diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18541 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 3086,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 5.6172,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1505376344086024e-05,
+      "loss": 5.3789,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 5.4805,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.301075268817205e-05,
+      "loss": 5.0391,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3763440860215054e-05,
+      "loss": 4.6797,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.451612903225807e-05,
+      "loss": 4.7969,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.526881720430108e-05,
+      "loss": 4.293,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.60215053763441e-05,
+      "loss": 4.0859,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.67741935483871e-05,
+      "loss": 3.9766,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010752688172043011,
+      "loss": 3.8613,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011827956989247312,
+      "loss": 3.7305,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012903225806451613,
+      "loss": 3.6973,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013978494623655913,
+      "loss": 3.6953,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015053763440860216,
+      "loss": 3.6504,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016129032258064516,
+      "loss": 3.4648,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001720430107526882,
+      "loss": 3.4824,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001827956989247312,
+      "loss": 3.5254,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001935483870967742,
+      "loss": 3.3438,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020430107526881721,
+      "loss": 3.3242,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021505376344086021,
+      "loss": 3.4258,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022580645161290321,
+      "loss": 3.3984,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023655913978494624,
+      "loss": 3.3789,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024731182795698927,
+      "loss": 3.2949,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025806451612903227,
+      "loss": 3.3379,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026881720430107527,
+      "loss": 3.252,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027956989247311827,
+      "loss": 3.1699,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002903225806451613,
+      "loss": 3.1465,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003010752688172043,
+      "loss": 3.2012,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003118279569892473,
+      "loss": 3.1426,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003225806451612903,
+      "loss": 3.1211,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 3.2344,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003440860215053764,
+      "loss": 3.1035,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003548387096774194,
+      "loss": 3.1719,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003655913978494624,
+      "loss": 3.1445,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003763440860215054,
+      "loss": 3.1289,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003870967741935484,
+      "loss": 3.0879,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003978494623655914,
+      "loss": 3.1875,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040860215053763443,
+      "loss": 3.1543,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041935483870967743,
+      "loss": 3.2832,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043010752688172043,
+      "loss": 3.1348,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044086021505376343,
+      "loss": 3.1602,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045161290322580643,
+      "loss": 3.1543,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046236559139784943,
+      "loss": 3.084,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004731182795698925,
+      "loss": 3.0996,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004838709677419355,
+      "loss": 3.1621,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004946236559139785,
+      "loss": 3.0488,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005053763440860215,
+      "loss": 3.1113,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005161290322580645,
+      "loss": 2.9688,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005268817204301075,
+      "loss": 3.0137,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005376344086021505,
+      "loss": 2.9883,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005483870967741935,
+      "loss": 3.0664,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005591397849462365,
+      "loss": 2.9941,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005698924731182796,
+      "loss": 3.0098,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005806451612903226,
+      "loss": 3.0215,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005913978494623656,
+      "loss": 3.041,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006021505376344086,
+      "loss": 3.0312,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006129032258064516,
+      "loss": 3.0254,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006236559139784946,
+      "loss": 3.0,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006344086021505376,
+      "loss": 2.9551,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006451612903225806,
+      "loss": 3.0156,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006559139784946236,
+      "loss": 2.9922,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 3.0391,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006774193548387097,
+      "loss": 3.0938,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006881720430107528,
+      "loss": 2.9668,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006989247311827958,
+      "loss": 2.9941,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007096774193548388,
+      "loss": 3.0508,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007204301075268818,
+      "loss": 2.9785,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007311827956989248,
+      "loss": 2.9434,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007419354838709678,
+      "loss": 2.9629,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007526881720430108,
+      "loss": 2.9102,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007634408602150538,
+      "loss": 2.8555,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007741935483870968,
+      "loss": 2.9766,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007849462365591398,
+      "loss": 2.9062,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007956989247311828,
+      "loss": 2.9277,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008064516129032258,
+      "loss": 2.9629,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008172043010752689,
+      "loss": 2.8301,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008279569892473119,
+      "loss": 2.8633,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008387096774193549,
+      "loss": 2.9824,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008494623655913979,
+      "loss": 2.9043,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008602150537634409,
+      "loss": 2.877,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008709677419354839,
+      "loss": 2.8613,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008817204301075269,
+      "loss": 2.9102,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008924731182795699,
+      "loss": 2.8887,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009032258064516129,
+      "loss": 2.9219,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009139784946236559,
+      "loss": 2.8262,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009247311827956989,
+      "loss": 2.8633,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009354838709677419,
+      "loss": 2.8066,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000946236559139785,
+      "loss": 2.8711,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000956989247311828,
+      "loss": 2.9375,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000967741935483871,
+      "loss": 2.8887,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000978494623655914,
+      "loss": 2.834,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000989247311827957,
+      "loss": 2.8047,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 2.8398,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999997245604625,
+      "loss": 2.8242,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999988982421534,
+      "loss": 2.9004,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999975210459831,
+      "loss": 2.9121,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999955929734688,
+      "loss": 2.7812,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999931140267351,
+      "loss": 2.8379,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999990084208513,
+      "loss": 2.877,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999865035221407,
+      "loss": 2.8613,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999982371971563,
+      "loss": 2.8184,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999977689561332,
+      "loss": 2.8066,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999972456296607,
+      "loss": 2.8965,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999666721831532,
+      "loss": 2.75,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999603372273437,
+      "loss": 2.8496,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999534514361578,
+      "loss": 2.7812,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999460148171822,
+      "loss": 2.7461,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00099993802737861,
+      "loss": 2.8223,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999294891292415,
+      "loss": 2.8027,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999920400078484,
+      "loss": 2.7539,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999107602363514,
+      "loss": 2.7051,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999005696134642,
+      "loss": 2.7402,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998898282210505,
+      "loss": 2.8047,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998785360709442,
+      "loss": 2.6895,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998666931755865,
+      "loss": 2.6738,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998542995480258,
+      "loss": 2.7266,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998413552019167,
+      "loss": 2.6777,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998278601515208,
+      "loss": 2.748,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998138144117063,
+      "loss": 2.7637,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997992179979483,
+      "loss": 2.7871,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997840709263283,
+      "loss": 2.793,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997683732135348,
+      "loss": 2.6504,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999752124876863,
+      "loss": 2.7852,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997353259342146,
+      "loss": 2.75,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999717976404098,
+      "loss": 2.6895,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997000763056281,
+      "loss": 2.7441,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996816256585266,
+      "loss": 2.7422,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996626244831214,
+      "loss": 2.7031,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996430728003475,
+      "loss": 2.6504,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996229706317457,
+      "loss": 2.6641,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996023179994645,
+      "loss": 2.7012,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995811149262574,
+      "loss": 2.666,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995593614354852,
+      "loss": 2.6895,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999537057551115,
+      "loss": 2.6953,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995142032977204,
+      "loss": 2.7754,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994907987004814,
+      "loss": 2.6191,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999466843785184,
+      "loss": 2.6797,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994423385782205,
+      "loss": 2.6699,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099941728310659,
+      "loss": 2.6367,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993916773978978,
+      "loss": 2.6621,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993655214803547,
+      "loss": 2.7559,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993388153827787,
+      "loss": 2.6133,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999311559134593,
+      "loss": 2.6074,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992837527658274,
+      "loss": 2.6348,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999255396307118,
+      "loss": 2.6523,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999226489789707,
+      "loss": 2.6934,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991970332454418,
+      "loss": 2.6602,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999167026706777,
+      "loss": 2.5781,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999136470206772,
+      "loss": 2.7109,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991053637790931,
+      "loss": 2.582,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999073707458012,
+      "loss": 2.6641,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999041501278406,
+      "loss": 2.5938,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990087452757587,
+      "loss": 2.6738,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989754394861593,
+      "loss": 2.6582,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989415839463027,
+      "loss": 2.5957,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989071786934897,
+      "loss": 2.6055,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988722237656265,
+      "loss": 2.5488,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988367192012247,
+      "loss": 2.6074,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998800665039402,
+      "loss": 2.5254,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987640613198813,
+      "loss": 2.6641,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998726908082991,
+      "loss": 2.5312,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986892053696652,
+      "loss": 2.6074,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998650953221443,
+      "loss": 2.5781,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986121516804688,
+      "loss": 2.5918,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998572800789493,
+      "loss": 2.5586,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985329005918703,
+      "loss": 2.6621,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984924511315613,
+      "loss": 2.6465,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984514524531314,
+      "loss": 2.623,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984099046017516,
+      "loss": 2.5508,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998367807623197,
+      "loss": 2.5859,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983251615638489,
+      "loss": 2.543,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982819664706925,
+      "loss": 2.5625,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982382223913185,
+      "loss": 2.5879,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998193929373922,
+      "loss": 2.5391,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981490874673038,
+      "loss": 2.6113,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981036967208685,
+      "loss": 2.5527,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980577571846257,
+      "loss": 2.5508,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980112689091897,
+      "loss": 2.6465,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979642319457793,
+      "loss": 2.5469,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979166463462178,
+      "loss": 2.5762,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978685121629332,
+      "loss": 2.5371,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978198294489575,
+      "loss": 2.5195,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977705982579276,
+      "loss": 2.6152,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997720818644084,
+      "loss": 2.4492,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976704906622723,
+      "loss": 2.5488,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997619614367941,
+      "loss": 2.5156,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975681898171443,
+      "loss": 2.6133,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997516217066539,
+      "loss": 2.5176,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974636961733868,
+      "loss": 2.5547,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974106271955529,
+      "loss": 2.4746,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973570101915066,
+      "loss": 2.5527,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973028452203209,
+      "loss": 2.4434,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997248132341672,
+      "loss": 2.4863,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997192871615841,
+      "loss": 2.4922,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971370631037114,
+      "loss": 2.4492,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970807068667709,
+      "loss": 2.5156,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970238029671102,
+      "loss": 2.5039,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996966351467424,
+      "loss": 2.4883,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969083524310096,
+      "loss": 2.6016,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968498059217679,
+      "loss": 2.373,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967907120042032,
+      "loss": 2.4961,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967310707434229,
+      "loss": 2.5,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966708822051367,
+      "loss": 2.5059,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966101464556584,
+      "loss": 2.5156,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965488635619035,
+      "loss": 2.5156,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964870335913913,
+      "loss": 2.4199,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964246566122434,
+      "loss": 2.5,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963617326931844,
+      "loss": 2.5,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962982619035408,
+      "loss": 2.5098,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962342443132423,
+      "loss": 2.498,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961696799928208,
+      "loss": 2.5039,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961045690134104,
+      "loss": 2.502,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960389114467478,
+      "loss": 2.5469,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959727073651718,
+      "loss": 2.4395,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959059568416233,
+      "loss": 2.4805,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995838659949645,
+      "loss": 2.4492,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957708167633822,
+      "loss": 2.4375,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957024273575812,
+      "loss": 2.5,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956334918075909,
+      "loss": 2.4863,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955640101893613,
+      "loss": 2.4551,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995493982579445,
+      "loss": 2.5137,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954234090549948,
+      "loss": 2.4434,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953522896937658,
+      "loss": 2.502,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952806245741145,
+      "loss": 2.4453,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952084137749989,
+      "loss": 2.4707,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951356573759769,
+      "loss": 2.4746,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950623554572091,
+      "loss": 2.5,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949885080994564,
+      "loss": 2.4277,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994914115384081,
+      "loss": 2.5723,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948391773930449,
+      "loss": 2.502,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947636942089122,
+      "loss": 2.3828,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994687665914847,
+      "loss": 2.4355,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946110925946143,
+      "loss": 2.5625,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945339743325792,
+      "loss": 2.5,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944563112137073,
+      "loss": 2.4648,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943781033235645,
+      "loss": 2.4395,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942993507483172,
+      "loss": 2.3887,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994220053574732,
+      "loss": 2.5801,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941402118901744,
+      "loss": 2.5488,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994059825782611,
+      "loss": 2.5449,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993978895340608,
+      "loss": 2.4785,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993897420653331,
+      "loss": 2.4609,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938154018105457,
+      "loss": 2.5293,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937328389026166,
+      "loss": 2.4668,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936497320205083,
+      "loss": 2.5449,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935660812557844,
+      "loss": 2.459,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934818867006082,
+      "loss": 2.4629,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993397148447741,
+      "loss": 2.5938,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933118665905443,
+      "loss": 2.5586,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932260412229782,
+      "loss": 2.5215,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993139672439601,
+      "loss": 2.4238,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930527603355709,
+      "loss": 2.375,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929653050066436,
+      "loss": 2.4395,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928773065491737,
+      "loss": 2.4883,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992788765060114,
+      "loss": 2.4023,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926996806370167,
+      "loss": 2.4316,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926100533780304,
+      "loss": 2.4238,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925198833819029,
+      "loss": 2.4082,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099242917074798,
+      "loss": 2.4453,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923379155762047,
+      "loss": 2.4805,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922461179671183,
+      "loss": 2.4883,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921537780218596,
+      "loss": 2.4434,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920608958421647,
+      "loss": 2.4785,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919674715303678,
+      "loss": 2.5781,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918735051893993,
+      "loss": 2.5527,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917789969227874,
+      "loss": 2.416,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991683946834658,
+      "loss": 2.5723,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915883550297327,
+      "loss": 2.4277,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914922216133308,
+      "loss": 2.541,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991395546691368,
+      "loss": 2.4512,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912983303703568,
+      "loss": 2.5176,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912005727574058,
+      "loss": 2.4316,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911022739602204,
+      "loss": 2.4062,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910034340871023,
+      "loss": 2.4609,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909040532469489,
+      "loss": 2.4551,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990804131549254,
+      "loss": 2.4473,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907036691041069,
+      "loss": 2.4453,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990602666022193,
+      "loss": 2.4707,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905011224147935,
+      "loss": 2.502,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903990383937846,
+      "loss": 2.4863,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902964140716384,
+      "loss": 2.4609,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901932495614222,
+      "loss": 2.4121,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900895449767978,
+      "loss": 2.4336,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899853004320233,
+      "loss": 2.4316,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898805160419503,
+      "loss": 2.4043,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897751919220265,
+      "loss": 2.4453,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989669328188293,
+      "loss": 2.4844,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895629249573863,
+      "loss": 2.3379,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894559823465372,
+      "loss": 2.541,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893485004735705,
+      "loss": 2.3711,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989240479456905,
+      "loss": 2.3809,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891319194155538,
+      "loss": 2.4453,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989022820469124,
+      "loss": 2.4688,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988913182737816,
+      "loss": 2.3965,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888030063424245,
+      "loss": 2.5195,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886922914043369,
+      "loss": 2.5117,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885810380455343,
+      "loss": 2.4902,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988469246388591,
+      "loss": 2.4609,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883569165566742,
+      "loss": 2.3555,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882440486735447,
+      "loss": 2.4883,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881306428635552,
+      "loss": 2.4922,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880166992516518,
+      "loss": 2.5332,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879022179633723,
+      "loss": 2.4355,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877871991248479,
+      "loss": 2.4375,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876716428628013,
+      "loss": 2.4648,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875555493045476,
+      "loss": 2.5918,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874389185779938,
+      "loss": 2.4004,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873217508116388,
+      "loss": 2.4668,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872040461345731,
+      "loss": 2.375,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870858046764788,
+      "loss": 2.4941,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869670265676294,
+      "loss": 2.3809,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868477119388895,
+      "loss": 2.3867,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867278609217154,
+      "loss": 2.4434,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866074736481535,
+      "loss": 2.4863,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864865502508414,
+      "loss": 2.4004,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863650908630078,
+      "loss": 2.4492,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862430956184714,
+      "loss": 2.3906,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861205646516415,
+      "loss": 2.3965,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859974980975174,
+      "loss": 2.4844,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985873896091689,
+      "loss": 2.4785,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857497587703356,
+      "loss": 2.3145,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856250862702266,
+      "loss": 2.4648,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854998787287208,
+      "loss": 2.5059,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853741362837668,
+      "loss": 2.3926,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852478590739022,
+      "loss": 2.4941,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851210472382542,
+      "loss": 2.4824,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849937009165385,
+      "loss": 2.4277,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848658202490602,
+      "loss": 2.3535,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847374053767125,
+      "loss": 2.4414,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846084564409778,
+      "loss": 2.5,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844789735839268,
+      "loss": 2.5684,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843489569482178,
+      "loss": 2.4082,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842184066770983,
+      "loss": 2.3828,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840873229144025,
+      "loss": 2.4434,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839557058045536,
+      "loss": 2.4219,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838235554925612,
+      "loss": 2.5039,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836908721240236,
+      "loss": 2.418,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835576558451253,
+      "loss": 2.3125,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834239068026388,
+      "loss": 2.4375,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983289625143923,
+      "loss": 2.4141,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831548110169237,
+      "loss": 2.4863,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830194645701737,
+      "loss": 2.4395,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828835859527919,
+      "loss": 2.4707,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827471753144836,
+      "loss": 2.4434,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826102328055406,
+      "loss": 2.4375,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824727585768403,
+      "loss": 2.502,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982334752779846,
+      "loss": 2.3984,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821962155666067,
+      "loss": 2.4727,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820571470897568,
+      "loss": 2.3789,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819175475025165,
+      "loss": 2.3848,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817774169586903,
+      "loss": 2.4277,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816367556126687,
+      "loss": 2.4258,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981495563619426,
+      "loss": 2.4199,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981353841134522,
+      "loss": 2.4258,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009812115883141004,
+      "loss": 2.3926,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810688053148893,
+      "loss": 2.4395,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809254922942014,
+      "loss": 2.373,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807816494099327,
+      "loss": 2.377,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806372768205633,
+      "loss": 2.3984,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980492374685157,
+      "loss": 2.3965,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803469431633606,
+      "loss": 2.418,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802009824154048,
+      "loss": 2.4688,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980054492602103,
+      "loss": 2.3223,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799074738848512,
+      "loss": 2.3789,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979759926425629,
+      "loss": 2.3867,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796118503869978,
+      "loss": 2.3516,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794632459321014,
+      "loss": 2.4531,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793141132246662,
+      "loss": 2.3809,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791644524290001,
+      "loss": 2.375,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790142637099936,
+      "loss": 2.4102,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788635472331177,
+      "loss": 2.4512,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787123031644261,
+      "loss": 2.3906,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978560531670553,
+      "loss": 2.375,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784082329187135,
+      "loss": 2.4141,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782554070767045,
+      "loss": 2.3555,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009781020543129027,
+      "loss": 2.4102,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779481747962662,
+      "loss": 2.3652,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777937686963327,
+      "loss": 2.3496,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776388361832204,
+      "loss": 2.3926,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009774833774276277,
+      "loss": 2.4238,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773273926008321,
+      "loss": 2.4023,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771708818746916,
+      "loss": 2.4902,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770138454216432,
+      "loss": 2.3965,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768562834147026,
+      "loss": 2.4844,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766981960274653,
+      "loss": 2.3828,
+      "step": 385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765395834341054,
+      "loss": 2.3652,
+      "step": 386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763804458093756,
+      "loss": 2.4004,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762207833286072,
+      "loss": 2.3828,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760605961677094,
+      "loss": 2.4414,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758998845031699,
+      "loss": 2.3555,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757386485120539,
+      "loss": 2.3672,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755768883720047,
+      "loss": 2.4492,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754146042612426,
+      "loss": 2.3906,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752517963585657,
+      "loss": 2.5039,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750884648433486,
+      "loss": 2.4199,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749246098955435,
+      "loss": 2.3438,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747602316956789,
+      "loss": 2.3711,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745953304248593,
+      "loss": 2.3652,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744299062647668,
+      "loss": 2.416,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742639593976581,
+      "loss": 2.3926,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740974900063669,
+      "loss": 2.4707,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739304982743023,
+      "loss": 2.4355,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737629843854485,
+      "loss": 2.3691,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735949485243653,
+      "loss": 2.3789,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734263908761879,
+      "loss": 2.4238,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732573116266258,
+      "loss": 2.3066,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730877109619635,
+      "loss": 2.4141,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097291758906906,
+      "loss": 2.4512,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727469461353483,
+      "loss": 2.3652,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725757823488358,
+      "loss": 2.416,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724040978981034,
+      "loss": 2.4141,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722318929723063,
+      "loss": 2.5225,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720591677611722,
+      "loss": 2.375,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718859224550027,
+      "loss": 2.3906,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717121572446722,
+      "loss": 2.3926,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971537872321628,
+      "loss": 2.4316,
+      "step": 416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713630678778898,
+      "loss": 2.3398,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00097118774410605,
+      "loss": 2.4609,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710119011992727,
+      "loss": 2.3047,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708355393512945,
+      "loss": 2.3574,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706586587564235,
