542,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 1.1078,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.51219512195122e-06,
+      "loss": 1.085,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 1.1363,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.078,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.024390243902439e-05,
+      "loss": 1.1724,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0487804878048782e-05,
+      "loss": 1.1019,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 1.0718,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0975609756097562e-05,
+      "loss": 1.1128,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1219512195121953e-05,
+      "loss": 1.1082,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1463414634146342e-05,
+      "loss": 1.0539,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 1.127,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1951219512195123e-05,
+      "loss": 1.0725,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 1.0472,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2439024390243903e-05,
+      "loss": 1.0672,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 1.0711,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2926829268292684e-05,
+      "loss": 1.075,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3170731707317076e-05,
+      "loss": 1.0539,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 1.0416,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 1.0795,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3902439024390244e-05,
+      "loss": 1.0639,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4146341463414635e-05,
+      "loss": 1.0585,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4390243902439025e-05,
+      "loss": 1.104,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 1.1137,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4878048780487806e-05,
+      "loss": 1.1071,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5121951219512196e-05,
+      "loss": 1.1571,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5365853658536586e-05,
+      "loss": 1.117,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 1.0514,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.585365853658537e-05,
+      "loss": 1.1197,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 1.0547,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6341463414634145e-05,
+      "loss": 1.0854,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 1.0879,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.682926829268293e-05,
+      "loss": 1.086,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 1.1575,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7317073170731708e-05,
+      "loss": 1.0845,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 1.1048,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7804878048780488e-05,
+      "loss": 1.062,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.804878048780488e-05,
+      "loss": 1.1517,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.829268292682927e-05,
+      "loss": 1.0576,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 1.1595,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.878048780487805e-05,
+      "loss": 1.0868,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.902439024390244e-05,
+      "loss": 1.1035,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.926829268292683e-05,
+      "loss": 1.092,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 1.091,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.975609756097561e-05,
+      "loss": 1.0893,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.1035,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992972870598e-05,
+      "loss": 1.1806,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997189149227e-05,
+      "loss": 1.0887,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993675589464e-05,
+      "loss": 1.1861,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988756612709e-05,
+      "loss": 1.1046,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999824322258754e-05,
+      "loss": 1.1361,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999747024378516e-05,
+      "loss": 1.1835,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999655672595014e-05,
+      "loss": 1.1122,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999550267036634e-05,
+      "loss": 1.1378,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999430807851516e-05,
+      "loss": 1.1251,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999929729520755e-05,
+      "loss": 1.1413,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999149729292383e-05,
+      "loss": 1.1583,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99989881103134e-05,
+      "loss": 1.1237,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999881243849775e-05,
+      "loss": 1.0559,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998622714092328e-05,
+      "loss": 1.1354,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999841893736377e-05,
+      "loss": 1.1404,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998201108598477e-05,
+      "loss": 1.0735,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999796922810259e-05,
+      "loss": 1.1255,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997723296201997e-05,
+      "loss": 1.1222,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999746331324234e-05,
+      "loss": 1.1448,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997189279589003e-05,
+      "loss": 1.0974,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996901195627122e-05,
+      "loss": 1.144,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996599061761575e-05,
+      "loss": 1.0863,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999628287841699e-05,
+      "loss": 1.1018,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995952646037743e-05,
+      "loss": 1.137,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995608365087945e-05,
+      "loss": 1.0721,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995250036051462e-05,
+      "loss": 1.0962,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994877659431898e-05,
+      "loss": 1.1612,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994491235752595e-05,
+      "loss": 1.0782,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999409076555665e-05,
+      "loss": 1.1189,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993676249406895e-05,
+      "loss": 1.1726,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993247687885894e-05,
+      "loss": 1.0916,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999280508159597e-05,
+      "loss": 1.087,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999234843115916e-05,
+      "loss": 1.1547,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999187773721726e-05,
+      "loss": 1.0905,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991393000431798e-05,
+      "loss": 1.0768,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990894221484027e-05,
+      "loss": 1.1383,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999038140107495e-05,
+      "loss": 1.1178,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989854539925296e-05,
+      "loss": 1.1143,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989313638775532e-05,
+      "loss": 1.1086,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988758698385854e-05,
+      "loss": 1.1666,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988189719536185e-05,
+      "loss": 1.2003,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987606703026187e-05,
+      "loss": 1.0057,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987009649675248e-05,
+      "loss": 1.1121,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986398560322476e-05,
+      "loss": 1.1512,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985773435826716e-05,
+      "loss": 1.1151,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985134277066533e-05,
+      "loss": 1.1012,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984481084940216e-05,
+      "loss": 1.1028,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998381386036578e-05,
+      "loss": 1.0318,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983132604280964e-05,
+      "loss": 1.0931,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 1.1455,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998172800142971e-05,
+      "loss": 1.1108,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981004656637344e-05,
+      "loss": 1.1149,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 1.0777,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979515885402156e-05,
+      "loss": 1.1605,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978750461051698e-05,
+      "loss": 1.0705,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977971012307085e-05,
+      "loss": 1.1395,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997717754026378e-05,
+      "loss": 1.1278,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976370046036947e-05,
+      "loss": 1.1358,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975548530761456e-05,
+      "loss": 1.0846,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974712995591887e-05,
+      "loss": 1.0539,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973863441702526e-05,
+      "loss": 1.127,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972999870287357e-05,
+      "loss": 1.1241,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972122282560066e-05,
+      "loss": 1.1132,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997123067975404e-05,
+      "loss": 1.1259,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970325063122348e-05,
+      "loss": 1.1144,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996940543393778e-05,
+      "loss": 1.1034,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968471793492805e-05,
+      "loss": 1.0919,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967524143099583e-05,
+      "loss": 1.1139,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966562484089964e-05,
+      "loss": 1.1776,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965586817815494e-05,
+      "loss": 1.113,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964597145647397e-05,
+      "loss": 1.0796,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963593468976583e-05,
+      "loss": 1.1055,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962575789213644e-05,
+      "loss": 1.0818,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961544107788855e-05,
+      "loss": 1.1057,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996049842615217e-05,
+      "loss": 1.1297,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959438745773216e-05,
+      "loss": 1.1115,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958365068141293e-05,
+      "loss": 1.1357,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957277394765377e-05,
+      "loss": 1.1151,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995617572717411e-05,
+      "loss": 1.0393,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995506006691581e-05,
+      "loss": 1.0867,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953930415558448e-05,
+      "loss": 1.075,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952786774689667e-05,
+      "loss": 1.0829,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951629145916773e-05,
+      "loss": 1.111,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950457530866726e-05,
+      "loss": 1.1061,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994927193118614e-05,
+      "loss": 1.0951,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948072348541294e-05,
+      "loss": 1.1392,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946858784618107e-05,
+      "loss": 1.1126,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945631241122158e-05,
+      "loss": 1.0881,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944389719778663e-05,
+      "loss": 1.0727,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943134222332493e-05,
+      "loss": 1.135,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941864750548154e-05,
+      "loss": 1.091,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994058130620979e-05,
+      "loss": 1.1087,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99392838911212e-05,
+      "loss": 1.1088,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937972507105793e-05,
+      "loss": 1.1094,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936647156006623e-05,
+      "loss": 1.1184,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993530783968638e-05,
+      "loss": 1.0667,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933954560027366e-05,
+      "loss": 1.0768,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993258731893152e-05,
+      "loss": 1.0949,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931206118320396e-05,
+      "loss": 1.1247,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 1.0883,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928401846336635e-05,
+      "loss": 1.0755,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926978778905193e-05,
+      "loss": 1.0908,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992554175984086e-05,
+      "loss": 1.0664,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992409079116326e-05,
+      "loss": 1.1475,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922625874911624e-05,
+      "loss": 1.1129,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 1.1089,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991965420794116e-05,
+      "loss": 1.1382,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918147461398796e-05,
+      "loss": 1.1025,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99166267756353e-05,
+      "loss": 1.1118,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915092152787888e-05,
+      "loss": 1.1067,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913543595013357e-05,
+      "loss": 1.101,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991198110448809e-05,
+      "loss": 1.1074,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910404683408058e-05,
+      "loss": 1.1041,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908814333988794e-05,
+      "loss": 1.098,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990721005846542e-05,
+      "loss": 1.0989,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990559185909263e-05,
+      "loss": 1.0952,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903959738144682e-05,
+      "loss": 1.0731,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902313697915395e-05,
+      "loss": 1.1407,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900653740718165e-05,
+      "loss": 1.0604,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898979868885933e-05,
+      "loss": 1.1036,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98972920847712e-05,
+      "loss": 1.0769,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989559039074603e-05,
+      "loss": 1.1807,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893874789202022e-05,
+      "loss": 1.1628,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989214528255033e-05,
+      "loss": 1.1359,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890401873221642e-05,
+      "loss": 1.1964,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888644563666194e-05,
+      "loss": 1.0955,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886873356353754e-05,
+      "loss": 1.1113,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885088253773623e-05,
+      "loss": 1.0621,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988328925843463e-05,
+      "loss": 1.0679,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988147637286513e-05,
+      "loss": 1.1269,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879649599612996e-05,
+      "loss": 1.0442,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987780894124563e-05,
+      "loss": 1.1234,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875954400349934e-05,
+      "loss": 1.0584,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987408597953233e-05,
+      "loss": 1.0997,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872203681418745e-05,
+      "loss": 1.1041,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987030750865461e-05,
+      "loss": 1.0757,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868397463904854e-05,
+      "loss": 1.0727,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866473549853904e-05,
+      "loss": 1.1608,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864535769205678e-05,
+      "loss": 1.1298,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862584124683587e-05,
+      "loss": 1.0993,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860618619030516e-05,
+      "loss": 1.1594,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858639255008844e-05,
+      "loss": 1.2043,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856646035400415e-05,
+      "loss": 1.0844,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854638963006552e-05,
+      "loss": 1.0523,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985261804064805e-05,
+      "loss": 1.0942,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850583271165166e-05,
+      "loss": 1.0945,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848534657417617e-05,
+      "loss": 1.0915,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846472202284574e-05,
+      "loss": 1.0651,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844395908664663e-05,
+      "loss": 1.1129,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 1.0754,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984020181765601e-05,
+      "loss": 1.0845,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838084026161746e-05,
+      "loss": 1.0834,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983595240796958e-05,
+      "loss": 1.1465,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833806966075343e-05,
+      "loss": 1.1841,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831647703494287e-05,
+      "loss": 1.045,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829474623261106e-05,
+      "loss": 1.0345,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827287728429897e-05,
+      "loss": 1.0949,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825087022074182e-05,
+      "loss": 1.0919,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982287250728689e-05,
+      "loss": 1.0556,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820644187180354e-05,
+      "loss": 1.1047,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818402064886314e-05,
+      "loss": 1.0754,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981614614355591e-05,
+      "loss": 1.0625,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813876426359666e-05,
+      "loss": 1.1032,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981159291648751e-05,
+      "loss": 1.1068,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980929561714874e-05,
+      "loss": 1.074,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806984531572038e-05,
+      "loss": 1.0986,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980465966300547e-05,
+      "loss": 1.0771,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802321014716465e-05,
+      "loss": 1.0703,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979996858999181e-05,
+      "loss": 1.1366,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797602392137678e-05,
+      "loss": 1.0638,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795222424479577e-05,
+      "loss": 1.0704,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792828690362377e-05,
+      "loss": 1.0903,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979042119315029e-05,
+      "loss": 1.0466,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787999936226877e-05,
+      "loss": 1.0823,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785564922995042e-05,
+      "loss": 1.1806,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783116156877008e-05,
+      "loss": 1.0942,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780653641314333e-05,
+      "loss": 1.0961,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778177379767903e-05,
+      "loss": 1.0888,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775687375717926e-05,
+      "loss": 1.0907,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773183632663907e-05,
+      "loss": 1.1134,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977066615412468e-05,
+      "loss": 1.0591,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768134943638367e-05,
+      "loss": 1.0788,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765590004762404e-05,
+      "loss": 1.0771,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763031341073512e-05,
+      "loss": 1.0677,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760458956167698e-05,
+      "loss": 1.0801,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757872853660265e-05,
+      "loss": 1.0685,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755273037185786e-05,
+      "loss": 1.1199,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975265951039811e-05,
+      "loss": 1.0869,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975003227697035e-05,
+      "loss": 1.0914,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97473913405949e-05,
+      "loss": 1.1234,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744736704983394e-05,
+      "loss": 1.0788,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974206837386672e-05,
+      "loss": 1.0986,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973938635099503e-05,
+      "loss": 1.1361,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736690640137696e-05,
+      "loss": 1.0872,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733981245083355e-05,
+      "loss": 1.0841,
+      "step": 277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731258169639846e-05,
+      "loss": 1.0824,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972852141763426e-05,
+      "loss": 1.0701,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725770992912893e-05,
+      "loss": 1.0872,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723006899341267e-05,
+      "loss": 1.0991,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 1.0657,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717437721205352e-05,
+      "loss": 1.0716,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714632644468135e-05,
+      "loss": 1.2292,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711813914534777e-05,
+      "loss": 1.1859,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708981535366797e-05,
+      "loss": 1.0637,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706135510944894e-05,
+      "loss": 1.0818,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970327584526895e-05,
+      "loss": 1.0991,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700402542358006e-05,
+      "loss": 1.0722,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697515606250276e-05,
+      "loss": 1.1184,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969461504100314e-05,
+      "loss": 1.0937,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691700850693126e-05,
+      "loss": 1.06,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688773039415905e-05,
+      "loss": 1.0525,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 1.0936,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682876570438298e-05,
+      "loss": 1.0699,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967990792102495e-05,
+      "loss": 1.1616,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676925667218498e-05,
+      "loss": 1.1211,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673929813210265e-05,
+      "loss": 1.0559,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967092036321071e-05,
+      "loss": 1.0886,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667897321449387e-05,
+      "loss": 1.1195,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664860692174958e-05,
+      "loss": 1.1285,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661810479655184e-05,
+      "loss": 1.1114,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965874668817691e-05,
+      "loss": 1.074,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655669322046068e-05,
+      "loss": 1.0716,
+      "step": 304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965257838558767e-05,
+      "loss": 1.1114,
+      "step": 305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649473883145792e-05,
+      "loss": 1.0893,
+      "step": 306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964635581908359e-05,
+      "loss": 1.1287,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643224197783265e-05,
+      "loss": 1.045,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640079023646085e-05,
+      "loss": 1.0937,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636920301092352e-05,
+      "loss": 1.088,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633748034561424e-05,
+      "loss": 1.0436,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630562228511682e-05,
+      "loss": 1.2809,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962736288742054e-05,
+      "loss": 1.1033,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962415001578444e-05,
+      "loss": 1.1039,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620923618118827e-05,
+      "loss": 1.0263,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617683698958168e-05,
+      "loss": 1.0905,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961443026285593e-05,
+      "loss": 1.0768,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611163314384574e-05,
+      "loss": 1.1159,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607882858135557e-05,
+      "loss": 1.1334,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604588898719314e-05,
+      "loss": 1.0357,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601281440765262e-05,
+      "loss": 1.0664,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597960488921785e-05,
+      "loss": 1.0784,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959462604785624e-05,
+      "loss": 1.0719,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591278122254938e-05,
+      "loss": 1.0818,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587916716823132e-05,
+      "loss": 1.0577,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584541836285035e-05,
+      "loss": 1.0794,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958115348538379e-05,
+      "loss": 1.1483,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957775166888147e-05,
+      "loss": 1.094,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574336391559084e-05,
+      "loss": 1.0709,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957090765821654e-05,
+      "loss": 1.098,
+      "step": 330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567465473672675e-05,
+      "loss": 1.0484,
+      "step": 331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 1.1144,
+      "step": 332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560540770350824e-05,
+      "loss": 1.0909,
+      "step": 333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955705826130499e-05,
+      "loss": 1.2324,
+      "step": 334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553562320522134e-05,
+      "loss": 1.0714,
+      "step": 335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550052952915545e-05,
+      "loss": 1.0964,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546530163417374e-05,
+      "loss": 1.1602,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542993956978647e-05,
+      "loss": 1.0811,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539444338569232e-05,
+      "loss": 1.0551,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535881313177864e-05,
+      "loss": 1.0471,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95323048858121e-05,
+      "loss": 1.0793,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528715061498355e-05,
+      "loss": 1.0652,
+      "step": 342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525111845281853e-05,
+      "loss": 1.0843,
+      "step": 343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521495242226648e-05,
+      "loss": 1.0436,
+      "step": 344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517865257415612e-05,
+      "loss": 1.0763,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514221895950416e-05,
+      "loss": 1.0836,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 1.0836,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950689506355824e-05,
+      "loss": 1.0486,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950321160292858e-05,
+      "loss": 1.0877,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949951478623938e-05,
+      "loss": 0.9892,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495804618686253e-05,
+      "loss": 1.0638,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949208110548356e-05,
+      "loss": 1.0813,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948834425186442e-05,
+      "loss": 1.0837,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948459406308071e-05,
+      "loss": 1.0649,
+      "step": 354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480830544403034e-05,
+      "loss": 1.0832,
+      "step": 355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477053701120746e-05,
+      "loss": 1.032,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473263538541916e-05,
+      "loss": 1.1723,
+      "step": 357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469460061993336e-05,
+      "loss": 1.0978,
+      "step": 358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946564327682051e-05,
+      "loss": 1.0485,
+      "step": 359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461813188387652e-05,
+      "loss": 1.0532,
+      "step": 360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945796980207766e-05,
+      "loss": 1.0788,
+      "step": 361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454113123292133e-05,
+      "loss": 1.0696,
+      "step": 362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450243157451346e-05,
+      "loss": 1.0709,
+      "step": 363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446359909994253e-05,
+      "loss": 1.0719,
+      "step": 364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442463386378464e-05,
+      "loss": 1.0363,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438553592080257e-05,
+      "loss": 1.0404,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434630532594555e-05,
+      "loss": 1.0741,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430694213434936e-05,
+      "loss": 1.0591,
+      "step": 368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426744640133598e-05,
+      "loss": 1.0843,
+      "step": 369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942278181824137e-05,
+      "loss": 1.1183,
+      "step": 370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418805753327713e-05,
+      "loss": 1.0583,
+      "step": 371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414816450980686e-05,
+      "loss": 1.0901,
+      "step": 372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941081391680696e-05,
+      "loss": 1.1072,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94067981564318e-05,
+      "loss": 1.0693,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402769175499062e-05,
+      "loss": 1.0742,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398726979671174e-05,
+      "loss": 1.0643,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394671574629147e-05,
+      "loss": 1.0685,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390602966072548e-05,
+      "loss": 1.0312,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386521159719513e-05,
+      "loss": 1.1454,
+      "step": 379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382426161306712e-05,
+      "loss": 1.0682,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937831797658936e-05,
+      "loss": 1.0508,
+      "step": 381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 1.071,
+      "step": 382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370062071354535e-05,
+      "loss": 1.076,
+      "step": 383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365914362440125e-05,
+      "loss": 1.1212,
+      "step": 384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361753490427272e-05,
+      "loss": 1.0692,
+      "step": 385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357579461163783e-05,
+      "loss": 1.0825,
+      "step": 386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353392280515938e-05,
+      "loss": 1.0677,
+      "step": 387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349191954368515e-05,
+      "loss": 1.0091,
+      "step": 388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344978488624754e-05,
+      "loss": 1.0553,
+      "step": 389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340751889206378e-05,
+      "loss": 1.0608,
+      "step": 390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933651216205355e-05,
+      "loss": 1.109,
+      "step": 391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93322593131249e-05,
+      "loss": 1.0399,
+      "step": 392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327993348397488e-05,
+      "loss": 1.0695,
+      "step": 393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932371427386681e-05,
+      "loss": 1.0441,
+      "step": 394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319422095546794e-05,
+      "loss": 1.1254,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931511681946977e-05,
+      "loss": 1.0399,
+      "step": 396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310798451686488e-05,
+      "loss": 1.0622,
+      "step": 397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306466998266102e-05,
+      "loss": 1.0886,
+      "step": 398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302122465296136e-05,
+      "loss": 1.0984,
+      "step": 399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 1.044,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929339418514953e-05,
+      "loss": 1.0402,
+      "step": 401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289010450239843e-05,
+      "loss": 1.0528,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284613660314465e-05,
+      "loss": 1.015,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928020382155276e-05,
+      "loss": 1.0338,
+      "step": 404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927578094015243e-05,
+      "loss": 1.0585,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271345022329502e-05,
+      "loss": 1.0602,
+      "step": 406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266896074318335e-05,
+      "loss": 1.1087,
+      "step": 407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262434102371596e-05,
+      "loss": 1.1035,
+      "step": 408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925795911276025e-05,
+      "loss": 1.0685,
+      "step": 409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253471111773572e-05,
+      "loss": 1.0824,
+      "step": 410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248970105719107e-05,
+      "loss": 1.1054,
+      "step": 411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924445610092269e-05,
+      "loss": 1.0684,
+      "step": 412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239929103728415e-05,
+      "loss": 1.0886,
+      "step": 413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235389120498645e-05,
+      "loss": 1.0063,
+      "step": 414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923083615761399e-05,
+      "loss": 1.1,
+      "step": 415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226270221473302e-05,
+      "loss": 1.0316,
+      "step": 416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221691318493666e-05,
+      "loss": 1.0804,
+      "step": 417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921709945511039e-05,
+      "loss": 1.0397,
+      "step": 418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212494637776995e-05,
+      "loss": 1.0148,
+      "step": 419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207876872965217e-05,
+      "loss": 1.0941,
+      "step": 420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203246167164976e-05,
+      "loss": 1.0496,
+      "step": 421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198602526884388e-05,
+      "loss": 1.0547,
+      "step": 422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193945958649744e-05,
+      "loss": 1.1048,
+      "step": 423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189276469005508e-05,
+      "loss": 1.0829,
+      "step": 424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184594064514298e-05,
+      "loss": 1.1032,
+      "step": 425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917989875175689e-05,
+      "loss": 1.0266,
+      "step": 426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175190537332198e-05,
+      "loss": 1.0815,
+      "step": 427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170469427857264e-05,
+      "loss": 1.0721,
