diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29706 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998989184271707,
+  "eval_steps": 500,
+  "global_step": 4946,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.7114093959731546e-06,
+      "loss": 6.5081,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3422818791946309e-05,
+      "loss": 6.5167,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.013422818791946e-05,
+      "loss": 6.4218,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6845637583892618e-05,
+      "loss": 5.3159,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3557046979865775e-05,
+      "loss": 5.0078,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.026845637583892e-05,
+      "loss": 4.4732,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.697986577181208e-05,
+      "loss": 3.6586,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3691275167785237e-05,
+      "loss": 3.3308,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.040268456375839e-05,
+      "loss": 3.0749,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.711409395973155e-05,
+      "loss": 2.9314,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.38255033557047e-05,
+      "loss": 2.8257,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.053691275167784e-05,
+      "loss": 2.7712,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.7248322147651e-05,
+      "loss": 2.6402,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.395973154362417e-05,
+      "loss": 2.5781,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010067114093959731,
+      "loss": 2.4845,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010738255033557047,
+      "loss": 2.4082,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011409395973154363,
+      "loss": 2.3542,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012080536912751678,
+      "loss": 2.3931,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012751677852348994,
+      "loss": 2.3453,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001342281879194631,
+      "loss": 2.2771,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014093959731543624,
+      "loss": 2.2842,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001476510067114094,
+      "loss": 2.2795,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015436241610738256,
+      "loss": 2.2922,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001610738255033557,
+      "loss": 2.2773,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016778523489932888,
+      "loss": 2.246,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000174496644295302,
+      "loss": 2.1612,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018120805369127517,
+      "loss": 2.1943,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018791946308724833,
+      "loss": 2.1769,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019463087248322146,
+      "loss": 2.2003,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020134228187919463,
+      "loss": 2.213,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020805369127516779,
+      "loss": 2.1629,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021476510067114095,
+      "loss": 2.1677,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002214765100671141,
+      "loss": 2.2235,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022818791946308727,
+      "loss": 2.1486,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002348993288590604,
+      "loss": 2.1535,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024161073825503356,
+      "loss": 2.111,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002483221476510067,
+      "loss": 2.1341,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002550335570469799,
+      "loss": 2.1604,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000261744966442953,
+      "loss": 2.1548,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002684563758389262,
+      "loss": 2.1062,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027516778523489934,
+      "loss": 2.1495,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028187919463087247,
+      "loss": 2.1688,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028859060402684566,
+      "loss": 2.0777,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002953020134228188,
+      "loss": 2.0693,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000302013422818792,
+      "loss": 2.0645,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003087248322147651,
+      "loss": 2.0279,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031543624161073825,
+      "loss": 2.1481,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003221476510067114,
+      "loss": 2.0765,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003288590604026846,
+      "loss": 2.0709,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00033557046979865775,
+      "loss": 2.0834,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003422818791946309,
+      "loss": 2.0597,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000348993288590604,
+      "loss": 2.0304,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035570469798657715,
+      "loss": 2.042,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036241610738255034,
+      "loss": 2.0856,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036912751677852353,
+      "loss": 2.1084,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037583892617449666,
+      "loss": 2.0576,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003825503355704698,
+      "loss": 2.072,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038926174496644293,
+      "loss": 2.0896,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003959731543624161,
+      "loss": 2.0074,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040268456375838925,
+      "loss": 2.0437,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040939597315436244,
+      "loss": 2.0408,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041610738255033557,
+      "loss": 2.0581,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042281879194630876,
+      "loss": 2.0622,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004295302013422819,
+      "loss": 2.0788,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000436241610738255,
+      "loss": 2.0516,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004429530201342282,
+      "loss": 2.062,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044966442953020135,
+      "loss": 2.0968,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045637583892617453,
+      "loss": 2.0695,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046308724832214767,
+      "loss": 2.0023,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004697986577181208,
+      "loss": 2.0383,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047651006711409394,
+      "loss": 2.0434,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004832214765100671,
+      "loss": 2.0139,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004899328859060403,
+      "loss": 2.065,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004966442953020134,
+      "loss": 2.0446,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005033557046979866,
+      "loss": 2.0108,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005100671140939598,
+      "loss": 2.0048,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005167785234899329,
+      "loss": 2.0606,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000523489932885906,
+      "loss": 2.0048,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005302013422818792,
+      "loss": 2.0257,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005369127516778524,
+      "loss": 1.9665,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005436241610738255,
+      "loss": 2.0459,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005503355704697987,
+      "loss": 2.024,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005570469798657718,
+      "loss": 2.0347,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005637583892617449,
+      "loss": 2.0128,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005704697986577181,
+      "loss": 1.9931,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005771812080536913,
+      "loss": 2.0367,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005838926174496644,
+      "loss": 2.035,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005906040268456376,
+      "loss": 2.0506,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005973154362416108,
+      "loss": 2.0223,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000604026845637584,
+      "loss": 2.0345,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006107382550335571,
+      "loss": 2.0479,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006174496644295302,
+      "loss": 2.0201,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006241610738255034,
+      "loss": 2.0141,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006308724832214765,
+      "loss": 2.0105,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006375838926174496,
+      "loss": 2.0691,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006442953020134228,
+      "loss": 2.0247,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006510067114093959,
+      "loss": 2.0333,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006577181208053692,
+      "loss": 2.041,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006644295302013424,
+      "loss": 2.0171,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006711409395973155,
+      "loss": 2.009,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006778523489932886,
+      "loss": 1.9819,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006845637583892618,
+      "loss": 1.9484,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006912751677852349,
+      "loss": 2.0053,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000697986577181208,
+      "loss": 2.031,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007046979865771812,
+      "loss": 2.013,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007114093959731543,
+      "loss": 2.0107,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007181208053691274,
+      "loss": 2.0004,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007248322147651007,
+      "loss": 1.957,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007315436241610739,
+      "loss": 2.0122,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007382550335570471,
+      "loss": 2.0139,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007449664429530202,
+      "loss": 1.9701,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007516778523489933,
+      "loss": 1.9604,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007583892617449665,
+      "loss": 2.0031,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007651006711409396,
+      "loss": 2.0365,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007718120805369127,
+      "loss": 1.9908,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007785234899328859,
+      "loss": 1.9759,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007852348993288591,
+      "loss": 1.9766,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007919463087248322,
+      "loss": 1.9981,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007986577181208054,
+      "loss": 2.0232,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008053691275167785,
+      "loss": 1.9637,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008120805369127517,
+      "loss": 1.9471,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008187919463087249,
+      "loss": 2.0101,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000825503355704698,
+      "loss": 1.9532,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008322147651006711,
+      "loss": 1.958,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008389261744966443,
+      "loss": 2.0223,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008456375838926175,
+      "loss": 1.9738,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008523489932885907,
+      "loss": 1.9769,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008590604026845638,
+      "loss": 1.9681,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008657718120805369,
+      "loss": 1.9992,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00087248322147651,
+      "loss": 2.0236,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008791946308724832,
+      "loss": 1.9902,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008859060402684564,
+      "loss": 1.9624,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008926174496644296,
+      "loss": 1.9921,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008993288590604027,
+      "loss": 1.9378,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009060402684563759,
+      "loss": 1.9522,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009127516778523491,
+      "loss": 2.0026,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009194630872483222,
+      "loss": 2.0072,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009261744966442953,
+      "loss": 1.9815,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009328859060402685,
+      "loss": 1.9662,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009395973154362416,
+      "loss": 1.9571,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009463087248322147,
+      "loss": 1.9718,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009530201342281879,
+      "loss": 1.9573,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000959731543624161,
+      "loss": 1.9602,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009664429530201342,
+      "loss": 1.9107,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009731543624161075,
+      "loss": 1.9404,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009798657718120805,
+      "loss": 1.9445,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009865771812080538,
+      "loss": 1.9775,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009932885906040268,
+      "loss": 1.9759,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 1.9526,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999998927739515,
+      "loss": 1.9265,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999995710958521,
+      "loss": 1.9413,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999990349658396,
+      "loss": 1.907,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999998284384144,
+      "loss": 1.935,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999997319351087,
+      "loss": 1.9273,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999961398670834,
+      "loss": 1.9635,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999947459326378,
+      "loss": 1.8989,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999993137548349,
+      "loss": 1.9327,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999913147149067,
+      "loss": 1.9149,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999892774330923,
+      "loss": 1.9641,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00099998702570378,
+      "loss": 1.9029,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999845595279355,
+      "loss": 1.9133,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999818789066164,
+      "loss": 1.8718,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999789838409726,
+      "loss": 1.9595,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999758743322457,
+      "loss": 1.9446,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999725503817693,
+      "loss": 1.8969,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999690119909693,
+      "loss": 1.9231,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999965259161363,
+      "loss": 1.9635,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999612918945606,
+      "loss": 1.8788,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999957110192263,
+      "loss": 1.8874,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999527140562641,
+      "loss": 1.9123,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999481034884494,
+      "loss": 1.8927,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999432784907963,
+      "loss": 1.8692,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999382390653744,
+      "loss": 1.8893,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999329852143448,
+      "loss": 1.9054,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999275169399614,
+      "loss": 1.8664,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999218342445692,
+      "loss": 1.936,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999159371306056,
+      "loss": 1.8473,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999098256006,
+      "loss": 1.8801,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999034996571736,
+      "loss": 1.9026,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998969593030397,
+      "loss": 1.8645,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998902045410033,
+      "loss": 1.8665,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998832353739614,
+      "loss": 1.9017,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998760518049037,
+      "loss": 1.8621,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999868653836911,
+      "loss": 1.9002,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999861041473156,
+      "loss": 1.8738,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999853214716904,
+      "loss": 1.8815,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998451735715116,
+      "loss": 1.8518,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998369180404282,
+      "loss": 1.8291,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998284481271943,