+      "loss": 2.3848,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704812596095394,
+      "loss": 2.3789,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970303342106093,
+      "loss": 2.3906,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701249064421063,
+      "loss": 2.3516,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699459528141723,
+      "loss": 2.3906,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697664814194549,
+      "loss": 2.3926,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695864924556878,
+      "loss": 2.3613,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694059861211755,
+      "loss": 2.4902,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692249626147922,
+      "loss": 2.4102,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690434221359821,
+      "loss": 2.3789,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968861364884759,
+      "loss": 2.3828,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686787910617059,
+      "loss": 2.3535,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684957008679749,
+      "loss": 2.4062,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968312094505287,
+      "loss": 2.4375,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681279721759324,
+      "loss": 2.4355,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967943334082769,
+      "loss": 2.4219,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677581804292236,
+      "loss": 2.4707,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675725114192906,
+      "loss": 2.459,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673863272575323,
+      "loss": 2.4434,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671996281490788,
+      "loss": 2.4199,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009670124142996273,
+      "loss": 2.3594,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668246859154419,
+      "loss": 2.4004,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666364432033543,
+      "loss": 2.3711,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664476863707621,
+      "loss": 2.3535,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662584156256299,
+      "loss": 2.4219,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660686311764884,
+      "loss": 2.4492,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658783332324339,
+      "loss": 2.2734,
+      "step": 447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656875220031287,
+      "loss": 2.4473,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654961976988008,
+      "loss": 2.3164,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653043605302432,
+      "loss": 2.3809,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965112010708814,
+      "loss": 2.3613,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009649191484464366,
+      "loss": 2.4141,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647257739555979,
+      "loss": 2.3164,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645318874493504,
+      "loss": 2.2949,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643374891413098,
+      "loss": 2.4004,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641425792456562,
+      "loss": 2.4316,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963947157977133,
+      "loss": 2.2891,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637512255510475,
+      "loss": 2.373,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635547821832693,
+      "loss": 2.3809,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963357828090232,
+      "loss": 2.2363,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631603634889311,
+      "loss": 2.4121,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000962962388596925,
+      "loss": 2.3027,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627639036323339,
+      "loss": 2.3535,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625649088138407,
+      "loss": 2.2734,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623654043606889,
+      "loss": 2.4551,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621653904926847,
+      "loss": 2.4824,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619648674301949,
+      "loss": 2.2969,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617638353941471,
+      "loss": 2.4395,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615622946060303,
+      "loss": 2.3574,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613602452878938,
+      "loss": 2.4551,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611576876623467,
+      "loss": 2.3398,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609546219525589,
+      "loss": 2.291,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607510483822595,
+      "loss": 2.3457,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605469671757372,
+      "loss": 2.3223,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603423785578404,
+      "loss": 2.3691,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601372827539763,
+      "loss": 2.3652,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599316799901106,
+      "loss": 2.3652,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009597255704927681,
+      "loss": 2.416,
+      "step": 478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009595189544890312,
+      "loss": 2.4785,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593118322065414,
+      "loss": 2.3477,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000959104203873497,
+      "loss": 2.3125,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588960697186539,
+      "loss": 2.3867,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586874299713262,
+      "loss": 2.2949,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584782848613842,
+      "loss": 2.3027,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582686346192551,
+      "loss": 2.4453,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580584794759229,
+      "loss": 2.3984,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578478196629277,
+      "loss": 2.3848,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576366554123656,
+      "loss": 2.4707,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574249869568887,
+      "loss": 2.4375,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572128145297044,
+      "loss": 2.373,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570001383645752,
+      "loss": 2.2539,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956786958695819,
+      "loss": 2.3262,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565732757583081,
+      "loss": 2.4336,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563590897874696,
+      "loss": 2.3848,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561444010192845,
+      "loss": 2.3594,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955929209690288,
+      "loss": 2.3672,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955713516037569,
+      "loss": 2.3047,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554973202987694,
+      "loss": 2.3789,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552806227120849,
+      "loss": 2.3652,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550634235162636,
+      "loss": 2.4238,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548457229506067,
+      "loss": 2.3594,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546275212549676,
+      "loss": 2.3574,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544088186697515,
+      "loss": 2.4824,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954189615435916,
+      "loss": 2.332,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539699117949701,
+      "loss": 2.4277,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537497079889738,
+      "loss": 2.4277,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535290042605387,
+      "loss": 2.3359,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533078008528269,
+      "loss": 2.4434,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009530860980095508,
+      "loss": 2.3457,
+      "step": 509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528638959749735,
+      "loss": 2.3672,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526411949939082,
+      "loss": 2.3789,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524179953117172,
+      "loss": 2.4121,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009521942971743124,
+      "loss": 2.3457,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519701008281554,
+      "loss": 2.3594,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951745406520256,
+      "loss": 2.373,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515202144981734,
+      "loss": 2.3203,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512945250100144,
+      "loss": 2.334,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510683383044345,
+      "loss": 2.4961,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508416546306364,
+      "loss": 2.4434,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506144742383711,
+      "loss": 2.375,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503867973779363,
+      "loss": 2.3887,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501586243001765,
+      "loss": 2.4336,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499299552564837,
+      "loss": 2.3027,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497007904987958,
+      "loss": 2.334,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494711302795967,
+      "loss": 2.3535,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492409748519166,
+      "loss": 2.4023,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490103244693311,
+      "loss": 2.2812,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487791793859612,
+      "loss": 2.4688,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485475398564728,
+      "loss": 2.3359,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483154061360766,
+      "loss": 2.4004,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480827784805278,
+      "loss": 2.3301,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478496571461259,
+      "loss": 2.2773,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476160423897143,
+      "loss": 2.3516,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473819344686798,
+      "loss": 2.4023,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471473336409528,
+      "loss": 2.3867,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469122401650068,
+      "loss": 2.3906,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466766542998576,
+      "loss": 2.3867,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464405763050641,
+      "loss": 2.4004,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462040064407272,
+      "loss": 2.4043,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009459669449674895,
+      "loss": 2.3926,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457293921465355,
+      "loss": 2.332,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945491348239591,
+      "loss": 2.3301,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452528135089228,
+      "loss": 2.4102,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450137882173384,
+      "loss": 2.2812,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447742726281859,
+      "loss": 2.334,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445342670053536,
+      "loss": 2.291,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442937716132696,
+      "loss": 2.3906,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009440527867169016,
+      "loss": 2.4004,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438113125817569,
+      "loss": 2.3242,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435693494738813,
+      "loss": 2.4121,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433268976598599,
+      "loss": 2.3613,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430839574068158,
+      "loss": 2.3887,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428405289824104,
+      "loss": 2.334,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425966126548431,
+      "loss": 2.3164,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423522086928505,
+      "loss": 2.3652,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421073173657066,
+      "loss": 2.3164,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418619389432228,
+      "loss": 2.3086,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416160736957464,
+      "loss": 2.4102,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413697218941617,
+      "loss": 2.3828,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411228838098886,
+      "loss": 2.3379,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408755597148833,
+      "loss": 2.3457,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406277498816366,
+      "loss": 2.4199,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009403794545831754,
+      "loss": 2.3828,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401306740930608,
+      "loss": 2.4395,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398814086853891,
+      "loss": 2.3945,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396316586347904,
+      "loss": 2.3477,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393814242164285,
+      "loss": 2.3223,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391307057060016,
+      "loss": 2.4434,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388795033797406,
+      "loss": 2.3633,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009386278175144099,
+      "loss": 2.3828,
+      "step": 570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009383756483873064,
+      "loss": 2.3652,
+      "step": 571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009381229962762594,
+      "loss": 2.377,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378698614596307,
+      "loss": 2.375,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009376162442163132,
+      "loss": 2.2715,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373621448257321,
+      "loss": 2.3418,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371075635678434,
+      "loss": 2.4277,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368525007231341,
+      "loss": 2.334,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365969565726217,
+      "loss": 2.377,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000936340931397854,
+      "loss": 2.4199,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000936084425480909,
+      "loss": 2.3691,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358274391043941,
+      "loss": 2.3398,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000935569972551446,
+      "loss": 2.2988,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353120261057309,
+      "loss": 2.3789,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350536000514429,
+      "loss": 2.3281,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347946946733055,
+      "loss": 2.4141,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009345353102565696,
+      "loss": 2.2148,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000934275447087014,
+      "loss": 2.2617,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340151054509453,
+      "loss": 2.3887,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337542856351968,
+      "loss": 2.3965,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334929879271291,
+      "loss": 2.4238,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332312126146289,
+      "loss": 2.3398,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009329689599861092,
+      "loss": 2.3047,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327062303305092,
+      "loss": 2.3359,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324430239372932,
+      "loss": 2.4102,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009321793410964512,
+      "loss": 2.4277,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009319151820984978,
+      "loss": 2.2871,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009316505472344723,
+      "loss": 2.3574,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009313854367959385,
+      "loss": 2.3652,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311198510749838,
+      "loss": 2.3848,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308537903642195,
+      "loss": 2.2891,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009305872549567802,
+      "loss": 2.3496,
+      "step": 601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009303202451463234,
+      "loss": 2.3145,
+      "step": 602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009300527612270293,
+      "loss": 2.3633,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297848034936007,
+      "loss": 2.3594,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009295163722412618,
+      "loss": 2.4043,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009292474677657593,
+      "loss": 2.373,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289780903633608,
+      "loss": 2.2891,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928708240330855,
+      "loss": 2.4082,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284379179655513,
+      "loss": 2.3223,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009281671235652798,
+      "loss": 2.2539,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278958574283902,
+      "loss": 2.2715,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009276241198537523,
+      "loss": 2.2539,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273519111407553,
+      "loss": 2.4004,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270792315893071,
+      "loss": 2.3945,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268060814998347,
+      "loss": 2.3672,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009265324611732835,
+      "loss": 2.3262,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009262583709111169,
+      "loss": 2.3848,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259838110153162,
+      "loss": 2.2832,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009257087817883798,
+      "loss": 2.3154,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009254332835333237,
+      "loss": 2.3633,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251573165536799,
+      "loss": 2.3027,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009248808811534976,
+      "loss": 2.2988,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246039776373416,
+      "loss": 2.2832,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243266063102926,
+      "loss": 2.3633,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000924048767477947,
+      "loss": 2.4238,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237704614464157,
+      "loss": 2.3164,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009234916885223246,
+      "loss": 2.2871,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232124490128142,
+      "loss": 2.332,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922932743225539,
+      "loss": 2.3301,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009226525714686669,
+      "loss": 2.3145,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009223719340508793,
+      "loss": 2.3125,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220908312813714,
+      "loss": 2.3281,
+      "step": 632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009218092634698499,
+      "loss": 2.416,
+      "step": 633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215272309265343,
+      "loss": 2.3652,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009212447339621568,
+      "loss": 2.3164,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209617728879604,
+      "loss": 2.373,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206783480156999,
+      "loss": 2.3535,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203944596576408,
+      "loss": 2.3691,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201101081265596,
+      "loss": 2.3691,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198252937357427,
+      "loss": 2.3027,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919540016798987,
+      "loss": 2.3066,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009192542776305983,
+      "loss": 2.2793,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009189680765453922,
+      "loss": 2.2324,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009186814138586933,
+      "loss": 2.4453,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183942898863343,
+      "loss": 2.3828,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181067049446565,
+      "loss": 2.4023,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178186593505088,
+      "loss": 2.293,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917530153421248,
+      "loss": 2.3555,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009172411874747377,
+      "loss": 2.3477,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169517618293486,
+      "loss": 2.3438,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009166618768039575,
+      "loss": 2.3848,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916371532717948,
+      "loss": 2.2852,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009160807298912087,
+      "loss": 2.3516,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009157894686441342,
+      "loss": 2.3223,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000915497749297624,
+      "loss": 2.2969,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000915205572173082,
+      "loss": 2.3574,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009149129375924169,
+      "loss": 2.2715,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009146198458780413,
+      "loss": 2.2656,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143262973528712,
+      "loss": 2.3809,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009140322923403262,
+      "loss": 2.4277,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137378311643287,
+      "loss": 2.3633,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009134429141493037,
+      "loss": 2.3398,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131475416201785,
+      "loss": 2.3828,
+      "step": 663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009128517139023822,
+      "loss": 2.2539,
+      "step": 664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125554313218451,
+      "loss": 2.4102,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009122586942049993,
+      "loss": 2.3789,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009119615028787771,
+      "loss": 2.2578,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116638576706117,
+      "loss": 2.4414,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113657589084359,
+      "loss": 2.2891,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110672069206825,
+      "loss": 2.3281,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009107682020362837,
+      "loss": 2.3066,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104687445846705,
+      "loss": 2.3184,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009101688348957726,
+      "loss": 2.332,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000909868473300018,
+      "loss": 2.29,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095676601283322,
+      "loss": 2.2148,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009092663957121391,
+      "loss": 2.3242,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009089646803833589,
+      "loss": 2.3105,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908662514474409,
+      "loss": 2.2559,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009083598983182029,
+      "loss": 2.293,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009080568322481508,
+      "loss": 2.25,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907753316598158,
+      "loss": 2.332,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074493517026255,
+      "loss": 2.3691,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907144937896449,
+      "loss": 2.3359,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068400755150189,
+      "loss": 2.248,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065347648942198,
+      "loss": 2.3125,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00090622900637043,
+      "loss": 2.3809,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009059228002805217,
+      "loss": 2.3418,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00090561614696186,
+      "loss": 2.373,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009053090467523025,
+      "loss": 2.2793,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009050014999901994,
+      "loss": 2.3184,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046935070143929,
+      "loss": 2.373,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009043850681642167,
+      "loss": 2.3711,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040761837794959,
+      "loss": 2.3066,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009037668542005464,
+      "loss": 2.3418,
+      "step": 694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034570797681745,
+      "loss": 2.3301,
+      "step": 695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009031468608236766,
+      "loss": 2.3555,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009028361977088393,
+      "loss": 2.3301,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025250907659379,
+      "loss": 2.291,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009022135403377371,
+      "loss": 2.3613,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009019015467674902,
+      "loss": 2.2812,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015891103989385,
+      "loss": 2.2344,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012762315763115,
+      "loss": 2.3828,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009629106443261,
+      "loss": 2.332,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009006491479481857,
+      "loss": 2.3398,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003349438335813,
+      "loss": 2.3535,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009000202986466899,
+      "loss": 2.332,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899705212734174,
+      "loss": 2.2832,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993896864431826,
+      "loss": 2.3203,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990737201213489,
+      "loss": 2.2529,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987573141167915,
+      "loss": 2.3135,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008984404687781135,
+      "loss": 2.3496,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008981231844544015,
+      "loss": 2.3789,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008978054614952265,
+      "loss": 2.2988,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897487300250642,
+      "loss": 2.3711,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971687010711851,
+      "loss": 2.2402,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008968496643078745,
+      "loss": 2.3086,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896530190312212,
+      "loss": 2.334,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008962102794361806,
+      "loss": 2.3447,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008958899320322447,
+      "loss": 2.3535,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955691484533496,
+      "loss": 2.3789,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008952479290529213,
+      "loss": 2.3516,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949262741848657,
+      "loss": 2.4238,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000894604184203569,
+      "loss": 2.2861,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008942816594638961,
+      "loss": 2.373,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939587003211915,
+      "loss": 2.2969,
+      "step": 725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008936353071312781,
+      "loss": 2.3301,
+      "step": 726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933114802504567,
+      "loss": 2.3164,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008929872200355067,
+      "loss": 2.3789,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008926625268436839,
+      "loss": 2.3203,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000892337401032722,
+      "loss": 2.3047,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008920118429608309,
+      "loss": 2.2773,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008916858529866969,
+      "loss": 2.3633,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891359431469482,
+      "loss": 2.377,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910325787688238,
+      "loss": 2.3691,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907052952448351,
+      "loss": 2.3301,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008903775812581031,
+      "loss": 2.4004,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008900494371696892,
+      "loss": 2.2344,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008897208633411291,
+      "loss": 2.3086,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008893918601344314,
+      "loss": 2.3438,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008890624279120782,
+      "loss": 2.3652,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887325670370244,
+      "loss": 2.2969,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008884022778726964,
+      "loss": 2.248,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880715607829932,
+      "loss": 2.2793,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008877404161322851,
+      "loss": 2.2754,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874088442854136,
+      "loss": 2.3359,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008870768456076903,
+      "loss": 2.2832,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867444204648977,
+      "loss": 2.3008,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864115692232877,
+      "loss": 2.3555,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008860782922495822,
+      "loss": 2.3027,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857445899109715,
+      "loss": 2.2832,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000885410462575115,
+      "loss": 2.293,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008850759106101403,
+      "loss": 2.3047,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008847409343846426,
+      "loss": 2.2754,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844055342676849,
+      "loss": 2.3516,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840697106287966,
+      "loss": 2.3047,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008837334638379746,
+      "loss": 2.3926,
+      "step": 756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833967942656813,
+      "loss": 2.3145,
+      "step": 757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008830597022828452,
+      "loss": 2.4082,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827221882608601,
+      "loss": 2.2539,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823842525715848,
+      "loss": 2.2793,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008820458955873428,
+      "loss": 2.2988,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817071176809217,
+      "loss": 2.3086,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008813679192255726,
+      "loss": 2.3574,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008810283005950104,
+      "loss": 2.3711,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008806882621634125,
+      "loss": 2.2969,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803478043054192,
+      "loss": 2.3359,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800069273961327,
+      "loss": 2.3086,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008796656318111167,
+      "loss": 2.2441,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008793239179263966,
+      "loss": 2.2734,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008789817861184582,
+      "loss": 2.3242,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008786392367642484,
+      "loss": 2.3223,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008782962702411734,
+      "loss": 2.2637,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008779528869270997,
+      "loss": 2.2871,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776090872003524,
+      "loss": 2.2812,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008772648714397155,
+      "loss": 2.3418,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008769202400244321,
+      "loss": 2.3105,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008765751933342022,
+      "loss": 2.3496,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762297317491837,
+      "loss": 2.3555,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875883855649992,
+      "loss": 2.291,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008755375654176989,
+      "loss": 2.3574,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008751908614338324,
+      "loss": 2.2979,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748437440803765,
+      "loss": 2.2891,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744962137397705,
+      "loss": 2.3711,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008741482707949089,
+      "loss": 2.4004,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008737999156291405,
+      "loss": 2.2656,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008734511486262687,
+      "loss": 2.2676,
+      "step": 786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008731019701705501,
+      "loss": 2.3379,
+      "step": 787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000872752380646695,
+      "loss": 2.4062,
+      "step": 788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008724023804398666,
+      "loss": 2.3418,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720519699356804,
+      "loss": 2.3477,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717011495202041,
+      "loss": 2.418,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008713499195799568,
+      "loss": 2.3203,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008709982805019092,
+      "loss": 2.2676,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008706462326734824,
+      "loss": 2.3262,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702937764825478,
+      "loss": 2.2256,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000869940912317427,
+      "loss": 2.2676,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008695876405668911,
+      "loss": 2.3301,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692339616201599,
+      "loss": 2.2383,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008688798758669024,
+      "loss": 2.2793,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008685253836972351,
+      "loss": 2.2695,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008681704855017227,
+      "loss": 2.2773,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008678151816713775,
+      "loss": 2.3672,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674594725976578,
+      "loss": 2.3008,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008671033586724697,
+      "loss": 2.3496,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008667468402881642,
+      "loss": 2.3301,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008663899178375381,
+      "loss": 2.3086,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000866032591713834,
+      "loss": 2.2588,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008656748623107391,
+      "loss": 2.2773,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008653167300223841,
+      "loss": 2.3711,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649581952433444,
+      "loss": 2.2754,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008645992583686386,
+      "loss": 2.373,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008642399197937285,
+      "loss": 2.2539,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008638801799145181,
+      "loss": 2.3164,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008635200391273538,
+      "loss": 2.3027,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631594978290236,
+      "loss": 2.3477,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862798556416757,
+      "loss": 2.3633,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862437215288224,
+      "loss": 2.3418,
+      "step": 817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000862075474841535,
+      "loss": 2.2324,
+      "step": 818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617133354752407,
+      "loss": 2.3223,
+      "step": 819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008613507975883312,
+      "loss": 2.3574,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008609878615802352,
+      "loss": 2.2383,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008606245278508207,
+      "loss": 2.418,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602607968003935,
+      "loss": 2.2344,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008598966688296972,
+      "loss": 2.2148,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008595321443399129,
+      "loss": 2.3203,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008591672237326585,
+      "loss": 2.3066,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588019074099877,
+      "loss": 2.293,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008584361957743913,
+      "loss": 2.3047,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008580700892287951,
+      "loss": 2.3086,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008577035881765596,
+      "loss": 2.2344,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008573366930214806,
+      "loss": 2.3262,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008569694041677875,
+      "loss": 2.3301,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008566017220201443,
+      "loss": 2.2324,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008562336469836474,
+      "loss": 2.2676,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008558651794638264,