+      "step": 428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165735429967264e-05,
+      "loss": 0.9926,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160988550315475e-05,
+      "loss": 1.071,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156228795573288e-05,
+      "loss": 1.0918,
+      "step": 431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 1.0959,
+      "step": 432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146670687593735e-05,
+      "loss": 1.1671,
+      "step": 433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914187234778958e-05,
+      "loss": 1.0147,
+      "step": 434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137061159761436e-05,
+      "loss": 1.0637,
+      "step": 435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913223713027106e-05,
+      "loss": 1.0457,
+      "step": 436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912740026609828e-05,
+      "loss": 1.0312,
+      "step": 437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122550574040942e-05,
+      "loss": 1.0897,
+      "step": 438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117688060914936e-05,
+      "loss": 1.0635,
+      "step": 439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112812733554155e-05,
+      "loss": 1.02,
+      "step": 440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107924598810517e-05,
+      "loss": 1.009,
+      "step": 441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910302366355393e-05,
+      "loss": 1.0835,
+      "step": 442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098109934672303e-05,
+      "loss": 1.0734,
+      "step": 443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90931834190715e-05,
+      "loss": 1.086,
+      "step": 444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088244123675394e-05,
+      "loss": 1.0353,
+      "step": 445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083292055425783e-05,
+      "loss": 1.0701,
+      "step": 446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907832722128244e-05,
+      "loss": 1.0489,
+      "step": 447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907334962822307e-05,
+      "loss": 1.062,
+      "step": 448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906835928324331e-05,
+      "loss": 0.9948,
+      "step": 449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906335619335672e-05,
+      "loss": 1.2917,
+      "step": 450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058340365594775e-05,
+      "loss": 1.0475,
+      "step": 451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053311807006845e-05,
+      "loss": 1.1154,
+      "step": 452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048270524660197e-05,
+      "loss": 1.0702,
+      "step": 453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904321652563998e-05,
+      "loss": 1.0343,
+      "step": 454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038149817049218e-05,
+      "loss": 1.0211,
+      "step": 455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903307040600879e-05,
+      "loss": 1.1,
+      "step": 456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027978299657436e-05,
+      "loss": 1.0829,
+      "step": 457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902287350515173e-05,
+      "loss": 1.1057,
+      "step": 458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017756029666085e-05,
+      "loss": 1.0291,
+      "step": 459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012625880392733e-05,
+      "loss": 1.0588,
+      "step": 460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900748306454172e-05,
+      "loss": 1.0697,
+      "step": 461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900232758934089e-05,
+      "loss": 1.0544,
+      "step": 462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899715946203588e-05,
+      "loss": 1.0155,
+      "step": 463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899197868989011e-05,
+      "loss": 0.9948,
+      "step": 464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898678528018478e-05,
+      "loss": 1.0513,
+      "step": 465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898157924021883e-05,
+      "loss": 1.0912,
+      "step": 466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897636057730897e-05,
+      "loss": 1.0651,
+      "step": 467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971129298789644e-05,
+      "loss": 1.078,
+      "step": 468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965885412013024e-05,
+      "loss": 1.0311,
+      "step": 469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960628924349006e-05,
+      "loss": 1.0709,
+      "step": 470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895535984318519e-05,
+      "loss": 1.0432,
+      "step": 471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950078175926886e-05,
+      "loss": 1.0588,
+      "step": 472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894478392999708e-05,
+      "loss": 1.0302,
+      "step": 473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939477112836445e-05,
+      "loss": 1.1202,
+      "step": 474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934157731903316e-05,
+      "loss": 1.0543,
+      "step": 475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892882579467369e-05,
+      "loss": 1.0584,
+      "step": 476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923481308641216e-05,
+      "loss": 1.0941,
+      "step": 477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918124281317162e-05,
+      "loss": 1.049,
+      "step": 478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912754720230442e-05,
+      "loss": 1.0494,
+      "step": 479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907372632927573e-05,
+      "loss": 1.1006,
+      "step": 480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901978026972676e-05,
+      "loss": 1.121,
+      "step": 481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 1.0894,
+      "step": 482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889115128945127e-05,
+      "loss": 1.0688,
+      "step": 483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885719173100937e-05,
+      "loss": 1.0146,
+      "step": 484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888027456853091e-05,
+      "loss": 1.0018,
+      "step": 485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887481748339318e-05,
+      "loss": 1.0387,
+      "step": 486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869347925357275e-05,
+      "loss": 1.0728,
+      "step": 487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863865902110253e-05,
+      "loss": 1.0484,
+      "step": 488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885837142135669e-05,
+      "loss": 1.0197,
+      "step": 489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852864490818678e-05,
+      "loss": 1.0909,
+      "step": 490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847345118235792e-05,
+      "loss": 1.0124,
+      "step": 491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841813311365105e-05,
+      "loss": 1.0325,
+      "step": 492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883626907798116e-05,
+      "loss": 1.0421,
+      "step": 493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830712425875964e-05,
+      "loss": 1.0368,
+      "step": 494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825143362858985e-05,
+      "loss": 1.1029,
+      "step": 495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819561896757124e-05,
+      "loss": 1.0175,
+      "step": 496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881396803541472e-05,
+      "loss": 1.0461,
+      "step": 497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808361786693533e-05,
+      "loss": 1.0403,
+      "step": 498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802743158472722e-05,
+      "loss": 1.0408,
+      "step": 499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879711215864886e-05,
+      "loss": 1.062,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791468795135896e-05,
+      "loss": 1.0944,
+      "step": 501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785813075865164e-05,
+      "loss": 1.0251,
+      "step": 502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878014500878535e-05,
+      "loss": 0.9976,
+      "step": 503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877446460186251e-05,
+      "loss": 1.0358,
+      "step": 504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768771863080033e-05,
+      "loss": 1.0562,
+      "step": 505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 1.0331,
+      "step": 506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875734942195637e-05,
+      "loss": 1.0223,
+      "step": 507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875161973566858e-05,
+      "loss": 1.0443,
+      "step": 508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874587774962792e-05,
+      "loss": 1.0179,
+      "step": 509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874012347190432e-05,
+      "loss": 1.0294,
+      "step": 510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734356910585e-05,
+      "loss": 1.0524,
+      "step": 511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728578073774427e-05,
+      "loss": 1.0332,
+      "step": 512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722786969594327e-05,
+      "loss": 1.0279,
+      "step": 513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716983606183673e-05,
+      "loss": 1.061,
+      "step": 514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871116799169866e-05,
+      "loss": 1.0505,
+      "step": 515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87053401343127e-05,
+      "loss": 1.0526,
+      "step": 516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699500042216423e-05,
+      "loss": 1.0568,
+      "step": 517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693647723617637e-05,
+      "loss": 1.0569,
+      "step": 518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687783186741347e-05,
+      "loss": 1.1177,
+      "step": 519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681906439829716e-05,
+      "loss": 1.0599,
+      "step": 520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676017491142092e-05,
+      "loss": 1.1051,
+      "step": 521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670116348954945e-05,
+      "loss": 0.9944,
+      "step": 522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866420302156189e-05,
+      "loss": 1.133,
+      "step": 523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865827751727368e-05,
+      "loss": 1.0136,
+      "step": 524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652339844418164e-05,
+      "loss": 1.0558,
+      "step": 525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864639001134031e-05,
+      "loss": 1.0485,
+      "step": 526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640428026402158e-05,
+      "loss": 1.036,
+      "step": 527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863445389798284e-05,
+      "loss": 1.0307,
+      "step": 528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862846763447855e-05,
+      "loss": 1.0815,
+      "step": 529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622469244302542e-05,
+      "loss": 1.0105,
+      "step": 530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86164587358851e-05,
+      "loss": 1.143,
+      "step": 531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 1.0273,
+      "step": 532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604401398132252e-05,
+      "loss": 1.0711,
+      "step": 533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598354585742537e-05,
+      "loss": 1.1484,
+      "step": 534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592295689002753e-05,
+      "loss": 1.0355,
+      "step": 535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858622471642824e-05,
+      "loss": 1.0198,
+      "step": 536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580141676551298e-05,
+      "loss": 1.0763,
+      "step": 537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574046577921182e-05,
+      "loss": 1.0611,
+      "step": 538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567939429104108e-05,
+      "loss": 1.1487,
+      "step": 539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561820238683216e-05,
+      "loss": 1.0554,
+      "step": 540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855568901525858e-05,
+      "loss": 1.0378,
+      "step": 541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549545767447174e-05,
+      "loss": 1.0866,
+      "step": 542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543390503882884e-05,
+      "loss": 1.0869,
+      "step": 543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853722323321647e-05,
+      "loss": 1.0257,
+      "step": 544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853104396411558e-05,
+      "loss": 1.0742,
+      "step": 545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524852705264716e-05,
+      "loss": 0.9967,
+      "step": 546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851864946536523e-05,
+      "loss": 1.0403,
+      "step": 547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512434253135324e-05,
+      "loss": 1.0857,
+      "step": 548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850620707731001e-05,
+      "loss": 1.0929,
+      "step": 549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499967946641127e-05,
+      "loss": 1.0334,
+      "step": 550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493716869897307e-05,
+      "loss": 1.0594,
+      "step": 551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848745385586398e-05,
+      "loss": 1.0311,
+      "step": 552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481178913343343e-05,
+      "loss": 1.0786,
+      "step": 553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474892051154366e-05,
+      "loss": 1.1368,
+      "step": 554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468593278132767e-05,
+      "loss": 1.0907,
+      "step": 555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462282603131005e-05,
+      "loss": 1.1033,
+      "step": 556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845596003501826e-05,
+      "loss": 1.011,
+      "step": 557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449625582680445e-05,
+      "loss": 1.095,
+      "step": 558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443279255020153e-05,
+      "loss": 1.0605,
+      "step": 559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843692106095668e-05,
+      "loss": 1.0403,
+      "step": 560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430551009425998e-05,
+      "loss": 1.0146,
+      "step": 561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842416910938074e-05,
+      "loss": 1.001,
+      "step": 562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417775369790197e-05,
+      "loss": 1.0511,
+      "step": 563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841136979964029e-05,
+      "loss": 1.0578,
+      "step": 564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840495240793358e-05,
+      "loss": 1.0574,
+      "step": 565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398523203689235e-05,
+      "loss": 1.0668,
+      "step": 566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392082195943017e-05,
+      "loss": 1.0331,
+      "step": 567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385629393747292e-05,
+      "loss": 1.071,
+      "step": 568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379164806170994e-05,
+      "loss": 1.0501,
+      "step": 569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837268844229962e-05,
+      "loss": 1.0135,
+      "step": 570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366200311235223e-05,
+      "loss": 1.0887,
+      "step": 571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359700422096385e-05,
+      "loss": 1.1284,
+      "step": 572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353188784018225e-05,
+      "loss": 1.0454,
+      "step": 573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346665406152362e-05,
+      "loss": 1.0563,
+      "step": 574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834013029766692e-05,
+      "loss": 1.1266,
+      "step": 575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333583467746515e-05,
+      "loss": 1.0134,
+      "step": 576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327024925592226e-05,
+      "loss": 1.0635,
+      "step": 577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83204546804216e-05,
+      "loss": 1.0461,
+      "step": 578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831387274146863e-05,
+      "loss": 1.0148,
+      "step": 579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307279117983744e-05,
+      "loss": 1.0341,
+      "step": 580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300673819233785e-05,
+      "loss": 1.0569,
+      "step": 581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 1.0386,
+      "step": 582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287428233088096e-05,
+      "loss": 1.078,
+      "step": 583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828078796430805e-05,
+      "loss": 1.0049,
+      "step": 584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274136057494293e-05,
+      "loss": 1.0436,
+      "step": 585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826747252199558e-05,
+      "loss": 1.0587,
+      "step": 586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826079736717702e-05,
+      "loss": 1.0616,
+      "step": 587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254110602420047e-05,
+      "loss": 1.0495,
+      "step": 588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247412237122412e-05,
+      "loss": 0.9778,
+      "step": 589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240702280698176e-05,
+      "loss": 0.9961,
+      "step": 590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823398074257768e-05,
+      "loss": 1.0202,
+      "step": 591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822724763220755e-05,
+      "loss": 0.9713,
+      "step": 592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822050295905067e-05,
+      "loss": 1.0563,
+      "step": 593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213746732586186e-05,
+      "loss": 1.0474,
+      "step": 594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206978962309466e-05,
+      "loss": 0.9904,
+      "step": 595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200199657732115e-05,
+      "loss": 1.0171,
+      "step": 596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819340882838194e-05,
+      "loss": 1.0812,
+      "step": 597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186606483802945e-05,
+      "loss": 1.0428,
+      "step": 598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179792633555328e-05,
+      "loss": 1.0455,
+      "step": 599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817296728721545e-05,
+      "loss": 1.0515,
+      "step": 600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816613045437582e-05,
+      "loss": 1.0566,
+      "step": 601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815928214464511e-05,
+      "loss": 1.0751,
+      "step": 602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815242236764811e-05,
+      "loss": 1.1078,
+      "step": 603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814555113302573e-05,
+      "loss": 1.0358,
+      "step": 604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813866845043497e-05,
+      "loss": 1.0393,
+      "step": 605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813177432954894e-05,
+      "loss": 1.056,
+      "step": 606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124868780056814e-05,
+      "loss": 1.1155,
+      "step": 607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811795181166383e-05,
+      "loss": 1.0431,
+      "step": 608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111023434091276e-05,
+      "loss": 1.0637,
+      "step": 609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104083657076466e-05,
+      "loss": 1.0943,
+      "step": 610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809713249037275e-05,
+      "loss": 1.1201,
+      "step": 611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.0767,
+      "step": 612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083196026991985e-05,
+      "loss": 1.0333,
+      "step": 613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80762107499016e-05,
+      "loss": 0.9673,
+      "step": 614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069214122295614e-05,
+      "loss": 1.0413,
+      "step": 615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062206154007267e-05,
+      "loss": 0.9789,
+      "step": 616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055186854885733e-05,
+      "loss": 1.048,
+      "step": 617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048156234796124e-05,
+      "loss": 1.0082,
+      "step": 618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041114303619455e-05,
+      "loss": 1.0595,
+      "step": 619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034061071252632e-05,
+      "loss": 1.0956,
+      "step": 620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802699654760846e-05,
+      "loss": 1.0602,
+      "step": 621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019920742615596e-05,
+      "loss": 1.0781,
+      "step": 622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801283366621856e-05,
+      "loss": 1.0777,
+      "step": 623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005735328377718e-05,
+      "loss": 1.0284,
+      "step": 624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799862573906925e-05,
+      "loss": 1.0573,
+      "step": 625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991504908285162e-05,
+      "loss": 1.0673,
+      "step": 626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984372846033252e-05,
+      "loss": 1.0566,
+      "step": 627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977229562337104e-05,
+      "loss": 1.0054,
+      "step": 628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797007506723608e-05,
+      "loss": 0.9886,
+      "step": 629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962909370785283e-05,
+      "loss": 1.019,
+      "step": 630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955732483055574e-05,
+      "loss": 1.0275,
+      "step": 631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 1.0344,
+      "step": 632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941345174121466e-05,
+      "loss": 1.0439,
+      "step": 633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934134773137364e-05,
+      "loss": 1.0582,
+      "step": 634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792691322131491e-05,
+      "loss": 1.0488,
+      "step": 635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919680528803468e-05,
+      "loss": 1.0428,
+      "step": 636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912436705768045e-05,
+      "loss": 1.0792,
+      "step": 637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905181762389298e-05,
+      "loss": 1.0341,
+      "step": 638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897915708863508e-05,
+      "loss": 1.028,
+      "step": 639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890638555402585e-05,
+      "loss": 1.0259,
+      "step": 640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883350312234022e-05,
+      "loss": 1.0765,
+      "step": 641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7876050989600908e-05,
+      "loss": 0.9725,
+      "step": 642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868740597761895e-05,
+      "loss": 1.0508,
+      "step": 643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861419146991204e-05,
+      "loss": 1.0197,
+      "step": 644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785408664757858e-05,
+      "loss": 1.022,
+      "step": 645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846743109829318e-05,
+      "loss": 1.0436,
+      "step": 646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839388544064215e-05,
+      "loss": 1.0065,
+      "step": 647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832022960619562e-05,
+      "loss": 1.0495,
+      "step": 648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824646369847146e-05,
+      "loss": 1.0087,
+      "step": 649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817258782114216e-05,
+      "loss": 1.0176,
+      "step": 650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780986020780348e-05,
+      "loss": 0.9969,
+      "step": 651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802450657313086e-05,
+      "loss": 0.9877,
+      "step": 652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795030141056607e-05,
+      "loss": 1.0405,
+      "step": 653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787598669463027e-05,
+      "loss": 1.0128,
+      "step": 654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780156252976736e-05,
+      "loss": 1.0301,
+      "step": 655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777270290205749e-05,
+      "loss": 1.0643,
+      "step": 656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765238627180424e-05,
+      "loss": 0.9858,
+      "step": 657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757763438836027e-05,
+      "loss": 1.0141,
+      "step": 658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750277347530116e-05,
+      "loss": 1.1415,
+      "step": 659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742780363783843e-05,
+      "loss": 1.0287,
+      "step": 660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773527249813366e-05,
+      "loss": 0.9818,
+      "step": 661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727753761131312e-05,
+      "loss": 1.0814,
+      "step": 662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720224163343835e-05,
+      "loss": 1.1168,
+      "step": 663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712683715353514e-05,
+      "loss": 1.0176,
+      "step": 664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705132427757895e-05,
+      "loss": 1.0753,
+      "step": 665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697570311169746e-05,
+      "loss": 1.0486,
+      "step": 666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768999737621707e-05,
+      "loss": 1.0559,
+      "step": 667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682413633543057e-05,
+      "loss": 1.0529,
+      "step": 668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76748190938061e-05,
+      "loss": 1.0479,
+      "step": 669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766721376767976e-05,
+      "loss": 0.9924,
+      "step": 670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659597665852767e-05,
+      "loss": 0.9629,
+      "step": 671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651970799028976e-05,
+      "loss": 1.0094,
+      "step": 672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764433317792739e-05,
+      "loss": 1.0511,
+      "step": 673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636684813282113e-05,
+      "loss": 1.0438,
+      "step": 674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629025715842362e-05,
+      "loss": 1.0277,
+      "step": 675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621355896372424e-05,
+      "loss": 1.0981,
+      "step": 676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761367536565167e-05,
+      "loss": 1.0646,
+      "step": 677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760598413447451e-05,
+      "loss": 1.0313,
+      "step": 678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598282213650405e-05,
+      "loss": 1.0543,
+      "step": 679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590569614003825e-05,
+      "loss": 1.0378,
+      "step": 680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582846346374266e-05,
+      "loss": 1.0337,
+      "step": 681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 1.0161,
+      "step": 682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567367850599095e-05,
+      "loss": 1.024,
+      "step": 683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559612644207364e-05,
+      "loss": 1.0226,
+      "step": 684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551846813340375e-05,
+      "loss": 1.1084,
+      "step": 685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544070368912435e-05,
+      "loss": 1.062,
+      "step": 686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753628332185275e-05,
+      "loss": 1.0685,
+      "step": 687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528485683105444e-05,
+      "loss": 1.0114,
+      "step": 688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520677463629518e-05,
+      "loss": 1.0061,
+      "step": 689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751285867439885e-05,
+      "loss": 1.0389,
+      "step": 690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505029326402163e-05,
+      "loss": 1.0467,
+      "step": 691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497189430643025e-05,
+      "loss": 1.0289,
+      "step": 692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489338998139833e-05,
+      "loss": 1.0735,
+      "step": 693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481478039925784e-05,
+      "loss": 1.0137,
+      "step": 694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473606567048874e-05,
+      "loss": 0.9736,
+      "step": 695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746572459057188e-05,
+      "loss": 1.0534,
+      "step": 696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457832121572323e-05,
+      "loss": 1.0426,
+      "step": 697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449929171142495e-05,
+      "loss": 1.0576,
+      "step": 698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442015750389403e-05,
+      "loss": 1.0076,
+      "step": 699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434091870434772e-05,
+      "loss": 1.0139,
+      "step": 700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742615754241503e-05,
+      "loss": 1.0228,
+      "step": 701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741821277748128e-05,
+      "loss": 1.0176,
+      "step": 702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410257586799314e-05,
+      "loss": 0.9923,
+      "step": 703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740229198154955e-05,
+      "loss": 1.0641,
+      "step": 704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394315972927063e-05,
+      "loss": 1.1274,
+      "step": 705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738632957214154e-05,
+      "loss": 0.9663,
+      "step": 706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378332790417275e-05,
+      "loss": 1.0839,
+      "step": 707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737032563899315e-05,
+      "loss": 1.0537,
+      "step": 708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362308129122627e-05,
+      "loss": 1.0296,
+      "step": 709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354280272073718e-05,
+      "loss": 1.0364,
+      "step": 710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346242079128986e-05,
+      "loss": 1.0211,
+      "step": 711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338193561585507e-05,
+      "loss": 1.0086,
+      "step": 712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330134730754885e-05,
+      "loss": 1.0476,
+      "step": 713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322065597963206e-05,
+      "loss": 1.0103,
+      "step": 714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313986174551036e-05,
+      "loss": 1.1996,
+      "step": 715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730589647187341e-05,
+      "loss": 1.071,
+      "step": 716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72977965012998e-05,
+      "loss": 1.0252,
+      "step": 717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 1.058,
+      "step": 718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281565802014683e-05,
+      "loss": 1.0712,
+      "step": 719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273435096114223e-05,
+      "loss": 1.0862,
+      "step": 720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265294167939843e-05,
+      "loss": 1.0263,
+      "step": 721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257143028933004e-05,
+      "loss": 1.0271,
+      "step": 722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7248981690549538e-05,
+      "loss": 1.1215,
+      "step": 723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7240810164259597e-05,
+      "loss": 1.0286,
+      "step": 724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232628461547655e-05,
+      "loss": 1.0791,
+      "step": 725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722443659391249e-05,
+      "loss": 1.061,
+      "step": 726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216234572867165e-05,
+      "loss": 0.9961,
+      "step": 727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208022409939012e-05,
+      "loss": 1.0797,
+      "step": 728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199800116669618e-05,
+      "loss": 1.0311,
+      "step": 729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191567704614806e-05,
+      "loss": 1.0269,
+      "step": 730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718332518534462e-05,
+      "loss": 1.0522,
+      "step": 731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 1.0133,
+      "step": 732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716680987150932e-05,
+      "loss": 1.0862,
+      "step": 733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158537100155256e-05,
+      "loss": 0.9933,
+      "step": 734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150254268007884e-05,
+      "loss": 1.0469,
+      "step": 735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714196138670811e-05,
+      "loss": 1.0247,
+      "step": 736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133658467910968e-05,
+      "loss": 0.9778,
+      "step": 737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125345523285598e-05,
+      "loss": 0.9937,
+      "step": 738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117022564515217e-05,
+      "loss": 1.008,
+      "step": 739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108689603297134e-05,
+      "loss": 1.0778,
+      "step": 740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100346651342707e-05,
+      "loss": 1.0787,
+      "step": 741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091993720377336e-05,
+      "loss": 0.9933,
+      "step": 742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083630822140444e-05,
+      "loss": 1.0164,
+      "step": 743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075257968385472e-05,
+      "loss": 1.0353,
+      "step": 744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066875170879836e-05,
+      "loss": 1.0507,
+      "step": 745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058482441404946e-05,
+      "loss": 1.0212,
+      "step": 746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050079791756157e-05,
+      "loss": 1.0461,
+      "step": 747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041667233742763e-05,
+      "loss": 1.0238,
+      "step": 748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033244779188002e-05,
+      "loss": 1.0986,
+      "step": 749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024812439929004e-05,
+      "loss": 1.0497,
+      "step": 750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70163702278168e-05,
+      "loss": 1.1047,
+      "step": 751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007918154716286e-05,
+      "loss": 1.0123,
+      "step": 752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999456232506236e-05,
+      "loss": 1.0596,
+      "step": 753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990984473079245e-05,
+      "loss": 1.0309,
+      "step": 754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698250288834175e-05,
+      "loss": 1.0554,
+      "step": 755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974011490213976e-05,
+      "loss": 0.9955,
+      "step": 756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965510290629973e-05,
+      "loss": 1.0596,
+      "step": 757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956999301537533e-05,
+      "loss": 0.9893,
+      "step": 758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948478534898225e-05,
+      "loss": 1.0287,
+      "step": 759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6939948002687352e-05,
+      "loss": 1.0422,
+      "step": 760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931407716893952e-05,
+      "loss": 1.0371,
+      "step": 761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692285768952076e-05,