+      "loss": 1.8793,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998197638354427,
+      "loss": 1.8673,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998108651688982,
+      "loss": 1.8596,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998017521313774,
+      "loss": 1.8752,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997924247267891,
+      "loss": 1.8468,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997828829591336,
+      "loss": 1.8513,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997731268325035,
+      "loss": 1.8084,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997631563510832,
+      "loss": 1.8522,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997529715191494,
+      "loss": 1.9125,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00099974257234107,
+      "loss": 1.8737,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997319588213054,
+      "loss": 1.8628,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997211309644078,
+      "loss": 1.8286,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997100887750214,
+      "loss": 1.8457,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996988322578822,
+      "loss": 1.8623,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999687361417818,
+      "loss": 1.8129,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996756762597487,
+      "loss": 1.8357,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996637767886866,
+      "loss": 1.8504,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996516630097348,
+      "loss": 1.8338,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996393349280893,
+      "loss": 1.8022,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996267925490375,
+      "loss": 1.8386,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999614035877959,
+      "loss": 1.8044,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996010649203251,
+      "loss": 1.7704,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995878796816991,
+      "loss": 1.8068,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995744801677363,
+      "loss": 1.833,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999560866384184,
+      "loss": 1.8079,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995470383368807,
+      "loss": 1.819,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995329960317575,
+      "loss": 1.801,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995187394748375,
+      "loss": 1.8745,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999504268672235,
+      "loss": 1.7976,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999489583630157,
+      "loss": 1.8153,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994746843549017,
+      "loss": 1.7774,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994595708528594,
+      "loss": 1.7987,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994442431305122,
+      "loss": 1.8244,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994287011944347,
+      "loss": 1.8026,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994129450512927,
+      "loss": 1.8118,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993969747078442,
+      "loss": 1.8418,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993807901709385,
+      "loss": 1.7652,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993643914475176,
+      "loss": 1.7513,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999347778544615,
+      "loss": 1.8353,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993309514693559,
+      "loss": 1.7968,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993139102289575,
+      "loss": 1.7863,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999296654830729,
+      "loss": 1.81,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992791852820708,
+      "loss": 1.7832,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992615015904764,
+      "loss": 1.7726,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992436037635302,
+      "loss": 1.7518,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992254918089083,
+      "loss": 1.7747,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992071657343793,
+      "loss": 1.7721,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991886255478033,
+      "loss": 1.7776,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991698712571323,
+      "loss": 1.7783,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991509028704098,
+      "loss": 1.7804,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991317203957718,
+      "loss": 1.7491,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991123238414455,
+      "loss": 1.7423,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990927132157504,
+      "loss": 1.8294,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999072888527097,
+      "loss": 1.7857,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990528497839891,
+      "loss": 1.7919,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990325969950207,
+      "loss": 1.7839,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990121301688786,
+      "loss": 1.772,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998991449314341,
+      "loss": 1.749,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998970554440278,
+      "loss": 1.7704,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989494455556517,
+      "loss": 1.7584,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989281226695154,
+      "loss": 1.8027,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989065857910148,
+      "loss": 1.7719,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988848349293874,
+      "loss": 1.7167,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988628700939616,
+      "loss": 1.7283,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998840691294159,
+      "loss": 1.7778,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988182985394916,
+      "loss": 1.7692,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998795691839564,
+      "loss": 1.7376,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998772871204072,
+      "loss": 1.7571,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998749836642804,
+      "loss": 1.7836,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987265881656392,
+      "loss": 1.7532,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998703125782549,
+      "loss": 1.7422,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986794495035968,
+      "loss": 1.7295,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986555593389373,
+      "loss": 1.7469,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998631455298817,
+      "loss": 1.7163,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998607137393574,
+      "loss": 1.7355,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998582605633639,
+      "loss": 1.7526,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998557860029533,
+      "loss": 1.7162,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985329005918703,
+      "loss": 1.7068,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985077273313553,
+      "loss": 1.7544,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984823402587857,
+      "loss": 1.7292,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984567393850497,
+      "loss": 1.7766,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984309247211276,
+      "loss": 1.7529,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984048962780913,
+      "loss": 1.7264,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998378654067105,
+      "loss": 1.7021,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983521980994237,
+      "loss": 1.7337,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983255283863945,
+      "loss": 1.7234,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982986449394561,
+      "loss": 1.8016,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982715477701394,
+      "loss": 1.6971,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982442368900656,
+      "loss": 1.6974,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982167123109495,
+      "loss": 1.6988,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981889740445957,
+      "loss": 1.6865,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981610221029016,
+      "loss": 1.686,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981328564978558,
+      "loss": 1.706,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998104477241539,
+      "loss": 1.7252,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980758843461223,
+      "loss": 1.7231,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980470778238703,
+      "loss": 1.7304,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980180576871379,
+      "loss": 1.7082,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979888239483716,
+      "loss": 1.7073,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979593766201102,
+      "loss": 1.6992,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997929715714984,
+      "loss": 1.7154,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978998412457142,
+      "loss": 1.6996,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978697532251144,
+      "loss": 1.6998,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978394516660894,
+      "loss": 1.684,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978089365816356,
+      "loss": 1.6753,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997778207984841,
+      "loss": 1.7114,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977472658888856,
+      "loss": 1.7357,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977161103070402,
+      "loss": 1.7019,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976847412526677,
+      "loss": 1.6701,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976531587392223,
+      "loss": 1.6703,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976213627802503,
+      "loss": 1.6886,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975893533893884,
+      "loss": 1.7077,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975571305803662,
+      "loss": 1.7072,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975246943670038,
+      "loss": 1.6897,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974920447632135,
+      "loss": 1.6816,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974591817829986,
+      "loss": 1.6706,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974261054404544,
+      "loss": 1.6649,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973928157497674,
+      "loss": 1.6646,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973593127252157,
+      "loss": 1.6807,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973255963811687,
+      "loss": 1.6763,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997291666732088,
+      "loss": 1.7083,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972575237925255,
+      "loss": 1.6756,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972231675771256,
+      "loss": 1.7091,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971885981006237,
+      "loss": 1.6445,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997153815377847,
+      "loss": 1.6516,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997118819423714,
+      "loss": 1.6648,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970836102532344,
+      "loss": 1.6639,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970481878815093,
+      "loss": 1.6534,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970125523237321,
+      "loss": 1.6773,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969767035951866,
+      "loss": 1.6888,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969406417112488,
+      "loss": 1.6818,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969043666873856,
+      "loss": 1.6691,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968678785391553,
+      "loss": 1.6745,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968311772822083,
+      "loss": 1.6725,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967942629322856,
+      "loss": 1.6692,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099675713550522,
+      "loss": 1.6329,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967197950169356,
+      "loss": 1.714,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966822414834479,
+      "loss": 1.627,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966444749208636,
+      "loss": 1.6394,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966064953453812,
+      "loss": 1.6461,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965683027732901,
+      "loss": 1.6282,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965298972209715,
+      "loss": 1.6257,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964912787048977,
+      "loss": 1.6376,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964524472416319,
+      "loss": 1.6908,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964134028478296,
+      "loss": 1.6317,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963741455402367,
+      "loss": 1.6345,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996334675335691,
+      "loss": 1.6571,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962949922511215,
+      "loss": 1.6057,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962550963035484,
+      "loss": 1.6405,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962149875100833,
+      "loss": 1.6664,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961746658879286,
+      "loss": 1.6316,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961341314543791,
+      "loss": 1.689,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960933842268198,
+      "loss": 1.6617,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960524242227274,
+      "loss": 1.6677,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960112514596697,
+      "loss": 1.6499,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959698659553061,
+      "loss": 1.6413,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959282677273868,
+      "loss": 1.6193,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958864567937537,
+      "loss": 1.6198,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958444331723394,
+      "loss": 1.669,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958021968811683,
+      "loss": 1.6407,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957597479383557,
+      "loss": 1.6771,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957170863621077,
+      "loss": 1.6439,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956742121707226,
+      "loss": 1.6256,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995631125382589,
+      "loss": 1.6186,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955878260161871,
+      "loss": 1.6559,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995544314090088,
+      "loss": 1.6252,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955005896229543,
+      "loss": 1.642,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954566526335395,
+      "loss": 1.6125,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954125031406886,
+      "loss": 1.6952,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953681411633375,
+      "loss": 1.6447,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995323566720513,
+      "loss": 1.6373,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952787798313332,
+      "loss": 1.6339,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952337805150078,
+      "loss": 1.6527,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951885687908368,
+      "loss": 1.6225,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951431446782117,
+      "loss": 1.6137,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950975081966156,
+      "loss": 1.6054,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950516593656214,
+      "loss": 1.6075,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950055982048945,
+      "loss": 1.6111,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949593247341904,
+      "loss": 1.6334,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949128389733562,
+      "loss": 1.6569,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948661409423297,
+      "loss": 1.6069,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948192306611398,
+      "loss": 1.6161,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947721081499067,
+      "loss": 1.5909,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009947247734288414,
+      "loss": 1.6224,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946772265182458,
+      "loss": 1.5997,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946294674385132,
+      "loss": 1.6496,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945814962101275,
+      "loss": 1.6125,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945333128536637,
+      "loss": 1.6568,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944849173897882,
+      "loss": 1.6211,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944363098392576,
+      "loss": 1.6334,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099438749022292,
+      "loss": 1.6538,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943384585617146,
+      "loss": 1.6011,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994289214876671,
+      "loss": 1.5974,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942397591889101,
+      "loss": 1.6235,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941900915196435,
+      "loss": 1.6655,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941402118901744,
+      "loss": 1.596,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940901203218957,