+      "loss": 2.3027,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008554963198666439,
+      "loss": 2.2979,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008551270685984934,
+      "loss": 2.334,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008547574260662009,
+      "loss": 2.4102,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008543873926770228,
+      "loss": 2.3281,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008540169688386467,
+      "loss": 2.3164,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008536461549591896,
+      "loss": 2.2949,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008532749514471993,
+      "loss": 2.2695,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529033587116519,
+      "loss": 2.3164,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008525313771619528,
+      "loss": 2.2637,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008521590072079359,
+      "loss": 2.3223,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008517862492598625,
+      "loss": 2.332,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008514131037284219,
+      "loss": 2.3262,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510395710247302,
+      "loss": 2.2656,
+      "step": 848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008506656515603301,
+      "loss": 2.3164,
+      "step": 849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008502913457471906,
+      "loss": 2.2344,
+      "step": 850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008499166539977057,
+      "loss": 2.377,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008495415767246957,
+      "loss": 2.2695,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008491661143414047,
+      "loss": 2.293,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008487902672615013,
+      "loss": 2.2188,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484140358990783,
+      "loss": 2.3809,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008480374206686517,
+      "loss": 2.2891,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008476604219851602,
+      "loss": 2.3008,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008472830402639656,
+      "loss": 2.1953,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008469052759208506,
+      "loss": 2.2598,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008465271293720209,
+      "loss": 2.3008,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846148601034102,
+      "loss": 2.2539,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008457696913241407,
+      "loss": 2.2695,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000845390400659604,
+      "loss": 2.332,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008450107294583785,
+      "loss": 2.2559,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008446306781387698,
+      "loss": 2.3281,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008442502471195027,
+      "loss": 2.2568,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008438694368197202,
+      "loss": 2.2715,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000843488247658983,
+      "loss": 2.2988,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431066800572695,
+      "loss": 2.2891,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008427247344349748,
+      "loss": 2.3008,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000842342411212911,
+      "loss": 2.2969,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008419597108123054,
+      "loss": 2.2715,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008415766336548012,
+      "loss": 2.334,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008411931801624571,
+      "loss": 2.3066,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008408093507577461,
+      "loss": 2.2949,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008404251458635551,
+      "loss": 2.3027,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008400405659031852,
+      "loss": 2.3066,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008396556113003504,
+      "loss": 2.3672,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392702824791778,
+      "loss": 2.248,
+      "step": 879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008388845798642062,
+      "loss": 2.3555,
+      "step": 880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008384985038803869,
+      "loss": 2.334,
+      "step": 881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838112054953082,
+      "loss": 2.2246,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837725233508065,
+      "loss": 2.1816,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008373380399715196,
+      "loss": 2.2568,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008369504747700391,
+      "loss": 2.3594,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008365625383306269,
+      "loss": 2.2598,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000836174231080695,
+      "loss": 2.2539,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008357855534480643,
+      "loss": 2.3301,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008353965058609632,
+      "loss": 2.334,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008350070887480286,
+      "loss": 2.3008,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008346173025383033,
+      "loss": 2.3535,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008342271476612377,
+      "loss": 2.2832,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008338366245466884,
+      "loss": 2.332,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008334457336249173,
+      "loss": 2.2891,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008330544753265913,
+      "loss": 2.3418,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008326628500827827,
+      "loss": 2.2852,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008322708583249678,
+      "loss": 2.3047,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008318785004850268,
+      "loss": 2.2988,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314857769952429,
+      "loss": 2.2871,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008310926882883026,
+      "loss": 2.2344,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008306992347972946,
+      "loss": 2.2832,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008303054169557094,
+      "loss": 2.3066,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008299112351974389,
+      "loss": 2.3691,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008295166899567764,
+      "loss": 2.2656,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008291217816684152,
+      "loss": 2.2969,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008287265107674484,
+      "loss": 2.293,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008283308776893692,
+      "loss": 2.2949,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008279348828700697,
+      "loss": 2.2637,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008275385267458404,
+      "loss": 2.3477,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271418097533696,
+      "loss": 2.2822,
+      "step": 910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008267447323297436,
+      "loss": 2.293,
+      "step": 911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826347294912446,
+      "loss": 2.165,
+      "step": 912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008259494979393563,
+      "loss": 2.3066,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008255513418487507,
+      "loss": 2.3027,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000825152827079301,
+      "loss": 2.3086,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000824753954070074,
+      "loss": 2.3301,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008243547232605313,
+      "loss": 2.293,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008239551350905287,
+      "loss": 2.334,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008235551900003158,
+      "loss": 2.3184,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008231548884305353,
+      "loss": 2.291,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008227542308222227,
+      "loss": 2.2988,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008223532176168058,
+      "loss": 2.2539,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219518492561043,
+      "loss": 2.3223,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215501261823288,
+      "loss": 2.1885,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008211480488380811,
+      "loss": 2.251,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008207456176663533,
+      "loss": 2.2559,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008203428331105269,
+      "loss": 2.3379,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008199396956143735,
+      "loss": 2.3086,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195362056220527,
+      "loss": 2.29,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008191323635781133,
+      "loss": 2.3184,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008187281699274915,
+      "loss": 2.2812,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008183236251155106,
+      "loss": 2.3223,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008179187295878813,
+      "loss": 2.3281,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175134837907006,
+      "loss": 2.3164,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008171078881704514,
+      "loss": 2.3301,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167019431740022,
+      "loss": 2.3184,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008162956492486056,
+      "loss": 2.2119,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008158890068418996,
+      "loss": 2.2363,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008154820164019058,
+      "loss": 2.3066,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008150746783770294,
+      "loss": 2.3164,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008146669932160581,
+      "loss": 2.2441,
+      "step": 941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008142589613681626,
+      "loss": 2.2578,
+      "step": 942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008138505832828949,
+      "loss": 2.3164,
+      "step": 943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134418594101893,
+      "loss": 2.3672,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008130327902003604,
+      "loss": 2.2266,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126233761041038,
+      "loss": 2.3008,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008122136175724946,
+      "loss": 2.2842,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118035150569876,
+      "loss": 2.3496,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113930690094167,
+      "loss": 2.3223,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008109822798819941,
+      "loss": 2.3047,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008105711481273102,
+      "loss": 2.2734,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008101596741983323,
+      "loss": 2.3008,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809747858548406,
+      "loss": 2.2422,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008093357016312517,
+      "loss": 2.2305,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008089232039009671,
+      "loss": 2.3262,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808510365812025,
+      "loss": 2.3105,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008080971878192729,
+      "loss": 2.3086,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807683670377933,
+      "loss": 2.332,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008072698139436016,
+      "loss": 2.3711,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008068556189722484,
+      "loss": 2.2695,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064410859202162,
+      "loss": 2.2734,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008060262152442201,
+      "loss": 2.3574,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056110074013469,
+      "loss": 2.2539,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008051954628490558,
+      "loss": 2.252,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000804779582045176,
+      "loss": 2.293,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008043633654479078,
+      "loss": 2.3867,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803946813515821,
+      "loss": 2.2812,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035299267078553,
+      "loss": 2.2891,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803112705483319,
+      "loss": 2.3398,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008026951503018891,
+      "loss": 2.2598,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008022772616236103,
+      "loss": 2.2734,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000801859039908895,
+      "loss": 2.2188,
+      "step": 972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014404856185221,
+      "loss": 2.2734,
+      "step": 973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008010215992136373,
+      "loss": 2.2266,
+      "step": 974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008006023811557526,
+      "loss": 2.3027,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008001828319067443,
+      "loss": 2.3086,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007997629519288545,
+      "loss": 2.2314,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007993427416846891,
+      "loss": 2.4141,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007989222016372185,
+      "loss": 2.2822,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000798501332249776,
+      "loss": 2.2812,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000798080133986058,
+      "loss": 2.3184,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007976586073101228,
+      "loss": 2.2812,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007972367526863911,
+      "loss": 2.2344,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007968145705796446,
+      "loss": 2.252,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796392061455026,
+      "loss": 2.2891,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007959692257780379,
+      "loss": 2.375,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007955460640145432,
+      "loss": 2.3086,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007951225766307637,
+      "loss": 2.3047,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007946987640932801,
+      "loss": 2.2461,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007942746268690314,
+      "loss": 2.2949,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007938501654253141,
+      "loss": 2.2539,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007934253802297822,
+      "loss": 2.2695,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007930002717504461,
+      "loss": 2.2012,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007925748404556727,
+      "loss": 2.2031,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007921490868141843,
+      "loss": 2.2598,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007917230112950585,
+      "loss": 2.3984,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007912966143677272,
+      "loss": 2.418,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007908698965019772,
+      "loss": 2.2314,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007904428581679479,
+      "loss": 2.3262,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007900154998361324,
+      "loss": 2.291,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007895878219773765,
+      "loss": 2.2539,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891598250628775,
+      "loss": 2.2031,
+      "step": 1002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007887315095641844,
+      "loss": 2.2852,
+      "step": 1003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007883028759531977,
+      "loss": 2.332,
+      "step": 1004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878739247021676,
+      "loss": 2.3301,
+      "step": 1005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007874446562836948,
+      "loss": 2.3828,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007870150711707291,
+      "loss": 2.2344,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007865851698365696,
+      "loss": 2.2891,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007861549527548636,
+      "loss": 2.2891,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007857244203996061,
+      "loss": 2.1689,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007852935732451398,
+      "loss": 2.3223,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000784862411766154,
+      "loss": 2.3281,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007844309364376844,
+      "loss": 2.1709,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007839991477351123,
+      "loss": 2.2832,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835670461341646,
+      "loss": 2.2969,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007831346321109128,
+      "loss": 2.291,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007827019061417724,
+      "loss": 2.2793,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000782268868703503,
+      "loss": 2.1875,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818355202732068,
+      "loss": 2.3301,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007814018613283293,
+      "loss": 2.2539,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007809678923466575,
+      "loss": 2.2939,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007805336138063204,
+      "loss": 2.2363,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007800990261857879,
+      "loss": 2.2441,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007796641299638706,
+      "loss": 2.25,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007792289256197186,
+      "loss": 2.3184,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007787934136328223,
+      "loss": 2.248,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007783575944830102,
+      "loss": 2.3105,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007779214686504497,
+      "loss": 2.252,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000777485036615646,
+      "loss": 2.2363,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007770482988594419,
+      "loss": 2.2422,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007766112558630163,
+      "loss": 2.2461,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007761739081078851,
+      "loss": 2.3359,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007757362560758999,
+      "loss": 2.3047,
+      "step": 1033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007752983002492473,
+      "loss": 2.2471,
+      "step": 1034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007748600411104487,
+      "loss": 2.2949,
+      "step": 1035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007744214791423596,
+      "loss": 2.2969,
+      "step": 1036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007739826148281694,
+      "loss": 2.3691,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007735434486514001,
+      "loss": 2.2559,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007731039810959071,
+      "loss": 2.2734,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000772664212645877,
+      "loss": 2.332,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007722241437858284,
+      "loss": 2.3125,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007717837750006106,
+      "loss": 2.2969,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007713431067754038,
+      "loss": 2.291,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007709021395957174,
+      "loss": 2.2246,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000770460873947391,
+      "loss": 2.2734,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007700193103165923,
+      "loss": 2.3281,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000769577449189818,
+      "loss": 2.2578,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007691352910538918,
+      "loss": 2.2246,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007686928363959651,
+      "loss": 2.2256,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682500857035162,
+      "loss": 2.2969,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767807039464349,
+      "loss": 2.2402,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007673636981665935,
+      "loss": 2.3184,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007669200622987044,
+      "loss": 2.2812,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007664761323494613,
+      "loss": 2.1992,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007660319088079675,
+      "loss": 2.2832,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00076558739216365,
+      "loss": 2.2949,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007651425829062589,
+      "loss": 2.2734,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007646974815258658,
+      "loss": 2.2383,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007642520885128652,
+      "loss": 2.2031,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007638064043579722,
+      "loss": 2.2148,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007633604295522234,
+      "loss": 2.3164,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007629141645869751,
+      "loss": 2.2324,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007624676099539028,
+      "loss": 2.2676,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007620207661450021,
+      "loss": 2.2891,
+      "step": 1064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007615736336525871,
+      "loss": 2.3379,
+      "step": 1065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007611262129692892,
+      "loss": 2.2656,
+      "step": 1066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007606785045880577,
+      "loss": 2.2783,
+      "step": 1067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007602305090021596,
+      "loss": 2.2676,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007597822267051773,
+      "loss": 2.2578,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593336581910093,
+      "loss": 2.291,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007588848039538698,
+      "loss": 2.2285,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007584356644882875,
+      "loss": 2.1895,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007579862402891058,
+      "loss": 2.2344,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007575365318514811,
+      "loss": 2.2578,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570865396708835,
+      "loss": 2.3496,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007566362642430955,
+      "loss": 2.3066,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007561857060642119,
+      "loss": 2.3184,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007557348656306384,
+      "loss": 2.2607,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007552837434390927,
+      "loss": 2.3184,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007548323399866019,
+      "loss": 2.3066,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007543806557705039,
+      "loss": 2.2891,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007539286912884447,
+      "loss": 2.2344,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007534764470383808,
+      "loss": 2.3301,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007530239235185753,
+      "loss": 2.3438,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007525711212275998,
+      "loss": 2.334,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007521180406643331,
+      "loss": 2.3066,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007516646823279603,
+      "loss": 2.2617,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007512110467179727,
+      "loss": 2.3242,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007507571343341668,
+      "loss": 2.2617,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007503029456766446,
+      "loss": 2.3086,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000749848481245812,
+      "loss": 2.2285,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007493937415423789,
+      "loss": 2.2148,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007489387270673584,
+      "loss": 2.1797,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007484834383220665,
+      "loss": 2.2529,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007480278758081213,
+      "loss": 2.168,
+      "step": 1095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007475720400274424,
+      "loss": 2.2598,
+      "step": 1096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007471159314822506,
+      "loss": 2.3027,
+      "step": 1097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007466595506750674,
+      "loss": 2.3008,
+      "step": 1098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007462028981087139,
+      "loss": 2.2832,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007457459742863108,
+      "loss": 2.3613,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007452887797112776,
+      "loss": 2.3398,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007448313148873325,
+      "loss": 2.2246,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007443735803184905,
+      "loss": 2.252,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007439155765090648,
+      "loss": 2.293,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007434573039636649,
+      "loss": 2.3457,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742998763187196,
+      "loss": 2.2754,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007425399546848593,
+      "loss": 2.2949,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007420808789621507,
+      "loss": 2.2207,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007416215365248608,
+      "loss": 2.3105,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000741161927879074,
+      "loss": 2.373,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007407020535311673,
+      "loss": 2.2695,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007402419139878116,
+      "loss": 2.2344,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007397815097559691,
+      "loss": 2.2793,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007393208413428941,
+      "loss": 2.2637,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007388599092561315,
+      "loss": 2.2832,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007383987140035171,
+      "loss": 2.3496,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007379372560931768,
+      "loss": 2.2051,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374755360335252,
+      "loss": 2.2617,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007370135543332663,
+      "loss": 2.1934,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365513115013923,
+      "loss": 2.3496,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000736088808047183,
+      "loss": 2.3301,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356260444802051,
+      "loss": 2.2988,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007351630213103123,
+      "loss": 2.3203,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007346997390476444,
+      "loss": 2.2227,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007342361982026261,
+      "loss": 2.209,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007337723992859673,
+      "loss": 2.2109,
+      "step": 1126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333083428086622,
+      "loss": 2.1895,
+      "step": 1127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007328440292819891,
+      "loss": 2.2676,
+      "step": 1128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007323794592175091,
+      "loss": 2.3281,
+      "step": 1129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007319146331270658,
+      "loss": 2.2188,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007314495515227854,
+      "loss": 2.3008,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007309842149170753,
+      "loss": 2.2598,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007305186238226238,
+      "loss": 2.2637,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007300527787523997,
+      "loss": 2.2617,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295866802196518,
+      "loss": 2.2139,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007291203287379078,
+      "loss": 2.3164,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007286537248209742,
+      "loss": 2.2852,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007281868689829357,
+      "loss": 2.3867,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007277197617381545,
+      "loss": 2.2773,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072725240360127,
+      "loss": 2.1846,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007267847950871975,
+      "loss": 2.2441,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007263169367111289,
+      "loss": 2.2891,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007258488289885307,
+      "loss": 2.2734,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007253804724351445,
+      "loss": 2.2598,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007249118675669857,
+      "loss": 2.3125,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007244430149003439,
+      "loss": 2.2305,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007239739149517813,
+      "loss": 2.252,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007235045682381325,
+      "loss": 2.2695,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230349752765038,
+      "loss": 2.3574,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007225651365842735,
+      "loss": 2.3008,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072209505267909,
+      "loss": 2.2227,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000721624724078872,
+      "loss": 2.2822,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007211541513018082,
+      "loss": 2.3027,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007206833348663557,
+      "loss": 2.3047,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007202122752912405,
+      "loss": 2.2529,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007197409730954561,
+      "loss": 2.2363,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007192694287982635,
+      "loss": 2.1855,
+      "step": 1157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007187976429191909,
+      "loss": 2.2109,
+      "step": 1158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007183256159780321,
+      "loss": 2.2539,
+      "step": 1159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007178533484948462,
+      "loss": 2.2598,
+      "step": 1160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007173808409899582,
+      "loss": 2.3496,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000716908093983957,
+      "loss": 2.2695,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007164351079976954,
+      "loss": 2.293,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159618835522895,
+      "loss": 2.2949,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007154884211691185,
+      "loss": 2.3066,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007150147213698232,
+      "loss": 2.1426,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000714540784676306,
+      "loss": 2.2012,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007140666116107309,
+      "loss": 2.3574,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007135922026955217,
+      "loss": 2.332,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007131175584533625,
+      "loss": 2.2852,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007126426794071963,
+      "loss": 2.2422,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007121675660802249,
+      "loss": 2.2676,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007116922189959085,
+      "loss": 2.2617,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007112166386779645,
+      "loss": 2.373,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007107408256503674,
+      "loss": 2.2559,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007102647804373479,
+      "loss": 2.293,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007097885035633932,
+      "loss": 2.2246,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007093119955532447,
+      "loss": 2.2734,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007088352569318991,
+      "loss": 2.2676,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007083582882246073,
+      "loss": 2.2793,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007078810899568733,
+      "loss": 2.3086,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007074036626544541,
+      "loss": 2.2461,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007069260068433592,
+      "loss": 2.2949,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007064481230498499,
+      "loss": 2.3555,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007059700118004384,
+      "loss": 2.2383,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007054916736218875,
+      "loss": 2.2598,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007050131090412106,
+      "loss": 2.3457,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00070453431858567,
+      "loss": 2.2656,
+      "step": 1188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007040553027827767,
+      "loss": 2.1914,
+      "step": 1189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007035760621602907,
+      "loss": 2.2656,
+      "step": 1190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000703096597246219,
+      "loss": 2.3145,
+      "step": 1191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007026169085688161,
+      "loss": 2.2656,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007021369966565827,
+      "loss": 2.2256,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007016568620382657,
+      "loss": 2.2949,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007011765052428576,
+      "loss": 2.2656,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007006959267995953,
+      "loss": 2.2949,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007002151272379598,
+      "loss": 2.3086,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000699734107087676,
+      "loss": 2.2129,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006992528668787121,
+      "loss": 2.3789,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006987714071412781,
+      "loss": 2.2891,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698289728405826,
+      "loss": 2.25,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006978078312030498,
+      "loss": 2.291,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006973257160638835,
+      "loss": 2.2871,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006968433835195011,
+      "loss": 2.3281,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006963608341013167,
+      "loss": 2.3105,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000695878068340983,
+      "loss": 2.2344,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006953950867703911,
+      "loss": 2.1016,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006949118899216698,
+      "loss": 2.2129,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006944284783271851,
+      "loss": 2.2178,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006939448525195399,
+      "loss": 2.2676,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006934610130315728,
+      "loss": 2.3027,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006929769603963578,
+      "loss": 2.2949,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000692492695147204,
+      "loss": 2.2363,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006920082178176545,
+      "loss": 2.3027,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006915235289414861,
+      "loss": 2.2773,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910386290527084,
+      "loss": 2.2422,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006905535186855644,
+      "loss": 2.2285,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006900681983745283,
+      "loss": 2.2793,
+      "step": 1218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006895826686543055,
+      "loss": 2.3008,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006890969300598324,
+      "loss": 2.2949,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006886109831262754,
+      "loss": 2.2803,
+      "step": 1221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006881248283890306,
+      "loss": 2.2422,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006876384663837227,
+      "loss": 2.3242,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006871518976462052,
+      "loss": 2.1836,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006866651227125593,
+      "loss": 2.3477,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006861781421190931,
+      "loss": 2.3086,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856909564023413,
+      "loss": 2.2656,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852035660990651,
+      "loss": 2.1504,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006847159717462504,
+      "loss": 2.2656,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006842281738811083,
+      "loss": 2.2852,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006837401730410743,
+      "loss": 2.248,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006832519697638071,
+      "loss": 2.2344,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006827635645871887,
+      "loss": 2.2031,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006822749580493233,
+      "loss": 2.2266,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006817861506885375,
+      "loss": 2.2451,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006812971430433786,
+      "loss": 2.2676,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006808079356526147,
+      "loss": 2.2432,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000680318529055234,