+      "loss": 1.0391,
+      "step": 762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69142979325842e-05,
+      "loss": 1.0096,
+      "step": 763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905728458114384e-05,
+      "loss": 1.049,
+      "step": 764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897149278155068e-05,
+      "loss": 1.0376,
+      "step": 765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888560404763656e-05,
+      "loss": 0.9909,
+      "step": 766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879961850011174e-05,
+      "loss": 1.021,
+      "step": 767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687135362598225e-05,
+      "loss": 1.0114,
+      "step": 768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862735744775104e-05,
+      "loss": 1.0131,
+      "step": 769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854108218501534e-05,
+      "loss": 1.1026,
+      "step": 770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684547105928689e-05,
+      "loss": 1.116,
+      "step": 771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836824279270053e-05,
+      "loss": 1.0367,
+      "step": 772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828167890603433e-05,
+      "loss": 1.0566,
+      "step": 773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819501905452945e-05,
+      "loss": 1.0076,
+      "step": 774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810826335997986e-05,
+      "loss": 1.0332,
+      "step": 775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680214119443143e-05,
+      "loss": 0.9698,
+      "step": 776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793446492959596e-05,
+      "loss": 1.0408,
+      "step": 777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784742243802242e-05,
+      "loss": 1.0489,
+      "step": 778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677602845919255e-05,
+      "loss": 1.0133,
+      "step": 779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676730515137709e-05,
+      "loss": 1.0765,
+      "step": 780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758572332615834e-05,
+      "loss": 1.0278,
+      "step": 781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 0.9831,
+      "step": 782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674107821136259e-05,
+      "loss": 1.0501,
+      "step": 783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673231693345729e-05,
+      "loss": 1.0081,
+      "step": 784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723546193779535e-05,
+      "loss": 1.0171,
+      "step": 785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714766004655952e-05,
+      "loss": 1.0259,
+      "step": 786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705976378426447e-05,
+      "loss": 0.9753,
+      "step": 787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697177327444185e-05,
+      "loss": 1.065,
+      "step": 788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688368864075582e-05,
+      "loss": 1.0075,
+      "step": 789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679551000700277e-05,
+      "loss": 0.9897,
+      "step": 790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667072374971113e-05,
+      "loss": 1.0364,
+      "step": 791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661887123514183e-05,
+      "loss": 1.0936,
+      "step": 792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665304113452866e-05,
+      "loss": 1.0367,
+      "step": 793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644185795186946e-05,
+      "loss": 0.9838,
+      "step": 794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635321117934556e-05,
+      "loss": 1.0106,
+      "step": 795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662644711523014e-05,
+      "loss": 1.0044,
+      "step": 796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617563799545462e-05,
+      "loss": 1.0589,
+      "step": 797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660867118336535e-05,
+      "loss": 1.0449,
+      "step": 798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659976927918772e-05,
+      "loss": 1.0168,
+      "step": 799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6590858099523545e-05,
+      "loss": 0.971,
+      "step": 800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581937656896818e-05,
+      "loss": 0.9831,
+      "step": 801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657300796384457e-05,
+      "loss": 1.0269,
+      "step": 802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564069032916813e-05,
+      "loss": 1.0219,
+      "step": 803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555120876676557e-05,
+      "loss": 1.0688,
+      "step": 804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654616350769977e-05,
+      "loss": 1.008,
+      "step": 805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6537196938575376e-05,
+      "loss": 1.0139,
+      "step": 806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528221181905217e-05,
+      "loss": 1.0315,
+      "step": 807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519236250304058e-05,
+      "loss": 1.0609,
+      "step": 808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651024215639955e-05,
+      "loss": 1.0747,
+      "step": 809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501238912832226e-05,
+      "loss": 1.0392,
+      "step": 810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492226532255486e-05,
+      "loss": 0.9975,
+      "step": 811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648320502733555e-05,
+      "loss": 0.9606,
+      "step": 812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647417441075148e-05,
+      "loss": 1.0208,
+      "step": 813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646513469519514e-05,
+      "loss": 0.9926,
+      "step": 814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456085893371184e-05,
+      "loss": 1.0671,
+      "step": 815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644702801799702e-05,
+      "loss": 1.0436,
+      "step": 816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437961081802835e-05,
+      "loss": 1.0754,
+      "step": 817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428885097531524e-05,
+      "loss": 1.0286,
+      "step": 818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419800077938715e-05,
+      "loss": 1.0145,
+      "step": 819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641070603579273e-05,
+      "loss": 1.0525,
+      "step": 820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640160298387457e-05,
+      "loss": 1.002,
+      "step": 821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63924909349779e-05,
+      "loss": 1.0349,
+      "step": 822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383369901909035e-05,
+      "loss": 1.0717,
+      "step": 823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 1.002,
+      "step": 824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365100934543042e-05,
+      "loss": 1.0279,
+      "step": 825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355953025921606e-05,
+      "loss": 0.9625,
+      "step": 826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6346796184479293e-05,
+      "loss": 0.9744,
+      "step": 827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633763042308536e-05,
+      "loss": 1.0305,
+      "step": 828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328455754621614e-05,
+      "loss": 1.0276,
+      "step": 829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319272191982364e-05,
+      "loss": 0.9905,
+      "step": 830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631007974807443e-05,
+      "loss": 1.0078,
+      "step": 831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 1.0206,
+      "step": 832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291668268142172e-05,
+      "loss": 1.0067,
+      "step": 833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282449257993814e-05,
+      "loss": 0.9907,
+      "step": 834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627322141832868e-05,
+      "loss": 1.0826,
+      "step": 835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626398476211581e-05,
+      "loss": 1.0028,
+      "step": 836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625473930233664e-05,
+      "loss": 0.9937,
+      "step": 837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624548505198498e-05,
+      "loss": 1.0254,
+      "step": 838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236222024066993e-05,
+      "loss": 1.0327,
+      "step": 839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622695023160117e-05,
+      "loss": 1.0545,
+      "step": 840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217669687618345e-05,
+      "loss": 0.988,
+      "step": 841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208380405161623e-05,
+      "loss": 1.0108,
+      "step": 842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199082397286404e-05,
+      "loss": 1.084,
+      "step": 843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189775677060347e-05,
+      "loss": 1.064,
+      "step": 844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618046025756336e-05,
+      "loss": 1.0635,
+      "step": 845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171136151887577e-05,
+      "loss": 0.9634,
+      "step": 846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616180337313733e-05,
+      "loss": 0.9776,
+      "step": 847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152461934429154e-05,
+      "loss": 1.0301,
+      "step": 848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614311184889175e-05,
+      "loss": 1.004,
+      "step": 849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133753129665968e-05,
+      "loss": 1.0309,
+      "step": 850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612438578990479e-05,
+      "loss": 1.0108,
+      "step": 851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115009842773322e-05,
+      "loss": 1.0831,
+      "step": 852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105625301448762e-05,
+      "loss": 1.0516,
+      "step": 853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096232179120388e-05,
+      "loss": 1.0321,
+      "step": 854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086830488989534e-05,
+      "loss": 0.9621,
+      "step": 855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077420244269585e-05,
+      "loss": 1.0325,
+      "step": 856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068001458185934e-05,
+      "loss": 1.0016,
+      "step": 857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058574143975995e-05,
+      "loss": 0.9874,
+      "step": 858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049138314889156e-05,
+      "loss": 1.0743,
+      "step": 859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603969398418677e-05,
+      "loss": 1.059,
+      "step": 860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6030241165142155e-05,
+      "loss": 1.0226,
+      "step": 861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020779871040538e-05,
+      "loss": 1.0079,
+      "step": 862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011310115179074e-05,
+      "loss": 1.0162,
+      "step": 863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001831910866795e-05,
+      "loss": 1.0247,
+      "step": 864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599234527142462e-05,
+      "loss": 1.0359,
+      "step": 865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982850210185313e-05,
+      "loss": 1.0156,
+      "step": 866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973346740493486e-05,
+      "loss": 1.0519,
+      "step": 867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963834875705556e-05,
+      "loss": 1.0329,
+      "step": 868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954314629189743e-05,
+      "loss": 0.9898,
+      "step": 869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944786014326053e-05,
+      "loss": 0.9819,
+      "step": 870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935249044506245e-05,
+      "loss": 1.0704,
+      "step": 871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925703733133823e-05,
+      "loss": 1.0305,
+      "step": 872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916150093624014e-05,
+      "loss": 0.9694,
+      "step": 873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906588139403752e-05,
+      "loss": 0.9865,
+      "step": 874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589701788391165e-05,
+      "loss": 0.9644,
+      "step": 875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887439340598002e-05,
+      "loss": 1.0582,
+      "step": 876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.9963,
+      "step": 877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868257444365408e-05,
+      "loss": 1.0092,
+      "step": 878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858654118405196e-05,
+      "loss": 0.997,
+      "step": 879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849042558540863e-05,
+      "loss": 1.0836,
+      "step": 880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839422778280743e-05,
+      "loss": 0.9931,
+      "step": 881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 1.0449,
+      "step": 882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820158610664226e-05,
+      "loss": 1.0273,
+      "step": 883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581051425038219e-05,
+      "loss": 0.9815,
+      "step": 884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800861723853045e-05,
+      "loss": 0.9967,
+      "step": 885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791201044642707e-05,
+      "loss": 0.9896,
+      "step": 886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781532226328544e-05,
+      "loss": 1.0204,
+      "step": 887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577185528249936e-05,
+      "loss": 0.9994,
+      "step": 888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762170226755383e-05,
+      "loss": 1.0591,
+      "step": 889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752477072708247e-05,
+      "loss": 1.0155,
+      "step": 890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574277583398095e-05,
+      "loss": 1.0331,
+      "step": 891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5733066524207875e-05,
+      "loss": 1.0004,
+      "step": 892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572334915703473e-05,
+      "loss": 1.0012,
+      "step": 893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713623746118558e-05,
+      "loss": 1.0215,
+      "step": 894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57038903051277e-05,
+      "loss": 0.9881,
+      "step": 895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5694148847741793e-05,
+      "loss": 0.9937,
+      "step": 896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684399387651725e-05,
+      "loss": 1.0057,
+      "step": 897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5674641938559644e-05,
+      "loss": 1.0463,
+      "step": 898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566487651417892e-05,
+      "loss": 0.9858,
+      "step": 899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655103128234134e-05,
+      "loss": 1.0466,
+      "step": 900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564532179446106e-05,
+      "loss": 1.0155,
+      "step": 901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635532526606625e-05,
+      "loss": 0.9616,
+      "step": 902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625735338428932e-05,
+      "loss": 1.03,
+      "step": 903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615930243697196e-05,
+      "loss": 0.9684,
+      "step": 904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560611725619175e-05,
+      "loss": 0.9796,
+      "step": 905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559629638970403e-05,
+      "loss": 1.0004,
+      "step": 906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586467658036526e-05,
+      "loss": 0.9901,
+      "step": 907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576631075002796e-05,
+      "loss": 1.055,
+      "step": 908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566786654427426e-05,
+      "loss": 1.0279,
+      "step": 909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556934410146024e-05,
+      "loss": 1.0223,
+      "step": 910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547074356005184e-05,
+      "loss": 1.0172,
+      "step": 911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5537206505862486e-05,
+      "loss": 1.0132,
+      "step": 912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527330873586456e-05,
+      "loss": 1.0073,
+      "step": 913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517447473056568e-05,
+      "loss": 1.0043,
+      "step": 914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507556318163207e-05,
+      "loss": 0.9717,
+      "step": 915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549765742280766e-05,
+      "loss": 0.9814,
+      "step": 916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487750800902094e-05,
+      "loss": 1.0009,
+      "step": 917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477836466369522e-05,
+      "loss": 1.0193,
+      "step": 918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546791443314381e-05,
+      "loss": 1.0142,
+      "step": 919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457984715169643e-05,
+      "loss": 1.0365,
+      "step": 920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448047326402503e-05,
+      "loss": 0.9744,
+      "step": 921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438102280808653e-05,
+      "loss": 1.0028,
+      "step": 922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542814959236512e-05,
+      "loss": 0.991,
+      "step": 923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541818927505966e-05,
+      "loss": 1.0273,
+      "step": 924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408221342890772e-05,
+      "loss": 0.9715,
+      "step": 925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398245809867643e-05,
+      "loss": 1.0275,
+      "step": 926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538826269001014e-05,
+      "loss": 0.9779,
+      "step": 927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537827199734881e-05,
+      "loss": 0.9929,
+      "step": 928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368273745924816e-05,
+      "loss": 1.004,
+      "step": 929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.9605,
+      "step": 930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534825462300667e-05,
+      "loss": 1.1039,
+      "step": 931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 1.0214,
+      "step": 932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328205433797238e-05,
+      "loss": 0.9745,
+      "step": 933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5318169599548755e-05,
+      "loss": 1.0044,
+      "step": 934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530812629100708e-05,
+      "loss": 0.9891,
+      "step": 935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529807552228734e-05,
+      "loss": 1.0369,
+      "step": 936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288017307515142e-05,
+      "loss": 1.0253,
+      "step": 937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277951660826568e-05,
+      "loss": 0.9973,
+      "step": 938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267878596368132e-05,
+      "loss": 1.025,
+      "step": 939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257798128296783e-05,
+      "loss": 1.0533,
+      "step": 940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247710270779867e-05,
+      "loss": 1.0167,
+      "step": 941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237615037995129e-05,
+      "loss": 1.0077,
+      "step": 942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227512444130663e-05,
+      "loss": 0.9703,
+      "step": 943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217402503384914e-05,
+      "loss": 1.0439,
+      "step": 944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207285229966662e-05,
+      "loss": 1.0524,
+      "step": 945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197160638094981e-05,
+      "loss": 1.015,
+      "step": 946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187028741999234e-05,
+      "loss": 0.9639,
+      "step": 947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517688955591905e-05,
+      "loss": 0.999,
+      "step": 948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166743094104304e-05,
+      "loss": 0.9469,
+      "step": 949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156589370815096e-05,
+      "loss": 1.0673,
+      "step": 950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146428400321736e-05,
+      "loss": 0.9979,
+      "step": 951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136260196904704e-05,
+      "loss": 0.9868,
+      "step": 952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126084774854666e-05,
+      "loss": 1.0743,
+      "step": 953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115902148472418e-05,
+      "loss": 0.973,
+      "step": 954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105712332068892e-05,
+      "loss": 0.9961,
+      "step": 955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095515339965117e-05,
+      "loss": 1.0613,
+      "step": 956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085311186492206e-05,
+      "loss": 0.9784,
+      "step": 957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075099885991345e-05,
+      "loss": 1.0078,
+      "step": 958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506488145281376e-05,
+      "loss": 1.0319,
+      "step": 959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054655901320697e-05,
+      "loss": 0.9971,
+      "step": 960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044423245883418e-05,
+      "loss": 0.97,
+      "step": 961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034183500883153e-05,
+      "loss": 0.9717,
+      "step": 962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023936680711112e-05,
+      "loss": 1.0012,
+      "step": 963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013682799768435e-05,
+      "loss": 1.0811,
+      "step": 964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50034218724662e-05,
+      "loss": 1.0192,
+      "step": 965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993153913225374e-05,
+      "loss": 1.0317,
+      "step": 966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982878936476808e-05,
+      "loss": 1.0133,
+      "step": 967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972596956661229e-05,
+      "loss": 1.0412,
+      "step": 968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962307988229197e-05,
+      "loss": 1.0306,
+      "step": 969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495201204564109e-05,
+      "loss": 0.9874,
+      "step": 970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941709143367092e-05,
+      "loss": 0.9855,
+      "step": 971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931399295887172e-05,
+      "loss": 1.0098,
+      "step": 972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921082517691055e-05,
+      "loss": 1.0127,
+      "step": 973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910758823278208e-05,
+      "loss": 1.0154,
+      "step": 974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900428227157816e-05,
+      "loss": 0.9916,
+      "step": 975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890090743848774e-05,
+      "loss": 1.0078,
+      "step": 976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487974638787964e-05,
+      "loss": 1.0437,
+      "step": 977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869395173788642e-05,
+      "loss": 1.0134,
+      "step": 978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859037116123642e-05,
+      "loss": 1.0074,
+      "step": 979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848672229442132e-05,
+      "loss": 1.0543,
+      "step": 980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483830052831118e-05,
+      "loss": 0.9704,
+      "step": 981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 1.0237,
+      "step": 982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817536741017153e-05,
+      "loss": 0.9768,
+      "step": 983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807144684036044e-05,
+      "loss": 1.0497,
+      "step": 984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479674587096938e-05,
+      "loss": 0.984,
+      "step": 985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786340316431931e-05,
+      "loss": 0.9929,
+      "step": 986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775928035047928e-05,
+      "loss": 0.9708,
+      "step": 987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476550904145106e-05,
+      "loss": 0.9407,
+      "step": 988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755083350284454e-05,
+      "loss": 1.0214,
+      "step": 989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744650976200643e-05,
+      "loss": 0.9805,
+      "step": 990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734211933861555e-05,
+      "loss": 0.972,
+      "step": 991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723766237938495e-05,
+      "loss": 1.027,
+      "step": 992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713313903112108e-05,
+      "loss": 1.0169,
+      "step": 993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702854944072383e-05,
+      "loss": 0.9563,
+      "step": 994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692389375518603e-05,
+      "loss": 1.0307,
+      "step": 995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681917212159358e-05,
+      "loss": 0.9887,
+      "step": 996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467143846871249e-05,
+      "loss": 1.031,
+      "step": 997
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46609531599051e-05,
+      "loss": 1.0334,
+      "step": 998
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650461300473513e-05,
+      "loss": 1.0341,
+      "step": 999
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639962905163258e-05,
+      "loss": 0.9977,
+      "step": 1000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629457988729052e-05,
+      "loss": 0.9377,
+      "step": 1001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4618946565934775e-05,
+      "loss": 0.9669,
+      "step": 1002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608428651553454e-05,
+      "loss": 0.9343,
+      "step": 1003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597904260367239e-05,
+      "loss": 1.0802,
+      "step": 1004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587373407167382e-05,
+      "loss": 1.0157,
+      "step": 1005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576836106754213e-05,
+      "loss": 1.0096,
+      "step": 1006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566292373937133e-05,
+      "loss": 1.0097,
+      "step": 1007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455574222353457e-05,
+      "loss": 0.9922,
+      "step": 1008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545185670373982e-05,
+      "loss": 0.9615,
+      "step": 1009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453462272929182e-05,
+      "loss": 1.0084,
+      "step": 1010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524053415133521e-05,
+      "loss": 1.0344,
+      "step": 1011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513477742753465e-05,
+      "loss": 0.9624,
+      "step": 1012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502895727014978e-05,
+      "loss": 0.9763,
+      "step": 1013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44923073827903e-05,
+      "loss": 0.9937,
+      "step": 1014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481712724960563e-05,
+      "loss": 1.0392,
+      "step": 1015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471111768415777e-05,
+      "loss": 0.9687,
+      "step": 1016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460504528054796e-05,
+      "loss": 0.9578,
+      "step": 1017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444989101878531e-05,
+      "loss": 1.0605,
+      "step": 1018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443927125552382e-05,
+      "loss": 1.0181,
+      "step": 1019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428645253195621e-05,
+      "loss": 1.0072,
+      "step": 1020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418013026734766e-05,
+      "loss": 1.0246,
+      "step": 1021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407374591084064e-05,
+      "loss": 0.9814,
+      "step": 1022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396729961195047e-05,
+      "loss": 0.9794,
+      "step": 1023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386079152027952e-05,
+      "loss": 1.0266,
+      "step": 1024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375422178551704e-05,
+      "loss": 0.9806,
+      "step": 1025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364759055743888e-05,
+      "loss": 1.0063,
+      "step": 1026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354089798590735e-05,
+      "loss": 0.9776,
+      "step": 1027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343414422087093e-05,
+      "loss": 0.9693,
+      "step": 1028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332732941236411e-05,
+      "loss": 1.0303,
+      "step": 1029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322045371050722e-05,
+      "loss": 1.0021,
+      "step": 1030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311351726550612e-05,
+      "loss": 1.0237,
+      "step": 1031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 0.9819,
+      "step": 1032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4289946274732142e-05,
+      "loss": 1.0129,
+      "step": 1033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4279234497497563e-05,
+      "loss": 0.9952,
+      "step": 1034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268516706116072e-05,
+      "loss": 0.9288,
+      "step": 1035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 0.9467,
+      "step": 1036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424706314117303e-05,
+      "loss": 0.9662,
+      "step": 1037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236327397762874e-05,
+      "loss": 0.9996,
+      "step": 1038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422558570050856e-05,
+      "loss": 0.931,
+      "step": 1039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214838064506738e-05,
+      "loss": 1.0415,
+      "step": 1040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420408450486242e-05,
+      "loss": 0.944,
+      "step": 1041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419332503668894e-05,
+      "loss": 1.0119,
+      "step": 1042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418255967510792e-05,
+      "loss": 0.9686,
+      "step": 1043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417178843524929e-05,
+      "loss": 0.9967,
+      "step": 1044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161011332251229e-05,
+      "loss": 1.032,
+      "step": 1045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415022838126015e-05,
+      "loss": 1.0647,
+      "step": 1046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139439597430693e-05,
+      "loss": 1.0443,
+      "step": 1047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128644995925696e-05,
+      "loss": 0.9923,
+      "step": 1048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117844591916172e-05,
+      "loss": 0.9684,
+      "step": 1049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107038400581288e-05,
+      "loss": 0.9924,
+      "step": 1050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096226437108342e-05,
+      "loss": 0.959,
+      "step": 1051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408540871669275e-05,
+      "loss": 1.0818,
+      "step": 1052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4074585254538015e-05,
+      "loss": 0.9845,
+      "step": 1053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063756065855714e-05,
+      "loss": 0.9457,
+      "step": 1054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052921165865461e-05,
+      "loss": 1.0115,
+      "step": 1055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042080569794916e-05,
+      "loss": 1.0004,
+      "step": 1056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031234292879726e-05,
+      "loss": 0.9455,
+      "step": 1057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020382350363528e-05,
+      "loss": 1.0386,
+      "step": 1058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009524757497925e-05,
+      "loss": 0.9851,
+      "step": 1059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998661529542463e-05,
+      "loss": 1.0405,
+      "step": 1060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39877926817646e-05,
+      "loss": 0.9354,
+      "step": 1061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976918229439698e-05,
+      "loss": 0.983,
+      "step": 1062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966038187850985e-05,
+      "loss": 0.9649,
+      "step": 1063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955152572289568e-05,
+      "loss": 0.9818,
+      "step": 1064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944261398054361e-05,
+      "loss": 1.0201,
+      "step": 1065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933364680452106e-05,
+      "loss": 0.9851,
+      "step": 1066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922462434797335e-05,
+      "loss": 0.9934,
+      "step": 1067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391155467641234e-05,
+      "loss": 0.9938,
+      "step": 1068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3900641420627172e-05,
+      "loss": 0.9728,
+      "step": 1069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889722682779598e-05,
+      "loss": 0.9807,
+      "step": 1070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878798478215103e-05,
+      "loss": 0.9824,
+      "step": 1071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3867868822286838e-05,
+      "loss": 1.0292,
+      "step": 1072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385693373035563e-05,
+      "loss": 1.0399,
+      "step": 1073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845993217789937e-05,
+      "loss": 0.9764,
+      "step": 1074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383504729996584e-05,
+      "loss": 0.9963,
+      "step": 1075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824095992267017e-05,
+      "loss": 0.9624,
+      "step": 1076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813139310084715e-05,
+      "loss": 1.0192,
+      "step": 1077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802177268817742e-05,
+      "loss": 0.963,
+      "step": 1078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3791209883872434e-05,
+      "loss": 0.9904,
+      "step": 1079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3780237170662638e-05,
+      "loss": 1.0082,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769259144609688e-05,
+      "loss": 0.9844,
+      "step": 1081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 0.9898,
+      "step": 1082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747287215696976e-05,
+      "loss": 0.9842,
+      "step": 1083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736293343717134e-05,
+      "loss": 1.0105,
+      "step": 1084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372529422065393e-05,
+      "loss": 0.9974,
+      "step": 1085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714289861965816e-05,
+      "loss": 0.9964,
+      "step": 1086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703280283118601e-05,
+      "loss": 1.0222,
+      "step": 1087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692265499585438e-05,
+      "loss": 1.0013,
+      "step": 1088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681245526846782e-05,
+      "loss": 0.9831,
+      "step": 1089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670220380390389e-05,
+      "loss": 1.0105,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365919007571129e-05,
+      "loss": 0.996,
+      "step": 1091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648154628311754e-05,
+      "loss": 0.9903,
+      "step": 1092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637114053701291e-05,
+      "loss": 0.9756,
+      "step": 1093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626068367396603e-05,