+      "loss": 1.618,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940398168362923,
+      "loss": 1.6549,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939893014549394,
+      "loss": 1.6573,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939385741995035,
+      "loss": 1.646,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938876350917415,
+      "loss": 1.6898,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938364841535013,
+      "loss": 1.632,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993785121406722,
+      "loss": 1.6056,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937335468734334,
+      "loss": 1.6083,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936817605757555,
+      "loss": 1.6061,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936297625359003,
+      "loss": 1.5867,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935775527761697,
+      "loss": 1.62,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935251313189565,
+      "loss": 1.6134,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934724981867447,
+      "loss": 1.6306,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934196534021086,
+      "loss": 1.6174,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933665969877141,
+      "loss": 1.6279,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993313328966317,
+      "loss": 1.5923,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993259849360764,
+      "loss": 1.6256,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932061581939928,
+      "loss": 1.5959,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993152255489032,
+      "loss": 1.6333,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930981412690006,
+      "loss": 1.6108,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930438155571085,
+      "loss": 1.6181,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929892783766558,
+      "loss": 1.6132,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992934529751034,
+      "loss": 1.6148,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928795697037255,
+      "loss": 1.6021,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992824398258302,
+      "loss": 1.6116,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927690154384271,
+      "loss": 1.621,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927134212678553,
+      "loss": 1.6273,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926576157704304,
+      "loss": 1.5907,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992601598970088,
+      "loss": 1.5878,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925453708908537,
+      "loss": 1.6383,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924889315568442,
+      "loss": 1.5887,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924322809922665,
+      "loss": 1.6573,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923754192214182,
+      "loss": 1.6493,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923183462686876,
+      "loss": 1.6291,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922610621585534,
+      "loss": 1.6488,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922035669155853,
+      "loss": 1.6103,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992145860564443,
+      "loss": 1.6298,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920879431298772,
+      "loss": 1.5927,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920298146367287,
+      "loss": 1.6031,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991971475109929,
+      "loss": 1.5964,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919129245745005,
+      "loss": 1.5631,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918541630555555,
+      "loss": 1.6171,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991795190578297,
+      "loss": 1.6381,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917360071680188,
+      "loss": 1.6313,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991676612850105,
+      "loss": 1.5553,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916170076500297,
+      "loss": 1.5707,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915571915933578,
+      "loss": 1.606,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914971647057452,
+      "loss": 1.5753,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914369270129372,
+      "loss": 1.5833,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913764785407701,
+      "loss": 1.6306,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913158193151706,
+      "loss": 1.5911,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912549493621555,
+      "loss": 1.6143,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911938687078323,
+      "loss": 1.6072,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991132577378399,
+      "loss": 1.5957,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910710754001432,
+      "loss": 1.6091,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910093627994439,
+      "loss": 1.6076,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909474396027694,
+      "loss": 1.6117,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990885305836679,
+      "loss": 1.6064,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908229615278224,
+      "loss": 1.5946,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990760406702939,
+      "loss": 1.5975,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990697641388859,
+      "loss": 1.6034,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906346656125024,
+      "loss": 1.6227,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905714794008803,
+      "loss": 1.5954,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905080827810931,
+      "loss": 1.594,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990444475780332,
+      "loss": 1.5796,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903806584258784,
+      "loss": 1.5983,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903166307451037,
+      "loss": 1.6189,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099025239276547,
+      "loss": 1.592,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901879445145284,
+      "loss": 1.6399,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901232860199218,
+      "loss": 1.6583,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900584173093824,
+      "loss": 1.6162,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899933384107324,
+      "loss": 1.5462,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899280493518847,
+      "loss": 1.6128,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898625501608418,
+      "loss": 1.6049,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897968408656966,
+      "loss": 1.6487,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897309214946322,
+      "loss": 1.6348,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896647920759217,
+      "loss": 1.5641,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989598452637928,
+      "loss": 1.5633,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895319032091048,
+      "loss": 1.5874,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894651438179954,
+      "loss": 1.568,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893981744932329,
+      "loss": 1.6366,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893309952635407,
+      "loss": 1.6235,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892636061577325,
+      "loss": 1.6132,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891960072047116,
+      "loss": 1.6143,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891281984334716,
+      "loss": 1.5894,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890601798730957,
+      "loss": 1.6046,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889919515527579,
+      "loss": 1.5593,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889235135017212,
+      "loss": 1.6162,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888548657493387,
+      "loss": 1.6182,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887860083250544,
+      "loss": 1.5807,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988716941258401,
+      "loss": 1.5741,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988647664579002,
+      "loss": 1.5593,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885781783165702,
+      "loss": 1.6052,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885084825009085,
+      "loss": 1.618,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098843857716191,
+      "loss": 1.6329,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883684623295573,
+      "loss": 1.5917,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988298138033923,
+      "loss": 1.6056,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988227604305169,
+      "loss": 1.6155,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881568611735483,
+      "loss": 1.6052,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880859086694022,
+      "loss": 1.5913,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880147468231632,
+      "loss": 1.5798,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879433756653524,
+      "loss": 1.5903,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878717952265812,
+      "loss": 1.5835,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878000055375512,
+      "loss": 1.6372,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877280066290528,
+      "loss": 1.6102,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987655798531967,
+      "loss": 1.5875,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875833812772639,
+      "loss": 1.5885,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875107548960036,
+      "loss": 1.6073,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987437919419336,
+      "loss": 1.5999,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873648748785005,
+      "loss": 1.5516,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872916213048261,
+      "loss": 1.6187,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872181587297318,
+      "loss": 1.5678,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871444871847256,
+      "loss": 1.5558,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987070606701406,
+      "loss": 1.601,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869965173114601,
+      "loss": 1.613,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869222190466659,
+      "loss": 1.5662,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868477119388895,
+      "loss": 1.5413,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867729960200877,
+      "loss": 1.6158,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866980713223066,
+      "loss": 1.5974,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866229378776813,
+      "loss": 1.6139,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865475957184372,
+      "loss": 1.6038,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864720448768886,
+      "loss": 1.6037,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863962853854397,
+      "loss": 1.5838,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863203172765843,
+      "loss": 1.6403,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986244140582905,
+      "loss": 1.5944,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861677553370744,
+      "loss": 1.6114,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860911615718548,
+      "loss": 1.6033,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860143593200972,
+      "loss": 1.5516,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859373486147426,
+      "loss": 1.6273,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858601294888212,
+      "loss": 1.5562,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857827019754525,
+      "loss": 1.582,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857050661078457,
+      "loss": 1.6031,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985627221919299,
+      "loss": 1.5598,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855491694432002,
+      "loss": 1.5933,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985470908713026,
+      "loss": 1.5677,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853924397623432,
+      "loss": 1.5778,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009853137626248073,
+      "loss": 1.5669,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985234877334163,
+      "loss": 1.585,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851557839242449,
+      "loss": 1.6115,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850764824289762,
+      "loss": 1.587,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098499697288237,
+      "loss": 1.6415,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984917255318528,
+      "loss": 1.5761,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848373297716413,
+      "loss": 1.5585,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847571962759906,
+      "loss": 1.5745,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846768548659455,
+      "loss": 1.6034,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845963055759646,
+      "loss": 1.5846,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845155484405957,
+      "loss": 1.5549,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844345834944762,
+      "loss": 1.5691,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984353410772332,
+      "loss": 1.5836,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842720303089784,
+      "loss": 1.5943,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841904421393204,
+      "loss": 1.6178,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841086462983508,
+      "loss": 1.5673,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840266428211525,
+      "loss": 1.6006,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839444317428971,
+      "loss": 1.577,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838620130988454,
+      "loss": 1.569,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837793869243467,
+      "loss": 1.546,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836965532548402,
+      "loss": 1.5808,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836135121258536,
+      "loss": 1.6005,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983530263573003,
+      "loss": 1.5904,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834468076319947,
+      "loss": 1.5779,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983363144338623,
+      "loss": 1.5887,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832792737287715,
+      "loss": 1.537,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831951958384126,
+      "loss": 1.5745,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831109107036075,
+      "loss": 1.589,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983026418360507,
+      "loss": 1.5868,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829417188453498,
+      "loss": 1.5573,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828568121944639,
+      "loss": 1.5669,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982771698444266,
+      "loss": 1.5984,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982686377631262,
+      "loss": 1.5937,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826008497920463,
+      "loss": 1.5783,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825151149633021,
+      "loss": 1.5782,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824291731818015,
+      "loss": 1.5631,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982343024484405,
+      "loss": 1.5475,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822566689080628,
+      "loss": 1.5481,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821701064898125,
+      "loss": 1.6072,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820833372667813,
+      "loss": 1.5396,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981996361276185,
+      "loss": 1.6054,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819091785553276,
+      "loss": 1.583,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818217891416027,
+      "loss": 1.5615,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817341930724915,
+      "loss": 1.612,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816463903855647,
+      "loss": 1.5659,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815583811184807,
+      "loss": 1.5818,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814701653089878,
+      "loss": 1.5405,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813817429949216,
+      "loss": 1.6069,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981293114214207,
+      "loss": 1.5678,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000981204279004857,
+      "loss": 1.5951,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811152374049736,
+      "loss": 1.5687,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810259894527473,
+      "loss": 1.5826,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809365351864565,
+      "loss": 1.5635,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980846874644469,
+      "loss": 1.5787,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00098075700786524,
+      "loss": 1.5725,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980666934887314,
+      "loss": 1.5828,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805766557493241,
+      "loss": 1.5687,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804861704899909,
+      "loss": 1.6073,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980395479148124,
+      "loss": 1.5666,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803045817626211,