+      "loss": 2.3633,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006798289237904444,
+      "loss": 2.3164,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006793391203976725,
+      "loss": 2.2793,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788491194165629,
+      "loss": 2.2266,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006783589213869782,
+      "loss": 2.2266,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006778685268489983,
+      "loss": 2.3184,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773779363429196,
+      "loss": 2.2852,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006768871504092536,
+      "loss": 2.125,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006763961695887279,
+      "loss": 2.2617,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000675904994422285,
+      "loss": 2.2061,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006754136254510807,
+      "loss": 2.2529,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006749220632164849,
+      "loss": 2.334,
+      "step": 1249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006744303082600802,
+      "loss": 2.2402,
+      "step": 1250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000673938361123662,
+      "loss": 2.2344,
+      "step": 1251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006734462223492367,
+      "loss": 2.2969,
+      "step": 1252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006729538924790221,
+      "loss": 2.3027,
+      "step": 1253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006724613720554473,
+      "loss": 2.2715,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00067196866162115,
+      "loss": 2.2363,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006714757617189782,
+      "loss": 2.2695,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006709826728919883,
+      "loss": 2.2441,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006704893956834452,
+      "loss": 2.334,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699959306368209,
+      "loss": 2.2559,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006695022782957945,
+      "loss": 2.2422,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006690084392042513,
+      "loss": 2.2305,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006685144139062831,
+      "loss": 2.2324,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006680202029461859,
+      "loss": 2.2754,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006675258068684606,
+      "loss": 2.3105,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006670312262178121,
+      "loss": 2.2637,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666536461539149,
+      "loss": 2.373,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006660415133775821,
+      "loss": 2.2539,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006655463822784243,
+      "loss": 2.2852,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000665051068787191,
+      "loss": 2.2402,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006645555734495974,
+      "loss": 2.3555,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006640598968115594,
+      "loss": 2.2676,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006635640394191931,
+      "loss": 2.2539,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006630680018188133,
+      "loss": 2.208,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006625717845569335,
+      "loss": 2.2344,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000662075388180265,
+      "loss": 2.293,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006615788132357166,
+      "loss": 2.3574,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006610820602703939,
+      "loss": 2.2002,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006605851298315987,
+      "loss": 2.291,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006600880224668275,
+      "loss": 2.2754,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006595907387237731,
+      "loss": 2.2363,
+      "step": 1280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006590932791503216,
+      "loss": 2.2207,
+      "step": 1281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006585956442945531,
+      "loss": 2.2676,
+      "step": 1282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006580978347047408,
+      "loss": 2.2754,
+      "step": 1283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006575998509293508,
+      "loss": 2.2598,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006571016935170406,
+      "loss": 2.3223,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000656603363016659,
+      "loss": 2.1904,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006561048599772459,
+      "loss": 2.2949,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000655606184948031,
+      "loss": 2.2012,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006551073384784338,
+      "loss": 2.2383,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006546083211180623,
+      "loss": 2.2881,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006541091334167126,
+      "loss": 2.2324,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006536097759243696,
+      "loss": 2.25,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006531102491912037,
+      "loss": 2.2168,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006526105537675729,
+      "loss": 2.2539,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006521106902040209,
+      "loss": 2.2656,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006516106590512763,
+      "loss": 2.2715,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006511104608602522,
+      "loss": 2.3027,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006506100961820463,
+      "loss": 2.2441,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006501095655679395,
+      "loss": 2.248,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006496088695693954,
+      "loss": 2.2656,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00064910800873806,
+      "loss": 2.1641,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006486069836257606,
+      "loss": 2.2158,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000648105794784506,
+      "loss": 2.2734,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006476044427664849,
+      "loss": 2.293,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006471029281240658,
+      "loss": 2.2598,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006466012514097967,
+      "loss": 2.248,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006460994131764044,
+      "loss": 2.2266,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006455974139767927,
+      "loss": 2.2148,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006450952543640433,
+      "loss": 2.2715,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006445929348914151,
+      "loss": 2.1924,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006440904561123425,
+      "loss": 2.2266,
+      "step": 1311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006435878185804354,
+      "loss": 2.1973,
+      "step": 1312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006430850228494789,
+      "loss": 2.3145,
+      "step": 1313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425820694734324,
+      "loss": 2.3008,
+      "step": 1314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006420789590064288,
+      "loss": 2.2227,
+      "step": 1315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641575692002774,
+      "loss": 2.334,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006410722690169469,
+      "loss": 2.3438,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006405686906035974,
+      "loss": 2.3496,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006400649573175477,
+      "loss": 2.207,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006395610697137893,
+      "loss": 2.1914,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006390570283474852,
+      "loss": 2.2441,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006385528337739668,
+      "loss": 2.2227,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380484865487346,
+      "loss": 2.3203,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006375439872274569,
+      "loss": 2.1777,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006370393363659706,
+      "loss": 2.2188,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006365345345202785,
+      "loss": 2.3086,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006360295822465502,
+      "loss": 2.165,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006355244801011208,
+      "loss": 2.2578,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006350192286404912,
+      "loss": 2.2383,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006345138284213261,
+      "loss": 2.1963,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000634008280000454,
+      "loss": 2.1992,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006335025839348671,
+      "loss": 2.1416,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006329967407817206,
+      "loss": 2.2217,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000632490751098331,
+      "loss": 2.2822,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006319846154421762,
+      "loss": 2.2383,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000631478334370896,
+      "loss": 2.3145,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006309719084422892,
+      "loss": 2.2168,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006304653382143148,
+      "loss": 2.1992,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006299586242450907,
+      "loss": 2.3047,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006294517670928934,
+      "loss": 2.3145,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006289447673161566,
+      "loss": 2.207,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006284376254734712,
+      "loss": 2.2207,
+      "step": 1342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006279303421235854,
+      "loss": 2.209,
+      "step": 1343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006274229178254025,
+      "loss": 2.2812,
+      "step": 1344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006269153531379814,
+      "loss": 2.3027,
+      "step": 1345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264076486205354,
+      "loss": 2.3789,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006258998048324324,
+      "loss": 2.2246,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006253918223331935,
+      "loss": 2.3672,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006248837016824923,
+      "loss": 2.3242,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006243754434401548,
+      "loss": 2.2715,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623867048166159,
+      "loss": 2.2002,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006233585164206332,
+      "loss": 2.3145,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006228498487638566,
+      "loss": 2.251,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006223410457562578,
+      "loss": 2.2012,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006218321079584149,
+      "loss": 2.2988,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006213230359310543,
+      "loss": 2.2637,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006208138302350497,
+      "loss": 2.2773,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000620304491431423,
+      "loss": 2.3066,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006197950200813424,
+      "loss": 2.2305,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006192854167461223,
+      "loss": 2.2617,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006187756819872219,
+      "loss": 2.2656,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006182658163662459,
+      "loss": 2.2148,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000617755820444943,
+      "loss": 2.2637,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006172456947852048,
+      "loss": 2.2119,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006167354399490671,
+      "loss": 2.1562,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000616225056498707,
+      "loss": 2.2051,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006157145449964438,
+      "loss": 2.2139,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006152039060047372,
+      "loss": 2.3398,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006146931400861883,
+      "loss": 2.2109,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006141822478035378,
+      "loss": 2.2715,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006136712297196652,
+      "loss": 2.2471,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006131600863975886,
+      "loss": 2.2676,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006126488184004646,
+      "loss": 2.3477,
+      "step": 1373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121374262915869,
+      "loss": 2.2266,
+      "step": 1374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006116259106343858,
+      "loss": 2.2256,
+      "step": 1375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006111142719924278,
+      "loss": 2.375,
+      "step": 1376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006106025109294151,
+      "loss": 2.2637,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006100906280091847,
+      "loss": 2.2617,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006095786237957075,
+      "loss": 2.1953,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006090664988530884,
+      "loss": 2.3105,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006085542537455654,
+      "loss": 2.25,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080418890375086,
+      "loss": 2.2051,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006075294052934199,
+      "loss": 2.3281,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006070168030779326,
+      "loss": 2.2578,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006065040829558104,
+      "loss": 2.3457,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006059912454919467,
+      "loss": 2.2734,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006054782912513643,
+      "loss": 2.2188,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000604965220799215,
+      "loss": 2.3574,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006044520347007783,
+      "loss": 2.2773,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000603938733521461,
+      "loss": 2.25,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006034253178267968,
+      "loss": 2.248,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006029117881824459,
+      "loss": 2.1738,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006023981451541935,
+      "loss": 2.2598,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006018843893079503,
+      "loss": 2.25,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006013705212097506,
+      "loss": 2.2832,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006008565414257532,
+      "loss": 2.2402,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006003424505222392,
+      "loss": 2.2578,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005998282490656124,
+      "loss": 2.3066,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005993139376223987,
+      "loss": 2.2441,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005987995167592448,
+      "loss": 2.2637,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005982849870429181,
+      "loss": 2.209,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005977703490403058,
+      "loss": 2.2363,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005972556033184146,
+      "loss": 2.2891,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005967407504443701,
+      "loss": 2.2676,
+      "step": 1404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000596225790985415,
+      "loss": 2.2754,
+      "step": 1405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005957107255089105,
+      "loss": 2.2246,
+      "step": 1406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005951955545823342,
+      "loss": 2.293,
+      "step": 1407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005946802787732798,
+      "loss": 2.2129,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005941648986494566,
+      "loss": 2.2305,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005936494147786887,
+      "loss": 2.207,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000593133827728915,
+      "loss": 2.2402,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926181380681876,
+      "loss": 2.3086,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005921023463646714,
+      "loss": 2.1445,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005915864531866446,
+      "loss": 2.2568,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910704591024967,
+      "loss": 2.2793,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005905543646807283,
+      "loss": 2.1836,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005900381704899503,
+      "loss": 2.2227,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005895218770988843,
+      "loss": 2.1885,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890054850763607,
+      "loss": 2.2266,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005884889949913183,
+      "loss": 2.3379,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005879724074128045,
+      "loss": 2.2051,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005874557229099739,
+      "loss": 2.2959,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000586938942052088,
+      "loss": 2.2012,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005864220654085139,
+      "loss": 2.1855,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005859050935487248,
+      "loss": 2.2676,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000585388027042299,
+      "loss": 2.2598,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005848708664589184,
+      "loss": 2.248,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005843536123683688,
+      "loss": 2.2461,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005838362653405395,
+      "loss": 2.1787,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005833188259454215,
+      "loss": 2.1924,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000582801294753108,
+      "loss": 2.3066,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000582283672333793,
+      "loss": 2.1602,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005817659592577716,
+      "loss": 2.3066,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005812481560954381,
+      "loss": 2.2051,
+      "step": 1434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005807302634172864,
+      "loss": 2.1895,
+      "step": 1435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000580212281793909,
+      "loss": 2.2656,
+      "step": 1436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005796942117959966,
+      "loss": 2.293,
+      "step": 1437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005791760539943367,
+      "loss": 2.2168,
+      "step": 1438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005786578089598138,
+      "loss": 2.3027,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005781394772634091,
+      "loss": 2.2754,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005776210594761984,
+      "loss": 2.2383,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005771025561693529,
+      "loss": 2.2715,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005765839679141375,
+      "loss": 2.3008,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005760652952819114,
+      "loss": 2.2344,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005755465388441265,
+      "loss": 2.252,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005750276991723264,
+      "loss": 2.2168,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005745087768381472,
+      "loss": 2.21,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005739897724133162,
+      "loss": 2.2344,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005734706864696503,
+      "loss": 2.2871,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005729515195790567,
+      "loss": 2.2891,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005724322723135317,
+      "loss": 2.2832,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005719129452451607,
+      "loss": 2.3066,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000571393538946116,
+      "loss": 2.2422,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005708740539886578,
+      "loss": 2.2422,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005703544909451331,
+      "loss": 2.2793,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005698348503879746,
+      "loss": 2.2451,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005693151328897006,
+      "loss": 2.1777,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005687953390229138,
+      "loss": 2.293,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005682754693603017,
+      "loss": 2.3145,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005677555244746348,
+      "loss": 2.2871,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005672355049387665,
+      "loss": 2.293,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005667154113256326,
+      "loss": 2.3301,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005661952442082508,
+      "loss": 2.1299,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000565675004159719,
+      "loss": 2.1914,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005651546917532161,
+      "loss": 2.1914,
+      "step": 1465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005646343075620006,
+      "loss": 2.3223,
+      "step": 1466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005641138521594099,
+      "loss": 2.2305,
+      "step": 1467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005635933261188601,
+      "loss": 2.2578,
+      "step": 1468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005630727300138448,
+      "loss": 2.2441,
+      "step": 1469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005625520644179354,
+      "loss": 2.2031,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005620313299047791,
+      "loss": 2.252,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005615105270480996,
+      "loss": 2.2598,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005609896564216954,
+      "loss": 2.1826,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005604687185994404,
+      "loss": 2.2822,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005599477141552818,
+      "loss": 2.1846,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005594266436632405,
+      "loss": 2.2734,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005589055076974101,
+      "loss": 2.2734,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005583843068319567,
+      "loss": 2.2656,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005578630416411174,
+      "loss": 2.1904,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005573417126992003,
+      "loss": 2.165,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005568203205805839,
+      "loss": 2.1523,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005562988658597163,
+      "loss": 2.2305,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005557773491111142,
+      "loss": 2.291,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005552557709093632,
+      "loss": 2.2334,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005547341318291164,
+      "loss": 2.2129,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005542124324450936,
+      "loss": 2.207,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536906733320815,
+      "loss": 2.2441,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005531688550649325,
+      "loss": 2.1641,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005526469782185643,
+      "loss": 2.2227,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005521250433679586,
+      "loss": 2.2578,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005516030510881616,
+      "loss": 2.2324,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005510810019542824,
+      "loss": 2.2227,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005505588965414931,
+      "loss": 2.209,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005500367354250275,
+      "loss": 2.2227,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005495145191801805,
+      "loss": 2.2432,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005489922483823087,
+      "loss": 2.2246,
+      "step": 1496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005484699236068281,
+      "loss": 2.2676,
+      "step": 1497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005479475454292139,
+      "loss": 2.2041,
+      "step": 1498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005474251144250008,
+      "loss": 2.2773,
+      "step": 1499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005469026311697813,
+      "loss": 2.3223,
+      "step": 1500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005463800962392059,
+      "loss": 2.2871,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005458575102089813,
+      "loss": 2.1426,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005453348736548709,
+      "loss": 2.2578,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005448121871526942,
+      "loss": 2.2598,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005442894512783251,
+      "loss": 2.2305,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000543766666607692,
+      "loss": 2.1953,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005432438337167773,
+      "loss": 2.2773,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005427209531816163,
+      "loss": 2.2734,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000542198025578297,
+      "loss": 2.1895,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005416750514829589,
+      "loss": 2.2246,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005411520314717935,
+      "loss": 2.2832,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005406289661210419,
+      "loss": 2.2012,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005401058560069957,
+      "loss": 2.1914,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005395827017059958,
+      "loss": 2.2305,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005390595037944317,
+      "loss": 2.2354,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000538536262848741,
+      "loss": 2.2656,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005380129794454086,
+      "loss": 2.3203,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005374896541609662,
+      "loss": 2.2285,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000536966287571992,
+      "loss": 2.1787,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000536442880255109,
+      "loss": 2.2432,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005359194327869857,
+      "loss": 2.2363,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005353959457443346,
+      "loss": 2.2598,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005348724197039117,
+      "loss": 2.2383,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005343488552425164,
+      "loss": 2.2549,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005338252529369895,
+      "loss": 2.1816,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005333016133642149,
+      "loss": 2.1543,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005327779371011163,
+      "loss": 2.2441,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005322542247246583,
+      "loss": 2.2012,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005317304768118451,
+      "loss": 2.2695,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005312066939397208,
+      "loss": 2.2734,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000530682876685367,
+      "loss": 2.2188,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005301590256259037,
+      "loss": 2.2871,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000529635141338488,
+      "loss": 2.2383,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005291112244003141,
+      "loss": 2.2383,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005285872753886113,
+      "loss": 2.1592,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005280632948806446,
+      "loss": 2.209,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005275392834537145,
+      "loss": 2.2812,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005270152416851542,
+      "loss": 2.2461,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005264911701523312,
+      "loss": 2.3066,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005259670694326455,
+      "loss": 2.2324,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005254429401035295,
+      "loss": 2.166,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000524918782742447,
+      "loss": 2.2441,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005243945979268922,
+      "loss": 2.1953,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005238703862343906,
+      "loss": 2.1934,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005233461482424963,
+      "loss": 2.1836,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000522821884528793,
+      "loss": 2.1504,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005222975956708923,
+      "loss": 2.2773,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005217732822464338,
+      "loss": 2.1182,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005212489448330843,
+      "loss": 2.1992,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005207245840085366,
+      "loss": 2.3086,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005202002003505094,
+      "loss": 2.2119,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005196757944367471,
+      "loss": 2.2871,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005191513668450177,
+      "loss": 2.2148,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005186269181531139,
+      "loss": 2.3242,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000518102448938851,
+      "loss": 2.3477,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005175779597800678,
+      "loss": 2.2793,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000517053451254624,
+      "loss": 2.2441,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005165289239404012,
+      "loss": 2.2109,
+      "step": 1558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005160043784153018,
+      "loss": 2.2227,
+      "step": 1559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005154798152572481,
+      "loss": 2.2012,
+      "step": 1560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005149552350441817,
+      "loss": 2.3262,
+      "step": 1561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005144306383540632,
+      "loss": 2.2363,
+      "step": 1562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005139060257648713,
+      "loss": 2.2949,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005133813978546023,
+      "loss": 2.2129,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005128567552012689,
+      "loss": 2.2246,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005123320983829009,
+      "loss": 2.2324,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005118074279775429,
+      "loss": 2.2656,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000511282744563255,
+      "loss": 2.2812,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005107580487181112,
+      "loss": 2.2441,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005102333410201995,
+      "loss": 2.2725,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000509708622047621,
+      "loss": 2.2578,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005091838923784892,
+      "loss": 2.252,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005086591525909289,
+      "loss": 2.3066,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005081344032630766,
+      "loss": 2.2842,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005076096449730794,
+      "loss": 2.2754,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005070848782990938,
+      "loss": 2.2207,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005065601038192855,
+      "loss": 2.1738,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005060353221118298,
+      "loss": 2.3848,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005055105337549085,
+      "loss": 2.2021,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005049857393267117,
+      "loss": 2.2236,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005044609394054361,
+      "loss": 2.1484,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005039361345692842,
+      "loss": 2.1904,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000503411325396464,
+      "loss": 2.2969,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000502886512465188,
+      "loss": 2.2402,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005023616963536735,
+      "loss": 2.2021,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005018368776401408,
+      "loss": 2.2637,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005013120569028132,
+      "loss": 2.3008,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000500787234719916,
+      "loss": 2.2734,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005002624116696768,
+      "loss": 2.3291,
+      "step": 1589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004997375883303234,
+      "loss": 2.1797,
+      "step": 1590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004992127652800841,
+      "loss": 2.1895,
+      "step": 1591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000498687943097187,
+      "loss": 2.1875,
+      "step": 1592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004981631223598595,
+      "loss": 2.2168,
+      "step": 1593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004976383036463265,
+      "loss": 2.2852,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004971134875348121,
+      "loss": 2.2148,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004965886746035361,
+      "loss": 2.2246,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000496063865430716,
+      "loss": 2.1934,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004955390605945639,
+      "loss": 2.1553,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004950142606732883,
+      "loss": 2.1689,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004944894662450917,
+      "loss": 2.2969,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004939646778881702,
+      "loss": 2.1328,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004934398961807144,
+      "loss": 2.1572,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004929151217009063,
+      "loss": 2.291,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004923903550269207,
+      "loss": 2.168,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004918655967369235,
+      "loss": 2.1748,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004913408474090712,
+      "loss": 2.2188,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000490816107621511,
+      "loss": 2.3809,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004902913779523791,
+      "loss": 2.2402,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004897666589798005,
+      "loss": 2.2637,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004892419512818889,
+      "loss": 2.2129,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004887172554367452,
+      "loss": 2.2188,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004881925720224572,
+      "loss": 2.3691,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004876679016170993,
+      "loss": 2.2793,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048714324479873114,
+      "loss": 2.2803,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004866186021453979,
+      "loss": 2.2715,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048609397423512866,
+      "loss": 2.3184,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004855693616459368,
+      "loss": 2.1377,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048504476495581827,
+      "loss": 2.2344,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000484520184742752,
+      "loss": 2.2852,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004839956215846983,
+      "loss": 2.2324,
+      "step": 1620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048347107605959885,
+      "loss": 2.2988,
+      "step": 1621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048294654874537617,
+      "loss": 2.2402,
+      "step": 1622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004824220402199324,
+      "loss": 2.1738,
+      "step": 1623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048189755106114887,
+      "loss": 2.2637,
+      "step": 1624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004813730818468862,
+      "loss": 2.2539,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048084863315498236,
+      "loss": 2.2305,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048032420556325307,
+      "loss": 2.1816,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047979979964949066,
+      "loss": 2.1973,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004792754159914636,
+      "loss": 2.2891,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004787510551669159,
+      "loss": 2.3379,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004782267177535661,
+      "loss": 2.1953,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004777024043291078,
+      "loss": 2.1426,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047717811547120706,
+      "loss": 2.2383,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004766538517575038,
+      "loss": 2.2412,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004761296137656095,
+      "loss": 2.2227,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004756054020731078,
+      "loss": 2.1963,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004750812172575533,
+      "loss": 2.248,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047455705989647045,