+      "loss": 0.998,
+      "step": 1094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361501758492159e-05,
+      "loss": 0.9654,
+      "step": 1095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603961721807304e-05,
+      "loss": 0.9847,
+      "step": 1096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592900793591941e-05,
+      "loss": 1.0355,
+      "step": 1097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581834815820817e-05,
+      "loss": 0.9759,
+      "step": 1098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570763804046348e-05,
+      "loss": 0.9726,
+      "step": 1099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559687773828012e-05,
+      "loss": 1.0153,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548606740732353e-05,
+      "loss": 1.0089,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537520720332943e-05,
+      "loss": 1.0228,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526429728210358e-05,
+      "loss": 0.976,
+      "step": 1103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515333779952169e-05,
+      "loss": 1.0421,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504232891152904e-05,
+      "loss": 0.9501,
+      "step": 1105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493127077414046e-05,
+      "loss": 0.9527,
+      "step": 1106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348201635434399e-05,
+      "loss": 1.0043,
+      "step": 1107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470900737558032e-05,
+      "loss": 0.9804,
+      "step": 1108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459780242678351e-05,
+      "loss": 0.9988,
+      "step": 1109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448654885333974e-05,
+      "loss": 0.9906,
+      "step": 1110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437524681160773e-05,
+      "loss": 0.9623,
+      "step": 1111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426389645801415e-05,
+      "loss": 0.9952,
+      "step": 1112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415249794905375e-05,
+      "loss": 1.0201,
+      "step": 1113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3404105144128885e-05,
+      "loss": 1.026,
+      "step": 1114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3392955709134925e-05,
+      "loss": 0.9485,
+      "step": 1115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381801505593201e-05,
+      "loss": 0.9631,
+      "step": 1116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370642549180117e-05,
+      "loss": 0.9979,
+      "step": 1117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359478855578764e-05,
+      "loss": 0.9631,
+      "step": 1118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3348310440478882e-05,
+      "loss": 1.0547,
+      "step": 1119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333713731957685e-05,
+      "loss": 0.9727,
+      "step": 1120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325959508575663e-05,
+      "loss": 1.015,
+      "step": 1121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314777023184907e-05,
+      "loss": 0.9563,
+      "step": 1122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3303589879120733e-05,
+      "loss": 0.9789,
+      "step": 1123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292398092105842e-05,
+      "loss": 0.9612,
+      "step": 1124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281201677869468e-05,
+      "loss": 0.9463,
+      "step": 1125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270000652147339e-05,
+      "loss": 0.9594,
+      "step": 1126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325879503068166e-05,
+      "loss": 0.9633,
+      "step": 1127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247584829221104e-05,
+      "loss": 1.0272,
+      "step": 1128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236370063520787e-05,
+      "loss": 0.946,
+      "step": 1129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225150749342222e-05,
+      "loss": 1.0103,
+      "step": 1130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213926902453322e-05,
+      "loss": 0.9869,
+      "step": 1131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 0.9478,
+      "step": 1132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319146567364802e-05,
+      "loss": 0.9986,
+      "step": 1133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318022832329921e-05,
+      "loss": 0.9906,
+      "step": 1134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168986503375205e-05,
+      "loss": 0.9933,
+      "step": 1135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157740229675557e-05,
+      "loss": 0.9809,
+      "step": 1136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146489518006065e-05,
+      "loss": 0.9974,
+      "step": 1137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135234384178772e-05,
+      "loss": 1.0082,
+      "step": 1138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123974844011934e-05,
+      "loss": 0.9415,
+      "step": 1139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311271091333e-05,
+      "loss": 0.9662,
+      "step": 1140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101442607963595e-05,
+      "loss": 0.9607,
+      "step": 1141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.9895,
+      "step": 1142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078892936530543e-05,
+      "loss": 1.0176,
+      "step": 1143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067611602155799e-05,
+      "loss": 0.9863,
+      "step": 1144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056325956480316e-05,
+      "loss": 0.9628,
+      "step": 1145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045036015365233e-05,
+      "loss": 0.9609,
+      "step": 1146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3033741794677728e-05,
+      "loss": 1.0385,
+      "step": 1147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022443310290993e-05,
+      "loss": 0.9581,
+      "step": 1148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011140578084207e-05,
+      "loss": 0.9772,
+      "step": 1149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299983361394252e-05,
+      "loss": 0.9657,
+      "step": 1150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988522433757039e-05,
+      "loss": 0.9474,
+      "step": 1151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977207053424781e-05,
+      "loss": 0.9312,
+      "step": 1152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965887488848682e-05,
+      "loss": 0.96,
+      "step": 1153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954563755937546e-05,
+      "loss": 0.9727,
+      "step": 1154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943235870606043e-05,
+      "loss": 0.991,
+      "step": 1155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931903848774676e-05,
+      "loss": 1.0234,
+      "step": 1156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292056770636976e-05,
+      "loss": 0.9791,
+      "step": 1157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909227459323403e-05,
+      "loss": 0.9922,
+      "step": 1158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897883123573485e-05,
+      "loss": 1.0439,
+      "step": 1159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2886534715063626e-05,
+      "loss": 0.9726,
+      "step": 1160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2875182249743174e-05,
+      "loss": 0.9938,
+      "step": 1161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2863825743567174e-05,
+      "loss": 0.9644,
+      "step": 1162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852465212496362e-05,
+      "loss": 0.9694,
+      "step": 1163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841100672497116e-05,
+      "loss": 0.9986,
+      "step": 1164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829732139541456e-05,
+      "loss": 1.0496,
+      "step": 1165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818359629607008e-05,
+      "loss": 0.975,
+      "step": 1166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806983158677e-05,
+      "loss": 0.9736,
+      "step": 1167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795602742740217e-05,
+      "loss": 0.9859,
+      "step": 1168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2784218397790983e-05,
+      "loss": 0.9762,
+      "step": 1169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277283013982916e-05,
+      "loss": 1.0292,
+      "step": 1170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761437984860096e-05,
+      "loss": 0.9816,
+      "step": 1171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750041948894621e-05,
+      "loss": 0.974,
+      "step": 1172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738642047949014e-05,
+      "loss": 0.9875,
+      "step": 1173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727238298045002e-05,
+      "loss": 1.0574,
+      "step": 1174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715830715209702e-05,
+      "loss": 0.9714,
+      "step": 1175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704419315475629e-05,
+      "loss": 1.0146,
+      "step": 1176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693004114880654e-05,
+      "loss": 0.9234,
+      "step": 1177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681585129468003e-05,
+      "loss": 0.9636,
+      "step": 1178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670162375286213e-05,
+      "loss": 0.8909,
+      "step": 1179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658735868389113e-05,
+      "loss": 0.9783,
+      "step": 1180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2647305624835816e-05,
+      "loss": 0.9674,
+      "step": 1181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 1.0178,
+      "step": 1182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624433992023293e-05,
+      "loss": 1.0173,
+      "step": 1183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612992634908454e-05,
+      "loss": 0.9797,
+      "step": 1184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601547605426138e-05,
+      "loss": 0.9628,
+      "step": 1185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259009891966149e-05,
+      "loss": 0.9532,
+      "step": 1186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578646593704786e-05,
+      "loss": 0.9537,
+      "step": 1187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567190643651426e-05,
+      "loss": 1.0071,
+      "step": 1188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555731085601892e-05,
+      "loss": 1.0005,
+      "step": 1189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544267935661751e-05,
+      "loss": 1.0071,
+      "step": 1190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532801209941606e-05,
+      "loss": 1.015,
+      "step": 1191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521330924557087e-05,
+      "loss": 0.9448,
+      "step": 1192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2509857095628832e-05,
+      "loss": 0.9254,
+      "step": 1193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498379739282465e-05,
+      "loss": 1.0334,
+      "step": 1194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486898871648552e-05,
+      "loss": 0.9627,
+      "step": 1195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475414508862598e-05,
+      "loss": 0.9926,
+      "step": 1196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463926667065031e-05,
+      "loss": 0.9901,
+      "step": 1197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452435362401161e-05,
+      "loss": 0.9532,
+      "step": 1198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440940611021161e-05,
+      "loss": 0.9963,
+      "step": 1199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429442429080054e-05,
+      "loss": 0.9864,
+      "step": 1200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417940832737682e-05,
+      "loss": 0.9727,
+      "step": 1201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406435838158686e-05,
+      "loss": 0.9998,
+      "step": 1202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2394927461512486e-05,
+      "loss": 0.958,
+      "step": 1203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383415718973245e-05,
+      "loss": 0.9696,
+      "step": 1204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371900626719873e-05,
+      "loss": 0.9612,
+      "step": 1205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360382200935971e-05,
+      "loss": 0.9979,
+      "step": 1206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234886045780984e-05,
+      "loss": 0.9654,
+      "step": 1207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337335413534428e-05,
+      "loss": 0.9813,
+      "step": 1208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325807084307338e-05,
+      "loss": 1.009,
+      "step": 1209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314275486330778e-05,
+      "loss": 1.0016,
+      "step": 1210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302740635811554e-05,
+      "loss": 0.9385,
+      "step": 1211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291202548961042e-05,
+      "loss": 0.9656,
+      "step": 1212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279661241995177e-05,
+      "loss": 0.9814,
+      "step": 1213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.22681167311344e-05,
+      "loss": 0.9515,
+      "step": 1214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256569032603671e-05,
+      "loss": 1.0317,
+      "step": 1215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245018162632421e-05,
+      "loss": 0.9633,
+      "step": 1216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233464137454542e-05,
+      "loss": 0.9532,
+      "step": 1217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221906973308365e-05,
+      "loss": 0.95,
+      "step": 1218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210346686436623e-05,
+      "loss": 0.9784,
+      "step": 1219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198783293086442e-05,
+      "loss": 0.9766,
+      "step": 1220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187216809509314e-05,
+      "loss": 0.9269,
+      "step": 1221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217564725196108e-05,
+      "loss": 0.9684,
+      "step": 1222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164074636701888e-05,
+      "loss": 0.9659,
+      "step": 1223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152498979996195e-05,
+      "loss": 1.0124,
+      "step": 1224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214092029811273e-05,
+      "loss": 0.9592,
+      "step": 1225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2129338607324468e-05,
+      "loss": 0.9424,
+      "step": 1226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117753923908617e-05,
+      "loss": 0.9917,
+      "step": 1227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106166264146598e-05,
+      "loss": 0.9662,
+      "step": 1228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094575644324e-05,
+      "loss": 0.9959,
+      "step": 1229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082982080730583e-05,
+      "loss": 1.0121,
+      "step": 1230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071385589660241e-05,
+      "loss": 0.9501,
+      "step": 1231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 1.0007,
+      "step": 1232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048183890284912e-05,
+      "loss": 0.9877,
+      "step": 1233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036578714588191e-05,
+      "loss": 1.0482,
+      "step": 1234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2024970676631039e-05,
+      "loss": 0.9421,
+      "step": 1235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013359792727688e-05,
+      "loss": 0.9968,
+      "step": 1236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001746079196381e-05,
+      "loss": 0.9591,
+      "step": 1237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990129552359326e-05,
+      "loss": 0.9678,
+      "step": 1238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978510228542694e-05,
+      "loss": 1.0357,
+      "step": 1239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966888124076584e-05,
+      "loss": 0.9406,
+      "step": 1240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955263255294996e-05,
+      "loss": 0.9471,
+      "step": 1241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943635638535827e-05,
+      "loss": 0.957,
+      "step": 1242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1932005290140832e-05,
+      "loss": 0.9862,
+      "step": 1243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19203722264556e-05,
+      "loss": 0.9633,
+      "step": 1244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190873646382954e-05,
+      "loss": 0.957,
+      "step": 1245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897098018615854e-05,
+      "loss": 1.0104,
+      "step": 1246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885456907171517e-05,
+      "loss": 0.976,
+      "step": 1247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 1.0392,
+      "step": 1248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862166751037485e-05,
+      "loss": 0.9699,
+      "step": 1249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850517739080381e-05,
+      "loss": 1.0754,
+      "step": 1250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838866126357749e-05,
+      "loss": 0.9503,
+      "step": 1251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827211929245075e-05,
+      "loss": 0.972,
+      "step": 1252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815555164121466e-05,
+      "loss": 0.9926,
+      "step": 1253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803895847369645e-05,
+      "loss": 1.0186,
+      "step": 1254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792233995375912e-05,
+      "loss": 0.9636,
+      "step": 1255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780569624530135e-05,
+      "loss": 0.9884,
+      "step": 1256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176890275122573e-05,
+      "loss": 0.9748,
+      "step": 1257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757233391859617e-05,
+      "loss": 0.9505,
+      "step": 1258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745561562832219e-05,
+      "loss": 0.9986,
+      "step": 1259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733887280547425e-05,
+      "loss": 0.9926,
+      "step": 1260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722210561412568e-05,
+      "loss": 1.0037,
+      "step": 1261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710531421838422e-05,
+      "loss": 1.0085,
+      "step": 1262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698849878239145e-05,
+      "loss": 0.9925,
+      "step": 1263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1687165947032285e-05,
+      "loss": 0.9881,
+      "step": 1264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675479644638732e-05,
+      "loss": 0.9449,
+      "step": 1265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663790987482729e-05,
+      "loss": 0.9708,
+      "step": 1266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165209999199181e-05,
+      "loss": 0.9886,
+      "step": 1267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640406674596807e-05,
+      "loss": 0.9943,
+      "step": 1268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162871105173181e-05,
+      "loss": 1.028,
+      "step": 1269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1617013139834148e-05,
+      "loss": 0.9821,
+      "step": 1270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605312955344368e-05,
+      "loss": 1.0112,
+      "step": 1271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593610514706217e-05,
+      "loss": 1.009,
+      "step": 1272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581905834366605e-05,
+      "loss": 1.0928,
+      "step": 1273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570198930775594e-05,
+      "loss": 0.9887,
+      "step": 1274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558489820386368e-05,
+      "loss": 1.0063,
+      "step": 1275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546778519655209e-05,
+      "loss": 0.9739,
+      "step": 1276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535065045041492e-05,
+      "loss": 0.9926,
+      "step": 1277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523349413007633e-05,
+      "loss": 0.9789,
+      "step": 1278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511631640019086e-05,
+      "loss": 0.9724,
+      "step": 1279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499911742544304e-05,
+      "loss": 0.96,
+      "step": 1280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488189737054744e-05,
+      "loss": 0.9893,
+      "step": 1281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 0.9537,
+      "step": 1282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464739467931863e-05,
+      "loss": 0.979,
+      "step": 1283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453011237256152e-05,
+      "loss": 0.9749,
+      "step": 1284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441280964480845e-05,
+      "loss": 0.9794,
+      "step": 1285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429548666091969e-05,
+      "loss": 1.0308,
+      "step": 1286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14178143585784e-05,
+      "loss": 0.9714,
+      "step": 1287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140607805843184e-05,
+      "loss": 0.994,
+      "step": 1288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394339782146787e-05,
+      "loss": 1.0148,
+      "step": 1289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382599546220516e-05,
+      "loss": 0.9627,
+      "step": 1290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137085736715306e-05,
+      "loss": 0.9665,
+      "step": 1291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359113261447183e-05,
+      "loss": 0.9952,
+      "step": 1292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347367245608356e-05,
+      "loss": 0.963,
+      "step": 1293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133561933614473e-05,
+      "loss": 0.9623,
+      "step": 1294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323869549567121e-05,
+      "loss": 0.9385,
+      "step": 1295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312117902388986e-05,
+      "loss": 0.9387,
+      "step": 1296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300364411126395e-05,
+      "loss": 1.0339,
+      "step": 1297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1288609092298004e-05,
+      "loss": 0.9838,
+      "step": 1298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276851962425043e-05,
+      "loss": 1.0197,
+      "step": 1299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265093038031294e-05,
+      "loss": 0.9975,
+      "step": 1300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253332335643043e-05,
+      "loss": 0.9589,
+      "step": 1301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241569871789096e-05,
+      "loss": 0.9532,
+      "step": 1302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229805663000712e-05,
+      "loss": 0.958,
+      "step": 1303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218039725811626e-05,
+      "loss": 0.9408,
+      "step": 1304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206272076757987e-05,
+      "loss": 0.961,
+      "step": 1305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194502732378349e-05,
+      "loss": 0.9497,
+      "step": 1306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182731709213658e-05,
+      "loss": 0.9996,
+      "step": 1307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170959023807216e-05,
+      "loss": 1.0047,
+      "step": 1308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159184692704655e-05,
+      "loss": 0.9717,
+      "step": 1309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147408732453926e-05,
+      "loss": 0.9688,
+      "step": 1310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135631159605267e-05,
+      "loss": 0.9537,
+      "step": 1311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112385199071119e-05,
+      "loss": 0.9413,
+      "step": 1312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1112071242326438e-05,
+      "loss": 0.9493,
+      "step": 1313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100288931007982e-05,
+      "loss": 0.8918,
+      "step": 1314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088505073314983e-05,
+      "loss": 0.9283,
+      "step": 1315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076719685808786e-05,
+      "loss": 1.0237,
+      "step": 1316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064932785052877e-05,
+      "loss": 0.9805,
+      "step": 1317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105314438761287e-05,
+      "loss": 0.9242,
+      "step": 1318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1041354510056483e-05,
+      "loss": 0.9411,
+      "step": 1319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102956316895352e-05,
+      "loss": 1.0392,
+      "step": 1320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017770380875832e-05,
+      "loss": 1.0161,
+      "step": 1321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005976162397309e-05,
+      "loss": 0.9261,
+      "step": 1322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994180530093856e-05,
+      "loss": 0.9702,
+      "step": 1323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982383500543355e-05,
+      "loss": 0.9872,
+      "step": 1324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0970585090325656e-05,
+      "loss": 0.9219,
+      "step": 1325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958785316022551e-05,
+      "loss": 0.9366,
+      "step": 1326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946984194217753e-05,
+      "loss": 0.9537,
+      "step": 1327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935181741496858e-05,
+      "loss": 0.9758,
+      "step": 1328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923377974447343e-05,
+      "loss": 0.9575,
+      "step": 1329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911572909658524e-05,
+      "loss": 0.9329,
+      "step": 1330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899766563721543e-05,
+      "loss": 0.9494,
+      "step": 1331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 0.9637,
+      "step": 1332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087615009477666e-05,
+      "loss": 0.9288,
+      "step": 1333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864340004959957e-05,
+      "loss": 0.9691,
+      "step": 1334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852528700377439e-05,
+      "loss": 0.9011,
+      "step": 1335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084071619762902e-05,
+      "loss": 0.9668,
+      "step": 1336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828902513316299e-05,
+      "loss": 0.9266,
+      "step": 1337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817087664042536e-05,
+      "loss": 0.9278,
+      "step": 1338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805271666412625e-05,
+      "loss": 0.9279,
+      "step": 1339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793454537033068e-05,
+      "loss": 0.9511,
+      "step": 1340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078163629251197e-05,
+      "loss": 1.0648,
+      "step": 1341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0769816949459002e-05,
+      "loss": 0.9404,
+      "step": 1342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757996524485368e-05,
+      "loss": 0.9509,
+      "step": 1343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746175034203799e-05,
+      "loss": 0.9479,
+      "step": 1344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734352495228526e-05,
+      "loss": 0.9947,
+      "step": 1345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722528924175254e-05,
+      "loss": 0.9326,
+      "step": 1346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0710704337661131e-05,
+      "loss": 0.955,
+      "step": 1347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698878752304738e-05,
+      "loss": 0.9767,
+      "step": 1348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687052184726061e-05,
+      "loss": 0.9589,
+      "step": 1349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675224651546459e-05,
+      "loss": 0.952,
+      "step": 1350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066339616938866e-05,
+      "loss": 0.9797,
+      "step": 1351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651566754876715e-05,
+      "loss": 0.9697,
+      "step": 1352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063973642463599e-05,
+      "loss": 0.9466,
+      "step": 1353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 0.9719,
+      "step": 1354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616073083476069e-05,
+      "loss": 0.9264,
+      "step": 1355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604240105813948e-05,
+      "loss": 0.962,
+      "step": 1356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592406278937143e-05,
+      "loss": 0.9941,
+      "step": 1357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580571619477225e-05,
+      "loss": 0.928,
+      "step": 1358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0568736144066927e-05,
+      "loss": 0.942,
+      "step": 1359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556899869340127e-05,
+      "loss": 1.0291,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545062811931844e-05,
+      "loss": 0.9594,
+      "step": 1361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533224988478176e-05,
+      "loss": 0.9702,
+      "step": 1362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521386415616313e-05,
+      "loss": 0.9498,
+      "step": 1363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509547109984484e-05,
+      "loss": 0.9466,
+      "step": 1364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497707088221958e-05,
+      "loss": 0.9628,
+      "step": 1365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485866366969012e-05,
+      "loss": 0.9659,
+      "step": 1366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04740249628669e-05,
+      "loss": 0.9534,
+      "step": 1367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462182892557834e-05,
+      "loss": 0.9591,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450340172684975e-05,
+      "loss": 0.9803,
+      "step": 1369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438496819892376e-05,
+      "loss": 0.9783,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426652850825004e-05,
+      "loss": 0.9536,
+      "step": 1371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414808282128668e-05,
+      "loss": 0.972,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402963130450039e-05,
+      "loss": 0.94,
+      "step": 1373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03911174124366e-05,
+      "loss": 0.9502,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0379271144736622e-05,
+      "loss": 0.9591,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367424343999164e-05,
+      "loss": 0.9655,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355577026874026e-05,
+      "loss": 1.0043,
+      "step": 1377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343729210011731e-05,
+      "loss": 0.9473,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331880910063505e-05,
+      "loss": 0.9408,
+      "step": 1379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320032143681262e-05,
+      "loss": 0.965,
+      "step": 1380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308182927517562e-05,
+      "loss": 0.9556,
+      "step": 1381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 0.9653,
+      "step": 1382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284483212459178e-05,
+      "loss": 0.9833,
+      "step": 1383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272632746872687e-05,
+      "loss": 0.9406,
+      "step": 1384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260781898121077e-05,
+      "loss": 0.9427,
+      "step": 1385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248930682859839e-05,
+      "loss": 0.9435,
+      "step": 1386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237079117744977e-05,
+      "loss": 0.9969,
+      "step": 1387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225227219432988e-05,
+      "loss": 0.9678,
+      "step": 1388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0213375004580837e-05,
+      "loss": 0.9968,
+      "step": 1389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201522489845927e-05,
+      "loss": 1.0141,
+      "step": 1390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189669691886097e-05,
+      "loss": 0.9637,
+      "step": 1391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177816627359575e-05,
+      "loss": 0.9397,
+      "step": 1392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0165963312924963e-05,
+      "loss": 0.9989,
+      "step": 1393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154109765241214e-05,
+      "loss": 0.9973,
+      "step": 1394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142256000967612e-05,
+      "loss": 1.0045,
+      "step": 1395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130402036763747e-05,
+      "loss": 0.9972,
+      "step": 1396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118547889289485e-05,
+      "loss": 0.9597,
+      "step": 1397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106693575204947e-05,
+      "loss": 0.9571,
+      "step": 1398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094839111170503e-05,
+      "loss": 0.9619,
+      "step": 1399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082984513846713e-05,
+      "loss": 0.9773,
+      "step": 1400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007112979989434e-05,
+      "loss": 0.9789,
+      "step": 1401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059274985974305e-05,
+      "loss": 0.9783,
+      "step": 1402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047420088747675e-05,
+      "loss": 0.9102,
+      "step": 1403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035565124875623e-05,
+      "loss": 0.9659,
+      "step": 1404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023710111019422e-05,
+      "loss": 0.9533,
+      "step": 1405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011855063840416e-05,
+      "loss": 0.9268,
+      "step": 1406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 1.0008,
+      "step": 1407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988144936159582e-06,
+      "loss": 0.9525,
+      "step": 1408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976289888980582e-06,
+      "loss": 0.968,
+      "step": 1409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96443487512438e-06,
+      "loss": 0.9623,
+      "step": 1410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95257991125233e-06,
+      "loss": 0.9616,
+      "step": 1411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940725014025696e-06,
+      "loss": 0.9798,
+      "step": 1412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92887020010566e-06,
+      "loss": 0.8679,
+      "step": 1413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91701548615329e-06,
+      "loss": 0.9541,
+      "step": 1414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905160888829502e-06,
+      "loss": 1.0193,
+      "step": 1415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893306424795055e-06,
+      "loss": 0.985,
+      "step": 1416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881452110710519e-06,
+      "loss": 1.0011,
+      "step": 1417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869597963236253e-06,
+      "loss": 0.9163,
+      "step": 1418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857743999032391e-06,
+      "loss": 1.0242,
+      "step": 1419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845890234758789e-06,
+      "loss": 0.9131,
+      "step": 1420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834036687075042e-06,
+      "loss": 0.9703,
+      "step": 1421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822183372640426e-06,
+      "loss": 0.9822,
+      "step": 1422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810330308113903e-06,
+      "loss": 0.9515,
+      "step": 1423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798477510154075e-06,
+      "loss": 0.9437,
+      "step": 1424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786624995419167e-06,
+      "loss": 0.9376,
+      "step": 1425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774772780567017e-06,