+      "loss": 1.5997,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980213478372469,
+      "loss": 1.5545,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980122169016742,
+      "loss": 1.5506,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980030653734603,
+      "loss": 1.6492,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799389325653033,
+      "loss": 1.6018,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798470055481828,
+      "loss": 1.5486,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797548727226689,
+      "loss": 1.5742,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796625341282781,
+      "loss": 1.5973,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795699898046148,
+      "loss": 1.5831,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794772397913715,
+      "loss": 1.5577,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793842841283292,
+      "loss": 1.5796,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979291122855357,
+      "loss": 1.5329,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979197756012412,
+      "loss": 1.5612,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791041836395395,
+      "loss": 1.595,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790104057768737,
+      "loss": 1.575,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789164224646356,
+      "loss": 1.5734,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788222337431354,
+      "loss": 1.584,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978727839652771,
+      "loss": 1.5928,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786332402340284,
+      "loss": 1.6086,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785384355274819,
+      "loss": 1.5609,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978443425573793,
+      "loss": 1.5385,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783482104137127,
+      "loss": 1.5641,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782527900880785,
+      "loss": 1.5038,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978157164637817,
+      "loss": 1.5868,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780613341039421,
+      "loss": 1.5801,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779652985275563,
+      "loss": 1.5794,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977869057949849,
+      "loss": 1.5929,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977772612412099,
+      "loss": 1.5958,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776759619556715,
+      "loss": 1.6075,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775791066220206,
+      "loss": 1.5543,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009774820464526878,
+      "loss": 1.5368,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977384781489303,
+      "loss": 1.5872,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977287311773583,
+      "loss": 1.5711,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771896373473334,
+      "loss": 1.5424,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977091758252447,
+      "loss": 1.6008,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769936745309047,
+      "loss": 1.5966,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768953862247748,
+      "loss": 1.5478,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767968933762138,
+      "loss": 1.5576,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766981960274653,
+      "loss": 1.5403,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009765992942208615,
+      "loss": 1.6011,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765001879988213,
+      "loss": 1.6004,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764008774038521,
+      "loss": 1.5474,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763013624785489,
+      "loss": 1.603,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762016432655935,
+      "loss": 1.5821,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761017198077562,
+      "loss": 1.5936,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760015921478944,
+      "loss": 1.5823,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759012603289534,
+      "loss": 1.561,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975800724393966,
+      "loss": 1.5832,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756999843860525,
+      "loss": 1.5702,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755990403484205,
+      "loss": 1.57,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754978923243655,
+      "loss": 1.573,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753965403572702,
+      "loss": 1.5224,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975294984490605,
+      "loss": 1.5477,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751932247679276,
+      "loss": 1.5391,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750912612328831,
+      "loss": 1.5499,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749890939292041,
+      "loss": 1.5455,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748867229007108,
+      "loss": 1.5425,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747841481913103,
+      "loss": 1.5404,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746813698449973,
+      "loss": 1.615,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745783879058541,
+      "loss": 1.5459,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744752024180499,
+      "loss": 1.5646,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743718134258415,
+      "loss": 1.5297,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742682209735727,
+      "loss": 1.5568,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741644251056751,
+      "loss": 1.6104,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740604258666668,
+      "loss": 1.5378,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739562233011536,
+      "loss": 1.568,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738518174538284,
+      "loss": 1.5814,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737472083694716,
+      "loss": 1.5552,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736423960929501,
+      "loss": 1.5531,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735373806692186,
+      "loss": 1.5356,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734321621433184,
+      "loss": 1.5112,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733267405603784,
+      "loss": 1.5518,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732211159656142,
+      "loss": 1.5558,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731152884043287,
+      "loss": 1.5407,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730092579219118,
+      "loss": 1.5803,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729030245638404,
+      "loss": 1.5468,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727965883756784,
+      "loss": 1.4974,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009726899494030768,
+      "loss": 1.5813,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725831076917733,
+      "loss": 1.5422,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972476063287593,
+      "loss": 1.5434,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723688162364477,
+      "loss": 1.5686,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722613665843358,
+      "loss": 1.5738,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721537143773433,
+      "loss": 1.6303,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720458596616426,
+      "loss": 1.5926,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719378024834927,
+      "loss": 1.5368,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718295428892402,
+      "loss": 1.5663,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717210809253178,
+      "loss": 1.5809,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716124166382456,
+      "loss": 1.5666,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715035500746298,
+      "loss": 1.565,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713944812811639,
+      "loss": 1.5708,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000971285210304628,
+      "loss": 1.5799,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711757371918889,
+      "loss": 1.5707,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710660619899,
+      "loss": 1.5847,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709561847457013,
+      "loss": 1.5699,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708461055064202,
+      "loss": 1.5973,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707358243192694,
+      "loss": 1.5755,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706253412315493,
+      "loss": 1.5512,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705146562906467,
+      "loss": 1.538,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704037695440346,
+      "loss": 1.5708,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702926810392729,
+      "loss": 1.5521,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970181390824008,
+      "loss": 1.5384,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700698989459726,
+      "loss": 1.5762,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969958205452986,
+      "loss": 1.5415,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698463103929542,
+      "loss": 1.569,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697342138138694,
+      "loss": 1.5677,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696219157638101,
+      "loss": 1.5078,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695094162909416,
+      "loss": 1.5129,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693967154435154,
+      "loss": 1.5462,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692838132698692,
+      "loss": 1.5644,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691707098184274,
+      "loss": 1.5391,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690574051377006,
+      "loss": 1.5655,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689438992762854,
+      "loss": 1.5744,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688301922828649,
+      "loss": 1.6157,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687162842062089,
+      "loss": 1.5501,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686021750951728,
+      "loss": 1.5131,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684878649986983,
+      "loss": 1.5717,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009683733539658139,
+      "loss": 1.5767,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682586420456336,
+      "loss": 1.549,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681437292873578,
+      "loss": 1.5036,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680286157402732,
+      "loss": 1.5856,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679133014537524,
+      "loss": 1.5627,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677977864772543,
+      "loss": 1.5764,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676820708603235,
+      "loss": 1.5514,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675661546525911,
+      "loss": 1.5797,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967450037903774,
+      "loss": 1.5418,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673337206636752,
+      "loss": 1.552,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672172029821835,
+      "loss": 1.5953,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671004849092741,
+      "loss": 1.5346,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669835664950077,
+      "loss": 1.5265,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668664477895311,
+      "loss": 1.5892,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966749128843077,
+      "loss": 1.5149,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966631609705964,
+      "loss": 1.5866,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665138904285966,
+      "loss": 1.5777,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663959710614652,
+      "loss": 1.529,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662778516551455,
+      "loss": 1.5282,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661595322602997,
+      "loss": 1.6049,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660410129276756,
+      "loss": 1.5273,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009659222937081065,
+      "loss": 1.5461,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658033746525115,
+      "loss": 1.5801,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656842558118955,
+      "loss": 1.5669,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655649372373491,
+      "loss": 1.5911,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654454189800486,
+      "loss": 1.5728,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653257010912559,
+      "loss": 1.5707,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652057836223183,
+      "loss": 1.5389,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650856666246693,
+      "loss": 1.5316,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964965350149827,
+      "loss": 1.5945,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648448342493962,
+      "loss": 1.5242,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647241189750664,
+      "loss": 1.5708,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646032043786128,
+      "loss": 1.5259,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644820905118965,
+      "loss": 1.5882,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643607774268635,
+      "loss": 1.5437,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642392651755456,
+      "loss": 1.5783,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641175538100597,
+      "loss": 1.5253,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639956433826087,
+      "loss": 1.5335,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00096387353394548,
+      "loss": 1.57,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637512255510475,
+      "loss": 1.5279,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963628718251769,
+      "loss": 1.5502,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635060121001891,
+      "loss": 1.5993,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633831071489365,
+      "loss": 1.5485,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632600034507259,
+      "loss": 1.5504,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631367010583569,
+      "loss": 1.5078,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009630132000247144,
+      "loss": 1.5566,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628895004027687,
+      "loss": 1.5195,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627656022455748,
+      "loss": 1.5658,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626415056062732,
+      "loss": 1.5764,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625172105380898,
+      "loss": 1.5726,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623927170943347,
+      "loss": 1.5362,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622680253284042,
+      "loss": 1.5629,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621431352937788,
+      "loss": 1.5682,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009620180470440245,
+      "loss": 1.5077,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618927606327922,
+      "loss": 1.5416,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617672761138177,
+      "loss": 1.5608,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616415935409218,
+      "loss": 1.5619,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615157129680102,
+      "loss": 1.5367,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613896344490738,
+      "loss": 1.5464,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612633580381883,
+      "loss": 1.5494,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611368837895138,
+      "loss": 1.5442,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610102117572958,
+      "loss": 1.542,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608833419958647,
+      "loss": 1.5348,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607562745596352,
+      "loss": 1.5689,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009606290095031071,
+      "loss": 1.5129,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605015468808651,
+      "loss": 1.5699,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603738867475783,
+      "loss": 1.5389,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602460291580007,
+      "loss": 1.5776,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601179741669707,
+      "loss": 1.5828,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599897218294122,
+      "loss": 1.5114,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598612722003326,
+      "loss": 1.6012,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009597326253348248,
+      "loss": 1.5492,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009596037812880658,
+      "loss": 1.618,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594747401153174,
+      "loss": 1.5399,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593455018719262,
+      "loss": 1.561,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592160666133225,
+      "loss": 1.5536,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590864343950221,
+      "loss": 1.5404,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589566052726244,
+      "loss": 1.523,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588265793018141,
+      "loss": 1.5126,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586963565383594,