+      "loss": 2.209,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004740329305673545,
+      "loss": 2.1797,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047350882984766897,
+      "loss": 2.1846,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004729847583148459,
+      "loss": 2.3184,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047246071654628564,
+      "loss": 2.252,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004719367051193554,
+      "loss": 2.2852,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004714127246113889,
+      "loss": 2.1875,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047088877559968613,
+      "loss": 2.2783,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004703648586615119,
+      "loss": 2.248,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004698409743740964,
+      "loss": 2.375,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004693171233146331,
+      "loss": 2.2598,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004687933060602793,
+      "loss": 2.2871,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046826952318815496,
+      "loss": 2.2549,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004677457752753419,
+      "loss": 2.1494,
+      "step": 1651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004672220628988839,
+      "loss": 2.1738,
+      "step": 1652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046669838663578506,
+      "loss": 2.2656,
+      "step": 1653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004661747470630104,
+      "loss": 2.2871,
+      "step": 1654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046565114475748365,
+      "loss": 2.2109,
+      "step": 1655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046512758029608833,
+      "loss": 2.2363,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004646040542556655,
+      "loss": 2.2617,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046408056721301453,
+      "loss": 2.2539,
+      "step": 1658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004635571197448911,
+      "loss": 2.3145,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046303371242800825,
+      "loss": 2.1699,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046251034583903375,
+      "loss": 2.248,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046198702055459147,
+      "loss": 2.248,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000461463737151259,
+      "loss": 2.2373,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046094049620556835,
+      "loss": 2.3594,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004604172982940043,
+      "loss": 2.1973,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045989414399300434,
+      "loss": 2.1963,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004593710338789582,
+      "loss": 2.251,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004588479685282065,
+      "loss": 2.1416,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000458324948517041,
+      "loss": 2.1572,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000457801974421703,
+      "loss": 2.1816,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004572790468183838,
+      "loss": 2.2695,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045675616628322287,
+      "loss": 2.3066,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004562333333923081,
+      "loss": 2.248,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045571054872167504,
+      "loss": 2.2051,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004551878128473059,
+      "loss": 2.2148,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000454665126345129,
+      "loss": 2.1924,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004541424897910189,
+      "loss": 2.2246,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004536199037607942,
+      "loss": 2.168,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045309736883021866,
+      "loss": 2.2744,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004525748855749993,
+      "loss": 2.2188,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004520524545707862,
+      "loss": 2.1816,
+      "step": 1681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004515300763931721,
+      "loss": 2.2148,
+      "step": 1682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004510077516176912,
+      "loss": 2.1855,
+      "step": 1683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004504854808198195,
+      "loss": 2.1484,
+      "step": 1684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044996326457497266,
+      "loss": 2.1846,
+      "step": 1685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044944110345850703,
+      "loss": 2.1973,
+      "step": 1686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004489189980457177,
+      "loss": 2.1885,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044839694891183853,
+      "loss": 2.1816,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004478749566320416,
+      "loss": 2.2285,
+      "step": 1689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004473530217814359,
+      "loss": 2.2441,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044683114493506744,
+      "loss": 2.2773,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004463093266679185,
+      "loss": 2.2207,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044578756755490645,
+      "loss": 2.2695,
+      "step": 1693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004452658681708838,
+      "loss": 2.2109,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044474422909063693,
+      "loss": 2.3281,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004442226508888859,
+      "loss": 2.2471,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000443701134140284,
+      "loss": 2.2539,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044317967941941616,
+      "loss": 2.2031,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004426582873007998,
+      "loss": 2.2617,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004421369583588827,
+      "loss": 2.2637,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044161569316804333,
+      "loss": 2.2246,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044109449230258997,
+      "loss": 2.1562,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004405733563367596,
+      "loss": 2.2402,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004400522858447184,
+      "loss": 2.2051,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004395312814005596,
+      "loss": 2.2285,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004390103435783046,
+      "loss": 2.2793,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043848947295190054,
+      "loss": 2.2012,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004379686700952209,
+      "loss": 2.1992,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004374479355820647,
+      "loss": 2.2285,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043692726998615527,
+      "loss": 2.2012,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043640667388114004,
+      "loss": 2.2852,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043588614784059027,
+      "loss": 2.209,
+      "step": 1712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043536569243799947,
+      "loss": 2.2598,
+      "step": 1713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043484530824678406,
+      "loss": 2.168,
+      "step": 1714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043432499584028104,
+      "loss": 2.25,
+      "step": 1715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043380475579174937,
+      "loss": 2.166,
+      "step": 1716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004332845886743674,
+      "loss": 2.2188,
+      "step": 1717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043276449506123363,
+      "loss": 2.2275,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043224447552536537,
+      "loss": 2.1777,
+      "step": 1719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004317245306396983,
+      "loss": 2.2695,
+      "step": 1720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043120466097708624,
+      "loss": 2.2578,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004306848671102995,
+      "loss": 2.0928,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043016514961202544,
+      "loss": 2.2158,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000429645509054867,
+      "loss": 2.1465,
+      "step": 1724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004291259460113423,
+      "loss": 2.3047,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004286064610538842,
+      "loss": 2.2402,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042808705475483956,
+      "loss": 2.251,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042756772768646817,
+      "loss": 2.2461,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004270484804209434,
+      "loss": 2.2734,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004265293135303498,
+      "loss": 2.2812,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042601022758668387,
+      "loss": 2.3516,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004254912231618528,
+      "loss": 2.332,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042497230082767367,
+      "loss": 2.2441,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004244534611558738,
+      "loss": 2.2949,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004239347047180885,
+      "loss": 2.1748,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004234160320858625,
+      "loss": 2.209,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004228974438306472,
+      "loss": 2.1816,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042237894052380164,
+      "loss": 2.25,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042186052273659097,
+      "loss": 2.1455,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004213421910401861,
+      "loss": 2.1572,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004208239460056635,
+      "loss": 2.1777,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004203057882040037,
+      "loss": 2.2539,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000419787718206091,
+      "loss": 2.3262,
+      "step": 1743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041926973658271366,
+      "loss": 2.2578,
+      "step": 1744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041875184390456193,
+      "loss": 2.2578,
+      "step": 1745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041823404074222846,
+      "loss": 2.2344,
+      "step": 1746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041771632766620704,
+      "loss": 2.209,
+      "step": 1747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004171987052468921,
+      "loss": 2.1934,
+      "step": 1748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041668117405457866,
+      "loss": 2.2324,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004161637346594605,
+      "loss": 2.2539,
+      "step": 1750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004156463876316312,
+      "loss": 2.1572,
+      "step": 1751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041512913354108165,
+      "loss": 2.2559,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041461197295770107,
+      "loss": 2.3203,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041409490645127526,
+      "loss": 2.2324,
+      "step": 1754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004135779345914862,
+      "loss": 2.2871,
+      "step": 1755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041306105794791225,
+      "loss": 2.1533,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041254427709002626,
+      "loss": 2.291,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004120275925871955,
+      "loss": 2.2559,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041151100500868177,
+      "loss": 2.2344,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041099451492363946,
+      "loss": 2.2578,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004104781229011158,
+      "loss": 2.2754,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040996182951004987,
+      "loss": 2.2656,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040944563531927194,
+      "loss": 2.2441,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004089295408975034,
+      "loss": 2.1592,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004084135468133553,
+      "loss": 2.1826,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004078976536353286,
+      "loss": 2.2109,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004073818619318126,
+      "loss": 2.209,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004068661722710851,
+      "loss": 2.2266,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004063505852213114,
+      "loss": 2.2246,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004058351013505436,
+      "loss": 2.2324,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040531972122672034,
+      "loss": 2.1982,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040480444541766573,
+      "loss": 2.2363,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004042892744910895,
+      "loss": 2.2236,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040377420901458507,
+      "loss": 2.1826,
+      "step": 1774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040325924955563006,
+      "loss": 2.3027,
+      "step": 1775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040274439668158533,
+      "loss": 2.2422,
+      "step": 1776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040222965095969434,
+      "loss": 2.166,
+      "step": 1777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000401715012957082,
+      "loss": 2.248,
+      "step": 1778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040120048324075535,
+      "loss": 2.1924,
+      "step": 1779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004006860623776013,
+      "loss": 2.2012,
+      "step": 1780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040017175093438764,
+      "loss": 2.2383,
+      "step": 1781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003996575494777609,
+      "loss": 2.2812,
+      "step": 1782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039914345857424696,
+      "loss": 2.3125,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003986294787902495,
+      "loss": 2.1387,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003981156106920498,
+      "loss": 2.2236,
+      "step": 1785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003976018548458066,
+      "loss": 2.1758,
+      "step": 1786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039708821181755403,
+      "loss": 2.1934,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003965746821732032,
+      "loss": 2.2461,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000396061266478539,
+      "loss": 2.1758,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003955479652992218,
+      "loss": 2.2422,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000395034779200785,
+      "loss": 2.2373,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003945217087486358,
+      "loss": 2.1895,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003940087545080535,
+      "loss": 2.1309,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003934959170441898,
+      "loss": 2.2148,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039298319692206737,
+      "loss": 2.2441,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039247059470658017,
+      "loss": 2.21,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039195811096249145,
+      "loss": 2.1895,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003914457462544347,
+      "loss": 2.1846,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003909335011469117,
+      "loss": 2.2285,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003904213762042926,
+      "loss": 2.25,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003899093719908155,
+      "loss": 2.1699,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038939748907058485,
+      "loss": 2.1875,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003888857280075722,
+      "loss": 2.2598,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003883740893656142,
+      "loss": 2.2383,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038786257370841324,
+      "loss": 2.1133,
+      "step": 1805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003873511815995355,
+      "loss": 2.1738,
+      "step": 1806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003868399136024115,
+      "loss": 2.2393,
+      "step": 1807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038632877028033503,
+      "loss": 2.1973,
+      "step": 1808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038581775219646234,
+      "loss": 2.1953,
+      "step": 1809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003853068599138115,
+      "loss": 2.3105,
+      "step": 1810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003847960939952628,
+      "loss": 2.2676,
+      "step": 1811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038428545500355637,
+      "loss": 2.1875,
+      "step": 1812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003837749435012931,
+      "loss": 2.1445,
+      "step": 1813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000383264560050933,
+      "loss": 2.2832,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003827543052147952,
+      "loss": 2.2676,
+      "step": 1815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003822441795550573,
+      "loss": 2.2148,
+      "step": 1816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038173418363375404,
+      "loss": 2.2246,
+      "step": 1817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003812243180127781,
+      "loss": 2.1787,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038071458325387775,
+      "loss": 2.2793,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038020497991865754,
+      "loss": 2.1641,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003796955085685771,
+      "loss": 2.2314,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037918616976495047,
+      "loss": 2.2012,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037867696406894603,
+      "loss": 2.2393,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037816789204158524,
+      "loss": 2.1582,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037765895424374214,
+      "loss": 2.0977,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003771501512361435,
+      "loss": 2.1465,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037664148357936686,
+      "loss": 2.2051,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003761329518338412,
+      "loss": 2.1455,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003756245565598453,
+      "loss": 2.1543,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003751162983175079,
+      "loss": 2.1406,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003746081776668067,
+      "loss": 2.249,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003741001951675675,
+      "loss": 2.1797,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037359235137946454,
+      "loss": 2.2617,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003730846468620187,
+      "loss": 2.2676,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037257708217459755,
+      "loss": 2.2461,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720696578764147,
+      "loss": 2.2852,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037156237452652875,
+      "loss": 2.1279,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003710552326838437,
+      "loss": 2.1797,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003705482329071066,
+      "loss": 2.2002,
+      "step": 1839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037004137575490917,
+      "loss": 2.248,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003695346617856852,
+      "loss": 2.2227,
+      "step": 1841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036902809155771077,
+      "loss": 2.1387,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003685216656291041,
+      "loss": 2.2793,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036801538455782386,
+      "loss": 2.2607,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036750924890166926,
+      "loss": 2.2441,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036700325921827957,
+      "loss": 2.1914,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003664974160651328,
+      "loss": 2.2871,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003659917199995461,
+      "loss": 2.2832,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036548617157867406,
+      "loss": 2.2578,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003649807713595088,
+      "loss": 2.2988,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036447551989887917,
+      "loss": 2.2441,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003639704177534499,
+      "loss": 2.2539,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003634654654797217,
+      "loss": 2.2695,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036296066363402936,
+      "loss": 2.1572,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036245601277254305,
+      "loss": 2.3398,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036195151345126556,
+      "loss": 2.2754,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003614471662260333,
+      "loss": 2.248,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003609429716525149,
+      "loss": 2.207,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003604389302862108,
+      "loss": 2.2266,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003599350426824526,
+      "loss": 2.2109,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003594313093964027,
+      "loss": 2.2383,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003589277309830532,
+      "loss": 2.1855,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000358424307997226,
+      "loss": 2.1729,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003579210409935713,
+      "loss": 2.2383,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003574179305265677,
+      "loss": 2.1445,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035691497715052126,
+      "loss": 2.2754,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003564121814195647,
+      "loss": 2.2461,
+      "step": 1867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035590954388765774,
+      "loss": 2.1855,
+      "step": 1868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035540706510858483,
+      "loss": 2.2373,
+      "step": 1869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035490474563595664,
+      "loss": 2.1641,
+      "step": 1870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003544025860232074,
+      "loss": 2.1494,
+      "step": 1871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003539005868235957,
+      "loss": 2.2598,
+      "step": 1872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003533987485902033,
+      "loss": 2.25,
+      "step": 1873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003528970718759342,
+      "loss": 2.1553,
+      "step": 1874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035239555723351533,
+      "loss": 2.2285,
+      "step": 1875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003518942052154942,
+      "loss": 2.124,
+      "step": 1876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003513930163742394,
+      "loss": 2.208,
+      "step": 1877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035089199126194003,
+      "loss": 2.2031,
+      "step": 1878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035039113043060455,
+      "loss": 2.2402,
+      "step": 1879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003498904344320606,
+      "loss": 2.252,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003493899038179538,
+      "loss": 2.2207,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000348889539139748,
+      "loss": 2.1816,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000348389340948724,
+      "loss": 2.2109,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003478893097959791,
+      "loss": 2.2324,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000347389446232427,
+      "loss": 2.1602,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034688975080879623,
+      "loss": 2.2559,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003463902240756306,
+      "loss": 2.1738,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003458908665832874,
+      "loss": 2.2168,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003453916788819379,
+      "loss": 2.1816,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003448926615215663,
+      "loss": 2.2188,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003443938150519691,
+      "loss": 2.1895,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003438951400227541,
+      "loss": 2.1738,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003433966369833411,
+      "loss": 2.2939,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003428983064829595,
+      "loss": 2.1992,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003424001490706493,
+      "loss": 2.2031,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003419021652952593,
+      "loss": 2.1924,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034140435570544704,
+      "loss": 2.1953,
+      "step": 1897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003409067208496786,
+      "loss": 2.2617,
+      "step": 1898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003404092612762269,
+      "loss": 2.2363,
+      "step": 1899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003399119775331725,
+      "loss": 2.1426,
+      "step": 1900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033941487016840143,
+      "loss": 2.2812,
+      "step": 1901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033891793972960603,
+      "loss": 2.2217,
+      "step": 1902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033842118676428345,
+      "loss": 2.1953,
+      "step": 1903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003379246118197351,
+      "loss": 2.1621,
+      "step": 1904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003374282154430667,
+      "loss": 2.2061,
+      "step": 1905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003369319981811867,
+      "loss": 2.1953,
+      "step": 1906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033643596058080695,
+      "loss": 2.1562,
+      "step": 1907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003359401031884407,
+      "loss": 2.2266,
+      "step": 1908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033544442655040276,
+      "loss": 2.1797,
+      "step": 1909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003349489312128091,
+      "loss": 2.3145,
+      "step": 1910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003344536177215757,
+      "loss": 2.2012,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033395848662241806,
+      "loss": 2.1914,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003334635384608512,
+      "loss": 2.248,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033296877378218785,
+      "loss": 2.3242,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003324741931315396,
+      "loss": 2.1953,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033197979705381427,
+      "loss": 2.2188,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033148558609371703,
+      "loss": 2.2148,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003309915607957487,
+      "loss": 2.2305,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003304977217042057,
+      "loss": 2.1992,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033000406936317927,
+      "loss": 2.2578,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032951060431655473,
+      "loss": 2.1699,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003290173271080116,
+      "loss": 2.1602,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003285242382810218,
+      "loss": 2.2568,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032803133837885,
+      "loss": 2.2754,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003275386279445528,
+      "loss": 2.3018,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032704610752097785,
+      "loss": 2.1621,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032655377765076343,
+      "loss": 2.1729,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032606163887633817,
+      "loss": 2.2324,
+      "step": 1928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003255696917399197,
+      "loss": 2.1758,
+      "step": 1929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003250779367835152,
+      "loss": 2.2871,
+      "step": 1930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032458637454891936,
+      "loss": 2.1855,
+      "step": 1931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003240950055777151,
+      "loss": 2.2617,
+      "step": 1932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032360383041127213,
+      "loss": 2.2012,
+      "step": 1933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003231128495907465,
+      "loss": 2.1904,
+      "step": 1934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032262206365708064,
+      "loss": 2.1514,
+      "step": 1935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032213147315100153,
+      "loss": 2.2383,
+      "step": 1936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003216410786130218,
+      "loss": 2.2051,
+      "step": 1937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003211508805834372,
+      "loss": 2.083,
+      "step": 1938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003206608796023277,
+      "loss": 2.2793,
+      "step": 1939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032017107620955576,
+      "loss": 2.168,
+      "step": 1940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003196814709447661,
+      "loss": 2.1816,
+      "step": 1941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003191920643473855,
+      "loss": 2.1699,
+      "step": 1942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031870285695662163,
+      "loss": 2.1768,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031821384931146255,
+      "loss": 2.1992,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031772504195067675,
+      "loss": 2.082,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031723643541281137,
+      "loss": 2.2148,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031674803023619305,
+      "loss": 2.1533,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031625982695892587,
+      "loss": 2.1943,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031577182611889173,
+      "loss": 2.1777,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031528402825374975,
+      "loss": 2.2383,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031479643390093493,
+      "loss": 2.1885,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031430904359765866,
+      "loss": 2.2793,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031382185788090687,
+      "loss": 2.1426,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003133348772874407,
+      "loss": 2.2012,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003128481023537948,
+      "loss": 2.2656,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003123615336162774,
+      "loss": 2.1816,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031187517161096966,
+      "loss": 2.207,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003113890168737248,
+      "loss": 2.2188,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003109030699401677,
+      "loss": 2.1543,
+      "step": 1959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003104173313456946,
+      "loss": 2.166,
+      "step": 1960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003099318016254717,
+      "loss": 2.1992,
+      "step": 1961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003094464813144355,
+      "loss": 2.2188,
+      "step": 1962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003089613709472916,
+      "loss": 2.2051,
+      "step": 1963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030847647105851415,
+      "loss": 2.1777,
+      "step": 1964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003079917821823457,
+      "loss": 2.2012,
+      "step": 1965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000307507304852796,
+      "loss": 2.1719,
+      "step": 1966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003070230396036422,
+      "loss": 2.2578,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003065389869684272,
+      "loss": 2.2148,
+      "step": 1968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003060551474804601,
+      "loss": 2.1914,
+      "step": 1969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000305571521672815,
+      "loss": 2.168,
+      "step": 1970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030508811007833037,
+      "loss": 2.2324,
+      "step": 1971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003046049132296091,
+      "loss": 2.1865,
+      "step": 1972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000304121931659017,
+      "loss": 2.248,
+      "step": 1973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030363916589868327,
+      "loss": 2.2598,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000303156616480499,
+      "loss": 2.1162,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030267428393611654,
+      "loss": 2.2617,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003021921687969502,
+      "loss": 2.1855,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003017102715941741,
+      "loss": 2.3086,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030122859285872214,
+      "loss": 2.1973,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030074713312128816,
+      "loss": 2.25,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030026589291232397,
+      "loss": 2.1396,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029978487276204036,
+      "loss": 2.2002,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029930407320040485,
+      "loss": 2.1777,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002988234947571424,
+      "loss": 2.2129,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029834313796173433,
+      "loss": 2.2656,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029786300334341744,
+      "loss": 2.1611,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029738309143118415,
+      "loss": 2.2051,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029690340275378096,
+      "loss": 2.1045,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002964239378397093,
+      "loss": 2.2051,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029594469721722327,
+      "loss": 2.1426,
+      "step": 1990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029546568141433004,
+      "loss": 2.2109,
+      "step": 1991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029498689095878945,
+      "loss": 2.2754,
+      "step": 1992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002945083263781126,
+      "loss": 2.2363,
+      "step": 1993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029402998819956185,
+      "loss": 2.3027,
+      "step": 1994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002935518769501504,
+      "loss": 2.2188,
+      "step": 1995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029307399315664083,
+      "loss": 2.2461,
+      "step": 1996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029259633734554603,
+      "loss": 2.2148,
+      "step": 1997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002921189100431268,
+      "loss": 2.291,
+      "step": 1998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002916417117753928,
+      "loss": 2.25,
+      "step": 1999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000291164743068101,
+      "loss": 2.2676,
+      "step": 2000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002906880044467555,
+      "loss": 2.2461,
+      "step": 2001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029021149643660706,
+      "loss": 2.1885,
+      "step": 2002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289735219562652,
+      "loss": 2.2578,
+      "step": 2003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028925917434963254,
+      "loss": 2.1807,
+      "step": 2004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028878336132203547,
+      "loss": 2.1211,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002883077810040915,
+      "loss": 2.1777,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002878324339197751,
+      "loss": 2.2129,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028735732059280384,
+      "loss": 2.1875,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028688244154663765,
+      "loss": 2.2148,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002864077973044783,
+      "loss": 2.2031,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028593338838926903,
+      "loss": 2.2949,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028545921532369394,
+      "loss": 2.1699,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000284985278630177,
+      "loss": 2.2559,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002845115788308816,
+      "loss": 2.1289,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002840381164477105,
+      "loss": 2.2441,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028356489200230474,
+      "loss": 2.2246,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002830919060160431,
+      "loss": 2.1426,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028261915901004185,
+      "loss": 2.1934,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028214665150515375,
+      "loss": 2.2305,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000281674384021968,
+      "loss": 2.1641,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002812023570808091,
+      "loss": 2.1787,