+      "loss": 0.9921,
+      "step": 1426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762920882255026e-06,
+      "loss": 0.9575,
+      "step": 1427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751069317140163e-06,
+      "loss": 0.9061,
+      "step": 1428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739218101878925e-06,
+      "loss": 0.9551,
+      "step": 1429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727367253127315e-06,
+      "loss": 0.9932,
+      "step": 1430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715516787540828e-06,
+      "loss": 0.9501,
+      "step": 1431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 0.9648,
+      "step": 1432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69181707248244e-06,
+      "loss": 0.9024,
+      "step": 1433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679967856318741e-06,
+      "loss": 1.0268,
+      "step": 1434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668119089936496e-06,
+      "loss": 0.9893,
+      "step": 1435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656270789988274e-06,
+      "loss": 0.9596,
+      "step": 1436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644422973125977e-06,
+      "loss": 0.975,
+      "step": 1437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632575656000837e-06,
+      "loss": 0.922,
+      "step": 1438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62072885526338e-06,
+      "loss": 0.9963,
+      "step": 1439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608882587563404e-06,
+      "loss": 0.9093,
+      "step": 1440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597036869549965e-06,
+      "loss": 0.9099,
+      "step": 1441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585191717871336e-06,
+      "loss": 0.9832,
+      "step": 1442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573347149175001e-06,
+      "loss": 0.9411,
+      "step": 1443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561503180107626e-06,
+      "loss": 0.9548,
+      "step": 1444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549659827315029e-06,
+      "loss": 0.9747,
+      "step": 1445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53781710744217e-06,
+      "loss": 0.9332,
+      "step": 1446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525975037133101e-06,
+      "loss": 0.9823,
+      "step": 1447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514133633030987e-06,
+      "loss": 0.9213,
+      "step": 1448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502292911778045e-06,
+      "loss": 0.967,
+      "step": 1449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49045289001552e-06,
+      "loss": 0.958,
+      "step": 1450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478613584383692e-06,
+      "loss": 0.9658,
+      "step": 1451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466775011521825e-06,
+      "loss": 0.9512,
+      "step": 1452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454937188068158e-06,
+      "loss": 0.9387,
+      "step": 1453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443100130659876e-06,
+      "loss": 0.9846,
+      "step": 1454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431263855933078e-06,
+      "loss": 0.9071,
+      "step": 1455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41942838052278e-06,
+      "loss": 0.9669,
+      "step": 1456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407593721062858e-06,
+      "loss": 0.9362,
+      "step": 1457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395759894186054e-06,
+      "loss": 0.9075,
+      "step": 1458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383926916523933e-06,
+      "loss": 0.9581,
+      "step": 1459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.9332,
+      "step": 1460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360263575364015e-06,
+      "loss": 0.926,
+      "step": 1461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34843324512329e-06,
+      "loss": 0.9155,
+      "step": 1462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336603830611341e-06,
+      "loss": 0.973,
+      "step": 1463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324775348453543e-06,
+      "loss": 0.9528,
+      "step": 1464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312947815273942e-06,
+      "loss": 0.9634,
+      "step": 1465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301121247695265e-06,
+      "loss": 0.965,
+      "step": 1466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28929566233887e-06,
+      "loss": 0.9566,
+      "step": 1467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277471075824747e-06,
+      "loss": 0.9326,
+      "step": 1468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265647504771477e-06,
+      "loss": 0.9974,
+      "step": 1469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253824965796203e-06,
+      "loss": 0.9401,
+      "step": 1470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242003475514638e-06,
+      "loss": 0.9308,
+      "step": 1471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230183050541001e-06,
+      "loss": 0.9283,
+      "step": 1472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21836370748803e-06,
+      "loss": 0.9568,
+      "step": 1473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206545462966935e-06,
+      "loss": 0.9063,
+      "step": 1474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194728333587379e-06,
+      "loss": 0.9644,
+      "step": 1475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18291233595747e-06,
+      "loss": 0.9206,
+      "step": 1476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171097486683703e-06,
+      "loss": 0.9241,
+      "step": 1477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159283802370981e-06,
+      "loss": 0.9857,
+      "step": 1478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147471299622566e-06,
+      "loss": 0.9422,
+      "step": 1479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.135659995040046e-06,
+      "loss": 0.9097,
+      "step": 1480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123849905223341e-06,
+      "loss": 0.9337,
+      "step": 1481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 0.9598,
+      "step": 1482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100233436278458e-06,
+      "loss": 0.9507,
+      "step": 1483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.088427090341483e-06,
+      "loss": 0.95,
+      "step": 1484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076622025552662e-06,
+      "loss": 0.91,
+      "step": 1485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064818258503145e-06,
+      "loss": 0.966,
+      "step": 1486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053015805782249e-06,
+      "loss": 1.0317,
+      "step": 1487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041214683977449e-06,
+      "loss": 0.9423,
+      "step": 1488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.029414909674346e-06,
+      "loss": 0.9768,
+      "step": 1489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017616499456647e-06,
+      "loss": 0.9355,
+      "step": 1490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00581946990615e-06,
+      "loss": 0.9704,
+      "step": 1491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994023837602694e-06,
+      "loss": 0.9751,
+      "step": 1492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98222961912417e-06,
+      "loss": 0.9052,
+      "step": 1493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970436831046484e-06,
+      "loss": 0.9488,
+      "step": 1494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958645489943519e-06,
+      "loss": 1.0113,
+      "step": 1495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946855612387134e-06,
+      "loss": 1.0079,
+      "step": 1496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935067214947126e-06,
+      "loss": 0.9656,
+      "step": 1497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923280314191215e-06,
+      "loss": 1.0022,
+      "step": 1498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91149492668502e-06,
+      "loss": 0.937,
+      "step": 1499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.899711068992023e-06,
+      "loss": 0.9967,
+      "step": 1500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887928757673567e-06,
+      "loss": 0.9761,
+      "step": 1501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876148009288813e-06,
+      "loss": 0.9066,
+      "step": 1502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864368840394733e-06,
+      "loss": 0.9504,
+      "step": 1503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852591267546077e-06,
+      "loss": 0.9224,
+      "step": 1504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.840815307295347e-06,
+      "loss": 0.9296,
+      "step": 1505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829040976192789e-06,
+      "loss": 0.925,
+      "step": 1506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817268290786343e-06,
+      "loss": 0.9271,
+      "step": 1507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805497267621653e-06,
+      "loss": 0.9398,
+      "step": 1508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793727923242017e-06,
+      "loss": 1.084,
+      "step": 1509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781960274188376e-06,
+      "loss": 0.9572,
+      "step": 1510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77019433699929e-06,
+      "loss": 0.9604,
+      "step": 1511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758430128210908e-06,
+      "loss": 0.9576,
+      "step": 1512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746667664356957e-06,
+      "loss": 0.9312,
+      "step": 1513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734906961968713e-06,
+      "loss": 0.9264,
+      "step": 1514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723148037574958e-06,
+      "loss": 0.9616,
+      "step": 1515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711390907702001e-06,
+      "loss": 0.9027,
+      "step": 1516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69963558887361e-06,
+      "loss": 0.9581,
+      "step": 1517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687882097611016e-06,
+      "loss": 0.9426,
+      "step": 1518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676130450432882e-06,
+      "loss": 0.9365,
+      "step": 1519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664380663855272e-06,
+      "loss": 0.9628,
+      "step": 1520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65263275439165e-06,
+      "loss": 0.9661,
+      "step": 1521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64088673855282e-06,
+      "loss": 0.9414,
+      "step": 1522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629142632846941e-06,
+      "loss": 0.9742,
+      "step": 1523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617400453779487e-06,
+      "loss": 0.9279,
+      "step": 1524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.605660217853217e-06,
+      "loss": 0.9138,
+      "step": 1525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593921941568165e-06,
+      "loss": 0.9818,
+      "step": 1526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582185641421601e-06,
+      "loss": 0.9492,
+      "step": 1527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570451333908033e-06,
+      "loss": 0.9068,
+      "step": 1528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55871903551916e-06,
+      "loss": 0.9502,
+      "step": 1529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546988762743852e-06,
+      "loss": 0.9343,
+      "step": 1530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.535260532068144e-06,
+      "loss": 0.9402,
+      "step": 1531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 0.9498,
+      "step": 1532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511810262945257e-06,
+      "loss": 0.9297,
+      "step": 1533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5000882574557e-06,
+      "loss": 0.9244,
+      "step": 1534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48836835998092e-06,
+      "loss": 0.8626,
+      "step": 1535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476650586992372e-06,
+      "loss": 0.9524,
+      "step": 1536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46493495495851e-06,
+      "loss": 0.9171,
+      "step": 1537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453221480344791e-06,
+      "loss": 0.9478,
+      "step": 1538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441510179613637e-06,
+      "loss": 0.942,
+      "step": 1539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429801069224411e-06,
+      "loss": 0.92,
+      "step": 1540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418094165633397e-06,
+      "loss": 1.0342,
+      "step": 1541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406389485293786e-06,
+      "loss": 0.9303,
+      "step": 1542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394687044655633e-06,
+      "loss": 0.9428,
+      "step": 1543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382986860165859e-06,
+      "loss": 0.9199,
+      "step": 1544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371288948268195e-06,
+      "loss": 0.9627,
+      "step": 1545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359593325403195e-06,
+      "loss": 0.9893,
+      "step": 1546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347900008008194e-06,
+      "loss": 0.9245,
+      "step": 1547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336209012517273e-06,
+      "loss": 0.9584,
+      "step": 1548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324520355361271e-06,
+      "loss": 0.9873,
+      "step": 1549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31283405296772e-06,
+      "loss": 0.9347,
+      "step": 1550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301150121760855e-06,
+      "loss": 0.9144,
+      "step": 1551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289468578161581e-06,
+      "loss": 0.9373,
+      "step": 1552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277789438587432e-06,
+      "loss": 0.9936,
+      "step": 1553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266112719452579e-06,
+      "loss": 0.9951,
+      "step": 1554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254438437167783e-06,
+      "loss": 0.9395,
+      "step": 1555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242766608140383e-06,
+      "loss": 0.9433,
+      "step": 1556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231097248774273e-06,
+      "loss": 0.9578,
+      "step": 1557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219430375469863e-06,
+      "loss": 0.8647,
+      "step": 1558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207766004624093e-06,
+      "loss": 0.9651,
+      "step": 1559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19610415263036e-06,
+      "loss": 0.9686,
+      "step": 1560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184444835878534e-06,
+      "loss": 0.9446,
+      "step": 1561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172788070754927e-06,
+      "loss": 0.9423,
+      "step": 1562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161133873642253e-06,
+      "loss": 0.9605,
+      "step": 1563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149482260919625e-06,
+      "loss": 0.8923,
+      "step": 1564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137833248962516e-06,
+      "loss": 0.911,
+      "step": 1565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 0.9403,
+      "step": 1566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114543092828485e-06,
+      "loss": 0.8857,
+      "step": 1567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102901981384146e-06,
+      "loss": 0.9268,
+      "step": 1568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091263536170463e-06,
+      "loss": 0.9073,
+      "step": 1569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079627773544403e-06,
+      "loss": 0.9143,
+      "step": 1570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.067994709859168e-06,
+      "loss": 0.9443,
+      "step": 1571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056364361464176e-06,
+      "loss": 0.9231,
+      "step": 1572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044736744705006e-06,
+      "loss": 0.9436,
+      "step": 1573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033111875923421e-06,
+      "loss": 0.9385,
+      "step": 1574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02148977145731e-06,
+      "loss": 0.9287,
+      "step": 1575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009870447640676e-06,
+      "loss": 0.9776,
+      "step": 1576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99825392080362e-06,
+      "loss": 0.9464,
+      "step": 1577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986640207272312e-06,
+      "loss": 0.9204,
+      "step": 1578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975029323368966e-06,
+      "loss": 0.9654,
+      "step": 1579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963421285411812e-06,
+      "loss": 0.9315,
+      "step": 1580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95181610971509e-06,
+      "loss": 0.983,
+      "step": 1581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 0.9634,
+      "step": 1582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92861441033976e-06,
+      "loss": 0.8995,
+      "step": 1583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91701791926942e-06,
+      "loss": 0.9507,
+      "step": 1584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905424355676002e-06,
+      "loss": 0.9496,
+      "step": 1585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893833735853404e-06,
+      "loss": 0.9501,
+      "step": 1586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882246076091385e-06,
+      "loss": 0.8886,
+      "step": 1587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870661392675535e-06,
+      "loss": 0.9216,
+      "step": 1588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859079701887275e-06,
+      "loss": 0.9896,
+      "step": 1589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847501020003806e-06,
+      "loss": 0.9044,
+      "step": 1590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835925363298114e-06,
+      "loss": 0.9279,
+      "step": 1591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824352748038924e-06,
+      "loss": 0.9412,
+      "step": 1592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812783190490686e-06,
+      "loss": 0.9108,
+      "step": 1593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801216706913563e-06,
+      "loss": 0.935,
+      "step": 1594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789653313563382e-06,
+      "loss": 0.9146,
+      "step": 1595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778093026691636e-06,
+      "loss": 0.8896,
+      "step": 1596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76653586254546e-06,
+      "loss": 0.9127,
+      "step": 1597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75498183736758e-06,
+      "loss": 0.942,
+      "step": 1598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743430967396332e-06,
+      "loss": 0.9355,
+      "step": 1599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731883268865601e-06,
+      "loss": 0.9547,
+      "step": 1600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720338758004824e-06,
+      "loss": 0.9639,
+      "step": 1601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70879745103896e-06,
+      "loss": 0.9653,
+      "step": 1602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697259364188449e-06,
+      "loss": 1.0047,
+      "step": 1603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685724513669227e-06,
+      "loss": 0.8861,
+      "step": 1604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674192915692665e-06,
+      "loss": 0.9646,
+      "step": 1605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662664586465574e-06,
+      "loss": 0.9586,
+      "step": 1606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651139542190164e-06,
+      "loss": 0.9406,
+      "step": 1607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63961779906403e-06,
+      "loss": 0.9198,
+      "step": 1608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628099373280133e-06,
+      "loss": 0.9368,
+      "step": 1609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616584281026759e-06,
+      "loss": 0.9557,
+      "step": 1610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605072538487519e-06,
+      "loss": 0.8875,
+      "step": 1611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593564161841318e-06,
+      "loss": 0.9212,
+      "step": 1612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5820591672623215e-06,
+      "loss": 0.9623,
+      "step": 1613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57055757091995e-06,
+      "loss": 0.9066,
+      "step": 1614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559059388978841e-06,
+      "loss": 0.93,
+      "step": 1615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5475646375988395e-06,
+      "loss": 0.9042,
+      "step": 1616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536073332934972e-06,
+      "loss": 0.9092,
+      "step": 1617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524585491137404e-06,
+      "loss": 0.8946,
+      "step": 1618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513101128351454e-06,
+      "loss": 0.9042,
+      "step": 1619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501620260717538e-06,
+      "loss": 0.919,
+      "step": 1620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490142904371167e-06,
+      "loss": 0.926,
+      "step": 1621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478669075442917e-06,
+      "loss": 0.924,
+      "step": 1622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.467198790058399e-06,
+      "loss": 0.9713,
+      "step": 1623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455732064338255e-06,
+      "loss": 0.9128,
+      "step": 1624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4442689143981094e-06,
+      "loss": 0.9368,
+      "step": 1625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432809356348576e-06,
+      "loss": 0.9255,
+      "step": 1626
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421353406295215e-06,
+      "loss": 1.02,
+      "step": 1627
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409901080338512e-06,
+      "loss": 0.9437,
+      "step": 1628
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398452394573868e-06,
+      "loss": 0.9346,
+      "step": 1629
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38700736509155e-06,
+      "loss": 0.9536,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375566007976709e-06,
+      "loss": 0.9194,
+      "step": 1631
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 0.9548,
+      "step": 1632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3526943751641875e-06,
+      "loss": 0.8814,
+      "step": 1633
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34126413161089e-06,
+      "loss": 0.9479,
+      "step": 1634
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32983762471379e-06,
+      "loss": 0.919,
+      "step": 1635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318414870531996e-06,
+      "loss": 0.8816,
+      "step": 1636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306995885119349e-06,
+      "loss": 0.9611,
+      "step": 1637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2955806845243734e-06,
+      "loss": 0.9196,
+      "step": 1638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2841692847903025e-06,
+      "loss": 0.9145,
+      "step": 1639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272761701955e-06,
+      "loss": 0.9495,
+      "step": 1640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261357952050985e-06,
+      "loss": 0.9203,
+      "step": 1641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249958051105383e-06,
+      "loss": 0.9294,
+      "step": 1642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.238562015139907e-06,
+      "loss": 0.9304,
+      "step": 1643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227169860170845e-06,
+      "loss": 0.9196,
+      "step": 1644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21578160220902e-06,
+      "loss": 0.9409,
+      "step": 1645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2043972572597855e-06,
+      "loss": 0.9211,
+      "step": 1646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1930168413230025e-06,
+      "loss": 0.9589,
+      "step": 1647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181640370392994e-06,
+      "loss": 0.933,
+      "step": 1648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170267860458551e-06,
+      "loss": 0.9294,
+      "step": 1649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1588993275028885e-06,
+      "loss": 0.8714,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1475347875036405e-06,
+      "loss": 0.9338,
+      "step": 1651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136174256432828e-06,
+      "loss": 0.979,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.124817750256828e-06,
+      "loss": 0.9253,
+      "step": 1653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113465284936378e-06,
+      "loss": 0.9023,
+      "step": 1654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102116876426518e-06,
+      "loss": 0.9213,
+      "step": 1655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090772540676598e-06,
+      "loss": 0.9235,
+      "step": 1656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079432293630244e-06,
+      "loss": 0.9441,
+      "step": 1657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0680961512253254e-06,
+      "loss": 0.9346,
+      "step": 1658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0567641293939615e-06,
+      "loss": 0.9339,
+      "step": 1659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045436244062458e-06,
+      "loss": 0.9425,
+      "step": 1660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0341125111513185e-06,
+      "loss": 0.9529,
+      "step": 1661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022792946575222e-06,
+      "loss": 0.9274,
+      "step": 1662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011477566242965e-06,
+      "loss": 0.8801,
+      "step": 1663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000166386057483e-06,
+      "loss": 0.9342,
+      "step": 1664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988859421915797e-06,
+      "loss": 0.9737,
+      "step": 1665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977556689709009e-06,
+      "loss": 0.9426,
+      "step": 1666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966258205322274e-06,
+      "loss": 0.9194,
+      "step": 1667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954963984634768e-06,
+      "loss": 0.9188,
+      "step": 1668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943674043519688e-06,
+      "loss": 0.9562,
+      "step": 1669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932388397844204e-06,
+      "loss": 0.8998,
+      "step": 1670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921107063469457e-06,
+      "loss": 0.9592,
+      "step": 1671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.9746,
+      "step": 1672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898557392036409e-06,
+      "loss": 0.9276,
+      "step": 1673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887289086670004e-06,
+      "loss": 0.8911,
+      "step": 1674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876025155988069e-06,
+      "loss": 0.9299,
+      "step": 1675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864765615821231e-06,
+      "loss": 0.8878,
+      "step": 1676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853510481993939e-06,
+      "loss": 0.9229,
+      "step": 1677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842259770324447e-06,
+      "loss": 0.9187,
+      "step": 1678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831013496624798e-06,
+      "loss": 0.9525,
+      "step": 1679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819771676700794e-06,
+      "loss": 0.9405,
+      "step": 1680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808534326351981e-06,
+      "loss": 0.8931,
+      "step": 1681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 0.9891,
+      "step": 1682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786073097546681e-06,
+      "loss": 0.8901,
+      "step": 1683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774849250657784e-06,
+      "loss": 0.9514,
+      "step": 1684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763629936479217e-06,
+      "loss": 0.9333,
+      "step": 1685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752415170778894e-06,
+      "loss": 0.9122,
+      "step": 1686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741204969318343e-06,
+      "loss": 0.8927,
+      "step": 1687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729999347852665e-06,
+      "loss": 0.9439,
+      "step": 1688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718798322130535e-06,
+      "loss": 1.0043,
+      "step": 1689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707601907894159e-06,
+      "loss": 0.9509,
+      "step": 1690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6964101208792696e-06,
+      "loss": 0.9266,
+      "step": 1691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6852229768150976e-06,
+      "loss": 0.8957,
+      "step": 1692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6740404914243385e-06,
+      "loss": 0.9532,
+      "step": 1693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.662862680423153e-06,
+      "loss": 1.0174,
+      "step": 1694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.651689559521119e-06,
+      "loss": 0.9177,
+      "step": 1695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640521144421237e-06,
+      "loss": 0.8962,
+      "step": 1696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629357450819885e-06,
+      "loss": 0.898,
+      "step": 1697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618198494406802e-06,
+      "loss": 0.8944,
+      "step": 1698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607044290865079e-06,
+      "loss": 0.9295,
+      "step": 1699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595894855871119e-06,
+      "loss": 0.9416,
+      "step": 1700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584750205094628e-06,
+      "loss": 0.9303,
+      "step": 1701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573610354198587e-06,
+      "loss": 0.9443,
+      "step": 1702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562475318839232e-06,
+      "loss": 0.9066,
+      "step": 1703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55134511466603e-06,
+      "loss": 0.9321,
+      "step": 1704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540219757321653e-06,
+      "loss": 0.8829,
+      "step": 1705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52909926244197e-06,
+      "loss": 0.9583,
+      "step": 1706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517983645656014e-06,
+      "loss": 0.9283,
+      "step": 1707
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506872922585956e-06,
+      "loss": 0.9362,
+      "step": 1708
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4957671088470976e-06,
+      "loss": 0.8978,
+      "step": 1709
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484666220047835e-06,
+      "loss": 0.9294,
+      "step": 1710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473570271789644e-06,
+      "loss": 0.906,
+      "step": 1711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4624792796670624e-06,
+      "loss": 0.9496,
+      "step": 1712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45139325926765e-06,
+      "loss": 0.9278,
+      "step": 1713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440312226171992e-06,
+      "loss": 0.8901,
+      "step": 1714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429236195953656e-06,
+      "loss": 0.8874,
+      "step": 1715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418165184179183e-06,
+      "loss": 0.9211,
+      "step": 1716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4070992064080606e-06,
+      "loss": 0.8977,
+      "step": 1717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396038278192698e-06,
+      "loss": 0.9143,
+      "step": 1718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384982415078416e-06,
+      "loss": 0.921,
+      "step": 1719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3739316326034005e-06,
+      "loss": 0.9039,
+      "step": 1720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362885946298714e-06,
+      "loss": 0.964,
+      "step": 1721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35184537168825e-06,
+      "loss": 0.9378,
+      "step": 1722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340809924288713e-06,
+      "loss": 0.9447,
+      "step": 1723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329779619609615e-06,
+      "loss": 0.9819,
+      "step": 1724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318754473153221e-06,
+      "loss": 0.9284,
+      "step": 1725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307734500414564e-06,
+      "loss": 0.9606,
+      "step": 1726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296719716881401e-06,
+      "loss": 0.9306,
+      "step": 1727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285710138034187e-06,
+      "loss": 0.9077,
+      "step": 1728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274705779346074e-06,
+      "loss": 0.9791,
+      "step": 1729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263706656282869e-06,
+      "loss": 0.9225,
+      "step": 1730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252712784303025e-06,
+      "loss": 0.9704,
+      "step": 1731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 0.893,
+      "step": 1732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230740855390316e-06,
+      "loss": 0.8992,
+      "step": 1733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219762829337367e-06,
+      "loss": 0.9218,
+      "step": 1734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.208790116127569e-06,
+      "loss": 0.9227,
+      "step": 1735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197822731182259e-06,
+      "loss": 0.9052,
+      "step": 1736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186860689915286e-06,
+      "loss": 0.893,
+      "step": 1737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1759040077329845e-06,
+      "loss": 0.9351,
+      "step": 1738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164952700034164e-06,
+      "loss": 0.9612,
+      "step": 1739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154006782210066e-06,
+      "loss": 0.926,
+      "step": 1740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143066269644371e-06,
+      "loss": 0.911,
+      "step": 1741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.132131177713165e-06,
+      "loss": 0.882,
+      "step": 1742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1212015217849e-06,
+      "loss": 0.9432,
+      "step": 1743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1102773172204034e-06,
+      "loss": 0.9302,
+      "step": 1744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099358579372832e-06,
+      "loss": 0.9658,
+      "step": 1745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0884453235876615e-06,
+      "loss": 1.0391,
+      "step": 1746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07753756520267e-06,
+      "loss": 0.9325,
+      "step": 1747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066635319547895e-06,
+      "loss": 0.8871,
+      "step": 1748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055738601945643e-06,
+      "loss": 0.9203,
+      "step": 1749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0448474277104365e-06,
+      "loss": 0.9325,
+      "step": 1750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033961812149014e-06,
+      "loss": 0.9129,
+      "step": 1751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023081770560307e-06,
+      "loss": 0.9184,
+      "step": 1752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012207318235401e-06,
+      "loss": 0.9137,
+      "step": 1753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0013384704575406e-06,
+      "loss": 0.9022,
+      "step": 1754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.990475242502077e-06,
+      "loss": 0.9456,
+      "step": 1755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9796176496364735e-06,
+      "loss": 0.893,