+      "loss": 1.5213,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585659370381136,
+      "loss": 1.5537,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584353208570145,
+      "loss": 1.5198,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583045080510833,
+      "loss": 1.5939,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009581734986764265,
+      "loss": 1.5222,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580422927892347,
+      "loss": 1.5258,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579108904457825,
+      "loss": 1.5692,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957779291702429,
+      "loss": 1.5466,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576474966156173,
+      "loss": 1.5396,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957515505241875,
+      "loss": 1.4974,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573833176378135,
+      "loss": 1.5288,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957250933860129,
+      "loss": 1.5423,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957118353965601,
+      "loss": 1.5318,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569855780110941,
+      "loss": 1.5417,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568526060535562,
+      "loss": 1.5157,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567194381500195,
+      "loss": 1.5141,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565860743576002,
+      "loss": 1.5244,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564525147334989,
+      "loss": 1.5689,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563187593349996,
+      "loss": 1.5691,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561848082194706,
+      "loss": 1.5323,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560506614443643,
+      "loss": 1.532,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559163190672165,
+      "loss": 1.5579,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557817811456472,
+      "loss": 1.5324,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556470477373606,
+      "loss": 1.5552,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555121189001443,
+      "loss": 1.5746,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553769946918697,
+      "loss": 1.5635,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552416751704923,
+      "loss": 1.5622,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551061603940512,
+      "loss": 1.5574,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954970450420669,
+      "loss": 1.5659,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548345453085528,
+      "loss": 1.4997,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546984451159924,
+      "loss": 1.5496,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545621499013619,
+      "loss": 1.5315,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544256597231189,
+      "loss": 1.5356,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542889746398045,
+      "loss": 1.519,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541520947100437,
+      "loss": 1.5459,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540150199925448,
+      "loss": 1.5235,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538777505460996,
+      "loss": 1.5602,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537402864295836,
+      "loss": 1.5278,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536026277019561,
+      "loss": 1.551,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009534647744222589,
+      "loss": 1.5249,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533267266496182,
+      "loss": 1.5253,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009531884844432432,
+      "loss": 1.559,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009530500478624267,
+      "loss": 1.5272,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529114169665443,
+      "loss": 1.5348,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527725918150558,
+      "loss": 1.5717,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526335724675039,
+      "loss": 1.6046,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524943589835144,
+      "loss": 1.5098,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523549514227965,
+      "loss": 1.5635,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522153498451427,
+      "loss": 1.5201,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520755543104289,
+      "loss": 1.5486,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519355648786138,
+      "loss": 1.551,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517953816097395,
+      "loss": 1.5375,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516550045639314,
+      "loss": 1.536,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515144338013975,
+      "loss": 1.5212,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513736693824293,
+      "loss": 1.5348,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512327113674012,
+      "loss": 1.5541,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510915598167709,
+      "loss": 1.5304,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009509502147910785,
+      "loss": 1.5625,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508086763509478,
+      "loss": 1.5453,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506669445570853,
+      "loss": 1.5342,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505250194702799,
+      "loss": 1.5314,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503829011514044,
+      "loss": 1.5004,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502405896614136,
+      "loss": 1.5197,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009500980850613457,
+      "loss": 1.5601,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499553874123212,
+      "loss": 1.5302,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498124967755442,
+      "loss": 1.5447,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496694132123007,
+      "loss": 1.5342,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00094952613678396,
+      "loss": 1.6032,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493826675519739,
+      "loss": 1.5281,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000949239005577877,
+      "loss": 1.5791,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490951509232866,
+      "loss": 1.5682,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009489511036499024,
+      "loss": 1.5534,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488068638195071,
+      "loss": 1.5709,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486624314939654,
+      "loss": 1.5293,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485178067352253,
+      "loss": 1.558,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483729896053168,
+      "loss": 1.5167,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482279801663525,
+      "loss": 1.5108,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480827784805278,
+      "loss": 1.4925,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009479373846101202,
+      "loss": 1.5475,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009477917986174895,
+      "loss": 1.5383,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476460205650785,
+      "loss": 1.5266,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475000505154118,
+      "loss": 1.5657,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473538885310967,
+      "loss": 1.5758,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009472075346748226,
+      "loss": 1.5779,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470609890093613,
+      "loss": 1.4989,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469142515975669,
+      "loss": 1.5669,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009467673225023756,
+      "loss": 1.5268,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466202017868058,
+      "loss": 1.5677,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464728895139585,
+      "loss": 1.5386,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463253857470164,
+      "loss": 1.5411,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461776905492445,
+      "loss": 1.5525,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460298039839899,
+      "loss": 1.5352,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009458817261146816,
+      "loss": 1.5756,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945733457004831,
+      "loss": 1.5498,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009455849967180313,
+      "loss": 1.5259,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454363453179577,
+      "loss": 1.5603,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452875028683677,
+      "loss": 1.5127,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451384694331,
+      "loss": 1.5346,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009449892450760758,
+      "loss": 1.5633,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448398298612981,
+      "loss": 1.5329,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009446902238528518,
+      "loss": 1.5565,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445404271149037,
+      "loss": 1.5164,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443904397117018,
+      "loss": 1.553,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442402617075765,
+      "loss": 1.4972,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094408989316694,
+      "loss": 1.5198,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439393341542857,
+      "loss": 1.5494,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000943788584734189,
+      "loss": 1.5552,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436376449713073,
+      "loss": 1.5868,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009434865149303788,
+      "loss": 1.5355,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433351946762243,
+      "loss": 1.5336,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009431836842737455,
+      "loss": 1.5511,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430319837879258,
+      "loss": 1.5668,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428800932838301,
+      "loss": 1.5692,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427280128266049,
+      "loss": 1.5522,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425757424814784,
+      "loss": 1.592,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424232823137598,
+      "loss": 1.5375,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009422706323888397,
+      "loss": 1.5428,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421177927721907,
+      "loss": 1.5227,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941964763529366,
+      "loss": 1.5331,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418115447260007,
+      "loss": 1.5362,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416581364278109,
+      "loss": 1.5178,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009415045387005942,
+      "loss": 1.5331,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413507516102291,
+      "loss": 1.5055,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411967752226757,
+      "loss": 1.5551,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941042609603975,
+      "loss": 1.531,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408882548202493,
+      "loss": 1.5015,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940733710937702,
+      "loss": 1.4979,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009405789780226177,
+      "loss": 1.5276,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940424056141362,
+      "loss": 1.5214,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009402689453603814,
+      "loss": 1.577,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401136457462037,
+      "loss": 1.5292,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399581573654376,
+      "loss": 1.5198,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398024802847726,
+      "loss": 1.5142,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396466145709792,
+      "loss": 1.5395,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000939490560290909,
+      "loss": 1.5181,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393343175114943,
+      "loss": 1.5515,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391778862997482,
+      "loss": 1.5179,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390212667227649,
+      "loss": 1.5338,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000938864458847719,
+      "loss": 1.5237,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387074627418661,
+      "loss": 1.5486,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385502784725425,
+      "loss": 1.5085,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009383929061071652,
+      "loss": 1.5723,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382353457132317,
+      "loss": 1.5354,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380775973583208,
+      "loss": 1.5494,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937919661110091,
+      "loss": 1.5152,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377615370362818,
+      "loss": 1.5371,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009376032252047136,
+      "loss": 1.5256,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374447256832867,
+      "loss": 1.5462,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009372860385399824,
+      "loss": 1.5325,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937127163842862,
+      "loss": 1.5975,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000936968101660068,
+      "loss": 1.5271,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368088520598223,
+      "loss": 1.5213,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366494151104284,
+      "loss": 1.4945,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364897908802687,
+      "loss": 1.5293,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363299794378072,
+      "loss": 1.541,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361699808515876,
+      "loss": 1.5564,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360097951902338,
+      "loss": 1.5465,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358494225224503,
+      "loss": 1.5428,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356888629170215,
+      "loss": 1.5531,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355281164428122,
+      "loss": 1.5314,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353671831687671,
+      "loss": 1.4989,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352060631639113,
+      "loss": 1.522,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350447564973498,
+      "loss": 1.5404,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009348832632382675,
+      "loss": 1.5624,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00093472158345593,
+      "loss": 1.5183,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009345597172196818,
+      "loss": 1.5245,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343976645989486,
+      "loss": 1.5531,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342354256632352,
+      "loss": 1.5665,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340730004821265,
+      "loss": 1.5169,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339103891252874,
+      "loss": 1.5621,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337475916624626,
+      "loss": 1.5286,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335846081634766,
+      "loss": 1.5452,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334214386982337,
+      "loss": 1.5513,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933258083336718,
+      "loss": 1.5649,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330945421489933,
+      "loss": 1.556,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009329308152052031,
+      "loss": 1.5373,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327669025755706,
+      "loss": 1.5732,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326028043303984,
+      "loss": 1.562,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324385205400694,
+      "loss": 1.5338,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322740512750452,
+      "loss": 1.5243,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009321093966058674,
+      "loss": 1.4532,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931944556603157,
+      "loss": 1.5375,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931779531337615,
+      "loss": 1.4972,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931614320880021,
+      "loss": 1.5477,
+      "step": 957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009314489253012346,
+      "loss": 1.5308,
+      "step": 958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312833446721946,
+      "loss": 1.5061,
+      "step": 959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311175790639194,
+      "loss": 1.5149,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309516285475063,
+      "loss": 1.526,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009307854931941324,
+      "loss": 1.5362,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306191730750538,
+      "loss": 1.5401,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304526682616059,
+      "loss": 1.5019,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009302859788252032,
+      "loss": 1.4978,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301191048373396,
+      "loss": 1.5329,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299520463695881,
+      "loss": 1.5577,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297848034936007,