+      "step": 2021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002807305712017365,
+      "loss": 2.2383,
+      "step": 2022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028025902690454417,
+      "loss": 2.1699,
+      "step": 2023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002797877247087599,
+      "loss": 2.1631,
+      "step": 2024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002793166651336444,
+      "loss": 2.2676,
+      "step": 2025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027884584869819175,
+      "loss": 2.165,
+      "step": 2026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002783752759211279,
+      "loss": 2.2197,
+      "step": 2027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027790494732091007,
+      "loss": 2.1904,
+      "step": 2028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002774348634157267,
+      "loss": 2.1182,
+      "step": 2029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027696502472349636,
+      "loss": 2.25,
+      "step": 2030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002764954317618679,
+      "loss": 2.2607,
+      "step": 2031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002760260850482188,
+      "loss": 2.2227,
+      "step": 2032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027555698509965613,
+      "loss": 2.1807,
+      "step": 2033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002750881324330142,
+      "loss": 2.0752,
+      "step": 2034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027461952756485566,
+      "loss": 2.2715,
+      "step": 2035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027415117101146946,
+      "loss": 2.1982,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027368306328887126,
+      "loss": 2.248,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002732152049128026,
+      "loss": 2.1953,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002727475963987303,
+      "loss": 2.1729,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027228023826184554,
+      "loss": 2.167,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002718131310170643,
+      "loss": 2.207,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002713462751790259,
+      "loss": 2.2402,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002708796712620923,
+      "loss": 2.2695,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002704133197803483,
+      "loss": 2.1836,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002699472212476004,
+      "loss": 2.2441,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026948137617737643,
+      "loss": 2.166,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002690157850829248,
+      "loss": 2.2266,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002685504484772147,
+      "loss": 2.2246,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026808536687293415,
+      "loss": 2.2578,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000267620540782491,
+      "loss": 2.2402,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002671559707180109,
+      "loss": 2.2168,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002666916571913378,
+      "loss": 2.2227,
+      "step": 2052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002662276007140329,
+      "loss": 2.2656,
+      "step": 2053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002657638017973742,
+      "loss": 2.1953,
+      "step": 2054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002653002609523557,
+      "loss": 2.3105,
+      "step": 2055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026483697868968746,
+      "loss": 2.1934,
+      "step": 2056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643739555197949,
+      "loss": 2.2207,
+      "step": 2057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002639111919528171,
+      "loss": 2.2197,
+      "step": 2058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002634486884986078,
+      "loss": 2.1523,
+      "step": 2059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002629864456667338,
+      "loss": 2.209,
+      "step": 2060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000262524463966475,
+      "loss": 2.209,
+      "step": 2061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002620627439068233,
+      "loss": 2.1738,
+      "step": 2062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002616012859964827,
+      "loss": 2.2773,
+      "step": 2063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026114009074386846,
+      "loss": 2.168,
+      "step": 2064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000260679158657106,
+      "loss": 2.25,
+      "step": 2065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026021849024403086,
+      "loss": 2.1504,
+      "step": 2066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002597580860121885,
+      "loss": 2.3027,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025929794646883275,
+      "loss": 2.1699,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002588380721209263,
+      "loss": 2.2617,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025837846347513915,
+      "loss": 2.2598,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025791912103784913,
+      "loss": 2.1475,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025746004531514074,
+      "loss": 2.1953,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002570012368128041,
+      "loss": 2.1426,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002565426960363353,
+      "loss": 2.1641,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002560844234909353,
+      "loss": 2.2197,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025562641968150966,
+      "loss": 2.2266,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025516868511266767,
+      "loss": 2.1924,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025471122028872224,
+      "loss": 2.2363,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002542540257136892,
+      "loss": 2.2031,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025379710189128615,
+      "loss": 2.1631,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002533404493249327,
+      "loss": 2.3105,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025288406851774946,
+      "loss": 2.1719,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002524279599725578,
+      "loss": 2.2168,
+      "step": 2083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025197212419187883,
+      "loss": 2.2646,
+      "step": 2084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025151656167793357,
+      "loss": 2.2041,
+      "step": 2085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025106127293264156,
+      "loss": 2.1797,
+      "step": 2086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025060625845762114,
+      "loss": 2.1504,
+      "step": 2087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025015151875418804,
+      "loss": 2.2617,
+      "step": 2088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002496970543233555,
+      "loss": 2.2559,
+      "step": 2089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024924286566583334,
+      "loss": 2.1328,
+      "step": 2090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024878895328202754,
+      "loss": 2.2773,
+      "step": 2091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024833531767203974,
+      "loss": 2.2324,
+      "step": 2092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002478819593356668,
+      "loss": 2.1572,
+      "step": 2093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002474288787724002,
+      "loss": 2.2598,
+      "step": 2094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002469760764814248,
+      "loss": 2.1641,
+      "step": 2095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002465235529616193,
+      "loss": 2.2246,
+      "step": 2096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002460713087115553,
+      "loss": 2.2012,
+      "step": 2097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002456193442294964,
+      "loss": 2.2441,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024516766001339807,
+      "loss": 2.2441,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002447162565609074,
+      "loss": 2.2188,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002442651343693614,
+      "loss": 2.1699,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002438142939357882,
+      "loss": 2.2207,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002433637357569045,
+      "loss": 2.1582,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024291346032911666,
+      "loss": 2.2324,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002424634681485191,
+      "loss": 2.1484,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024201375971089444,
+      "loss": 2.1504,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024156433551171254,
+      "loss": 2.1016,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002411151960461302,
+      "loss": 2.1904,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024066634180899078,
+      "loss": 2.25,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024021777329482285,
+      "loss": 2.2461,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023976949099784045,
+      "loss": 2.1914,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023932149541194226,
+      "loss": 2.1777,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023887378703071104,
+      "loss": 2.1924,
+      "step": 2113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000238426366347413,
+      "loss": 2.2754,
+      "step": 2114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023797923385499793,
+      "loss": 2.2695,
+      "step": 2115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023753239004609716,
+      "loss": 2.2871,
+      "step": 2116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023708583541302508,
+      "loss": 2.1113,
+      "step": 2117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023663957044777663,
+      "loss": 2.2305,
+      "step": 2118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002361935956420278,
+      "loss": 2.2383,
+      "step": 2119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023574791148713503,
+      "loss": 2.2344,
+      "step": 2120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023530251847413448,
+      "loss": 2.1533,
+      "step": 2121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023485741709374132,
+      "loss": 2.1719,
+      "step": 2122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023441260783634983,
+      "loss": 2.1289,
+      "step": 2123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023396809119203245,
+      "loss": 2.2021,
+      "step": 2124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023352386765053873,
+      "loss": 2.1895,
+      "step": 2125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023307993770129565,
+      "loss": 2.1787,
+      "step": 2126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023263630183340667,
+      "loss": 2.1738,
+      "step": 2127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023219296053565114,
+      "loss": 2.1846,
+      "step": 2128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002317499142964839,
+      "loss": 2.1943,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023130716360403477,
+      "loss": 2.1797,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023086470894610823,
+      "loss": 2.207,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002304225508101821,
+      "loss": 2.2812,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022998068968340768,
+      "loss": 2.2402,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022953912605260912,
+      "loss": 2.1426,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022909786040428275,
+      "loss": 2.1143,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022865689322459654,
+      "loss": 2.2422,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022821622499938948,
+      "loss": 2.2061,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022777585621417163,
+      "loss": 2.1914,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000227335787354123,
+      "loss": 2.2422,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022689601890409295,
+      "loss": 2.166,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022645655134859987,
+      "loss": 2.1875,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022601738517183078,
+      "loss": 2.2549,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022557852085764053,
+      "loss": 2.2129,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022513995888955135,
+      "loss": 2.2178,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002247016997507526,
+      "loss": 2.2695,
+      "step": 2145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022426374392410005,
+      "loss": 2.1221,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002238260918921149,
+      "loss": 2.1455,
+      "step": 2147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022338874413698384,
+      "loss": 2.2559,
+      "step": 2148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022295170114055836,
+      "loss": 2.1836,
+      "step": 2149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002225149633843541,
+      "loss": 2.1816,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022207853134955036,
+      "loss": 2.2168,
+      "step": 2151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002216424055169899,
+      "loss": 2.1973,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022120658636717766,
+      "loss": 2.2402,
+      "step": 2153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022077107438028126,
+      "loss": 2.1914,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022033587003612948,
+      "loss": 2.125,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021990097381421208,
+      "loss": 2.2012,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002194663861936797,
+      "loss": 2.1914,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002190321076533427,
+      "loss": 2.1934,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021859813867167077,
+      "loss": 2.1592,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002181644797267931,
+      "loss": 2.1758,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000217731131296497,
+      "loss": 2.2461,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021729809385822758,
+      "loss": 2.25,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021686536788908728,
+      "loss": 2.166,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021643295386583546,
+      "loss": 2.2227,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021600085226488786,
+      "loss": 2.1426,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002155690635623157,
+      "loss": 2.1123,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002151375882338461,
+      "loss": 2.2344,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021470642675486018,
+      "loss": 2.1611,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002142755796003939,
+      "loss": 2.1973,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021384504724513648,
+      "loss": 2.1514,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002134148301634305,
+      "loss": 2.2129,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000212984928829271,
+      "loss": 2.2451,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021255534371630546,
+      "loss": 2.1699,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021212607529783246,
+      "loss": 2.1748,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021169712404680224,
+      "loss": 2.1836,
+      "step": 2175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021126849043581546,
+      "loss": 2.2793,
+      "step": 2176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021084017493712255,
+      "loss": 2.1934,
+      "step": 2177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021041217802262357,
+      "loss": 2.1777,
+      "step": 2178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020998450016386765,
+      "loss": 2.2969,
+      "step": 2179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020955714183205231,
+      "loss": 2.2197,
+      "step": 2180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020913010349802298,
+      "loss": 2.3047,
+      "step": 2181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020870338563227286,
+      "loss": 2.2148,
+      "step": 2182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020827698870494162,
+      "loss": 2.1514,
+      "step": 2183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002078509131858158,
+      "loss": 2.1855,
+      "step": 2184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002074251595443274,
+      "loss": 2.1387,
+      "step": 2185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020699972824955398,
+      "loss": 2.2051,
+      "step": 2186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020657461977021795,
+      "loss": 2.3008,
+      "step": 2187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020614983457468611,
+      "loss": 2.2305,
+      "step": 2188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020572537313096867,
+      "loss": 2.1982,
+      "step": 2189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020530123590671985,
+      "loss": 2.1738,
+      "step": 2190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020487742336923633,
+      "loss": 2.2559,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020445393598545687,
+      "loss": 2.209,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002040307742219622,
+      "loss": 2.2266,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020360793854497423,
+      "loss": 2.2148,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020318542942035557,
+      "loss": 2.1953,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020276324731360895,
+      "loss": 2.1992,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020234139268987718,
+      "loss": 2.2715,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020191986601394207,
+      "loss": 2.3145,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020149866775022397,
+      "loss": 2.3105,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002010777983627815,
+      "loss": 2.167,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000200657258315311,
+      "loss": 2.2168,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020023704807114578,
+      "loss": 2.2227,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019981716809325595,
+      "loss": 2.2305,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001993976188442475,
+      "loss": 2.1855,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019897840078636248,
+      "loss": 2.21,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001985595143814779,
+      "loss": 2.1992,
+      "step": 2206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019814096009110517,
+      "loss": 2.1748,
+      "step": 2207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019772273837638982,
+      "loss": 2.1797,
+      "step": 2208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001973048496981111,
+      "loss": 2.2109,
+      "step": 2209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019688729451668114,
+      "loss": 2.0674,
+      "step": 2210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019647007329214473,
+      "loss": 2.1641,
+      "step": 2211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019605318648417886,
+      "loss": 2.1543,
+      "step": 2212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001956366345520922,
+      "loss": 2.1689,
+      "step": 2213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019522041795482398,
+      "loss": 2.2969,
+      "step": 2214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019480453715094425,
+      "loss": 2.1318,
+      "step": 2215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001943889925986531,
+      "loss": 2.1992,
+      "step": 2216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001939737847557802,
+      "loss": 2.2188,
+      "step": 2217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019355891407978382,
+      "loss": 2.2275,
+      "step": 2218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001931443810277516,
+      "loss": 2.0195,
+      "step": 2219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019273018605639835,
+      "loss": 2.2051,
+      "step": 2220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019231632962206702,
+      "loss": 2.1875,
+      "step": 2221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019190281218072725,
+      "loss": 2.21,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001914896341879751,
+      "loss": 2.252,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019107679609903295,
+      "loss": 2.1406,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019066429836874844,
+      "loss": 2.2129,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019025214145159414,
+      "loss": 2.1416,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001898403258016675,
+      "loss": 2.2051,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018942885187268993,
+      "loss": 2.2197,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000189017720118006,
+      "loss": 2.1113,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018860693099058342,
+      "loss": 2.1318,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018819648494301257,
+      "loss": 2.1797,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018778638242750563,
+      "loss": 2.2344,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018737662389589627,
+      "loss": 2.2148,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018696720979963967,
+      "loss": 2.1934,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018655814058981068,
+      "loss": 2.1631,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018614941671710506,
+      "loss": 2.1787,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018574103863183757,
+      "loss": 2.2734,
+      "step": 2237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018533300678394193,
+      "loss": 2.2246,
+      "step": 2238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018492532162297072,
+      "loss": 2.2559,
+      "step": 2239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018451798359809423,
+      "loss": 2.1465,
+      "step": 2240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018411099315810038,
+      "loss": 2.3008,
+      "step": 2241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018370435075139435,
+      "loss": 2.1396,
+      "step": 2242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018329805682599793,
+      "loss": 2.2314,
+      "step": 2243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018289211182954856,
+      "loss": 2.1426,
+      "step": 2244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018248651620929941,
+      "loss": 2.2012,
+      "step": 2245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018208127041211887,
+      "loss": 2.1602,
+      "step": 2246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018167637488448964,
+      "loss": 2.1504,
+      "step": 2247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018127183007250863,
+      "loss": 2.207,
+      "step": 2248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018086763642188668,
+      "loss": 2.2754,
+      "step": 2249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018046379437794709,
+      "loss": 2.209,
+      "step": 2250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001800603043856265,
+      "loss": 2.2715,
+      "step": 2251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017965716688947308,
+      "loss": 2.1846,
+      "step": 2252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001792543823336469,
+      "loss": 2.2617,
+      "step": 2253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017885195116191906,
+      "loss": 2.1182,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017844987381767142,
+      "loss": 2.3027,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001780481507438958,
+      "loss": 2.2822,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017764678238319404,
+      "loss": 2.2305,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017724576917777725,
+      "loss": 2.1836,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001768451115694647,
+      "loss": 2.2305,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001764448099996842,
+      "loss": 2.2188,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017604486490947136,
+      "loss": 2.2109,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017564527673946884,
+      "loss": 2.2324,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017524604592992604,
+      "loss": 2.1846,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001748471729206989,
+      "loss": 2.1533,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017444865815124922,
+      "loss": 2.1211,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001740505020606437,
+      "loss": 2.1914,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001736527050875541,
+      "loss": 2.1572,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001732552676702564,
+      "loss": 2.1533,
+      "step": 2268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001728581902466306,
+      "loss": 2.1758,
+      "step": 2269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017246147325415985,
+      "loss": 2.1387,
+      "step": 2270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017206511712993035,
+      "loss": 2.1416,
+      "step": 2271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001716691223106307,
+      "loss": 2.2188,
+      "step": 2272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017127348923255166,
+      "loss": 2.1299,
+      "step": 2273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000170878218331585,
+      "loss": 2.1582,
+      "step": 2274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001704833100432237,
+      "loss": 2.1494,
+      "step": 2275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001700887648025612,
+      "loss": 2.1729,
+      "step": 2276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016969458304429082,
+      "loss": 2.3438,
+      "step": 2277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016930076520270548,
+      "loss": 2.2051,
+      "step": 2278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016890731171169727,
+      "loss": 2.25,
+      "step": 2279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016851422300475706,
+      "loss": 2.1611,
+      "step": 2280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016812149951497325,
+      "loss": 2.209,
+      "step": 2281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016772914167503223,
+      "loss": 2.2363,
+      "step": 2282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016733714991721738,
+      "loss": 2.2129,
+      "step": 2283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001669455246734089,
+      "loss": 2.1367,
+      "step": 2284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016655426637508287,
+      "loss": 2.2441,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001661633754533116,
+      "loss": 2.1602,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001657728523387621,
+      "loss": 2.1992,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001653826974616967,
+      "loss": 2.2285,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016499291125197157,
+      "loss": 2.2031,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001646034941390368,
+      "loss": 2.2412,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016421444655193585,
+      "loss": 2.2422,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001638257689193051,
+      "loss": 2.25,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016343746166937317,
+      "loss": 2.0977,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016304952522996091,
+      "loss": 2.1729,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001626619600284805,
+      "loss": 2.2031,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000162274766491935,
+      "loss": 2.0928,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016188794504691807,
+      "loss": 2.1816,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016150149611961328,
+      "loss": 2.2539,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016111542013579394,
+      "loss": 2.1738,
+      "step": 2299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016072971752082227,
+      "loss": 2.1865,
+      "step": 2300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016034438869964962,
+      "loss": 2.2383,
+      "step": 2301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015995943409681475,
+      "loss": 2.2266,
+      "step": 2302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001595748541364449,
+      "loss": 2.2188,
+      "step": 2303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000159190649242254,
+      "loss": 2.2168,
+      "step": 2304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001588068198375429,
+      "loss": 2.1973,
+      "step": 2305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015842336634519887,
+      "loss": 2.2559,
+      "step": 2306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015804028918769487,
+      "loss": 2.1973,
+      "step": 2307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015765758878708908,
+      "loss": 2.209,
+      "step": 2308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015727526556502498,
+      "loss": 2.1543,
+      "step": 2309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015689331994273042,
+      "loss": 2.207,
+      "step": 2310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015651175234101706,
+      "loss": 2.2734,
+      "step": 2311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015613056318027996,
+      "loss": 2.2461,
+      "step": 2312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015574975288049748,
+      "loss": 2.2188,
+      "step": 2313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015536932186123044,
+      "loss": 2.2012,
+      "step": 2314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015498927054162165,
+      "loss": 2.1816,
+      "step": 2315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001546095993403961,
+      "loss": 2.1875,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015423030867585925,
+      "loss": 2.1406,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015385139896589807,
+      "loss": 2.1953,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001534728706279792,
+      "loss": 2.2129,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015309472407914936,
+      "loss": 2.1152,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015271695973603466,
+      "loss": 2.1875,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015233957801483988,
+      "loss": 2.2656,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001519625793313484,
+      "loss": 2.1777,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001515859641009217,
+      "loss": 2.2734,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015120973273849875,
+      "loss": 2.2646,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015083388565859546,
+      "loss": 2.1875,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001504584232753044,
+      "loss": 2.1758,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015008334600229435,
+      "loss": 2.1914,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014970865425280967,
+      "loss": 2.0918,
+      "step": 2329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014933434843966993,
+      "loss": 2.1992,
+      "step": 2330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014896042897526973,
+      "loss": 2.2793,
+      "step": 2331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014858689627157812,
+      "loss": 2.2158,
+      "step": 2332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001482137507401376,
+      "loss": 2.1807,
+      "step": 2333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014784099279206426,
+      "loss": 2.1641,
+      "step": 2334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001474686228380473,
+      "loss": 2.1592,
+      "step": 2335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014709664128834826,
+      "loss": 2.1719,
+      "step": 2336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001467250485528009,
+      "loss": 2.1338,
+      "step": 2337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014635384504081038,
+      "loss": 2.1855,
+      "step": 2338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014598303116135337,
+      "loss": 2.1797,
+      "step": 2339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014561260732297716,
+      "loss": 2.165,
+      "step": 2340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014524257393379918,
+      "loss": 2.1621,
+      "step": 2341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014487293140150664,
+      "loss": 2.1709,
+      "step": 2342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001445036801333563,
+      "loss": 2.1465,
+      "step": 2343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014413482053617373,
+      "loss": 2.2246,
+      "step": 2344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014376635301635278,
+      "loss": 2.084,
+      "step": 2345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014339827797985573,
+      "loss": 2.1602,
+      "step": 2346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014303059583221244,
+      "loss": 2.2461,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014266330697851955,
+      "loss": 2.1855,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001422964118234405,
+      "loss": 2.1895,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014192991077120504,
+      "loss": 2.0977,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014156380422560873,
+      "loss": 2.207,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014119809259001225,
+      "loss": 2.293,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001408327762673417,
+      "loss": 2.1162,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014046785566008702,
+      "loss": 2.207,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014010333117030273,
+      "loss": 2.2363,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013973920319960653,
+      "loss": 2.2617,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013937547214917933,
+      "loss": 2.2148,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001390121384197649,
+      "loss": 2.2617,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013864920241166902,
+      "loss": 2.1924,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013828666452475925,
+      "loss": 2.1396,
+      "step": 2360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013792452515846488,
+      "loss": 2.2324,
+      "step": 2361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013756278471177604,
+      "loss": 2.248,
+      "step": 2362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013720144358324304,
+      "loss": 2.2656,
+      "step": 2363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013684050217097643,
+      "loss": 2.1914,
+      "step": 2364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013647996087264635,
+      "loss": 2.2422,
+      "step": 2365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013611982008548208,
+      "loss": 2.2031,
+      "step": 2366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013576008020627157,
+      "loss": 2.126,
+      "step": 2367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013540074163136145,
+      "loss": 2.1377,
+      "step": 2368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013504180475665563,
+      "loss": 2.2539,
+      "step": 2369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000134683269977616,
+      "loss": 2.1934,
+      "step": 2370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013432513768926103,
+      "loss": 2.2617,
+      "step": 2371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013396740828616594,
+      "loss": 2.1387,
+      "step": 2372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013361008216246196,
+      "loss": 2.2324,
+      "step": 2373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013325315971183605,
+      "loss": 2.2383,
+      "step": 2374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013289664132753033,
+      "loss": 2.2344,
+      "step": 2375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013254052740234202,
+      "loss": 2.2012,
+      "step": 2376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013218481832862256,
+      "loss": 2.1318,
+      "step": 2377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001318295144982773,
+      "loss": 2.2275,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013147461630276503,
+      "loss": 2.1875,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013112012413309777,
+      "loss": 2.2676,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001307660383798402,
+      "loss": 2.1484,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013041235943310898,
+      "loss": 2.2363,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013005908768257309,
+      "loss": 2.248,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012970622351745226,
+      "loss": 2.1631,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012935376732651778,
+      "loss": 2.1689,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012900171949809092,
+      "loss": 2.1699,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012865008042004327,
+      "loss": 2.2422,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012829885047979612,
+      "loss": 2.2012,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001279480300643198,
+      "loss": 2.1729,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012759761956013355,
+      "loss": 2.1426,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012724761935330503,
+      "loss": 2.2295,
+      "step": 2391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012689802982945,
+      "loss": 2.1211,
+      "step": 2392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012654885137373145,