+      "step": 1756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96876570712028e-06,
+      "loss": 0.9106,
+      "step": 1757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957919430205088e-06,
+      "loss": 0.8934,
+      "step": 1758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947078834134541e-06,
+      "loss": 0.9199,
+      "step": 1759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93624393414429e-06,
+      "loss": 0.9252,
+      "step": 1760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.925414745461986e-06,
+      "loss": 0.9191,
+      "step": 1761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9145912833072535e-06,
+      "loss": 0.9153,
+      "step": 1762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903773562891661e-06,
+      "loss": 0.9426,
+      "step": 1763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892961599418716e-06,
+      "loss": 0.9188,
+      "step": 1764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882155408083829e-06,
+      "loss": 1.0398,
+      "step": 1765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871355004074304e-06,
+      "loss": 0.9184,
+      "step": 1766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860560402569308e-06,
+      "loss": 0.8871,
+      "step": 1767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849771618739852e-06,
+      "loss": 1.0124,
+      "step": 1768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838988667748777e-06,
+      "loss": 0.9233,
+      "step": 1769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82821156475071e-06,
+      "loss": 0.9273,
+      "step": 1770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81744032489208e-06,
+      "loss": 0.9026,
+      "step": 1771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8066749633110675e-06,
+      "loss": 0.9126,
+      "step": 1772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795915495137582e-06,
+      "loss": 0.9841,
+      "step": 1773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785161935493266e-06,
+      "loss": 0.8776,
+      "step": 1774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7744142994914464e-06,
+      "loss": 0.8947,
+      "step": 1775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.763672602237129e-06,
+      "loss": 0.9337,
+      "step": 1776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7529368588269745e-06,
+      "loss": 0.9135,
+      "step": 1777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 0.902,
+      "step": 1778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7314832938839335e-06,
+      "loss": 0.9386,
+      "step": 1779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72076550250244e-06,
+      "loss": 0.8954,
+      "step": 1780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710053725267857e-06,
+      "loss": 0.8983,
+      "step": 1781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 0.8859,
+      "step": 1782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688648273449391e-06,
+      "loss": 0.8995,
+      "step": 1783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677954628949281e-06,
+      "loss": 0.9292,
+      "step": 1784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6672670587635905e-06,
+      "loss": 0.8901,
+      "step": 1785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656585577912908e-06,
+      "loss": 0.9177,
+      "step": 1786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645910201409268e-06,
+      "loss": 0.9249,
+      "step": 1787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635240944256113e-06,
+      "loss": 0.9468,
+      "step": 1788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6245778214483e-06,
+      "loss": 0.9026,
+      "step": 1789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613920847972052e-06,
+      "loss": 0.9564,
+      "step": 1790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603270038804958e-06,
+      "loss": 0.9277,
+      "step": 1791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592625408915939e-06,
+      "loss": 0.9428,
+      "step": 1792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581986973265234e-06,
+      "loss": 0.9361,
+      "step": 1793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571354746804383e-06,
+      "loss": 0.9044,
+      "step": 1794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560728744476182e-06,
+      "loss": 0.8643,
+      "step": 1795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550108981214692e-06,
+      "loss": 0.8755,
+      "step": 1796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53949547194521e-06,
+      "loss": 0.9325,
+      "step": 1797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5288882315842265e-06,
+      "loss": 0.9619,
+      "step": 1798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518287275039438e-06,
+      "loss": 0.8965,
+      "step": 1799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507692617209701e-06,
+      "loss": 0.8906,
+      "step": 1800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497104272985022e-06,
+      "loss": 0.927,
+      "step": 1801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486522257246538e-06,
+      "loss": 0.9224,
+      "step": 1802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4759465848664825e-06,
+      "loss": 0.9084,
+      "step": 1803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465377270708183e-06,
+      "loss": 0.9073,
+      "step": 1804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454814329626021e-06,
+      "loss": 0.9264,
+      "step": 1805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4442577764654334e-06,
+      "loss": 0.929,
+      "step": 1806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43370762606287e-06,
+      "loss": 0.9247,
+      "step": 1807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423163893245786e-06,
+      "loss": 0.9186,
+      "step": 1808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412626592832617e-06,
+      "loss": 0.9152,
+      "step": 1809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402095739632763e-06,
+      "loss": 0.9298,
+      "step": 1810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391571348446547e-06,
+      "loss": 0.9022,
+      "step": 1811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.381053434065229e-06,
+      "loss": 0.892,
+      "step": 1812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370542011270952e-06,
+      "loss": 0.8692,
+      "step": 1813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360037094836745e-06,
+      "loss": 0.9309,
+      "step": 1814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349538699526489e-06,
+      "loss": 0.9102,
+      "step": 1815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339046840094899e-06,
+      "loss": 0.921,
+      "step": 1816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328561531287513e-06,
+      "loss": 0.9023,
+      "step": 1817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318082787840646e-06,
+      "loss": 0.9342,
+      "step": 1818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307610624481397e-06,
+      "loss": 0.9281,
+      "step": 1819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297145055927622e-06,
+      "loss": 0.8769,
+      "step": 1820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286686096887894e-06,
+      "loss": 0.9202,
+      "step": 1821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276233762061507e-06,
+      "loss": 0.8706,
+      "step": 1822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265788066138444e-06,
+      "loss": 0.8705,
+      "step": 1823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255349023799357e-06,
+      "loss": 0.907,
+      "step": 1824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244916649715549e-06,
+      "loss": 0.9249,
+      "step": 1825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234490958548941e-06,
+      "loss": 0.8775,
+      "step": 1826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224071964952078e-06,
+      "loss": 0.9256,
+      "step": 1827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213659683568073e-06,
+      "loss": 0.9384,
+      "step": 1828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203254129030622e-06,
+      "loss": 0.8979,
+      "step": 1829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192855315963959e-06,
+      "loss": 0.9131,
+      "step": 1830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1824632589828465e-06,
+      "loss": 0.9393,
+      "step": 1831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172077972692553e-06,
+      "loss": 0.9372,
+      "step": 1832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161699471688821e-06,
+      "loss": 0.9457,
+      "step": 1833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15132777055787e-06,
+      "loss": 0.936,
+      "step": 1834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1409628838763595e-06,
+      "loss": 0.913,
+      "step": 1835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130604826211361e-06,
+      "loss": 0.9536,
+      "step": 1836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120253612120363e-06,
+      "loss": 0.9244,
+      "step": 1837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109909256151227e-06,
+      "loss": 0.9182,
+      "step": 1838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099571772842181e-06,
+      "loss": 0.8649,
+      "step": 1839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089241176721794e-06,
+      "loss": 0.8639,
+      "step": 1840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078917482308946e-06,
+      "loss": 0.9035,
+      "step": 1841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068600704112832e-06,
+      "loss": 0.9402,
+      "step": 1842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05829085663291e-06,
+      "loss": 0.9186,
+      "step": 1843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047987954358912e-06,
+      "loss": 0.9393,
+      "step": 1844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037692011770805e-06,
+      "loss": 0.893,
+      "step": 1845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02740304333877e-06,
+      "loss": 0.91,
+      "step": 1846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017121063523194e-06,
+      "loss": 0.9291,
+      "step": 1847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006846086774631e-06,
+      "loss": 0.899,
+      "step": 1848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996578127533802e-06,
+      "loss": 0.8887,
+      "step": 1849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9863172002315675e-06,
+      "loss": 0.8316,
+      "step": 1850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976063319288892e-06,
+      "loss": 0.9439,
+      "step": 1851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965816499116849e-06,
+      "loss": 0.9126,
+      "step": 1852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9555767541165845e-06,
+      "loss": 0.9067,
+      "step": 1853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945344098679302e-06,
+      "loss": 0.9124,
+      "step": 1854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935118547186243e-06,
+      "loss": 0.8697,
+      "step": 1855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924900114008656e-06,
+      "loss": 0.9655,
+      "step": 1856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914688813507798e-06,
+      "loss": 0.9235,
+      "step": 1857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904484660034887e-06,
+      "loss": 0.9032,
+      "step": 1858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8942876679311104e-06,
+      "loss": 0.9095,
+      "step": 1859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8840978515275816e-06,
+      "loss": 0.9252,
+      "step": 1860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8739152251453344e-06,
+      "loss": 0.925,
+      "step": 1861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863739803095299e-06,
+      "loss": 0.9529,
+      "step": 1862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853571599678269e-06,
+      "loss": 0.8663,
+      "step": 1863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8434106291849035e-06,
+      "loss": 0.9196,
+      "step": 1864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833256905895699e-06,
+      "loss": 0.9074,
+      "step": 1865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8231104440809524e-06,
+      "loss": 0.908,
+      "step": 1866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8129712580007725e-06,
+      "loss": 0.8598,
+      "step": 1867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802839361905021e-06,
+      "loss": 0.956,
+      "step": 1868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792714770033337e-06,
+      "loss": 0.9297,
+      "step": 1869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782597496615088e-06,
+      "loss": 0.9217,
+      "step": 1870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772487555869341e-06,
+      "loss": 0.8574,
+      "step": 1871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762384962004877e-06,
+      "loss": 0.8703,
+      "step": 1872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752289729220135e-06,
+      "loss": 0.923,
+      "step": 1873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74220187170322e-06,
+      "loss": 0.9601,
+      "step": 1874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732121403631873e-06,
+      "loss": 0.9092,
+      "step": 1875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7220483391734325e-06,
+      "loss": 0.9297,
+      "step": 1876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71198269248486e-06,
+      "loss": 0.9046,
+      "step": 1877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701924477712663e-06,
+      "loss": 0.9016,
+      "step": 1878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6918737089929215e-06,
+      "loss": 0.911,
+      "step": 1879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681830400451249e-06,
+      "loss": 0.9357,
+      "step": 1880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671794566202764e-06,
+      "loss": 0.8883,
+      "step": 1881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661766220352098e-06,
+      "loss": 0.8712,
+      "step": 1882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.651745376993335e-06,
+      "loss": 0.8922,
+      "step": 1883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 0.9023,
+      "step": 1884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6317262540751875e-06,
+      "loss": 0.903,
+      "step": 1885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621728002651194e-06,
+      "loss": 0.8992,
+      "step": 1886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611737309989861e-06,
+      "loss": 0.8829,
+      "step": 1887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6017541901323605e-06,
+      "loss": 0.9606,
+      "step": 1888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591778657109229e-06,
+      "loss": 0.9184,
+      "step": 1889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581810724940343e-06,
+      "loss": 0.9756,
+      "step": 1890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571850407634884e-06,
+      "loss": 0.8859,
+      "step": 1891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561897719191349e-06,
+      "loss": 0.9241,
+      "step": 1892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551952673597498e-06,
+      "loss": 0.9734,
+      "step": 1893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542015284830358e-06,
+      "loss": 0.9014,
+      "step": 1894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532085566856193e-06,
+      "loss": 0.838,
+      "step": 1895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5221635336304825e-06,
+      "loss": 0.8846,
+      "step": 1896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512249199097914e-06,
+      "loss": 0.8999,
+      "step": 1897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502342577192342e-06,
+      "loss": 0.9033,
+      "step": 1898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.49244368183679e-06,
+      "loss": 0.8871,
+      "step": 1899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482552526943432e-06,
+      "loss": 0.8992,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472669126413545e-06,
+      "loss": 0.8981,
+      "step": 1901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4627934941375185e-06,
+      "loss": 0.9476,
+      "step": 1902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452925643994819e-06,
+      "loss": 0.8997,
+      "step": 1903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443065589853977e-06,
+      "loss": 0.8875,
+      "step": 1904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433213345572578e-06,
+      "loss": 0.9631,
+      "step": 1905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423368924997208e-06,
+      "loss": 0.9118,
+      "step": 1906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413532341963477e-06,
+      "loss": 0.8819,
+      "step": 1907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403703610295972e-06,
+      "loss": 0.9407,
+      "step": 1908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3938827438082475e-06,
+      "loss": 0.8607,
+      "step": 1909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3840697563028076e-06,
+      "loss": 0.9018,
+      "step": 1910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3742646615710715e-06,
+      "loss": 0.9161,
+      "step": 1911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36446747339338e-06,
+      "loss": 0.9204,
+      "step": 1912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354678205538947e-06,
+      "loss": 0.9196,
+      "step": 1913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344896871765868e-06,
+      "loss": 0.9427,
+      "step": 1914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335123485821081e-06,
+      "loss": 0.8889,
+      "step": 1915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325358061440356e-06,
+      "loss": 0.91,
+      "step": 1916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315600612348278e-06,
+      "loss": 0.8881,
+      "step": 1917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30585115225821e-06,
+      "loss": 0.9346,
+      "step": 1918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296109694872299e-06,
+      "loss": 0.9123,
+      "step": 1919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2863762538814465e-06,
+      "loss": 0.8954,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276650842965274e-06,
+      "loss": 0.8772,
+      "step": 1921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2669334757921284e-06,
+      "loss": 0.9135,
+      "step": 1922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257224166019051e-06,
+      "loss": 0.9606,
+      "step": 1923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2475229272917565e-06,
+      "loss": 0.8752,
+      "step": 1924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237829773244619e-06,
+      "loss": 0.9297,
+      "step": 1925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228144717500642e-06,
+      "loss": 0.9275,
+      "step": 1926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218467773671461e-06,
+      "loss": 0.8985,
+      "step": 1927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208798955357295e-06,
+      "loss": 0.8993,
+      "step": 1928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1991382761469565e-06,
+      "loss": 0.955,
+      "step": 1929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189485749617813e-06,
+      "loss": 0.946,
+      "step": 1930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179841389335776e-06,
+      "loss": 0.8976,
+      "step": 1931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170205208855281e-06,
+      "loss": 0.876,
+      "step": 1932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1605772217192604e-06,
+      "loss": 0.9706,
+      "step": 1933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150957441459139e-06,
+      "loss": 0.8662,
+      "step": 1934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141345881594808e-06,
+      "loss": 0.869,
+      "step": 1935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131742555634597e-06,
+      "loss": 0.9288,
+      "step": 1936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.8655,
+      "step": 1937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112560659401999e-06,
+      "loss": 0.9232,
+      "step": 1938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102982116088348e-06,
+      "loss": 0.8622,
+      "step": 1939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093411860596253e-06,
+      "loss": 0.857,
+      "step": 1940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083849906375989e-06,
+      "loss": 0.9203,
+      "step": 1941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0742962668661826e-06,
+      "loss": 0.9073,
+      "step": 1942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064750955493759e-06,
+      "loss": 0.8662,
+      "step": 1943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055213985673949e-06,
+      "loss": 0.9065,
+      "step": 1944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0456853708102575e-06,
+      "loss": 0.903,
+      "step": 1945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036165124294445e-06,
+      "loss": 0.903,
+      "step": 1946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026653259506518e-06,
+      "loss": 0.8953,
+      "step": 1947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017149789814689e-06,
+      "loss": 0.8628,
+      "step": 1948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007654728575383e-06,
+      "loss": 0.9336,
+      "step": 1949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998168089133211e-06,
+      "loss": 0.868,
+      "step": 1950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988689884820932e-06,
+      "loss": 0.9431,
+      "step": 1951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979220128959463e-06,
+      "loss": 0.8993,
+      "step": 1952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969758834857846e-06,
+      "loss": 0.853,
+      "step": 1953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960306015813228e-06,
+      "loss": 0.8616,
+      "step": 1954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950861685110848e-06,
+      "loss": 0.8831,
+      "step": 1955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941425856024007e-06,
+      "loss": 0.8484,
+      "step": 1956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931998541814069e-06,
+      "loss": 0.8823,
+      "step": 1957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92257975573042e-06,
+      "loss": 0.9187,
+      "step": 1958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913169511010469e-06,
+      "loss": 0.8621,
+      "step": 1959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9037678208796144e-06,
+      "loss": 0.8938,
+      "step": 1960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8943746985512396e-06,
+      "loss": 0.9169,
+      "step": 1961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884990157226683e-06,
+      "loss": 0.9243,
+      "step": 1962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875614210095214e-06,
+      "loss": 0.9145,
+      "step": 1963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866246870334036e-06,
+      "loss": 0.9226,
+      "step": 1964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8568881511082535e-06,
+      "loss": 0.9297,
+      "step": 1965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847538065570847e-06,
+      "loss": 0.909,
+      "step": 1966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83819662686267e-06,
+      "loss": 0.8907,
+      "step": 1967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828863848112425e-06,
+      "loss": 0.9545,
+      "step": 1968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819539742436639e-06,
+      "loss": 0.9388,
+      "step": 1969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810224322939655e-06,
+      "loss": 0.8824,
+      "step": 1970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8009176027135986e-06,
+      "loss": 0.9694,
+      "step": 1971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7916195948383817e-06,
+      "loss": 0.8868,
+      "step": 1972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782330312381658e-06,
+      "loss": 0.8841,
+      "step": 1973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7730497683988287e-06,
+      "loss": 0.8498,
+      "step": 1974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7637779759330107e-06,
+      "loss": 0.8551,
+      "step": 1975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7545149480150224e-06,
+      "loss": 0.8853,
+      "step": 1976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7452606976633644e-06,
+      "loss": 0.8675,
+      "step": 1977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736015237884193e-06,
+      "loss": 0.8979,
+      "step": 1978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726778581671322e-06,
+      "loss": 0.9227,
+      "step": 1979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7175507420061885e-06,
+      "loss": 0.8964,
+      "step": 1980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7083317318578315e-06,
+      "loss": 0.8845,
+      "step": 1981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6991215641828903e-06,
+      "loss": 0.9188,
+      "step": 1982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6899202519255706e-06,
+      "loss": 0.9095,
+      "step": 1983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680727808017638e-06,
+      "loss": 0.9034,
+      "step": 1984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.671544245378391e-06,
+      "loss": 0.9239,
+      "step": 1985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662369576914642e-06,
+      "loss": 0.9225,
+      "step": 1986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.653203815520714e-06,
+      "loss": 0.9208,
+      "step": 1987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.644046974078397e-06,
+      "loss": 0.9088,
+      "step": 1988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6348990654569574e-06,
+      "loss": 0.9025,
+      "step": 1989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 0.8734,
+      "step": 1990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6166300980909687e-06,
+      "loss": 0.8781,
+      "step": 1991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607509065022101e-06,
+      "loss": 0.8977,
+      "step": 1992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5983970161254324e-06,
+      "loss": 0.9086,
+      "step": 1993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589293964207271e-06,
+      "loss": 0.8786,
+      "step": 1994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580199922061289e-06,
+      "loss": 0.8951,
+      "step": 1995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57111490246848e-06,
+      "loss": 0.8763,
+      "step": 1996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.562038918197168e-06,
+      "loss": 0.8716,
+      "step": 1997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5529719820029785e-06,
+      "loss": 0.861,
+      "step": 1998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5439141066288176e-06,
+      "loss": 0.8861,
+      "step": 1999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5348653048048598e-06,
+      "loss": 0.9046,
+      "step": 2000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5258255892485205e-06,
+      "loss": 0.9256,
+      "step": 2001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5167949726644545e-06,
+      "loss": 0.9276,
+      "step": 2002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5077734677445196e-06,
+      "loss": 0.9697,
+      "step": 2003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4987610871677746e-06,
+      "loss": 0.852,
+      "step": 2004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489757843600452e-06,
+      "loss": 0.903,
+      "step": 2005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4807637496959433e-06,
+      "loss": 0.9026,
+      "step": 2006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4717788180947855e-06,
+      "loss": 0.8969,
+      "step": 2007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4628030614246266e-06,
+      "loss": 0.8955,
+      "step": 2008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45383649230023e-06,
+      "loss": 0.894,
+      "step": 2009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4448791233234467e-06,
+      "loss": 0.9125,
+      "step": 2010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4359309670831907e-06,
+      "loss": 0.8666,
+      "step": 2011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4269920361554342e-06,
+      "loss": 0.8972,
+      "step": 2012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4180623431031824e-06,
+      "loss": 0.8952,
+      "step": 2013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409141900476457e-06,
+      "loss": 0.8771,
+      "step": 2014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.400230720812282e-06,
+      "loss": 0.8967,
+      "step": 2015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3913288166346525e-06,
+      "loss": 0.9003,
+      "step": 2016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382436200454543e-06,
+      "loss": 0.8639,
+      "step": 2017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3735528847698597e-06,
+      "loss": 0.8667,
+      "step": 2018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3646788820654473e-06,
+      "loss": 0.8881,
+      "step": 2019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355814204813058e-06,
+      "loss": 0.9218,
+      "step": 2020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3469588654713404e-06,
+      "loss": 0.8991,
+      "step": 2021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.338112876485821e-06,
+      "loss": 0.8718,
+      "step": 2022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329276250288872e-06,
+      "loss": 0.907,
+      "step": 2023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3204489992997226e-06,
+      "loss": 0.8467,
+      "step": 2024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3116311359244223e-06,
+      "loss": 0.9277,
+      "step": 2025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302822672555819e-06,
+      "loss": 0.8807,
+      "step": 2026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2940236215735554e-06,
+      "loss": 0.9257,
+      "step": 2027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285233995344049e-06,
+      "loss": 0.897,
+      "step": 2028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2764538062204664e-06,
+      "loss": 0.9031,
+      "step": 2029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267683066542715e-06,
+      "loss": 0.9066,
+      "step": 2030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.258921788637415e-06,
+      "loss": 0.9034,
+      "step": 2031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250169984817897e-06,
+      "loss": 0.916,
+      "step": 2032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241427667384168e-06,
+      "loss": 0.8907,
+      "step": 2033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2326948486229105e-06,
+      "loss": 0.8966,
+      "step": 2034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2239715408074533e-06,
+      "loss": 0.9007,
+      "step": 2035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215257756197758e-06,
+      "loss": 0.9385,
+      "step": 2036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2065535070404085e-06,
+      "loss": 0.9208,
+      "step": 2037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1978588055685733e-06,
+      "loss": 0.9399,
+      "step": 2038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1891736640020156e-06,
+      "loss": 0.8822,
+      "step": 2039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.18049809454706e-06,
+      "loss": 0.8099,
+      "step": 2040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171832109396571e-06,
+      "loss": 0.8793,
+      "step": 2041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.163175720729954e-06,
+      "loss": 0.9058,
+      "step": 2042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1545289407131128e-06,
+      "loss": 0.8984,
+      "step": 2043
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1458917814984657e-06,
+      "loss": 0.9068,
+      "step": 2044
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137264255224898e-06,
+      "loss": 0.9348,
+      "step": 2045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128646374017754e-06,
+      "loss": 0.8943,
+      "step": 2046
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.120038149988832e-06,
+      "loss": 0.8736,
+      "step": 2047
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1114395952363486e-06,
+      "loss": 0.9338,
+      "step": 2048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1028507218449355e-06,
+      "loss": 0.8831,
+      "step": 2049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0942715418856184e-06,
+      "loss": 0.8853,
+      "step": 2050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085702067415801e-06,
+      "loss": 0.9516,
+      "step": 2051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0771423104792454e-06,
+      "loss": 0.8677,
+      "step": 2052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0685922831060487e-06,
+      "loss": 0.8908,
+      "step": 2053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060051997312646e-06,
+      "loss": 0.9032,
+      "step": 2054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.051521465101779e-06,
+      "loss": 0.9156,
+      "step": 2055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0430006984624704e-06,
+      "loss": 0.8922,
+      "step": 2056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0344897093700333e-06,
+      "loss": 0.906,
+      "step": 2057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025988509786023e-06,
+      "loss": 0.8903,
+      "step": 2058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017497111658254e-06,
+      "loss": 0.9143,
+      "step": 2059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0090155269207575e-06,
+      "loss": 0.9086,
+      "step": 2060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0005437674937655e-06,
+      "loss": 0.9195,
+      "step": 2061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.992081845283715e-06,
+      "loss": 0.9132,
+      "step": 2062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.983629772183204e-06,
+      "loss": 0.9212,
+      "step": 2063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975187560070998e-06,
+      "loss": 0.9576,
+      "step": 2064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966755220812002e-06,
+      "loss": 0.903,
+      "step": 2065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958332766257237e-06,
+      "loss": 0.8824,
+      "step": 2066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9499202082438493e-06,
+      "loss": 0.9051,
+      "step": 2067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941517558595056e-06,
+      "loss": 0.8635,
+      "step": 2068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933124829120163e-06,
+      "loss": 0.9017,
+      "step": 2069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9247420316145324e-06,
+      "loss": 0.9177,
+      "step": 2070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9163691778595572e-06,
+      "loss": 0.8909,
+      "step": 2071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.908006279622667e-06,
+      "loss": 0.8724,
+      "step": 2072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8996533486572985e-06,
+      "loss": 0.8566,
+      "step": 2073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8913103967028664e-06,
+      "loss": 0.884,
+      "step": 2074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8829774354847863e-06,
+      "loss": 0.878,
+      "step": 2075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8746544767144056e-06,
+      "loss": 0.8536,
+      "step": 2076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866341532089031e-06,
+      "loss": 0.8903,
+      "step": 2077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8580386132918916e-06,
+      "loss": 0.9017,
+      "step": 2078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8497457319921187e-06,
+      "loss": 0.8667,
+      "step": 2079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841462899844749e-06,
+      "loss": 0.9116,
+      "step": 2080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.833190128490683e-06,
+      "loss": 0.8495,
+      "step": 2081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8249274295566863e-06,
+      "loss": 0.8882,
+      "step": 2082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816674814655381e-06,
+      "loss": 0.8965,
+      "step": 2083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8084322953851963e-06,