+      "loss": 1.5639,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296173762811085,
+      "loss": 1.549,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294497648039217,
+      "loss": 1.5036,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009292819691339298,
+      "loss": 1.4939,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009291139893431011,
+      "loss": 1.5092,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289458255034824,
+      "loss": 1.562,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287774776872003,
+      "loss": 1.5209,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009286089459664597,
+      "loss": 1.5563,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284402304135447,
+      "loss": 1.5455,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282713311008179,
+      "loss": 1.5584,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928102248100721,
+      "loss": 1.4955,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279329814857746,
+      "loss": 1.5771,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277635313285777,
+      "loss": 1.5736,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275938977018082,
+      "loss": 1.4925,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009274240806782226,
+      "loss": 1.5316,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009272540803306562,
+      "loss": 1.5348,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270838967320227,
+      "loss": 1.5583,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269135299553149,
+      "loss": 1.5195,
+      "step": 985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009267429800736036,
+      "loss": 1.5298,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009265722471600384,
+      "loss": 1.5507,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009264013312878473,
+      "loss": 1.5,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009262302325303369,
+      "loss": 1.5177,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009260589509608923,
+      "loss": 1.5127,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258874866529767,
+      "loss": 1.5096,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009257158396801319,
+      "loss": 1.4997,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925544010115978,
+      "loss": 1.5351,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253719980342135,
+      "loss": 1.5372,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925199803508615,
+      "loss": 1.5054,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009250274266130375,
+      "loss": 1.5297,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009248548674214141,
+      "loss": 1.5119,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246821260077564,
+      "loss": 1.5224,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009245092024461537,
+      "loss": 1.5088,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243360968107737,
+      "loss": 1.5051,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000924162809175862,
+      "loss": 1.5495,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239893396157426,
+      "loss": 1.5672,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009238156882048173,
+      "loss": 1.508,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236418550175659,
+      "loss": 1.5187,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009234678401285461,
+      "loss": 1.5318,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232936436123936,
+      "loss": 1.5384,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009231192655438221,
+      "loss": 1.5119,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009229447059976231,
+      "loss": 1.5641,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227699650486657,
+      "loss": 1.5305,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225950427718975,
+      "loss": 1.5143,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224199392423429,
+      "loss": 1.5335,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222446545351047,
+      "loss": 1.5581,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220691887253634,
+      "loss": 1.5739,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000921893541888377,
+      "loss": 1.4851,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217177140994809,
+      "loss": 1.5038,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215417054340886,
+      "loss": 1.5111,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921365515967691,
+      "loss": 1.4995,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211891457758563,
+      "loss": 1.5229,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009210125949342306,
+      "loss": 1.5188,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009208358635185372,
+      "loss": 1.5311,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206589516045769,
+      "loss": 1.5514,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009204818592682281,
+      "loss": 1.5376,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203045865854465,
+      "loss": 1.5229,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201271336322647,
+      "loss": 1.536,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009199495004847936,
+      "loss": 1.536,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009197716872192201,
+      "loss": 1.5352,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009195936939118096,
+      "loss": 1.5409,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009194155206389042,
+      "loss": 1.5355,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009192371674769229,
+      "loss": 1.5429,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919058634502362,
+      "loss": 1.5352,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188799217917955,
+      "loss": 1.5217,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009187010294218736,
+      "loss": 1.5531,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185219574693243,
+      "loss": 1.5191,
+      "step": 1033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183427060109521,
+      "loss": 1.5396,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181632751236389,
+      "loss": 1.4891,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179836648843429,
+      "loss": 1.4974,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178038753701003,
+      "loss": 1.5471,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009176239066580232,
+      "loss": 1.546,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009174437588253011,
+      "loss": 1.5328,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009172634319492001,
+      "loss": 1.5827,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009170829261070632,
+      "loss": 1.5425,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169022413763099,
+      "loss": 1.5044,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916721377834437,
+      "loss": 1.5474,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009165403355590172,
+      "loss": 1.4664,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009163591146277006,
+      "loss": 1.5351,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009161777151182136,
+      "loss": 1.5222,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159961371083591,
+      "loss": 1.5411,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009158143806760169,
+      "loss": 1.5167,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156324458991427,
+      "loss": 1.5576,
+      "step": 1049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009154503328557693,
+      "loss": 1.5419,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009152680416240059,
+      "loss": 1.5402,
+      "step": 1051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150855722820377,
+      "loss": 1.5495,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009149029249081266,
+      "loss": 1.5513,
+      "step": 1053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009147200995806112,
+      "loss": 1.5552,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009145370963779056,
+      "loss": 1.5163,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143539153785009,
+      "loss": 1.5324,
+      "step": 1056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914170556660964,
+      "loss": 1.5646,
+      "step": 1057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139870203039384,
+      "loss": 1.5462,
+      "step": 1058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009138033063861435,
+      "loss": 1.5129,
+      "step": 1059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000913619414986375,
+      "loss": 1.5088,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009134353461835048,
+      "loss": 1.522,
+      "step": 1061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009132511000564806,
+      "loss": 1.526,
+      "step": 1062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009130666766843265,
+      "loss": 1.5755,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009128820761461423,
+      "loss": 1.5195,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009126972985211042,
+      "loss": 1.4926,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125123438884637,
+      "loss": 1.5273,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009123272123275489,
+      "loss": 1.5342,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121419039177634,
+      "loss": 1.4899,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009119564187385869,
+      "loss": 1.5394,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009117707568695748,
+      "loss": 1.5359,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009115849183903579,
+      "loss": 1.5304,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113989033806433,
+      "loss": 1.518,
+      "step": 1072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112127119202139,
+      "loss": 1.4891,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110263440889276,
+      "loss": 1.5434,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009108397999667184,
+      "loss": 1.5117,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009106530796335961,
+      "loss": 1.5387,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104661831696455,
+      "loss": 1.4914,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009102791106550276,
+      "loss": 1.5653,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100918621699786,
+      "loss": 1.4948,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009099044377948097,
+      "loss": 1.5192,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009097168376099085,
+      "loss": 1.5156,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095290616957372,
+      "loss": 1.4912,
+      "step": 1082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009093411101328338,
+      "loss": 1.5318,
+      "step": 1083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009091529830018114,
+      "loss": 1.5553,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009089646803833589,
+      "loss": 1.5058,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087762023582396,
+      "loss": 1.5483,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009085875490072928,
+      "loss": 1.5444,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009083987204114326,
+      "loss": 1.5525,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082097166516483,
+      "loss": 1.5201,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009080205378090046,
+      "loss": 1.4924,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907831183964641,
+      "loss": 1.5182,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009076416551997722,
+      "loss": 1.5628,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074519515956879,
+      "loss": 1.473,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009072620732337527,
+      "loss": 1.5232,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009070720201954061,
+      "loss": 1.4854,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068817925621628,
+      "loss": 1.5378,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009066913904156124,
+      "loss": 1.4954,
+      "step": 1097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065008138374189,
+      "loss": 1.4579,
+      "step": 1098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009063100629093215,
+      "loss": 1.499,
+      "step": 1099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000906119137713134,
+      "loss": 1.4974,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009059280383307452,
+      "loss": 1.5348,
+      "step": 1101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057367648441182,
+      "loss": 1.4805,
+      "step": 1102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055453173352912,
+      "loss": 1.4976,
+      "step": 1103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009053536958863769,
+      "loss": 1.5339,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009051619005795621,
+      "loss": 1.5049,
+      "step": 1105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009049699314971091,
+      "loss": 1.5516,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000904777788721354,
+      "loss": 1.4986,
+      "step": 1107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009045854723347075,
+      "loss": 1.4972,
+      "step": 1108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009043929824196552,
+      "loss": 1.527,
+      "step": 1109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000904200319058757,
+      "loss": 1.4693,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040074823346465,
+      "loss": 1.4911,
+      "step": 1111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009038144723300326,
+      "loss": 1.5184,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009036212891276978,
+      "loss": 1.5229,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034279328104992,
+      "loss": 1.5006,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009032344034613684,
+      "loss": 1.528,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009030407011633108,
+      "loss": 1.5385,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009028468259994063,
+      "loss": 1.4954,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009026527780528085,
+      "loss": 1.5152,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009024585574067452,
+      "loss": 1.5318,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009022641641445191,
+      "loss": 1.5366,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009020695983495057,
+      "loss": 1.5764,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018748601051554,
+      "loss": 1.5669,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009016799494949921,
+      "loss": 1.5256,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009014848666026138,
+      "loss": 1.539,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012896115116924,
+      "loss": 1.4922,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009010941843059737,
+      "loss": 1.5467,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008985850692772,
+      "loss": 1.536,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009007028138854961,
+      "loss": 1.5013,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009005068708385978,
+      "loss": 1.5335,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003107560126226,
+      "loss": 1.5042,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001144694916854,
+      "loss": 1.5326,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899918011359974,
+      "loss": 1.4926,
+      "step": 1132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008997213817017506,
+      "loss": 1.5025,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00089952458060135,
+      "loss": 1.5231,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993276081431811,
+      "loss": 1.4875,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008991304644117266,
+      "loss": 1.5273,
+      "step": 1136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008989331494915417,
+      "loss": 1.5184,
+      "step": 1137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987356634672559,
+      "loss": 1.546,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008985380064235719,
+      "loss": 1.5249,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008983401784452654,
+      "loss": 1.495,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008981421796171857,
+      "loss": 1.515,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008979440100242554,
+      "loss": 1.5199,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008977456697514702,
+      "loss": 1.5179,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008975471588838992,
+      "loss": 1.4999,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008973484775066843,
+      "loss": 1.4845,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971496257050411,
+      "loss": 1.5212,
+      "step": 1146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008969506035642577,
+      "loss": 1.5285,
+      "step": 1147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008967514111696957,
+      "loss": 1.5294,
+      "step": 1148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008965520486067895,
+      "loss": 1.4975,
+      "step": 1149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008963525159610463,
+      "loss": 1.5226,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961528133180471,
+      "loss": 1.5158,
+      "step": 1151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959529407634446,
+      "loss": 1.5457,