+      "loss": 2.1895,
+      "step": 2393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012620008437085956,
+      "loss": 2.2188,
+      "step": 2394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012585172920509126,
+      "loss": 2.21,
+      "step": 2395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012550378626022962,
+      "loss": 2.2344,
+      "step": 2396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012515625591962354,
+      "loss": 2.1211,
+      "step": 2397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012480913856616755,
+      "loss": 2.2754,
+      "step": 2398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012446243458230104,
+      "loss": 2.0781,
+      "step": 2399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000124116144350008,
+      "loss": 2.1436,
+      "step": 2400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012377026825081638,
+      "loss": 2.2559,
+      "step": 2401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012342480666579809,
+      "loss": 2.1953,
+      "step": 2402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012307975997556813,
+      "loss": 2.085,
+      "step": 2403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012273512856028458,
+      "loss": 2.085,
+      "step": 2404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012239091279964777,
+      "loss": 2.1992,
+      "step": 2405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012204711307290028,
+      "loss": 2.1191,
+      "step": 2406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012170372975882648,
+      "loss": 2.2402,
+      "step": 2407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012136076323575162,
+      "loss": 2.1631,
+      "step": 2408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012101821388154183,
+      "loss": 2.2246,
+      "step": 2409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012067608207360359,
+      "loss": 2.209,
+      "step": 2410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012033436818888349,
+      "loss": 2.1875,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011999307260386743,
+      "loss": 2.2051,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011965219569458068,
+      "loss": 2.2227,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011931173783658738,
+      "loss": 2.1426,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011897169940498958,
+      "loss": 2.2041,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011863208077442739,
+      "loss": 2.248,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011829288231907842,
+      "loss": 2.1631,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001179541044126573,
+      "loss": 2.2324,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011761574742841519,
+      "loss": 2.2002,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011727781173913998,
+      "loss": 2.2607,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001169402977171548,
+      "loss": 2.125,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011660320573431871,
+      "loss": 2.0752,
+      "step": 2422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011626653616202542,
+      "loss": 2.1689,
+      "step": 2423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011593028937120343,
+      "loss": 2.1855,
+      "step": 2424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011559446573231531,
+      "loss": 2.1875,
+      "step": 2425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011525906561535754,
+      "loss": 2.1367,
+      "step": 2426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011492408938985977,
+      "loss": 2.1895,
+      "step": 2427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011458953742488493,
+      "loss": 2.0537,
+      "step": 2428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011425541008902851,
+      "loss": 2.2305,
+      "step": 2429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011392170775041788,
+      "loss": 2.1846,
+      "step": 2430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011358843077671233,
+      "loss": 2.1338,
+      "step": 2431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011325557953510246,
+      "loss": 2.1953,
+      "step": 2432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011292315439230988,
+      "loss": 2.1777,
+      "step": 2433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001125911557145865,
+      "loss": 2.2148,
+      "step": 2434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001122595838677149,
+      "loss": 2.2793,
+      "step": 2435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011192843921700673,
+      "loss": 2.1992,
+      "step": 2436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011159772212730368,
+      "loss": 2.1309,
+      "step": 2437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011126743296297575,
+      "loss": 2.1348,
+      "step": 2438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011093757208792182,
+      "loss": 2.2207,
+      "step": 2439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011060813986556872,
+      "loss": 2.2334,
+      "step": 2440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011027913665887107,
+      "loss": 2.2197,
+      "step": 2441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001099505628303108,
+      "loss": 2.2393,
+      "step": 2442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010962241874189689,
+      "loss": 2.1602,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010929470475516484,
+      "loss": 2.2295,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010896742123117615,
+      "loss": 2.1953,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010864056853051806,
+      "loss": 2.1641,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010831414701330327,
+      "loss": 2.1416,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010798815703916925,
+      "loss": 2.1104,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010766259896727804,
+      "loss": 2.1777,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010733747315631615,
+      "loss": 2.25,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010701277996449333,
+      "loss": 2.209,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010668851974954319,
+      "loss": 2.2598,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010636469286872196,
+      "loss": 2.1895,
+      "step": 2453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010604129967880854,
+      "loss": 2.126,
+      "step": 2454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010571834053610401,
+      "loss": 2.2812,
+      "step": 2455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001053958157964312,
+      "loss": 2.1738,
+      "step": 2456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010507372581513436,
+      "loss": 2.1562,
+      "step": 2457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010475207094707873,
+      "loss": 2.1445,
+      "step": 2458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010443085154665039,
+      "loss": 2.1553,
+      "step": 2459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010411006796775529,
+      "loss": 2.1465,
+      "step": 2460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001037897205638194,
+      "loss": 2.1914,
+      "step": 2461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010346980968778802,
+      "loss": 2.2188,
+      "step": 2462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010315033569212556,
+      "loss": 2.2412,
+      "step": 2463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010283129892881504,
+      "loss": 2.1777,
+      "step": 2464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010251269974935785,
+      "loss": 2.1973,
+      "step": 2465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010219453850477345,
+      "loss": 2.2559,
+      "step": 2466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001018768155455984,
+      "loss": 2.085,
+      "step": 2467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010155953122188654,
+      "loss": 2.1934,
+      "step": 2468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010124268588320856,
+      "loss": 2.2168,
+      "step": 2469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010092627987865128,
+      "loss": 2.2305,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010061031355681765,
+      "loss": 2.1699,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010029478726582603,
+      "loss": 2.1309,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.99797013533102e-05,
+      "loss": 2.1201,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.966505616641869e-05,
+      "loss": 2.1719,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.935085205181432e-05,
+      "loss": 2.1758,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.903708935567407e-05,
+      "loss": 2.1367,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.872376842368851e-05,
+      "loss": 2.2148,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.84108896010616e-05,
+      "loss": 2.2832,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.809845323250988e-05,
+      "loss": 2.1729,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.778645966226284e-05,
+      "loss": 2.1035,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.747490923406215e-05,
+      "loss": 2.2832,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.716380229116078e-05,
+      "loss": 2.2012,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.685313917632344e-05,
+      "loss": 2.1758,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.65429202318257e-05,
+      "loss": 2.0918,
+      "step": 2484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.623314579945381e-05,
+      "loss": 2.208,
+      "step": 2485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.592381622050412e-05,
+      "loss": 2.1621,
+      "step": 2486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.561493183578335e-05,
+      "loss": 2.1992,
+      "step": 2487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.530649298560706e-05,
+      "loss": 2.1523,
+      "step": 2488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.499850000980059e-05,
+      "loss": 2.1387,
+      "step": 2489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.469095324769755e-05,
+      "loss": 2.1748,
+      "step": 2490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.438385303814006e-05,
+      "loss": 2.252,
+      "step": 2491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.407719971947832e-05,
+      "loss": 2.168,
+      "step": 2492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.37709936295701e-05,
+      "loss": 2.2549,
+      "step": 2493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.346523510578032e-05,
+      "loss": 2.2715,
+      "step": 2494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.315992448498106e-05,
+      "loss": 2.2539,
+      "step": 2495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.285506210355098e-05,
+      "loss": 2.2207,
+      "step": 2496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.255064829737447e-05,
+      "loss": 2.1445,
+      "step": 2497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.224668340184195e-05,
+      "loss": 2.1641,
+      "step": 2498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.194316775184925e-05,
+      "loss": 2.2539,
+      "step": 2499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.16401016817972e-05,
+      "loss": 2.1084,
+      "step": 2500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.133748552559112e-05,
+      "loss": 2.1582,
+      "step": 2501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.103531961664119e-05,
+      "loss": 2.1699,
+      "step": 2502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.073360428786087e-05,
+      "loss": 2.1748,
+      "step": 2503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.04323398716677e-05,
+      "loss": 2.1758,
+      "step": 2504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.013152669998209e-05,
+      "loss": 2.1494,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.983116510422745e-05,
+      "loss": 2.126,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.953125541532959e-05,
+      "loss": 2.1738,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.923179796371638e-05,
+      "loss": 2.1582,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.89327930793175e-05,
+      "loss": 2.1953,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.863424109156409e-05,
+      "loss": 2.1758,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.833614232938836e-05,
+      "loss": 2.2578,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.803849712122292e-05,
+      "loss": 2.1914,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.774130579500078e-05,
+      "loss": 2.1953,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.744456867815498e-05,
+      "loss": 2.1748,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.7148286097618e-05,
+      "loss": 2.1689,
+      "step": 2515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.685245837982147e-05,
+      "loss": 2.1006,
+      "step": 2516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.655708585069627e-05,
+      "loss": 2.1592,
+      "step": 2517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.626216883567123e-05,
+      "loss": 2.2031,
+      "step": 2518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.596770765967376e-05,
+      "loss": 2.2012,
+      "step": 2519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.567370264712887e-05,
+      "loss": 2.1504,
+      "step": 2520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.538015412195882e-05,
+      "loss": 2.1777,
+      "step": 2521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.508706240758318e-05,
+      "loss": 2.167,
+      "step": 2522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.479442782691815e-05,
+      "loss": 2.2217,
+      "step": 2523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.450225070237611e-05,
+      "loss": 2.1523,
+      "step": 2524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.42105313558657e-05,
+      "loss": 2.1855,
+      "step": 2525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.391927010879124e-05,
+      "loss": 2.1562,
+      "step": 2526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.362846728205203e-05,
+      "loss": 2.1895,
+      "step": 2527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.33381231960425e-05,
+      "loss": 2.251,
+      "step": 2528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.30482381706516e-05,
+      "loss": 2.1699,
+      "step": 2529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.275881252526245e-05,
+      "loss": 2.1914,
+      "step": 2530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.246984657875206e-05,
+      "loss": 2.2412,
+      "step": 2531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.218134064949112e-05,
+      "loss": 2.1602,
+      "step": 2532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189329505534354e-05,
+      "loss": 2.1797,
+      "step": 2533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.160571011366575e-05,
+      "loss": 2.2275,
+      "step": 2534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131858614130677e-05,
+      "loss": 2.2344,
+      "step": 2535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.10319234546078e-05,
+      "loss": 2.1592,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.074572236940186e-05,
+      "loss": 2.1084,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.045998320101322e-05,
+      "loss": 2.2207,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.017470626425732e-05,
+      "loss": 2.1855,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.988989187344037e-05,
+      "loss": 2.2188,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.960554034235918e-05,
+      "loss": 2.0664,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.932165198430013e-05,
+      "loss": 2.2129,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.903822711203962e-05,
+      "loss": 2.1738,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.875526603784327e-05,
+      "loss": 2.2695,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.847276907346574e-05,
+      "loss": 2.1875,
+      "step": 2545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.819073653015025e-05,
+      "loss": 2.1445,
+      "step": 2546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.790916871862863e-05,
+      "loss": 2.1592,
+      "step": 2547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.762806594912052e-05,
+      "loss": 2.1719,
+      "step": 2548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.73474285313332e-05,
+      "loss": 2.1709,
+      "step": 2549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.706725677446114e-05,
+      "loss": 2.1064,
+      "step": 2550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.678755098718587e-05,
+      "loss": 2.1807,
+      "step": 2551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.650831147767557e-05,
+      "loss": 2.25,
+      "step": 2552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.622953855358456e-05,
+      "loss": 2.2324,
+      "step": 2553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.59512325220531e-05,
+      "loss": 2.2119,
+      "step": 2554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.567339368970733e-05,
+      "loss": 2.1133,
+      "step": 2555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.539602236265842e-05,
+      "loss": 2.2617,
+      "step": 2556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.511911884650251e-05,
+      "loss": 2.208,
+      "step": 2557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.484268344632017e-05,
+      "loss": 2.1982,
+      "step": 2558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.45667164666765e-05,
+      "loss": 2.1523,
+      "step": 2559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.429121821162027e-05,
+      "loss": 2.2422,
+      "step": 2560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.401618898468382e-05,
+      "loss": 2.2109,
+      "step": 2561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3741629088883e-05,
+      "loss": 2.1943,
+      "step": 2562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.346753882671647e-05,
+      "loss": 2.1611,
+      "step": 2563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.319391850016532e-05,
+      "loss": 2.1572,
+      "step": 2564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.292076841069301e-05,
+      "loss": 2.1924,
+      "step": 2565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.264808885924485e-05,
+      "loss": 2.1348,
+      "step": 2566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.237588014624774e-05,
+      "loss": 2.084,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.21041425716098e-05,
+      "loss": 2.1719,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.183287643472025e-05,
+      "loss": 2.2539,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.156208203444864e-05,
+      "loss": 2.1719,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.129175966914503e-05,
+      "loss": 2.2207,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.102190963663924e-05,
+      "loss": 2.1738,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.075253223424072e-05,
+      "loss": 2.2168,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.048362775873828e-05,
+      "loss": 2.1836,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.021519650639951e-05,
+      "loss": 2.1816,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.994723877297071e-05,
+      "loss": 2.2344,
+      "step": 2576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.967975485367656e-05,
+      "loss": 2.1895,
+      "step": 2577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.941274504321982e-05,
+      "loss": 2.2324,
+      "step": 2578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.91462096357805e-05,
+      "loss": 2.1777,
+      "step": 2579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.888014892501626e-05,
+      "loss": 2.0938,
+      "step": 2580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.861456320406163e-05,
+      "loss": 2.2207,
+      "step": 2581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.834945276552778e-05,
+      "loss": 2.2578,
+      "step": 2582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.808481790150228e-05,
+      "loss": 2.1562,
+      "step": 2583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.782065890354893e-05,
+      "loss": 2.1895,
+      "step": 2584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.755697606270683e-05,
+      "loss": 2.1855,
+      "step": 2585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.729376966949091e-05,
+      "loss": 2.2021,
+      "step": 2586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.703104001389087e-05,
+      "loss": 2.1807,
+      "step": 2587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.676878738537123e-05,
+      "loss": 2.2109,
+      "step": 2588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.650701207287102e-05,
+      "loss": 2.207,
+      "step": 2589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.624571436480326e-05,
+      "loss": 2.1758,
+      "step": 2590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.598489454905471e-05,
+      "loss": 2.2695,
+      "step": 2591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.57245529129859e-05,
+      "loss": 2.2168,
+      "step": 2592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.546468974343039e-05,
+      "loss": 2.1973,
+      "step": 2593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.520530532669449e-05,
+      "loss": 2.1836,
+      "step": 2594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.494639994855711e-05,
+      "loss": 2.1641,
+      "step": 2595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.468797389426928e-05,
+      "loss": 2.1172,
+      "step": 2596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.443002744855408e-05,
+      "loss": 2.1592,
+      "step": 2597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.417256089560597e-05,
+      "loss": 2.1553,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.391557451909097e-05,
+      "loss": 2.2217,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.365906860214593e-05,
+      "loss": 2.209,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.340304342737835e-05,
+      "loss": 2.2227,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.314749927686597e-05,
+      "loss": 2.25,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.289243643215664e-05,
+      "loss": 2.3105,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.263785517426795e-05,
+      "loss": 2.1406,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.238375578368688e-05,
+      "loss": 2.1934,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.213013854036942e-05,
+      "loss": 2.1621,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.18770037237405e-05,
+      "loss": 2.1523,
+      "step": 2607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.162435161269358e-05,
+      "loss": 2.1123,
+      "step": 2608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.137218248559012e-05,
+      "loss": 2.1328,
+      "step": 2609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.112049662025943e-05,
+      "loss": 2.1748,
+      "step": 2610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.086929429399857e-05,
+      "loss": 2.2207,
+      "step": 2611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.061857578357161e-05,
+      "loss": 2.1309,
+      "step": 2612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.036834136520974e-05,
+      "loss": 2.1533,
+      "step": 2613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.011859131461084e-05,
+      "loss": 2.1328,
+      "step": 2614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.986932590693911e-05,
+      "loss": 2.207,
+      "step": 2615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9620545416824655e-05,
+      "loss": 2.207,
+      "step": 2616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.93722501183635e-05,
+      "loss": 2.1943,
+      "step": 2617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9124440285116876e-05,
+      "loss": 2.2441,
+      "step": 2618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.887711619011143e-05,
+      "loss": 2.126,
+      "step": 2619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.863027810583837e-05,
+      "loss": 2.123,
+      "step": 2620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.838392630425354e-05,
+      "loss": 2.1826,
+      "step": 2621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.813806105677716e-05,
+      "loss": 2.167,
+      "step": 2622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.789268263429332e-05,
+      "loss": 2.2402,
+      "step": 2623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7647791307149586e-05,
+      "loss": 2.168,
+      "step": 2624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7403387345156986e-05,
+      "loss": 2.1846,
+      "step": 2625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.715947101758967e-05,
+      "loss": 2.1758,
+      "step": 2626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.691604259318434e-05,
+      "loss": 2.0977,
+      "step": 2627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.667310234014012e-05,
+      "loss": 2.125,
+      "step": 2628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.643065052611862e-05,
+      "loss": 2.1074,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.618868741824312e-05,
+      "loss": 2.1494,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5947213283098375e-05,
+      "loss": 2.1387,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5706228386730506e-05,
+      "loss": 2.1631,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5465732994646525e-05,
+      "loss": 2.1846,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.522572737181425e-05,
+      "loss": 2.1318,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4986211782661664e-05,
+      "loss": 2.2637,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.47471864910773e-05,
+      "loss": 2.1992,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4508651760408976e-05,
+      "loss": 2.2129,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4270607853464496e-05,
+      "loss": 2.0928,
+      "step": 2638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.403305503251055e-05,
+      "loss": 2.1602,
+      "step": 2639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3795993559272896e-05,
+      "loss": 2.1182,
+      "step": 2640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3559423694935995e-05,
+      "loss": 2.1719,
+      "step": 2641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.332334570014252e-05,
+      "loss": 2.1904,
+      "step": 2642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.308775983499331e-05,
+      "loss": 2.2305,
+      "step": 2643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.285266635904712e-05,
+      "loss": 2.1895,
+      "step": 2644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.261806553132015e-05,
+      "loss": 2.1895,
+      "step": 2645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2383957610285694e-05,
+      "loss": 2.2148,
+      "step": 2646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.21503428538741e-05,
+      "loss": 2.2363,
+      "step": 2647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191722151947226e-05,
+      "loss": 2.207,
+      "step": 2648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.168459386392355e-05,
+      "loss": 2.2422,
+      "step": 2649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.145246014352728e-05,
+      "loss": 2.0684,
+      "step": 2650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.122082061403882e-05,
+      "loss": 2.209,
+      "step": 2651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.09896755306688e-05,
+      "loss": 2.1523,
+      "step": 2652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0759025148083336e-05,
+      "loss": 2.2871,
+      "step": 2653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.052886972040332e-05,
+      "loss": 2.1455,
+      "step": 2654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0299209501204315e-05,
+      "loss": 2.1895,
+      "step": 2655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.007004474351634e-05,
+      "loss": 2.1553,
+      "step": 2656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.984137569982355e-05,
+      "loss": 2.207,
+      "step": 2657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9613202622063815e-05,
+      "loss": 2.1719,
+      "step": 2658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9385525761628857e-05,
+      "loss": 2.3203,
+      "step": 2659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.915834536936348e-05,
+      "loss": 2.2676,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8931661695565565e-05,
+      "loss": 2.1416,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.870547498998562e-05,
+      "loss": 2.165,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.847978550182669e-05,
+      "loss": 2.1992,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8254593479744036e-05,
+      "loss": 2.2227,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.802989917184469e-05,
+      "loss": 2.2461,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.780570282568758e-05,
+      "loss": 2.1299,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.758200468828289e-05,
+      "loss": 2.1797,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.73588050060918e-05,
+      "loss": 2.1689,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.713610402502638e-05,
+      "loss": 2.1338,
+      "step": 2669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.691390199044926e-05,
+      "loss": 2.1758,
+      "step": 2670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.669219914717332e-05,
+      "loss": 2.1855,
+      "step": 2671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.647099573946145e-05,
+      "loss": 2.168,
+      "step": 2672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.625029201102626e-05,
+      "loss": 2.1973,
+      "step": 2673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.603008820502996e-05,
+      "loss": 2.1777,
+      "step": 2674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5810384564083984e-05,
+      "loss": 2.1348,
+      "step": 2675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.559118133024853e-05,
+      "loss": 2.1279,
+      "step": 2676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.537247874503253e-05,
+      "loss": 2.1562,
+      "step": 2677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.515427704939334e-05,
+      "loss": 2.0918,
+      "step": 2678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4936576483736444e-05,
+      "loss": 2.1582,
+      "step": 2679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.471937728791514e-05,
+      "loss": 2.2656,
+      "step": 2680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.450267970123051e-05,
+      "loss": 2.165,
+      "step": 2681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.428648396243101e-05,
+      "loss": 2.1885,
+      "step": 2682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.407079030971184e-05,
+      "loss": 2.2012,
+      "step": 2683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3855598980715396e-05,
+      "loss": 2.1387,
+      "step": 2684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.364091021253036e-05,
+      "loss": 2.0928,
+      "step": 2685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3426724241691915e-05,
+      "loss": 2.1299,
+      "step": 2686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.321304130418113e-05,
+      "loss": 2.2227,
+      "step": 2687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.299986163542491e-05,
+      "loss": 2.2139,
+      "step": 2688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.27871854702957e-05,
+      "loss": 2.1641,
+      "step": 2689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.257501304311134e-05,
+      "loss": 2.1572,
+      "step": 2690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2363344587634415e-05,
+      "loss": 2.1641,
+      "step": 2691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.215218033707241e-05,
+      "loss": 2.1357,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1941520524077234e-05,
+      "loss": 2.1758,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.173136538074501e-05,
+      "loss": 2.04,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.152171513861586e-05,
+      "loss": 2.1055,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1312570028673725e-05,
+      "loss": 2.1201,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.110393028134601e-05,
+      "loss": 2.1416,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.089579612650313e-05,
+      "loss": 2.2637,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.068816779345863e-05,
+      "loss": 2.1934,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.048104551096876e-05,
+      "loss": 2.166,
+      "step": 2700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.027442950723209e-05,
+      "loss": 2.1738,
+      "step": 2701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.006832000988941e-05,
+      "loss": 2.125,
+      "step": 2702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.98627172460238e-05,
+      "loss": 2.1797,
+      "step": 2703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965762144215956e-05,
+      "loss": 2.2266,
+      "step": 2704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.945303282426277e-05,
+      "loss": 2.2139,
+      "step": 2705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9248951617740616e-05,
+      "loss": 2.2637,
+      "step": 2706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.904537804744118e-05,
+      "loss": 2.1699,
+      "step": 2707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.884231233765334e-05,
+      "loss": 2.1104,
+      "step": 2708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.863975471210635e-05,
+      "loss": 2.1367,
+      "step": 2709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.843770539396968e-05,
+      "loss": 2.1592,
+      "step": 2710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8236164605852885e-05,
+      "loss": 2.1816,
+      "step": 2711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.803513256980523e-05,
+      "loss": 2.2266,
+      "step": 2712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.783460950731532e-05,
+      "loss": 2.2441,
+      "step": 2713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7634595639311155e-05,
+      "loss": 2.1914,
+      "step": 2714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.743509118615951e-05,
+      "loss": 2.1348,
+      "step": 2715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.723609636766612e-05,
+      "loss": 2.2012,
+      "step": 2716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.70376114030751e-05,
+      "loss": 2.1572,
+      "step": 2717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6839636511068974e-05,
+      "loss": 2.2168,
+      "step": 2718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.664217190976804e-05,
+      "loss": 2.2383,
+      "step": 2719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.644521781673071e-05,
+      "loss": 2.1113,
+      "step": 2720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.624877444895269e-05,
+      "loss": 2.0986,
+      "step": 2721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.605284202286702e-05,
+      "loss": 2.2734,
+      "step": 2722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5857420754343895e-05,
+      "loss": 2.1934,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.566251085869032e-05,
+      "loss": 2.2305,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546811255064969e-05,
+      "loss": 2.0996,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5274226044402e-05,
+      "loss": 2.2197,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.50808515535635e-05,
+      "loss": 2.1758,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.488798929118592e-05,
+      "loss": 2.1289,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.469563946975684e-05,
+      "loss": 2.166,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.450380230119926e-05,
+      "loss": 2.1797,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.431247799687137e-05,
+      "loss": 2.2812,
+      "step": 2731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4121666767566175e-05,
+      "loss": 2.1602,
+      "step": 2732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.393136882351161e-05,
+      "loss": 2.1387,
+      "step": 2733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.374158437436991e-05,
+      "loss": 2.166,
+      "step": 2734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.355231362923783e-05,
+      "loss": 2.1816,
+      "step": 2735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.336355679664577e-05,
+      "loss": 2.3027,
+      "step": 2736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.317531408455809e-05,
+      "loss": 2.123,
+      "step": 2737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.298758570037286e-05,
+      "loss": 2.1445,
+      "step": 2738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2800371850921275e-05,
+      "loss": 2.1797,
+      "step": 2739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2613672742467685e-05,
+      "loss": 2.0547,
+      "step": 2740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.242748858070937e-05,
+      "loss": 2.291,
+      "step": 2741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2241819570776376e-05,
+      "loss": 2.1924,
+      "step": 2742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2056665917231e-05,
+      "loss": 2.1924,
+      "step": 2743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.187202782406767e-05,
+      "loss": 2.1973,
+      "step": 2744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.168790549471301e-05,
+      "loss": 2.1445,
+      "step": 2745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.150429913202529e-05,
+      "loss": 2.1758,