+      "loss": 0.8942,
+      "step": 2084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8001998833303856e-06,
+      "loss": 0.856,
+      "step": 2085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79197759006099e-06,
+      "loss": 0.8673,
+      "step": 2086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783765427132837e-06,
+      "loss": 0.8799,
+      "step": 2087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7755634060875135e-06,
+      "loss": 0.8471,
+      "step": 2088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.767371538452348e-06,
+      "loss": 0.942,
+      "step": 2089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7591898357404066e-06,
+      "loss": 0.8654,
+      "step": 2090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.751018309450464e-06,
+      "loss": 0.8302,
+      "step": 2091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742856971066996e-06,
+      "loss": 0.8918,
+      "step": 2092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.734705832060163e-06,
+      "loss": 0.8375,
+      "step": 2093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7265649038857776e-06,
+      "loss": 0.9606,
+      "step": 2094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7184341979853203e-06,
+      "loss": 0.9164,
+      "step": 2095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 0.8861,
+      "step": 2096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702203498700201e-06,
+      "loss": 0.9253,
+      "step": 2097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6941035281265936e-06,
+      "loss": 0.8694,
+      "step": 2098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686013825448964e-06,
+      "loss": 0.9043,
+      "step": 2099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677934402036797e-06,
+      "loss": 0.9068,
+      "step": 2100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6698652692451154e-06,
+      "loss": 0.9342,
+      "step": 2101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6618064384144925e-06,
+      "loss": 0.9205,
+      "step": 2102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6537579208710185e-06,
+      "loss": 0.8571,
+      "step": 2103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6457197279262835e-06,
+      "loss": 0.8586,
+      "step": 2104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.637691870877376e-06,
+      "loss": 0.917,
+      "step": 2105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629674361006851e-06,
+      "loss": 0.8844,
+      "step": 2106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6216672095827267e-06,
+      "loss": 0.8608,
+      "step": 2107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6136704278584624e-06,
+      "loss": 0.9069,
+      "step": 2108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605684027072939e-06,
+      "loss": 0.8921,
+      "step": 2109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597708018450453e-06,
+      "loss": 0.9462,
+      "step": 2110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58974241320069e-06,
+      "loss": 0.9338,
+      "step": 2111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58178722251872e-06,
+      "loss": 0.8803,
+      "step": 2112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573842457584974e-06,
+      "loss": 0.8998,
+      "step": 2113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5659081295652298e-06,
+      "loss": 0.8529,
+      "step": 2114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5579842496106e-06,
+      "loss": 0.8781,
+      "step": 2115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.550070828857506e-06,
+      "loss": 0.8686,
+      "step": 2116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5421678784276772e-06,
+      "loss": 0.8397,
+      "step": 2117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5342754094281253e-06,
+      "loss": 0.8756,
+      "step": 2118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5263934329511265e-06,
+      "loss": 0.88,
+      "step": 2119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518521960074217e-06,
+      "loss": 0.8806,
+      "step": 2120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5106610018601686e-06,
+      "loss": 0.8775,
+      "step": 2121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502810569356976e-06,
+      "loss": 0.8795,
+      "step": 2122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494970673597842e-06,
+      "loss": 0.9576,
+      "step": 2123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4871413256011534e-06,
+      "loss": 0.8819,
+      "step": 2124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4793225363704843e-06,
+      "loss": 0.8794,
+      "step": 2125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.471514316894559e-06,
+      "loss": 0.869,
+      "step": 2126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.463716678147251e-06,
+      "loss": 0.8809,
+      "step": 2127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455929631087568e-06,
+      "loss": 0.8872,
+      "step": 2128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.448153186659623e-06,
+      "loss": 0.8991,
+      "step": 2129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440387355792638e-06,
+      "loss": 0.8528,
+      "step": 2130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4326321494009073e-06,
+      "loss": 0.8826,
+      "step": 2131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424887578383799e-06,
+      "loss": 0.9102,
+      "step": 2132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417153653625739e-06,
+      "loss": 0.8787,
+      "step": 2133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4094303859961774e-06,
+      "loss": 0.8721,
+      "step": 2134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401717786349599e-06,
+      "loss": 0.8758,
+      "step": 2135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394015865525491e-06,
+      "loss": 0.8404,
+      "step": 2136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3863246343483306e-06,
+      "loss": 0.8754,
+      "step": 2137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3786441036275764e-06,
+      "loss": 0.8619,
+      "step": 2138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.370974284157641e-06,
+      "loss": 0.8781,
+      "step": 2139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3633151867178915e-06,
+      "loss": 0.9136,
+      "step": 2140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3556668220726164e-06,
+      "loss": 0.8895,
+      "step": 2141
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3480292009710282e-06,
+      "loss": 0.8861,
+      "step": 2142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3404023341472358e-06,
+      "loss": 0.859,
+      "step": 2143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3327862323202377e-06,
+      "loss": 0.8867,
+      "step": 2144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3251809061939025e-06,
+      "loss": 0.8272,
+      "step": 2145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3175863664569454e-06,
+      "loss": 0.8808,
+      "step": 2146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.310002623782933e-06,
+      "loss": 0.918,
+      "step": 2147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3024296888302565e-06,
+      "loss": 0.9498,
+      "step": 2148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2948675722421086e-06,
+      "loss": 0.8712,
+      "step": 2149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2873162846464868e-06,
+      "loss": 0.8663,
+      "step": 2150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279775836656166e-06,
+      "loss": 0.9229,
+      "step": 2151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272246238868687e-06,
+      "loss": 0.8912,
+      "step": 2152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.264727501866344e-06,
+      "loss": 0.9014,
+      "step": 2153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2572196362161592e-06,
+      "loss": 0.9102,
+      "step": 2154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2497226524698867e-06,
+      "loss": 0.8744,
+      "step": 2155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242236561163976e-06,
+      "loss": 0.8463,
+      "step": 2156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234761372819577e-06,
+      "loss": 0.8503,
+      "step": 2157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227297097942511e-06,
+      "loss": 0.9242,
+      "step": 2158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.219843747023265e-06,
+      "loss": 0.9255,
+      "step": 2159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212401330536973e-06,
+      "loss": 0.8926,
+      "step": 2160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2049698589433954e-06,
+      "loss": 0.8978,
+      "step": 2161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1975493426869155e-06,
+      "loss": 0.843,
+      "step": 2162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190139792196523e-06,
+      "loss": 0.8914,
+      "step": 2163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1827412178857866e-06,
+      "loss": 0.8441,
+      "step": 2164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1753536301528557e-06,
+      "loss": 0.8781,
+      "step": 2165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167977039380439e-06,
+      "loss": 0.8528,
+      "step": 2166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160611455935786e-06,
+      "loss": 0.8833,
+      "step": 2167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153256890170683e-06,
+      "loss": 0.8623,
+      "step": 2168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1459133524214203e-06,
+      "loss": 0.8457,
+      "step": 2169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1385808530088024e-06,
+      "loss": 0.9077,
+      "step": 2170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131259402238107e-06,
+      "loss": 0.9117,
+      "step": 2171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1239490103990946e-06,
+      "loss": 0.9092,
+      "step": 2172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116649687765978e-06,
+      "loss": 0.8689,
+      "step": 2173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109361444597414e-06,
+      "loss": 0.8893,
+      "step": 2174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1020842911364926e-06,
+      "loss": 0.9087,
+      "step": 2175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0948182376107063e-06,
+      "loss": 0.8988,
+      "step": 2176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087563294231958e-06,
+      "loss": 0.8679,
+      "step": 2177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0803194711965356e-06,
+      "loss": 0.8624,
+      "step": 2178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0730867786850915e-06,
+      "loss": 0.8979,
+      "step": 2179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0658652268626402e-06,
+      "loss": 0.8873,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0586548258785364e-06,
+      "loss": 0.9053,
+      "step": 2181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0514555858664663e-06,
+      "loss": 0.8943,
+      "step": 2182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0442675169444303e-06,
+      "loss": 0.8026,
+      "step": 2183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037090629214721e-06,
+      "loss": 0.9051,
+      "step": 2184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.029924932763927e-06,
+      "loss": 0.8711,
+      "step": 2185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0227704376628987e-06,
+      "loss": 0.8429,
+      "step": 2186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0156271539667517e-06,
+      "loss": 0.8774,
+      "step": 2187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0084950917148403e-06,
+      "loss": 0.9027,
+      "step": 2188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.001374260930752e-06,
+      "loss": 0.8927,
+      "step": 2189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9942646716222867e-06,
+      "loss": 0.8927,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987166333781441e-06,
+      "loss": 0.8181,
+      "step": 2191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980079257384405e-06,
+      "loss": 0.8838,
+      "step": 2192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973003452391543e-06,
+      "loss": 0.8988,
+      "step": 2193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9659389287473675e-06,
+      "loss": 0.8505,
+      "step": 2194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958885696380547e-06,
+      "loss": 0.8989,
+      "step": 2195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9518437652038757e-06,
+      "loss": 0.8669,
+      "step": 2196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944813145114266e-06,
+      "loss": 0.8449,
+      "step": 2197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937793845992737e-06,
+      "loss": 0.8659,
+      "step": 2198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9307858777043876e-06,
+      "loss": 0.876,
+      "step": 2199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9237892500984022e-06,
+      "loss": 0.9142,
+      "step": 2200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916803973008018e-06,
+      "loss": 0.9137,
+      "step": 2201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.8447,
+      "step": 2202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9028675096272508e-06,
+      "loss": 0.8914,
+      "step": 2203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.895916342923534e-06,
+      "loss": 0.9081,
+      "step": 2204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.888976565908729e-06,
+      "loss": 0.8555,
+      "step": 2205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.882048188336172e-06,
+      "loss": 0.8669,
+      "step": 2206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875131219943187e-06,
+      "loss": 0.8481,
+      "step": 2207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8682256704510625e-06,
+      "loss": 0.9077,
+      "step": 2208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.861331549565033e-06,
+      "loss": 0.9004,
+      "step": 2209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8544488669742755e-06,
+      "loss": 0.8489,
+      "step": 2210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8475776323518902e-06,
+      "loss": 0.8939,
+      "step": 2211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8407178553548876e-06,
+      "loss": 0.8289,
+      "step": 2212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8338695456241806e-06,
+      "loss": 0.8522,
+      "step": 2213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8270327127845534e-06,
+      "loss": 0.8127,
+      "step": 2214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8202073664446751e-06,
+      "loss": 0.874,
+      "step": 2215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8133935161970561e-06,
+      "loss": 0.8616,
+      "step": 2216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8065911716180639e-06,
+      "loss": 0.9012,
+      "step": 2217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7998003422678867e-06,
+      "loss": 0.8637,
+      "step": 2218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7930210376905332e-06,
+      "loss": 0.8871,
+      "step": 2219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7862532674138166e-06,
+      "loss": 0.8847,
+      "step": 2220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7794970409493295e-06,
+      "loss": 0.9461,
+      "step": 2221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772752367792452e-06,
+      "loss": 0.9006,
+      "step": 2222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7660192574223234e-06,
+      "loss": 0.8906,
+      "step": 2223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7592977193018268e-06,
+      "loss": 0.8939,
+      "step": 2224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7525877628775878e-06,
+      "loss": 0.8517,
+      "step": 2225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745889397579954e-06,
+      "loss": 0.8841,
+      "step": 2226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7392026328229804e-06,
+      "loss": 0.8768,
+      "step": 2227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732527478004422e-06,
+      "loss": 0.8829,
+      "step": 2228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7258639425057089e-06,
+      "loss": 0.8595,
+      "step": 2229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7192120356919517e-06,
+      "loss": 0.9171,
+      "step": 2230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7125717669119068e-06,
+      "loss": 0.8842,
+      "step": 2231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7059431454979825e-06,
+      "loss": 0.8539,
+      "step": 2232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6993261807662153e-06,
+      "loss": 0.83,
+      "step": 2233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6927208820162589e-06,
+      "loss": 0.8806,
+      "step": 2234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6861272585313737e-06,
+      "loss": 0.8654,
+      "step": 2235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6795453195784017e-06,
+      "loss": 0.8649,
+      "step": 2236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6729750744077755e-06,
+      "loss": 0.8535,
+      "step": 2237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6664165322534887e-06,
+      "loss": 0.8674,
+      "step": 2238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6598697023330823e-06,
+      "loss": 0.8494,
+      "step": 2239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.653334593847643e-06,
+      "loss": 0.8504,
+      "step": 2240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6468112159817772e-06,
+      "loss": 0.8445,
+      "step": 2241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6402995779036146e-06,
+      "loss": 0.8771,
+      "step": 2242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.63379968876478e-06,
+      "loss": 0.8369,
+      "step": 2243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6273115577003806e-06,
+      "loss": 0.9261,
+      "step": 2244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620835193829009e-06,
+      "loss": 0.9113,
+      "step": 2245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6143706062527108e-06,
+      "loss": 0.9109,
+      "step": 2246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6079178040569853e-06,
+      "loss": 0.8836,
+      "step": 2247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6014767963107715e-06,
+      "loss": 0.8719,
+      "step": 2248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59504759206642e-06,
+      "loss": 0.9192,
+      "step": 2249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.588630200359711e-06,
+      "loss": 0.881,
+      "step": 2250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5822246302098055e-06,
+      "loss": 0.8539,
+      "step": 2251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.575830890619261e-06,
+      "loss": 0.9026,
+      "step": 2252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5694489905740052e-06,
+      "loss": 0.8584,
+      "step": 2253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.563078939043322e-06,
+      "loss": 0.8963,
+      "step": 2254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5567207449798517e-06,
+      "loss": 0.8858,
+      "step": 2255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5503744173195568e-06,
+      "loss": 0.9044,
+      "step": 2256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5440399649817384e-06,
+      "loss": 0.8591,
+      "step": 2257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5377173968689985e-06,
+      "loss": 0.9203,
+      "step": 2258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5314067218672347e-06,
+      "loss": 0.872,
+      "step": 2259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5251079488456367e-06,
+      "loss": 0.8787,
+      "step": 2260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.518821086656659e-06,
+      "loss": 0.8436,
+      "step": 2261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5125461441360223e-06,
+      "loss": 0.8669,
+      "step": 2262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5062831301026958e-06,
+      "loss": 0.8735,
+      "step": 2263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.500032053358874e-06,
+      "loss": 0.8787,
+      "step": 2264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4937929226899928e-06,
+      "loss": 0.8267,
+      "step": 2265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4875657468646788e-06,
+      "loss": 0.8536,
+      "step": 2266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4813505346347701e-06,
+      "loss": 0.8976,
+      "step": 2267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4751472947352884e-06,
+      "loss": 0.8996,
+      "step": 2268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4689560358844224e-06,
+      "loss": 0.8736,
+      "step": 2269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4627767667835336e-06,
+      "loss": 0.8688,
+      "step": 2270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4566094961171206e-06,
+      "loss": 0.8869,
+      "step": 2271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.450454232552826e-06,
+      "loss": 0.8896,
+      "step": 2272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444310984741424e-06,
+      "loss": 0.8866,
+      "step": 2273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4381797613167859e-06,
+      "loss": 0.8636,
+      "step": 2274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4320605708958967e-06,
+      "loss": 0.8759,
+      "step": 2275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4259534220788207e-06,
+      "loss": 0.8821,
+      "step": 2276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4198583234487052e-06,
+      "loss": 0.8822,
+      "step": 2277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4137752835717622e-06,
+      "loss": 0.8794,
+      "step": 2278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4077043109972488e-06,
+      "loss": 0.8333,
+      "step": 2279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4016454142574676e-06,
+      "loss": 0.8644,
+      "step": 2280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3955986018677515e-06,
+      "loss": 0.8303,
+      "step": 2281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3895638823264447e-06,
+      "loss": 0.8978,
+      "step": 2282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3835412641149014e-06,
+      "loss": 0.8621,
+      "step": 2283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3775307556974616e-06,
+      "loss": 0.8625,
+      "step": 2284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.371532365521453e-06,
+      "loss": 0.8543,
+      "step": 2285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3655461020171635e-06,
+      "loss": 0.9112,
+      "step": 2286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3595719735978451e-06,
+      "loss": 0.8753,
+      "step": 2287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3536099886596933e-06,
+      "loss": 0.876,
+      "step": 2288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3476601555818358e-06,
+      "loss": 0.8908,
+      "step": 2289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3417224827263232e-06,
+      "loss": 0.9235,
+      "step": 2290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.335796978438112e-06,
+      "loss": 0.8434,
+      "step": 2291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3298836510450597e-06,
+      "loss": 0.8702,
+      "step": 2292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3239825088579116e-06,
+      "loss": 0.838,
+      "step": 2293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3180935601702838e-06,
+      "loss": 0.8933,
+      "step": 2294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3122168132586576e-06,
+      "loss": 0.8078,
+      "step": 2295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3063522763823655e-06,
+      "loss": 0.8943,
+      "step": 2296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3004999577835786e-06,
+      "loss": 0.8563,
+      "step": 2297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2946598656873e-06,
+      "loss": 0.8719,
+      "step": 2298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2888320083013416e-06,
+      "loss": 0.8616,
+      "step": 2299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2830163938163298e-06,
+      "loss": 0.8959,
+      "step": 2300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.277213030405675e-06,
+      "loss": 0.8452,
+      "step": 2301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2714219262255777e-06,
+      "loss": 0.9268,
+      "step": 2302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2656430894150029e-06,
+      "loss": 0.9101,
+      "step": 2303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598765280956793e-06,
+      "loss": 0.8831,
+      "step": 2304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.254122250372084e-06,
+      "loss": 0.8472,
+      "step": 2305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2483802643314224e-06,
+      "loss": 0.8348,
+      "step": 2306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2426505780436326e-06,
+      "loss": 0.8837,
+      "step": 2307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 0.819,
+      "step": 2308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2312281369199707e-06,
+      "loss": 0.845,
+      "step": 2309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2255353981374906e-06,
+      "loss": 0.8947,
+      "step": 2310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2198549912146496e-06,
+      "loss": 0.8825,
+      "step": 2311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.214186924134838e-06,
+      "loss": 0.8987,
+      "step": 2312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2085312048641052e-06,
+      "loss": 0.8471,
+      "step": 2313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2028878413511413e-06,
+      "loss": 0.891,
+      "step": 2314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1972568415272812e-06,
+      "loss": 0.851,
+      "step": 2315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1916382133064707e-06,
+      "loss": 0.888,
+      "step": 2316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1860319645852814e-06,
+      "loss": 0.89,
+      "step": 2317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180438103242877e-06,
+      "loss": 0.8831,
+      "step": 2318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1748566371410165e-06,
+      "loss": 0.8745,
+      "step": 2319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1692875741240384e-06,
+      "loss": 0.8871,
+      "step": 2320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1637309220188443e-06,
+      "loss": 0.8543,
+      "step": 2321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158186688634898e-06,
+      "loss": 0.8882,
+      "step": 2322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1526548817642115e-06,
+      "loss": 0.8628,
+      "step": 2323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1471355091813251e-06,
+      "loss": 0.8798,
+      "step": 2324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1416285786433102e-06,
+      "loss": 0.8658,
+      "step": 2325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1361340978897483e-06,
+      "loss": 0.8865,
+      "step": 2326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1306520746427263e-06,
+      "loss": 0.8579,
+      "step": 2327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.125182516606823e-06,
+      "loss": 0.8467,
+      "step": 2328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.119725431469093e-06,
+      "loss": 0.8454,
+      "step": 2329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1142808268990691e-06,
+      "loss": 0.8329,
+      "step": 2330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1088487105487344e-06,
+      "loss": 0.903,
+      "step": 2331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1034290900525279e-06,
+      "loss": 0.8403,
+      "step": 2332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0980219730273256e-06,
+      "loss": 0.8115,
+      "step": 2333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0926273670724296e-06,
+      "loss": 0.8837,
+      "step": 2334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0872452797695588e-06,
+      "loss": 0.8579,
+      "step": 2335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0818757186828388e-06,
+      "loss": 0.882,
+      "step": 2336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0765186913587866e-06,
+      "loss": 0.8708,
+      "step": 2337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0711742053263107e-06,
+      "loss": 0.9149,
+      "step": 2338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0658422680966874e-06,
+      "loss": 0.9092,
+      "step": 2339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0605228871635586e-06,
+      "loss": 0.8521,
+      "step": 2340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0552160700029224e-06,
+      "loss": 0.9283,
+      "step": 2341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0499218240731157e-06,
+      "loss": 0.9041,
+      "step": 2342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0446401568148123e-06,
+      "loss": 0.86,
+      "step": 2343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.039371075650998e-06,
+      "loss": 0.8474,
+      "step": 2344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0341145879869785e-06,
+      "loss": 0.8928,
+      "step": 2345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0288707012103593e-06,
+      "loss": 0.8926,
+      "step": 2346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0236394226910329e-06,
+      "loss": 0.8651,
+      "step": 2347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0184207597811724e-06,
+      "loss": 0.884,
+      "step": 2348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013214719815223e-06,
+      "loss": 0.9218,
+      "step": 2349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0080213101098891e-06,
+      "loss": 0.8699,
+      "step": 2350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0028405379641227e-06,
+      "loss": 0.8717,
+      "step": 2351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.976724106591128e-07,
+      "loss": 0.864,
+      "step": 2352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.92516935458283e-07,
+      "loss": 0.8721,
+      "step": 2353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.873741196072683e-07,
+      "loss": 0.8146,
+      "step": 2354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82243970333915e-07,
+      "loss": 0.8796,
+      "step": 2355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.771264948482695e-07,
+      "loss": 0.8479,
+      "step": 2356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720217003425648e-07,
+      "loss": 0.8772,
+      "step": 2357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.669295939912106e-07,
+      "loss": 0.8887,
+      "step": 2358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.618501829507832e-07,
+      "loss": 0.8618,
+      "step": 2359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.567834743600202e-07,
+      "loss": 0.8523,
+      "step": 2360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.517294753398066e-07,
+      "loss": 0.8474,
+      "step": 2361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.466881929931582e-07,
+      "loss": 0.8635,
+      "step": 2362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.416596344052276e-07,
+      "loss": 0.8904,
+      "step": 2363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.366438066432804e-07,
+      "loss": 0.9251,
+      "step": 2364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.316407167566899e-07,
+      "loss": 0.8854,
+      "step": 2365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.266503717769315e-07,
+      "loss": 0.8898,
+      "step": 2366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.216727787175605e-07,
+      "loss": 0.9029,
+      "step": 2367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.167079445742188e-07,
+      "loss": 0.8585,
+      "step": 2368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.117558763246104e-07,
+      "loss": 0.881,
+      "step": 2369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.06816580928499e-07,
+      "loss": 0.8852,
+      "step": 2370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.018900653277019e-07,
+      "loss": 0.9018,
+      "step": 2371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.969763364460682e-07,
+      "loss": 0.813,
+      "step": 2372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.920754011894838e-07,
+      "loss": 0.8906,
+      "step": 2373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.871872664458459e-07,
+      "loss": 0.8679,
+      "step": 2374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.823119390850665e-07,
+      "loss": 0.852,
+      "step": 2375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.774494259590594e-07,
+      "loss": 0.8788,
+      "step": 2376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.725997339017233e-07,
+      "loss": 0.8464,
+      "step": 2377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.677628697289408e-07,
+      "loss": 0.8538,
+      "step": 2378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.629388402385685e-07,
+      "loss": 0.8788,
+      "step": 2379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.581276522104198e-07,
+      "loss": 0.8992,
+      "step": 2380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.533293124062669e-07,
+      "loss": 0.8522,
+      "step": 2381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.485438275698154e-07,
+      "loss": 0.8895,
+      "step": 2382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.437712044267144e-07,
+      "loss": 0.8915,
+      "step": 2383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.39011449684527e-07,
+      "loss": 0.8458,
+      "step": 2384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34264570032739e-07,
+      "loss": 0.8793,
+      "step": 2385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.295305721427371e-07,
+      "loss": 0.8875,
+      "step": 2386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.24809462667805e-07,
+      "loss": 0.8684,
+      "step": 2387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.201012482431125e-07,
+      "loss": 0.8631,
+      "step": 2388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.154059354857047e-07,
+      "loss": 0.8742,
+      "step": 2389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.10723530994496e-07,
+      "loss": 0.9021,
+      "step": 2390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.060540413502604e-07,
+      "loss": 0.9008,
+      "step": 2391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.01397473115616e-07,
+      "loss": 0.8889,
+      "step": 2392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.967538328350277e-07,
+      "loss": 0.8907,
+      "step": 2393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.921231270347851e-07,
+      "loss": 0.8413,
+      "step": 2394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.875053622230045e-07,
+      "loss": 0.8786,
+      "step": 2395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82900544889612e-07,
+      "loss": 0.8463,
+      "step": 2396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.783086815063346e-07,
+      "loss": 0.8662,
+      "step": 2397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.737297785266995e-07,
+      "loss": 0.8765,
+      "step": 2398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.691638423860104e-07,
+      "loss": 0.8434,
+      "step": 2399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.646108795013563e-07,
+      "loss": 0.8874,
+      "step": 2400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.600708962715875e-07,
+      "loss": 0.878,
+      "step": 2401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.555438990773134e-07,
+      "loss": 0.8653,
+      "step": 2402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.510298942808969e-07,
+      "loss": 0.86,
+      "step": 2403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46528888226431e-07,
+      "loss": 0.8359,
+      "step": 2404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4204088723975e-07,
+      "loss": 0.8954,
+      "step": 2405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.375658976284073e-07,
+      "loss": 0.8679,
+      "step": 2406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.331039256816664e-07,
+      "loss": 0.8811,
+      "step": 2407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286549776704987e-07,
+      "loss": 0.8804,
+      "step": 2408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.242190598475717e-07,
+      "loss": 0.8707,
+      "step": 2409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.197961784472396e-07,