+      "step": 1152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008957528983829652,
+      "loss": 1.5063,
+      "step": 1153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955526862624079,
+      "loss": 1.5455,
+      "step": 1154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008953523044876446,
+      "loss": 1.5223,
+      "step": 1155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008951517531446198,
+      "loss": 1.5219,
+      "step": 1156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949510323193506,
+      "loss": 1.5435,
+      "step": 1157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947501420979275,
+      "loss": 1.4784,
+      "step": 1158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008945490825665128,
+      "loss": 1.5444,
+      "step": 1159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008943478538113419,
+      "loss": 1.5349,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008941464559187224,
+      "loss": 1.5463,
+      "step": 1161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939448889750352,
+      "loss": 1.5358,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008937431530667328,
+      "loss": 1.4947,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008935412482803408,
+      "loss": 1.5275,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933391747024569,
+      "loss": 1.5647,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931369324197511,
+      "loss": 1.4956,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008929345215189663,
+      "loss": 1.5262,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927319420869174,
+      "loss": 1.5164,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008925291942104914,
+      "loss": 1.504,
+      "step": 1169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923262779766477,
+      "loss": 1.5093,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008921231934724179,
+      "loss": 1.4834,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891919940784906,
+      "loss": 1.533,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008917165200012877,
+      "loss": 1.4903,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008915129312088111,
+      "loss": 1.5505,
+      "step": 1174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913091744947964,
+      "loss": 1.5545,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008911052499466357,
+      "loss": 1.5355,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008909011576517928,
+      "loss": 1.5369,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008906968976978041,
+      "loss": 1.5198,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008904924701722774,
+      "loss": 1.5613,
+      "step": 1179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008902878751628925,
+      "loss": 1.5459,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000890083112757401,
+      "loss": 1.5305,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008898781830436267,
+      "loss": 1.5018,
+      "step": 1182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008896730861094642,
+      "loss": 1.4891,
+      "step": 1183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889467822042881,
+      "loss": 1.4861,
+      "step": 1184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892623909319156,
+      "loss": 1.5247,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889056792864678,
+      "loss": 1.5209,
+      "step": 1186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008888510279293502,
+      "loss": 1.5175,
+      "step": 1187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008886450962141859,
+      "loss": 1.5281,
+      "step": 1188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008884389978075098,
+      "loss": 1.5492,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882327327977184,
+      "loss": 1.5534,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880263012732796,
+      "loss": 1.5692,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887819703322733,
+      "loss": 1.5493,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008876129390346891,
+      "loss": 1.5163,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874060084978299,
+      "loss": 1.5508,
+      "step": 1194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008871989118009089,
+      "loss": 1.5291,
+      "step": 1195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008869916490327509,
+      "loss": 1.5448,
+      "step": 1196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867842202822515,
+      "loss": 1.5496,
+      "step": 1197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008865766256383778,
+      "loss": 1.5361,
+      "step": 1198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000886368865190168,
+      "loss": 1.5325,
+      "step": 1199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861609390267318,
+      "loss": 1.5361,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000885952847237249,
+      "loss": 1.5167,
+      "step": 1201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857445899109715,
+      "loss": 1.4999,
+      "step": 1202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008855361671372215,
+      "loss": 1.5507,
+      "step": 1203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853275790053926,
+      "loss": 1.5073,
+      "step": 1204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851188256049489,
+      "loss": 1.466,
+      "step": 1205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849099070254258,
+      "loss": 1.5174,
+      "step": 1206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008847008233564294,
+      "loss": 1.5649,
+      "step": 1207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844915746876362,
+      "loss": 1.5204,
+      "step": 1208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008842821611087941,
+      "loss": 1.5305,
+      "step": 1209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840725827097216,
+      "loss": 1.5229,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008838628395803074,
+      "loss": 1.5148,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008836529318105115,
+      "loss": 1.504,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834428594903641,
+      "loss": 1.5108,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883232622709966,
+      "loss": 1.5131,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883022221559489,
+      "loss": 1.5021,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008828116561291747,
+      "loss": 1.5506,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008826009265093355,
+      "loss": 1.4866,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823900327903545,
+      "loss": 1.5356,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821789750626845,
+      "loss": 1.5297,
+      "step": 1219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819677534168493,
+      "loss": 1.5134,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817563679434427,
+      "loss": 1.4883,
+      "step": 1221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008815448187331289,
+      "loss": 1.5354,
+      "step": 1222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008813331058766421,
+      "loss": 1.537,
+      "step": 1223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008811212294647868,
+      "loss": 1.5176,
+      "step": 1224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809091895884379,
+      "loss": 1.482,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008806969863385401,
+      "loss": 1.5227,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008804846198061081,
+      "loss": 1.531,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802720900822269,
+      "loss": 1.5003,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800593972580515,
+      "loss": 1.5113,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008798465414248067,
+      "loss": 1.5499,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008796335226737872,
+      "loss": 1.5379,
+      "step": 1231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008794203410963577,
+      "loss": 1.5124,
+      "step": 1232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008792069967839525,
+      "loss": 1.4912,
+      "step": 1233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000878993489828076,
+      "loss": 1.5345,
+      "step": 1234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008787798203203024,
+      "loss": 1.5343,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008785659883522751,
+      "loss": 1.543,
+      "step": 1236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008783519940157076,
+      "loss": 1.5349,
+      "step": 1237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008781378374023834,
+      "loss": 1.4955,
+      "step": 1238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008779235186041545,
+      "loss": 1.5416,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008777090377129437,
+      "loss": 1.5329,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008774943948207425,
+      "loss": 1.4976,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008772795900196122,
+      "loss": 1.5396,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008770646234016833,
+      "loss": 1.5041,
+      "step": 1243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008768494950591561,
+      "loss": 1.5204,
+      "step": 1244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766342050843,
+      "loss": 1.5308,
+      "step": 1245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008764187535694537,
+      "loss": 1.4922,
+      "step": 1246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762031406070255,
+      "loss": 1.5072,
+      "step": 1247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008759873662894922,
+      "loss": 1.4866,
+      "step": 1248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008757714307094008,
+      "loss": 1.5309,
+      "step": 1249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008755553339593668,
+      "loss": 1.4993,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875339076132075,
+      "loss": 1.5334,
+      "step": 1251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875122657320279,
+      "loss": 1.5267,
+      "step": 1252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008749060776168023,
+      "loss": 1.536,
+      "step": 1253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746893371145366,
+      "loss": 1.5173,
+      "step": 1254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744724359064427,
+      "loss": 1.5258,
+      "step": 1255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008742553740855505,
+      "loss": 1.5115,
+      "step": 1256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008740381517449588,
+      "loss": 1.4964,
+      "step": 1257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008738207689778352,
+      "loss": 1.5229,
+      "step": 1258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008736032258774158,
+      "loss": 1.5129,
+      "step": 1259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873385522537006,
+      "loss": 1.507,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008731676590499796,
+      "loss": 1.496,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729496355097793,
+      "loss": 1.5157,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008727314520099161,
+      "loss": 1.4978,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008725131086439699,
+      "loss": 1.4692,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008722946055055892,
+      "loss": 1.4887,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720759426884906,
+      "loss": 1.5221,
+      "step": 1266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008718571202864598,
+      "loss": 1.5223,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008716381383933507,
+      "loss": 1.4995,
+      "step": 1268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008714189971030852,
+      "loss": 1.4754,
+      "step": 1269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711996965096542,
+      "loss": 1.5254,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008709802367071166,
+      "loss": 1.5065,
+      "step": 1271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707606177895996,
+      "loss": 1.5385,
+      "step": 1272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008705408398512986,
+      "loss": 1.4868,
+      "step": 1273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008703209029864774,
+      "loss": 1.5124,
+      "step": 1274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008701008072894679,
+      "loss": 1.5089,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008698805528546699,
+      "loss": 1.5711,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008696601397765513,
+      "loss": 1.5459,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008694395681496486,
+      "loss": 1.5022,
+      "step": 1278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692188380685658,
+      "loss": 1.503,
+      "step": 1279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689979496279746,
+      "loss": 1.4963,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008687769029226155,
+      "loss": 1.5313,
+      "step": 1281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000868555698047296,
+      "loss": 1.5202,
+      "step": 1282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008683343350968918,
+      "loss": 1.5024,
+      "step": 1283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008681128141663465,
+      "loss": 1.5352,
+      "step": 1284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008678911353506715,
+      "loss": 1.54,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008676692987449455,
+      "loss": 1.5283,
+      "step": 1286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674473044443154,
+      "loss": 1.5335,
+      "step": 1287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008672251525439952,
+      "loss": 1.4637,
+      "step": 1288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008670028431392671,
+      "loss": 1.4664,
+      "step": 1289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008667803763254804,
+      "loss": 1.5025,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665577521980519,
+      "loss": 1.5001,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008663349708524662,
+      "loss": 1.5387,
+      "step": 1292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661120323842751,
+      "loss": 1.5087,
+      "step": 1293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008658889368890979,
+      "loss": 1.5225,
+      "step": 1294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865665684462621,
+      "loss": 1.5144,
+      "step": 1295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008654422752005985,
+      "loss": 1.5001,
+      "step": 1296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008652187091988516,
+      "loss": 1.5855,
+      "step": 1297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649949865532686,
+      "loss": 1.519,
+      "step": 1298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864771107359805,
+      "loss": 1.498,
+      "step": 1299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008645470717144837,
+      "loss": 1.4779,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008643228797133944,
+      "loss": 1.4834,
+      "step": 1301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864098531452694,
+      "loss": 1.5587,
+      "step": 1302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008638740270286065,
+      "loss": 1.4837,
+      "step": 1303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008636493665374228,
+      "loss": 1.5317,
+      "step": 1304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008634245500755005,
+      "loss": 1.5093,
+      "step": 1305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631995777392645,
+      "loss": 1.489,
+      "step": 1306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008629744496252064,
+      "loss": 1.5174,
+      "step": 1307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008627491658298848,
+      "loss": 1.5172,
+      "step": 1308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008625237264499241,
+      "loss": 1.4986,
+      "step": 1309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862298131582017,
+      "loss": 1.4696,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000862072381322922,
+      "loss": 1.5479,
+      "step": 1311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000861846475769464,
+      "loss": 1.4935,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008616204150185348,
+      "loss": 1.4968,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000861394199167093,
+      "loss": 1.506,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008611678283121636,
+      "loss": 1.4529,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860941302550838,
+      "loss": 1.5121,
+      "step": 1316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008607146219802738,
+      "loss": 1.5005,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008604877866976954,
+      "loss": 1.5514,
+      "step": 1318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602607968003935,
+      "loss": 1.5014,
+      "step": 1319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008600336523857249,
+      "loss": 1.5086,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008598063535511129,
+      "loss": 1.5111,
+      "step": 1321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008595789003940467,
+      "loss": 1.4832,