+      "step": 2746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.132120893829421e-05,
+      "loss": 2.1924,
+      "step": 2747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1138635115240956e-05,
+      "loss": 2.1934,
+      "step": 2748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.095657786401784e-05,
+      "loss": 2.1543,
+      "step": 2749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.077503738520781e-05,
+      "loss": 2.2363,
+      "step": 2750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.059401387882455e-05,
+      "loss": 2.1279,
+      "step": 2751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0413507544312268e-05,
+      "loss": 2.2031,
+      "step": 2752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.023351858054524e-05,
+      "loss": 2.1602,
+      "step": 2753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0054047185827715e-05,
+      "loss": 2.2061,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.987509355789375e-05,
+      "loss": 2.0977,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9696657893907075e-05,
+      "loss": 2.2656,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9518740390460652e-05,
+      "loss": 2.0967,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.934134124357646e-05,
+      "loss": 2.1025,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9164460648705538e-05,
+      "loss": 2.1807,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8988098800727393e-05,
+      "loss": 2.1426,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8812255893950257e-05,
+      "loss": 2.2139,
+      "step": 2761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8636932122110327e-05,
+      "loss": 2.124,
+      "step": 2762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8462127678372086e-05,
+      "loss": 2.1475,
+      "step": 2763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.828784275532781e-05,
+      "loss": 2.1465,
+      "step": 2764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8114077544997342e-05,
+      "loss": 2.2188,
+      "step": 2765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7940832238827874e-05,
+      "loss": 2.208,
+      "step": 2766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7768107027693778e-05,
+      "loss": 2.2178,
+      "step": 2767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7595902101896552e-05,
+      "loss": 2.251,
+      "step": 2768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7424217651164206e-05,
+      "loss": 2.1797,
+      "step": 2769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7253053864651768e-05,
+      "loss": 2.1582,
+      "step": 2770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.708241093094005e-05,
+      "loss": 2.1445,
+      "step": 2771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.69122890380365e-05,
+      "loss": 2.2109,
+      "step": 2772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6742688373374235e-05,
+      "loss": 2.1504,
+      "step": 2773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6573609123812182e-05,
+      "loss": 2.2324,
+      "step": 2774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.640505147563471e-05,
+      "loss": 2.1875,
+      "step": 2775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6237015614551664e-05,
+      "loss": 2.1484,
+      "step": 2776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.606950172569783e-05,
+      "loss": 2.2119,
+      "step": 2777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5902509993633028e-05,
+      "loss": 2.1289,
+      "step": 2778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5736040602341858e-05,
+      "loss": 2.2021,
+      "step": 2779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.557009373523328e-05,
+      "loss": 2.1934,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5404669575140592e-05,
+      "loss": 2.1709,
+      "step": 2781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5239768304321233e-05,
+      "loss": 2.124,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5075390104456476e-05,
+      "loss": 2.125,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.491153515665129e-05,
+      "loss": 2.1406,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.474820364143432e-05,
+      "loss": 2.1113,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4585395738757342e-05,
+      "loss": 2.1367,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.442311162799532e-05,
+      "loss": 2.1562,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.426135148794606e-05,
+      "loss": 2.1992,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.410011549683011e-05,
+      "loss": 2.1377,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3939403832290594e-05,
+      "loss": 2.127,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3779216671392878e-05,
+      "loss": 2.1309,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.361955419062434e-05,
+      "loss": 2.2334,
+      "step": 2792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3460416565894537e-05,
+      "loss": 2.2422,
+      "step": 2793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.330180397253473e-05,
+      "loss": 2.1582,
+      "step": 2794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3143716585297514e-05,
+      "loss": 2.1504,
+      "step": 2795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.298615457835701e-05,
+      "loss": 2.1777,
+      "step": 2796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.282911812530841e-05,
+      "loss": 2.1514,
+      "step": 2797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2672607399167878e-05,
+      "loss": 2.127,
+      "step": 2798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2516622572372415e-05,
+      "loss": 2.1807,
+      "step": 2799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2361163816779608e-05,
+      "loss": 2.1143,
+      "step": 2800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2206231303667335e-05,
+      "loss": 2.1387,
+      "step": 2801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2051825203733832e-05,
+      "loss": 2.0918,
+      "step": 2802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.18979456870973e-05,
+      "loss": 2.1895,
+      "step": 2803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.174459292329567e-05,
+      "loss": 2.1982,
+      "step": 2804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.159176708128663e-05,
+      "loss": 2.2969,
+      "step": 2805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1439468329447264e-05,
+      "loss": 2.1201,
+      "step": 2806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1287696835573966e-05,
+      "loss": 2.1426,
+      "step": 2807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1136452766882196e-05,
+      "loss": 2.1572,
+      "step": 2808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0985736290006497e-05,
+      "loss": 2.168,
+      "step": 2809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0835547570999813e-05,
+      "loss": 2.123,
+      "step": 2810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0685886775333895e-05,
+      "loss": 2.127,
+      "step": 2811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0536754067898677e-05,
+      "loss": 2.0312,
+      "step": 2812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0388149613002338e-05,
+      "loss": 2.2344,
+      "step": 2813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.024007357437102e-05,
+      "loss": 2.1934,
+      "step": 2814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.009252611514867e-05,
+      "loss": 2.166,
+      "step": 2815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9945507397897132e-05,
+      "loss": 2.1582,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9799017584595236e-05,
+      "loss": 2.1211,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9653056836639483e-05,
+      "loss": 2.0576,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.950762531484318e-05,
+      "loss": 2.1348,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9362723179436825e-05,
+      "loss": 2.1348,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9218350590067436e-05,
+      "loss": 2.1836,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9074507705798603e-05,
+      "loss": 2.2295,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8931194685110564e-05,
+      "loss": 2.1738,
+      "step": 2823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878841168589962e-05,
+      "loss": 2.1777,
+      "step": 2824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.864615886547799e-05,
+      "loss": 2.2598,
+      "step": 2825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.850443638057392e-05,
+      "loss": 2.167,
+      "step": 2826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8363244387331336e-05,
+      "loss": 2.2422,
+      "step": 2827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.822258304130964e-05,
+      "loss": 2.1836,
+      "step": 2828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8082452497483582e-05,
+      "loss": 2.0928,
+      "step": 2829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7942852910243112e-05,
+      "loss": 2.1162,
+      "step": 2830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7803784433393423e-05,
+      "loss": 2.1504,
+      "step": 2831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7665247220154112e-05,
+      "loss": 2.166,
+      "step": 2832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.752724142315981e-05,
+      "loss": 2.2148,
+      "step": 2833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.738976719445945e-05,
+      "loss": 2.1816,
+      "step": 2834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7252824685516423e-05,
+      "loss": 2.1621,
+      "step": 2835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7116414047208205e-05,
+      "loss": 2.2012,
+      "step": 2836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6980535429826415e-05,
+      "loss": 2.0977,
+      "step": 2837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6845188983076298e-05,
+      "loss": 2.1855,
+      "step": 2838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.671037485607707e-05,
+      "loss": 2.2129,
+      "step": 2839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.657609319736125e-05,
+      "loss": 2.1426,
+      "step": 2840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6442344154874666e-05,
+      "loss": 2.1758,
+      "step": 2841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6309127875976493e-05,
+      "loss": 2.1445,
+      "step": 2842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.617644450743877e-05,
+      "loss": 2.2539,
+      "step": 2843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.604429419544651e-05,
+      "loss": 2.2188,
+      "step": 2844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.591267708559746e-05,
+      "loss": 2.1309,
+      "step": 2845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.57815933229018e-05,
+      "loss": 2.123,
+      "step": 2846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5651043051782164e-05,
+      "loss": 2.2139,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5521026416073335e-05,
+      "loss": 2.1777,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.539154355902217e-05,
+      "loss": 2.1357,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.526259462328755e-05,
+      "loss": 2.1162,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5134179750939947e-05,
+      "loss": 2.2227,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5006299083461517e-05,
+      "loss": 2.252,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4878952761745835e-05,
+      "loss": 2.1631,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4752140926097723e-05,
+      "loss": 2.1328,
+      "step": 2854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.46258637162332e-05,
+      "loss": 2.0723,
+      "step": 2855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4500121271279253e-05,
+      "loss": 2.1562,
+      "step": 2856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4374913729773508e-05,
+      "loss": 2.1641,
+      "step": 2857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.425024122966445e-05,
+      "loss": 2.0791,
+      "step": 2858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.412610390831104e-05,
+      "loss": 2.2061,
+      "step": 2859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4002501902482534e-05,
+      "loss": 2.21,
+      "step": 2860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.387943534835856e-05,
+      "loss": 2.1641,
+      "step": 2861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3756904381528657e-05,
+      "loss": 2.1953,
+      "step": 2862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.363490913699228e-05,
+      "loss": 2.1562,
+      "step": 2863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.351344974915869e-05,
+      "loss": 2.0586,
+      "step": 2864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3392526351846734e-05,
+      "loss": 2.1973,
+      "step": 2865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3272139078284784e-05,
+      "loss": 2.1172,
+      "step": 2866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3152288061110518e-05,
+      "loss": 2.1455,
+      "step": 2867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3032973432370699e-05,
+      "loss": 2.1758,
+      "step": 2868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2914195323521284e-05,
+      "loss": 2.1152,
+      "step": 2869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.279595386542698e-05,
+      "loss": 2.127,
+      "step": 2870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2678249188361246e-05,
+      "loss": 2.1729,
+      "step": 2871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2561081422006238e-05,
+      "loss": 2.1465,
+      "step": 2872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.244445069545247e-05,
+      "loss": 2.1885,
+      "step": 2873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.232835713719871e-05,
+      "loss": 2.1895,
+      "step": 2874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2212800875152086e-05,
+      "loss": 2.1123,
+      "step": 2875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2097782036627646e-05,
+      "loss": 2.1318,
+      "step": 2876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1983300748348302e-05,
+      "loss": 2.1445,
+      "step": 2877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1869357136444769e-05,
+      "loss": 2.0322,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.175595132645535e-05,
+      "loss": 2.0811,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1643083443325819e-05,
+      "loss": 2.2227,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.153075361140915e-05,
+      "loss": 2.1064,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1418961954465845e-05,
+      "loss": 2.1885,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1307708595663213e-05,
+      "loss": 2.1211,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1196993657575538e-05,
+      "loss": 2.2051,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1086817262183912e-05,
+      "loss": 2.2266,
+      "step": 2885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.097717953087607e-05,
+      "loss": 2.1426,
+      "step": 2886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0868080584446271e-05,
+      "loss": 2.1816,
+      "step": 2887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0759520543095148e-05,
+      "loss": 2.1582,
+      "step": 2888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0651499526429632e-05,
+      "loss": 2.2441,
+      "step": 2889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0544017653462745e-05,
+      "loss": 2.1338,
+      "step": 2890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0437075042613598e-05,
+      "loss": 2.3203,
+      "step": 2891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0330671811706993e-05,
+      "loss": 2.0557,
+      "step": 2892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0224808077973602e-05,
+      "loss": 2.1436,
+      "step": 2893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0119483958049624e-05,
+      "loss": 2.0986,
+      "step": 2894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0014699567976792e-05,
+      "loss": 2.165,
+      "step": 2895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.910455023202092e-06,
+      "loss": 2.1484,
+      "step": 2896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.806750438577872e-06,
+      "loss": 2.0986,
+      "step": 2897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.703585928361514e-06,
+      "loss": 2.2168,
+      "step": 2898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.600961606215319e-06,
+      "loss": 2.207,
+      "step": 2899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.498877585206512e-06,
+      "loss": 2.2598,
+      "step": 2900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.39733397780701e-06,
+      "loss": 2.1445,
+      "step": 2901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.296330895893213e-06,
+      "loss": 2.1719,
+      "step": 2902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.195868450746103e-06,
+      "loss": 2.2109,
+      "step": 2903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.095946753051087e-06,
+      "loss": 2.1836,
+      "step": 2904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.996565912897659e-06,
+      "loss": 2.0986,
+      "step": 2905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.897726039779507e-06,
+      "loss": 2.1543,
+      "step": 2906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.799427242594193e-06,
+      "loss": 2.209,
+      "step": 2907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.701669629643305e-06,
+      "loss": 2.2051,
+      "step": 2908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.604453308631965e-06,
+      "loss": 2.123,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.507778386669219e-06,
+      "loss": 2.0713,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.411644970267252e-06,
+      "loss": 2.1357,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.316053165341952e-06,
+      "loss": 2.1992,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.221003077212464e-06,
+      "loss": 2.1797,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.126494810600848e-06,
+      "loss": 2.1182,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.032528469632372e-06,
+      "loss": 2.1914,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.939104157835275e-06,
+      "loss": 2.0645,
+      "step": 2916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.846221978140556e-06,
+      "loss": 2.1777,
+      "step": 2917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.753882032881798e-06,
+      "loss": 2.1035,
+      "step": 2918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.662084423795456e-06,
+      "loss": 2.1992,
+      "step": 2919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.570829252020128e-06,
+      "loss": 2.1211,
+      "step": 2920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.48011661809711e-06,
+      "loss": 2.1689,
+      "step": 2921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.389946621969679e-06,
+      "loss": 2.166,
+      "step": 2922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.3003193629834204e-06,
+      "loss": 2.248,
+      "step": 2923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.211234939885847e-06,
+      "loss": 2.1562,
+      "step": 2924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.122693450826445e-06,
+      "loss": 2.1914,
+      "step": 2925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.034694993356517e-06,
+      "loss": 2.1221,
+      "step": 2926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.947239664429117e-06,
+      "loss": 2.1621,
+      "step": 2927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8603275603988934e-06,
+      "loss": 2.1875,
+      "step": 2928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.773958777021916e-06,
+      "loss": 2.0938,
+      "step": 2929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.6881334094557325e-06,
+      "loss": 2.1104,
+      "step": 2930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.6028515522590925e-06,
+      "loss": 2.1855,
+      "step": 2931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.518113299392003e-06,
+      "loss": 2.2246,
+      "step": 2932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.433918744215505e-06,
+      "loss": 2.2695,
+      "step": 2933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3502679794916755e-06,
+      "loss": 2.2031,
+      "step": 2934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.267161097383345e-06,
+      "loss": 2.0898,
+      "step": 2935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.184598189454327e-06,
+      "loss": 2.1367,
+      "step": 2936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.102579346668913e-06,
+      "loss": 2.1885,
+      "step": 2937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.021104659391985e-06,
+      "loss": 2.0791,
+      "step": 2938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.940174217389016e-06,
+      "loss": 2.1289,
+      "step": 2939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.859788109825792e-06,
+      "loss": 2.1504,
+      "step": 2940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.779946425268246e-06,
+      "loss": 2.1133,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7006492516826765e-06,
+      "loss": 2.1904,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.621896676435423e-06,
+      "loss": 2.1064,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5436887862927445e-06,
+      "loss": 2.168,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.466025667420826e-06,
+      "loss": 2.1494,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3889074053856124e-06,
+      "loss": 2.1523,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.312334085152859e-06,
+      "loss": 2.1055,
+      "step": 2947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.236305791087747e-06,
+      "loss": 2.2949,
+      "step": 2948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1608226069551066e-06,
+      "loss": 2.1973,
+      "step": 2949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.08588461591919e-06,
+      "loss": 2.207,
+      "step": 2950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.011491900543508e-06,
+      "loss": 2.2051,
+      "step": 2951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9376445427908314e-06,
+      "loss": 2.1328,
+      "step": 2952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.864342624023077e-06,
+      "loss": 2.1523,
+      "step": 2953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.79158622500131e-06,
+      "loss": 2.1016,
+      "step": 2954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.719375425885408e-06,
+      "loss": 2.1738,
+      "step": 2955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.647710306234176e-06,
+      "loss": 2.124,
+      "step": 2956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.576590945005288e-06,
+      "loss": 2.123,
+      "step": 2957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.50601742055512e-06,
+      "loss": 2.2852,
+      "step": 2958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.435989810638641e-06,
+      "loss": 2.2793,
+      "step": 2959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.366508192409246e-06,
+      "loss": 2.1523,
+      "step": 2960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.297572642418923e-06,
+      "loss": 2.1445,
+      "step": 2961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.229183236618028e-06,
+      "loss": 2.2051,
+      "step": 2962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1613400503550115e-06,
+      "loss": 2.2324,
+      "step": 2963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.094043158376748e-06,
+      "loss": 2.2266,
+      "step": 2964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.027292634828151e-06,
+      "loss": 2.2305,
+      "step": 2965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.961088553252168e-06,
+      "loss": 2.1699,
+      "step": 2966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.895430986589621e-06,
+      "loss": 2.2109,
+      "step": 2967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.830320007179311e-06,
+      "loss": 2.1064,
+      "step": 2968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.765755686757744e-06,
+      "loss": 2.1875,
+      "step": 2969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7017380964592394e-06,
+      "loss": 2.1953,
+      "step": 2970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.638267306815657e-06,
+      "loss": 2.2363,
+      "step": 2971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.575343387756447e-06,
+      "loss": 2.1025,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.512966408608653e-06,
+      "loss": 2.2109,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.451136438096525e-06,
+      "loss": 2.165,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3898535443417367e-06,
+      "loss": 2.1426,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.329117794863279e-06,
+      "loss": 2.1289,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.26892925657718e-06,
+      "loss": 2.1816,
+      "step": 2977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.209287995796728e-06,
+      "loss": 2.1855,
+      "step": 2978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1501940782320826e-06,
+      "loss": 2.1758,
+      "step": 2979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0916475689904966e-06,
+      "loss": 2.1445,
+      "step": 2980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.033648532576094e-06,
+      "loss": 2.1533,
+      "step": 2981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9761970328897605e-06,
+      "loss": 2.2188,
+      "step": 2982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.919293133229195e-06,
+      "loss": 2.1738,
+      "step": 2983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8629368962886372e-06,
+      "loss": 2.1602,
+      "step": 2984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8071283841590303e-06,
+      "loss": 2.1973,
+      "step": 2985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7518676583279666e-06,
+      "loss": 2.1934,
+      "step": 2986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6971547796793007e-06,
+      "loss": 2.2051,
+      "step": 2987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.642989808493479e-06,
+      "loss": 2.1367,
+      "step": 2988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5893728044471544e-06,
+      "loss": 2.1094,
+      "step": 2989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5363038266132953e-06,
+      "loss": 2.25,
+      "step": 2990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.483782933461076e-06,
+      "loss": 2.1934,
+      "step": 2991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4318101828558205e-06,
+      "loss": 2.1719,
+      "step": 2992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3803856320589455e-06,
+      "loss": 2.1914,
+      "step": 2993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.329509337727853e-06,
+      "loss": 2.1484,
+      "step": 2994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2791813559159267e-06,
+      "loss": 2.1768,
+      "step": 2995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2294017420724234e-06,
+      "loss": 2.2754,
+      "step": 2996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1801705510424706e-06,
+      "loss": 2.2422,
+      "step": 2997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.131487837066848e-06,
+      "loss": 2.1338,
+      "step": 2998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0833536537822605e-06,
+      "loss": 2.1396,
+      "step": 2999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035768054220788e-06,
+      "loss": 2.168,
+      "step": 3000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9887310908103806e-06,
+      "loss": 2.2324,
+      "step": 3001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9422428153743066e-06,
+      "loss": 2.2871,
+      "step": 3002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8963032791314837e-06,
+      "loss": 2.0928,
+      "step": 3003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8509125326960918e-06,
+      "loss": 2.1299,
+      "step": 3004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8060706260778492e-06,
+      "loss": 2.1514,
+      "step": 3005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7617776086815696e-06,
+      "loss": 2.1992,
+      "step": 3006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.71803352930755e-06,
+      "loss": 2.1562,
+      "step": 3007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6748384361511271e-06,
+      "loss": 2.0957,
+      "step": 3008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.632192376802899e-06,
+      "loss": 2.2207,
+      "step": 3009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5900953982484477e-06,
+      "loss": 2.1338,
+      "step": 3010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5485475468685063e-06,
+      "loss": 2.2617,
+      "step": 3011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5075488684387351e-06,
+      "loss": 2.1719,
+      "step": 3012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4670994081297795e-06,
+      "loss": 2.1562,
+      "step": 3013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4271992105071573e-06,
+      "loss": 2.0811,
+      "step": 3014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3878483195312041e-06,
+      "loss": 2.1855,
+      "step": 3015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.349046778557128e-06,
+      "loss": 2.2451,
+      "step": 3016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.310794630334844e-06,
+      "loss": 2.1758,
+      "step": 3017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2730919170089728e-06,
+      "loss": 2.1006,
+      "step": 3018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2359386801187312e-06,
+      "loss": 2.2637,
+      "step": 3019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.199334960598042e-06,
+      "loss": 2.1836,
+      "step": 3020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.163280798775368e-06,
+      "loss": 2.2129,
+      "step": 3021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1277762343736008e-06,
+      "loss": 2.0879,
+      "step": 3022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0928213065102833e-06,
+      "loss": 2.2754,
+      "step": 3023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.05841605369722e-06,
+      "loss": 2.1836,
+      "step": 3024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0245605138407e-06,
+      "loss": 2.1367,
+      "step": 3025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.91254724241386e-07,
+      "loss": 2.2031,
+      "step": 3026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.584987215941476e-07,
+      "loss": 2.1387,
+      "step": 3027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.262925419881718e-07,
+      "loss": 2.1758,
+      "step": 3028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.946362209069081e-07,
+      "loss": 2.2012,
+      "step": 3029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.635297932279018e-07,
+      "loss": 2.1729,
+      "step": 3030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.329732932230161e-07,
+      "loss": 2.0996,
+      "step": 3031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.029667545580988e-07,
+      "loss": 2.2598,
+      "step": 3032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.73510210293038e-07,
+      "loss": 2.1602,
+      "step": 3033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.446036928818733e-07,
+      "loss": 2.2129,
+      "step": 3034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.16247234172629e-07,
+      "loss": 2.2812,
+      "step": 3035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.884408654071472e-07,
+      "loss": 2.2363,
+      "step": 3036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.611846172214219e-07,
+      "loss": 2.2324,
+      "step": 3037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.344785196452096e-07,
+      "loss": 2.126,
+      "step": 3038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.083226021021959e-07,
+      "loss": 2.1523,
+      "step": 3039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.827168934098293e-07,
+      "loss": 2.1084,
+      "step": 3040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.576614217794318e-07,
+      "loss": 2.1787,
+      "step": 3041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.331562148160884e-07,
+      "loss": 2.0889,
+      "step": 3042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.092012995186468e-07,
+      "loss": 2.1504,
+      "step": 3043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.857967022795506e-07,
+      "loss": 2.1797,
+      "step": 3044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6294244888500647e-07,
+      "loss": 2.2266,
+      "step": 3045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.40638564514928e-07,
+      "loss": 2.1484,
+      "step": 3046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1888507374276963e-07,
+      "loss": 2.251,
+      "step": 3047
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9768200053563743e-07,
+      "loss": 2.1025,
+      "step": 3048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.770293682542336e-07,
+      "loss": 2.1328,
+      "step": 3049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5692719965263466e-07,
+      "loss": 2.168,
+      "step": 3050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3737551687867964e-07,
+      "loss": 2.1543,
+      "step": 3051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.183743414735263e-07,
+      "loss": 2.1777,
+      "step": 3052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9992369437198406e-07,
+      "loss": 2.1074,
+      "step": 3053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.820235959020701e-07,
+      "loss": 2.1836,
+      "step": 3054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6467406578539746e-07,
+      "loss": 2.1777,
+      "step": 3055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.478751231370091e-07,
+      "loss": 2.1621,
+      "step": 3056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3162678646526657e-07,
+      "loss": 2.1182,
+      "step": 3057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1592907367185e-07,
+      "loss": 2.1777,
+      "step": 3058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0078200205186913e-07,
+      "loss": 2.1562,
+      "step": 3059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8618558829375244e-07,
+      "loss": 2.1719,
+      "step": 3060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.721398484792469e-07,
+      "loss": 2.1299,
+      "step": 3061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5864479808325173e-07,
+      "loss": 2.2207,
+      "step": 3062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4570045197415116e-07,
+      "loss": 2.1416,
+      "step": 3063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3330682441348164e-07,
+      "loss": 2.2461,
+      "step": 3064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.214639290559316e-07,
+      "loss": 2.1836,
+      "step": 3065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1017177894961927e-07,
+      "loss": 2.2422,
+      "step": 3066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.94303865357038e-08,
+      "loss": 2.1582,
+      "step": 3067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.923976364855202e-08,
+      "loss": 2.0762,
+      "step": 3068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.959992151590489e-08,
+      "loss": 2.0957,
+      "step": 3069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.051087075843343e-08,
+      "loss": 2.168,
+      "step": 3070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.197262139007176e-08,
+      "loss": 2.1982,
+      "step": 3071
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.39851828179061e-08,
+      "loss": 2.0801,
+      "step": 3072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6548563842230276e-08,
+      "loss": 2.1523,
+      "step": 3073
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.966277265637919e-08,
+      "loss": 2.2285,
+      "step": 3074
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.332781684672881e-08,
+      "loss": 2.1279,
+      "step": 3075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.754370339297374e-08,
+      "loss": 2.2148,
+      "step": 3076
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.231043866779414e-08,
+      "loss": 2.1406,
+      "step": 3077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7628028436966758e-08,
+      "loss": 2.2012,
+      "step": 3078
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.349647785942043e-08,
+      "loss": 2.1455,
+      "step": 3079
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.915791487014048e-09,
+      "loss": 2.2109,
+      "step": 3080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.885973264869616e-09,
+      "loss": 2.2588,
+      "step": 3081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.407026531150215e-09,
+      "loss": 2.1934,
+      "step": 3082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4789540169489755e-09,
+      "loss": 2.124,
+      "step": 3083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1017578466776535e-09,
+      "loss": 2.1582,
+      "step": 3084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7543953751152375e-10,
+      "loss": 2.1777,
+      "step": 3085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.2461,
+      "step": 3086
+    },
+    {
+      "epoch": 1.0,
+      "step": 3086,
+      "total_flos": 3.3197425508304814e+18,
+      "train_loss": 2.298177821715003,
+      "train_runtime": 32742.7411,
+      "train_samples_per_second": 24.128,
+      "train_steps_per_second": 0.094
+    }
+  ],
+  "max_steps": 3086,
+  "num_train_epochs": 1,
+  "total_flos": 3.3197425508304814e+18,
+  "trial_name": null,
+  "trial_params": null
+}