+      "loss": 0.8626,
+      "step": 2410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.153863396855354e-07,
+      "loss": 0.8494,
+      "step": 2411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109895497601571e-07,
+      "loss": 0.8563,
+      "step": 2412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.066058148504707e-07,
+      "loss": 0.8629,
+      "step": 2413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 0.8689,
+      "step": 2414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.978775347038658e-07,
+      "loss": 0.8923,
+      "step": 2415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.935330017339003e-07,
+      "loss": 0.8757,
+      "step": 2416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.892015483135095e-07,
+      "loss": 0.8922,
+      "step": 2417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.848831805302314e-07,
+      "loss": 0.9172,
+      "step": 2418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.80577904453209e-07,
+      "loss": 0.8228,
+      "step": 2419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.762857261331901e-07,
+      "loss": 0.8899,
+      "step": 2420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.720066516025147e-07,
+      "loss": 0.8742,
+      "step": 2421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.677406868751013e-07,
+      "loss": 0.911,
+      "step": 2422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.634878379464505e-07,
+      "loss": 0.86,
+      "step": 2423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.592481107936243e-07,
+      "loss": 0.8284,
+      "step": 2424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.550215113752456e-07,
+      "loss": 0.8578,
+      "step": 2425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.50808045631488e-07,
+      "loss": 0.8472,
+      "step": 2426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.466077194840637e-07,
+      "loss": 0.8501,
+      "step": 2427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.424205388362203e-07,
+      "loss": 0.8933,
+      "step": 2428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.382465095727286e-07,
+      "loss": 0.869,
+      "step": 2429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.340856375598781e-07,
+      "loss": 0.9002,
+      "step": 2430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.299379286454666e-07,
+      "loss": 0.8796,
+      "step": 2431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.258033886587911e-07,
+      "loss": 0.861,
+      "step": 2432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.216820234106424e-07,
+      "loss": 0.8543,
+      "step": 2433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.175738386932917e-07,
+      "loss": 0.817,
+      "step": 2434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.134788402804881e-07,
+      "loss": 0.8888,
+      "step": 2435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.093970339274513e-07,
+      "loss": 0.8632,
+      "step": 2436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.053284253708547e-07,
+      "loss": 0.8438,
+      "step": 2437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.012730203288286e-07,
+      "loss": 0.8288,
+      "step": 2438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.972308245009395e-07,
+      "loss": 0.8633,
+      "step": 2439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.932018435681985e-07,
+      "loss": 0.8553,
+      "step": 2440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891860831930396e-07,
+      "loss": 0.8782,
+      "step": 2441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.851835490193136e-07,
+      "loss": 0.9099,
+      "step": 2442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.811942466722898e-07,
+      "loss": 0.8682,
+      "step": 2443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.772181817586309e-07,
+      "loss": 0.8598,
+      "step": 2444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.732553598664059e-07,
+      "loss": 0.8763,
+      "step": 2445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.693057865650676e-07,
+      "loss": 0.8612,
+      "step": 2446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.653694674054444e-07,
+      "loss": 0.8807,
+      "step": 2447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.614464079197457e-07,
+      "loss": 0.8815,
+      "step": 2448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.575366136215388e-07,
+      "loss": 0.8391,
+      "step": 2449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.536400900057493e-07,
+      "loss": 0.8572,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.497568425486543e-07,
+      "loss": 0.88,
+      "step": 2451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.458868767078673e-07,
+      "loss": 0.8502,
+      "step": 2452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420301979223419e-07,
+      "loss": 0.8911,
+      "step": 2453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.381868116123512e-07,
+      "loss": 0.8698,
+      "step": 2454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.343567231794889e-07,
+      "loss": 0.8443,
+      "step": 2455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.305399380066656e-07,
+      "loss": 0.9253,
+      "step": 2456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.267364614580861e-07,
+      "loss": 0.8645,
+      "step": 2457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229462988792566e-07,
+      "loss": 0.8932,
+      "step": 2458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.191694555969673e-07,
+      "loss": 0.886,
+      "step": 2459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.154059369192932e-07,
+      "loss": 0.8672,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116557481355811e-07,
+      "loss": 0.8214,
+      "step": 2461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.079188945164426e-07,
+      "loss": 0.9074,
+      "step": 2462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.041953813137479e-07,
+      "loss": 0.8731,
+      "step": 2463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004852137606198e-07,
+      "loss": 0.8848,
+      "step": 2464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.967883970714238e-07,
+      "loss": 0.8683,
+      "step": 2465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.931049364417628e-07,
+      "loss": 0.8654,
+      "step": 2466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.8744,
+      "step": 2467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.857781040495857e-07,
+      "loss": 0.8997,
+      "step": 2468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821347425843892e-07,
+      "loss": 0.8691,
+      "step": 2469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.785047577733515e-07,
+      "loss": 0.9008,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7488815471814943e-07,
+      "loss": 0.8661,
+      "step": 2471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7128493850164715e-07,
+      "loss": 0.8325,
+      "step": 2472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.676951141879005e-07,
+      "loss": 0.8784,
+      "step": 2473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6411868682213923e-07,
+      "loss": 0.8893,
+      "step": 2474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6055566143076827e-07,
+      "loss": 0.8706,
+      "step": 2475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5700604302135633e-07,
+      "loss": 0.8646,
+      "step": 2476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.534698365826273e-07,
+      "loss": 0.8583,
+      "step": 2477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4994704708445804e-07,
+      "loss": 0.8972,
+      "step": 2478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.46437679477868e-07,
+      "loss": 0.8421,
+      "step": 2479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42941738695013e-07,
+      "loss": 0.8839,
+      "step": 2480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3945922964918066e-07,
+      "loss": 0.902,
+      "step": 2481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.359901572347758e-07,
+      "loss": 0.8631,
+      "step": 2482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3253452632732617e-07,
+      "loss": 0.8606,
+      "step": 2483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.290923417834625e-07,
+      "loss": 0.8224,
+      "step": 2484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.256636084409194e-07,
+      "loss": 0.8886,
+      "step": 2485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222483311185299e-07,
+      "loss": 0.8515,
+      "step": 2486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.188465146162113e-07,
+      "loss": 0.8739,
+      "step": 2487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1545816371496685e-07,
+      "loss": 0.9414,
+      "step": 2488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.120832831768695e-07,
+      "loss": 0.8976,
+      "step": 2489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.087218777450652e-07,
+      "loss": 0.8581,
+      "step": 2490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.053739521437605e-07,
+      "loss": 0.8519,
+      "step": 2491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.02039511078216e-07,
+      "loss": 0.8459,
+      "step": 2492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.98718559234742e-07,
+      "loss": 0.894,
+      "step": 2493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.954111012806894e-07,
+      "loss": 0.8685,
+      "step": 2494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.921171418644465e-07,
+      "loss": 0.872,
+      "step": 2495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8883668561542907e-07,
+      "loss": 0.8488,
+      "step": 2496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8556973714407366e-07,
+      "loss": 0.881,
+      "step": 2497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8231630104183514e-07,
+      "loss": 0.8609,
+      "step": 2498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7907638188117714e-07,
+      "loss": 0.8546,
+      "step": 2499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7584998421556387e-07,
+      "loss": 0.876,
+      "step": 2500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7263711257946036e-07,
+      "loss": 0.8616,
+      "step": 2501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6943777148831907e-07,
+      "loss": 0.8762,
+      "step": 2502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6625196543857764e-07,
+      "loss": 0.9025,
+      "step": 2503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6307969890764907e-07,
+      "loss": 0.8486,
+      "step": 2504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.59920976353918e-07,
+      "loss": 0.8556,
+      "step": 2505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.567758022167378e-07,
+      "loss": 0.8254,
+      "step": 2506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5364418091641374e-07,
+      "loss": 0.8539,
+      "step": 2507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.505261168542107e-07,
+      "loss": 0.8739,
+      "step": 2508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4742161441233437e-07,
+      "loss": 0.8889,
+      "step": 2509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.443306779539335e-07,
+      "loss": 0.8522,
+      "step": 2510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4125331182309205e-07,
+      "loss": 0.8727,
+      "step": 2511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.381895203448182e-07,
+      "loss": 0.9037,
+      "step": 2512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3513930782504423e-07,
+      "loss": 0.8705,
+      "step": 2513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.321026785506165e-07,
+      "loss": 0.8438,
+      "step": 2514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2907963678929345e-07,
+      "loss": 0.8511,
+      "step": 2515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2607018678973646e-07,
+      "loss": 0.8948,
+      "step": 2516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2307433278150445e-07,
+      "loss": 0.8528,
+      "step": 2517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2009207897504945e-07,
+      "loss": 0.8671,
+      "step": 2518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1712342956170426e-07,
+      "loss": 0.8507,
+      "step": 2519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 0.8326,
+      "step": 2520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1122696058409454e-07,
+      "loss": 0.8663,
+      "step": 2521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0829914930687767e-07,
+      "loss": 0.8545,
+      "step": 2522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.053849589968616e-07,
+      "loss": 0.8602,
+      "step": 2523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.024843937497246e-07,
+      "loss": 0.8924,
+      "step": 2524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9959745764199575e-07,
+      "loss": 0.8411,
+      "step": 2525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9672415473105286e-07,
+      "loss": 0.8403,
+      "step": 2526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.938644890551057e-07,
+      "loss": 0.8882,
+      "step": 2527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9101846463320483e-07,
+      "loss": 0.8655,
+      "step": 2528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.881860854652263e-07,
+      "loss": 0.8583,
+      "step": 2529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8536735553186814e-07,
+      "loss": 0.9324,
+      "step": 2530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8256227879464805e-07,
+      "loss": 0.8715,
+      "step": 2531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7977085919589253e-07,
+      "loss": 0.8994,
+      "step": 2532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7699310065873563e-07,
+      "loss": 0.8378,
+      "step": 2533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7422900708710896e-07,
+      "loss": 0.921,
+      "step": 2534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.714785823657429e-07,
+      "loss": 0.8599,
+      "step": 2535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.687418303601563e-07,
+      "loss": 0.8568,
+      "step": 2536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.660187549166504e-07,
+      "loss": 0.8579,
+      "step": 2537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.633093598623049e-07,
+      "loss": 0.8669,
+      "step": 2538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.606136490049738e-07,
+      "loss": 0.8745,
+      "step": 2539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5793162613328095e-07,
+      "loss": 0.8745,
+      "step": 2540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5526329501661006e-07,
+      "loss": 0.8373,
+      "step": 2541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5260865940510027e-07,
+      "loss": 0.8597,
+      "step": 2542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4996772302964934e-07,
+      "loss": 0.8539,
+      "step": 2543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4734048960189385e-07,
+      "loss": 0.8701,
+      "step": 2544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.447269628142168e-07,
+      "loss": 0.8212,
+      "step": 2545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.421271463397368e-07,
+      "loss": 0.8748,
+      "step": 2546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3954104383230316e-07,
+      "loss": 0.861,
+      "step": 2547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.369686589264919e-07,
+      "loss": 0.8573,
+      "step": 2548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3440999523759777e-07,
+      "loss": 0.9003,
+      "step": 2549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3186505636163316e-07,
+      "loss": 0.831,
+      "step": 2550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2933384587532358e-07,
+      "loss": 0.8929,
+      "step": 2551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2681636733609457e-07,
+      "loss": 0.872,
+      "step": 2552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24312624282077e-07,
+      "loss": 0.8375,
+      "step": 2553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2182262023209612e-07,
+      "loss": 0.8415,
+      "step": 2554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1934635868566813e-07,
+      "loss": 0.8522,
+      "step": 2555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.168838431229958e-07,
+      "loss": 0.8497,
+      "step": 2556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1443507700495968e-07,
+      "loss": 0.8468,
+      "step": 2557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1200006377312232e-07,
+      "loss": 0.8882,
+      "step": 2558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0957880684971178e-07,
+      "loss": 0.8866,
+      "step": 2559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.071713096376271e-07,
+      "loss": 0.8859,
+      "step": 2560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.047775755204251e-07,
+      "loss": 0.8698,
+      "step": 2561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0239760786232355e-07,
+      "loss": 0.8546,
+      "step": 2562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0003141000819015e-07,
+      "loss": 0.8671,
+      "step": 2563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9767898528353923e-07,
+      "loss": 0.8627,
+      "step": 2564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953403369945317e-07,
+      "loss": 0.8283,
+      "step": 2565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.930154684279628e-07,
+      "loss": 0.8593,
+      "step": 2566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9070438285126337e-07,
+      "loss": 0.8333,
+      "step": 2567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8840708351249182e-07,
+      "loss": 0.876,
+      "step": 2568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.861235736403344e-07,
+      "loss": 0.8871,
+      "step": 2569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8385385644409282e-07,
+      "loss": 0.8746,
+      "step": 2570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8159793511368872e-07,
+      "loss": 0.8783,
+      "step": 2571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.793558128196493e-07,
+      "loss": 0.8646,
+      "step": 2572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7712749271311392e-07,
+      "loss": 0.8246,
+      "step": 2573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7491297792581963e-07,
+      "loss": 0.8688,
+      "step": 2574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7271227157010351e-07,
+      "loss": 0.8882,
+      "step": 2575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.705253767388948e-07,
+      "loss": 0.9165,
+      "step": 2576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6835229650571272e-07,
+      "loss": 0.8883,
+      "step": 2577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.661930339246609e-07,
+      "loss": 0.8469,
+      "step": 2578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6404759203042075e-07,
+      "loss": 0.8492,
+      "step": 2579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6191597383825473e-07,
+      "loss": 0.8026,
+      "step": 2580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5979818234399203e-07,
+      "loss": 0.8817,
+      "step": 2581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5769422052403172e-07,
+      "loss": 0.8848,
+      "step": 2582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5560409133533739e-07,
+      "loss": 0.8667,
+      "step": 2583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5352779771543037e-07,
+      "loss": 0.859,
+      "step": 2584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5146534258238643e-07,
+      "loss": 0.8922,
+      "step": 2585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.494167288348347e-07,
+      "loss": 0.8656,
+      "step": 2586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4738195935194987e-07,
+      "loss": 0.8774,
+      "step": 2587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4536103699344884e-07,
+      "loss": 0.871,
+      "step": 2588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.433539645995885e-07,
+      "loss": 0.8629,
+      "step": 2589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4136074499115914e-07,
+      "loss": 0.8233,
+      "step": 2590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.393813809694855e-07,
+      "loss": 0.8663,
+      "step": 2591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3741587531641566e-07,
+      "loss": 0.8764,
+      "step": 2592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3546423079432213e-07,
+      "loss": 0.8477,
+      "step": 2593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3352645014609756e-07,
+      "loss": 0.8604,
+      "step": 2594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3160253609514785e-07,
+      "loss": 0.8797,
+      "step": 2595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.296924913453923e-07,
+      "loss": 0.8874,
+      "step": 2596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2779631858125696e-07,
+      "loss": 0.8785,
+      "step": 2597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.259140204676712e-07,
+      "loss": 0.8466,
+      "step": 2598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.240455996500678e-07,
+      "loss": 0.8714,
+      "step": 2599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2219105875437176e-07,
+      "loss": 0.8549,
+      "step": 2600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2035040038700375e-07,
+      "loss": 0.8622,
+      "step": 2601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.185236271348722e-07,
+      "loss": 0.8605,
+      "step": 2602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.167107415653712e-07,
+      "loss": 0.8579,
+      "step": 2603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1491174622637934e-07,
+      "loss": 0.9276,
+      "step": 2604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1312664364624749e-07,
+      "loss": 0.8725,
+      "step": 2605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1135543633380764e-07,
+      "loss": 0.8489,
+      "step": 2606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0959812677835968e-07,
+      "loss": 0.8438,
+      "step": 2607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0785471744967247e-07,
+      "loss": 0.8994,
+      "step": 2608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0612521079797821e-07,
+      "loss": 0.8872,
+      "step": 2609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0440960925396925e-07,
+      "loss": 0.8426,
+      "step": 2610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0270791522879908e-07,
+      "loss": 0.8612,
+      "step": 2611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0102013111406905e-07,
+      "loss": 0.849,
+      "step": 2612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.934625928183728e-08,
+      "loss": 0.8481,
+      "step": 2613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.768630208460528e-08,
+      "loss": 0.8609,
+      "step": 2614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.604026185532022e-08,
+      "loss": 0.8803,
+      "step": 2615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.440814090737049e-08,
+      "loss": 0.9431,
+      "step": 2616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.278994153457898e-08,
+      "loss": 0.856,
+      "step": 2617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.11856660112076e-08,
+      "loss": 0.8451,
+      "step": 2618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.959531659194498e-08,
+      "loss": 0.853,
+      "step": 2619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.80188955119099e-08,
+      "loss": 0.8495,
+      "step": 2620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.645640498664454e-08,
+      "loss": 0.8825,
+      "step": 2621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.490784721211454e-08,
+      "loss": 0.868,
+      "step": 2622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.33732243647023e-08,
+      "loss": 0.8628,
+      "step": 2623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.185253860120701e-08,
+      "loss": 0.8383,
+      "step": 2624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.034579205884019e-08,
+      "loss": 0.8995,
+      "step": 2625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 0.863,
+      "step": 2626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.73741250883786e-08,
+      "loss": 0.8338,
+      "step": 2627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.590920883674192e-08,
+      "loss": 0.8416,
+      "step": 2628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.445824015914316e-08,
+      "loss": 0.8478,
+      "step": 2629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.302122109481002e-08,
+      "loss": 0.8698,
+      "step": 2630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.159815366336697e-08,
+      "loss": 0.8241,
+      "step": 2631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.018903986483083e-08,
+      "loss": 0.9111,
+      "step": 2632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.87938816796041e-08,
+      "loss": 0.8811,
+      "step": 2633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.741268106848164e-08,
+      "loss": 0.8341,
+      "step": 2634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.604543997263513e-08,
+      "loss": 0.8568,
+      "step": 2635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.469216031362302e-08,
+      "loss": 0.8698,
+      "step": 2636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.335284399337726e-08,
+      "loss": 0.844,
+      "step": 2637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.202749289420994e-08,
+      "loss": 0.8816,
+      "step": 2638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.071610887880331e-08,
+      "loss": 0.8649,
+      "step": 2639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9418693790209705e-08,
+      "loss": 0.8664,
+      "step": 2640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.813524945185056e-08,
+      "loss": 0.8448,
+      "step": 2641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.686577766751078e-08,
+      "loss": 0.8458,
+      "step": 2642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5610280221338745e-08,
+      "loss": 0.8863,
+      "step": 2643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4368758877845204e-08,
+      "loss": 0.8468,
+      "step": 2644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.314121538189332e-08,
+      "loss": 0.8578,
+      "step": 2645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.192765145870748e-08,
+      "loss": 0.8608,
+      "step": 2646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.072806881386005e-08,
+      "loss": 0.8647,
+      "step": 2647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.954246913327576e-08,
+      "loss": 0.8824,
+      "step": 2648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8370854083227284e-08,
+      "loss": 0.8473,
+      "step": 2649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.721322531033301e-08,
+      "loss": 0.8345,
+      "step": 2650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.606958444155485e-08,
+      "loss": 0.8633,
+      "step": 2651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4939933084192646e-08,
+      "loss": 0.8841,
+      "step": 2652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3824272825890856e-08,
+      "loss": 0.8779,
+      "step": 2653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2722605234625236e-08,
+      "loss": 0.8557,
+      "step": 2654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1634931858709485e-08,
+      "loss": 0.8698,
+      "step": 2655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0561254226786365e-08,
+      "loss": 0.8608,
+      "step": 2656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.950157384783104e-08,
+      "loss": 0.8438,
+      "step": 2657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.845589221114554e-08,
+      "loss": 0.8323,
+      "step": 2658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.742421078635761e-08,
+      "loss": 0.8927,
+      "step": 2659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6406531023420735e-08,
+      "loss": 0.8664,
+      "step": 2660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.540285435260527e-08,
+      "loss": 0.8453,
+      "step": 2661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4413182184507285e-08,
+      "loss": 0.8439,
+      "step": 2662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.343751591003641e-08,
+      "loss": 0.9101,
+      "step": 2663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.24758569004191e-08,
+      "loss": 0.8607,
+      "step": 2664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1528206507196456e-08,
+      "loss": 0.8907,
+      "step": 2665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0594566062219776e-08,
+      "loss": 0.8995,
+      "step": 2666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.967493687765277e-08,
+      "loss": 0.8692,
+      "step": 2667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8769320245966014e-08,
+      "loss": 0.871,
+      "step": 2668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7877717439934727e-08,
+      "loss": 0.8298,
+      "step": 2669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7000129712643208e-08,
+      "loss": 0.8602,
+      "step": 2670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6136558297474857e-08,
+      "loss": 0.8577,
+      "step": 2671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.528700440811438e-08,
+      "loss": 0.8678,
+      "step": 2672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4451469238547797e-08,
+      "loss": 0.879,
+      "step": 2673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3629953963058007e-08,
+      "loss": 0.8547,
+      "step": 2674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2822459736222547e-08,
+      "loss": 0.8663,
+      "step": 2675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2028987692915836e-08,
+      "loss": 0.8688,
+      "step": 2676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1249538948304726e-08,
+      "loss": 0.8603,
+      "step": 2677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.048411459784516e-08,
+      "loss": 0.879,
+      "step": 2678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.973271571728441e-08,
+      "loss": 0.8709,
+      "step": 2679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8995343362658846e-08,
+      "loss": 0.8398,
+      "step": 2680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8271998570290607e-08,
+      "loss": 0.8358,
+      "step": 2681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7562682356786488e-08,
+      "loss": 0.8543,
+      "step": 2682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.686739571903795e-08,
+      "loss": 0.8786,
+      "step": 2683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.618613963421889e-08,
+      "loss": 0.8345,
+      "step": 2684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5518915059784533e-08,
+      "loss": 0.8304,
+      "step": 2685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4865722933469218e-08,
+      "loss": 0.8571,
+      "step": 2686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4226564173286383e-08,
+      "loss": 0.8079,
+      "step": 2687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3601439677526363e-08,
+      "loss": 0.8835,
+      "step": 2688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2990350324755263e-08,
+      "loss": 0.8904,
+      "step": 2689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2393296973812751e-08,
+      "loss": 0.876,
+      "step": 2690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1810280463815382e-08,
+      "loss": 0.8661,
+      "step": 2691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1241301614147715e-08,
+      "loss": 0.8719,
+      "step": 2692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0686361224466757e-08,
+      "loss": 0.8556,
+      "step": 2693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0145460074703073e-08,
+      "loss": 0.9005,
+      "step": 2694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.618598925050792e-09,
+      "loss": 0.8883,
+      "step": 2695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.105778515974273e-09,
+      "loss": 0.8918,
+      "step": 2696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.606999568204766e-09,
+      "loss": 0.877,
+      "step": 2697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.12226278274042e-09,
+      "loss": 0.8504,
+      "step": 2698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.651568840840729e-09,
+      "loss": 0.8472,
+      "step": 2699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.194918404033191e-09,
+      "loss": 0.8876,
+      "step": 2700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.752312114105541e-09,
+      "loss": 0.8919,
+      "step": 2701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.323750593106859e-09,
+      "loss": 0.8851,
+      "step": 2702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.909234443349788e-09,
+      "loss": 0.8311,
+      "step": 2703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.508764247406096e-09,
+      "loss": 0.8494,
+      "step": 2704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.122340568105566e-09,
+      "loss": 0.8857,
+      "step": 2705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.749963948540437e-09,
+      "loss": 0.8858,
+      "step": 2706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.39163491205652e-09,
+      "loss": 0.8466,
+      "step": 2707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.047353962259859e-09,
+      "loss": 0.8781,
+      "step": 2708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7171215830100748e-09,
+      "loss": 0.8442,
+      "step": 2709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4009382384270206e-09,
+      "loss": 0.934,
+      "step": 2710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0988043728807925e-09,
+      "loss": 0.8666,
+      "step": 2711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.810720410998391e-09,
+      "loss": 0.8632,
+      "step": 2712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5366867576614995e-09,
+      "loss": 0.8346,
+      "step": 2713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2767037980031548e-09,
+      "loss": 0.8715,
+      "step": 2714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0307718974099666e-09,
+      "loss": 0.8637,
+      "step": 2715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7988914015221182e-09,
+      "loss": 0.8662,
+      "step": 2716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5810626362300351e-09,
+      "loss": 0.8402,
+      "step": 2717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3772859076754962e-09,
+      "loss": 0.858,
+      "step": 2718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1875615022516328e-09,
+      "loss": 0.8335,
+      "step": 2719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0118896866018192e-09,
+      "loss": 0.9077,
+      "step": 2720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.502707076207817e-10,
+      "loss": 0.8536,
+      "step": 2721
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.027047924512698e-10,
+      "loss": 0.8546,
+      "step": 2722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.69192148486275e-10,
+      "loss": 0.8659,
+      "step": 2723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.497329633679215e-10,
+      "loss": 0.9096,
+      "step": 2724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.443274049863554e-10,
+      "loss": 0.8334,
+      "step": 2725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.529756214841861e-10,
+      "loss": 0.8668,
+      "step": 2726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7567774124649384e-10,
+      "loss": 0.8627,
+      "step": 2727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1243387291082208e-10,
+      "loss": 0.8487,
+      "step": 2728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.324410536162618e-11,
+      "loss": 0.8642,
+      "step": 2729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.810850773249385e-11,
+      "loss": 0.844,
+      "step": 2730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.027129401704358e-12,
+      "loss": 0.8224,
+      "step": 2731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8651,
+      "step": 2732
+    },
+    {
+      "epoch": 1.0,
+      "step": 2732,
+      "total_flos": 5278279635828736.0,
+      "train_loss": 0.9729856379210163,
+      "train_runtime": 79682.8453,
+      "train_samples_per_second": 4.389,
+      "train_steps_per_second": 0.034
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2732,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 5278279635828736.0,
+  "trial_name": null,
+  "trial_params": null
+}