+      "step": 1322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593512930120821,
+      "loss": 1.5295,
+      "step": 1323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008591235315028409,
+      "loss": 1.4947,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588956159640108,
+      "loss": 1.5361,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008586675464933459,
+      "loss": 1.5155,
+      "step": 1326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000858439323188666,
+      "loss": 1.5497,
+      "step": 1327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008582109461478572,
+      "loss": 1.5046,
+      "step": 1328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579824154688711,
+      "loss": 1.5156,
+      "step": 1329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008577537312497258,
+      "loss": 1.4995,
+      "step": 1330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008575248935885048,
+      "loss": 1.4933,
+      "step": 1331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008572959025833573,
+      "loss": 1.4649,
+      "step": 1332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008570667583324988,
+      "loss": 1.468,
+      "step": 1333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008568374609342101,
+      "loss": 1.5311,
+      "step": 1334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008566080104868379,
+      "loss": 1.5195,
+      "step": 1335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008563784070887943,
+      "loss": 1.5136,
+      "step": 1336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008561486508385574,
+      "loss": 1.5411,
+      "step": 1337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008559187418346703,
+      "loss": 1.5358,
+      "step": 1338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008556886801757422,
+      "loss": 1.5096,
+      "step": 1339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008554584659604474,
+      "loss": 1.5319,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008552280992875259,
+      "loss": 1.4654,
+      "step": 1341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549975802557828,
+      "loss": 1.5293,
+      "step": 1342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008547669089640885,
+      "loss": 1.4819,
+      "step": 1343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854536085511379,
+      "loss": 1.5234,
+      "step": 1344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008543051099966557,
+      "loss": 1.5607,
+      "step": 1345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008540739825189848,
+      "loss": 1.5287,
+      "step": 1346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008538427031774978,
+      "loss": 1.5465,
+      "step": 1347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008536112720713913,
+      "loss": 1.4925,
+      "step": 1348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008533796892999273,
+      "loss": 1.5225,
+      "step": 1349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008531479549624324,
+      "loss": 1.51,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529160691582987,
+      "loss": 1.501,
+      "step": 1351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008526840319869827,
+      "loss": 1.5051,
+      "step": 1352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008524518435480061,
+      "loss": 1.5067,
+      "step": 1353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008522195039409557,
+      "loss": 1.4834,
+      "step": 1354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000851987013265483,
+      "loss": 1.5387,
+      "step": 1355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000851754371621304,
+      "loss": 1.5265,
+      "step": 1356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008515215791081998,
+      "loss": 1.5132,
+      "step": 1357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008512886358260161,
+      "loss": 1.5221,
+      "step": 1358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510555418746631,
+      "loss": 1.5309,
+      "step": 1359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008508222973541157,
+      "loss": 1.4964,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000850588902364414,
+      "loss": 1.4933,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008503553570056615,
+      "loss": 1.4931,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000850121661378027,
+      "loss": 1.5339,
+      "step": 1363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008498878155817437,
+      "loss": 1.5083,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008496538197171087,
+      "loss": 1.4963,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849419673884484,
+      "loss": 1.4908,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008491853781842958,
+      "loss": 1.4619,
+      "step": 1367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008489509327170344,
+      "loss": 1.487,
+      "step": 1368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008487163375832545,
+      "loss": 1.5207,
+      "step": 1369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484815928835748,
+      "loss": 1.5172,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008482466987186785,
+      "loss": 1.537,
+      "step": 1371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008480116551893125,
+      "loss": 1.5004,
+      "step": 1372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008477764623962881,
+      "loss": 1.4711,
+      "step": 1373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008475411204404803,
+      "loss": 1.5092,
+      "step": 1374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008473056294228285,
+      "loss": 1.509,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008470699894443357,
+      "loss": 1.5351,
+      "step": 1376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008468342006060687,
+      "loss": 1.4922,
+      "step": 1377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008465982630091586,
+      "loss": 1.4967,
+      "step": 1378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008463621767547997,
+      "loss": 1.4791,
+      "step": 1379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008461259419442507,
+      "loss": 1.5126,
+      "step": 1380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008458895586788334,
+      "loss": 1.4815,
+      "step": 1381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008456530270599339,
+      "loss": 1.5334,
+      "step": 1382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008454163471890013,
+      "loss": 1.5005,
+      "step": 1383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008451795191675487,
+      "loss": 1.5202,
+      "step": 1384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008449425430971529,
+      "loss": 1.5378,
+      "step": 1385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008447054190794534,
+      "loss": 1.4925,
+      "step": 1386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008444681472161542,
+      "loss": 1.5343,
+      "step": 1387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008442307276090219,
+      "loss": 1.496,
+      "step": 1388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008439931603598868,
+      "loss": 1.5178,
+      "step": 1389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008437554455706425,
+      "loss": 1.517,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000843517583343246,
+      "loss": 1.4971,
+      "step": 1391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008432795737797172,
+      "loss": 1.508,
+      "step": 1392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008430414169821396,
+      "loss": 1.4963,
+      "step": 1393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008428031130526594,
+      "loss": 1.4978,
+      "step": 1394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008425646620934864,
+      "loss": 1.5354,
+      "step": 1395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008423260642068932,
+      "loss": 1.5375,
+      "step": 1396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008420873194952153,
+      "loss": 1.5326,
+      "step": 1397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008418484280608513,
+      "loss": 1.472,
+      "step": 1398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416093900062629,
+      "loss": 1.4948,
+      "step": 1399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008413702054339742,
+      "loss": 1.4949,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841130874446573,
+      "loss": 1.4821,
+      "step": 1401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008408913971467089,
+      "loss": 1.4922,
+      "step": 1402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008406517736370949,
+      "loss": 1.5556,
+      "step": 1403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008404120040205066,
+      "loss": 1.5269,
+      "step": 1404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000840172088399782,
+      "loss": 1.4665,
+      "step": 1405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000839932026877822,
+      "loss": 1.4987,
+      "step": 1406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00083969181955759,
+      "loss": 1.4958,
+      "step": 1407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000839451466542112,
+      "loss": 1.5257,
+      "step": 1408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392109679344763,
+      "loss": 1.4944,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008389703238378339,
+      "loss": 1.4839,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008387295343553979,
+      "loss": 1.474,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838488599590444,
+      "loss": 1.4919,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008382475196463102,
+      "loss": 1.55,
+      "step": 1413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008380062946263964,
+      "loss": 1.5054,
+      "step": 1414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008377649246341654,
+      "loss": 1.4616,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008375234097731418,
+      "loss": 1.4969,
+      "step": 1416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837281750146912,
+      "loss": 1.5027,
+      "step": 1417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008370399458591251,
+      "loss": 1.5209,
+      "step": 1418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008367979970134917,
+      "loss": 1.5168,
+      "step": 1419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008365559037137851,
+      "loss": 1.4843,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008363136660638397,
+      "loss": 1.4916,
+      "step": 1421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008360712841675526,
+      "loss": 1.5108,
+      "step": 1422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008358287581288823,
+      "loss": 1.4948,
+      "step": 1423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355860880518489,
+      "loss": 1.4986,
+      "step": 1424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008353432740405353,
+      "loss": 1.5156,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008351003161990847,
+      "loss": 1.4962,
+      "step": 1426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008348572146317033,
+      "loss": 1.5429,
+      "step": 1427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000834613969442658,
+      "loss": 1.4964,
+      "step": 1428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008343705807362778,
+      "loss": 1.4906,
+      "step": 1429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341270486169534,
+      "loss": 1.5303,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008338833731891364,
+      "loss": 1.4837,
+      "step": 1431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008336395545573403,
+      "loss": 1.5395,
+      "step": 1432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008333955928261401,
+      "loss": 1.4742,
+      "step": 1433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008331514881001718,
+      "loss": 1.5006,
+      "step": 1434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329072404841332,
+      "loss": 1.5224,
+      "step": 1435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008326628500827827,
+      "loss": 1.513,
+      "step": 1436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008324183170009407,
+      "loss": 1.4728,
+      "step": 1437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008321736413434884,
+      "loss": 1.4935,
+      "step": 1438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008319288232153684,
+      "loss": 1.4888,
+      "step": 1439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008316838627215837,
+      "loss": 1.5183,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314387599671994,
+      "loss": 1.5253,
+      "step": 1441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008311935150573409,
+      "loss": 1.4955,
+      "step": 1442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008309481280971947,
+      "loss": 1.5268,
+      "step": 1443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008307025991920085,
+      "loss": 1.5043,
+      "step": 1444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008304569284470904,
+      "loss": 1.4785,
+      "step": 1445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008302111159678099,
+      "loss": 1.5116,
+      "step": 1446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008299651618595967,
+      "loss": 1.4906,
+      "step": 1447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008297190662279419,
+      "loss": 1.4774,
+      "step": 1448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294728291783967,
+      "loss": 1.5044,
+      "step": 1449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008292264508165733,
+      "loss": 1.5501,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008289799312481442,
+      "loss": 1.5291,
+      "step": 1451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008287332705788431,
+      "loss": 1.4934,
+      "step": 1452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008284864689144634,
+      "loss": 1.5073,
+      "step": 1453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008282395263608596,
+      "loss": 1.5367,
+      "step": 1454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008279924430239462,
+      "loss": 1.5328,
+      "step": 1455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008277452190096985,
+      "loss": 1.4924,
+      "step": 1456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008274978544241517,
+      "loss": 1.4566,
+      "step": 1457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008272503493734018,
+      "loss": 1.4803,
+      "step": 1458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008270027039636044,
+      "loss": 1.5072,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008267549183009759,
+      "loss": 1.4776,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008265069924917925,
+      "loss": 1.5367,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008262589266423908,
+      "loss": 1.5141,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826010720859167,
+      "loss": 1.4596,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008257623752485779,
+      "loss": 1.5285,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008255138899171397,
+      "loss": 1.5066,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000825265264971429,
+      "loss": 1.5328,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008250165005180819,
+      "loss": 1.5011,
+      "step": 1467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008247675966637947,
+      "loss": 1.5323,
+      "step": 1468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008245185535153232,
+      "loss": 1.5173,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008242693711794831,
+      "loss": 1.5011,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008240200497631497,
+      "loss": 1.5305,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237705893732581,
+      "loss": 1.5069,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008235209901168026,
+      "loss": 1.477,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008232712521008379,
+      "loss": 1.523,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008230213754324773,
+      "loss": 1.5092,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822771360218894,
+      "loss": 1.5432,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008225212065673205,
+      "loss": 1.5212,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822270914585049,
+      "loss": 1.5133,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008220204843794306,
+      "loss": 1.4819,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000821769916057876,
+      "loss": 1.4365,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215192097278548,
+      "loss": 1.4892,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000821268365496896,
+      "loss": 1.4891,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000821017383472588,
+      "loss": 1.5007,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008207662637625779,
+      "loss": 1.5186,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205150064745719,
+      "loss": 1.5141,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008202636117163356,
+      "loss": 1.471,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008200120795956928,
+      "loss": 1.5047,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000819760410220527,
+      "loss": 1.4888,
+      "step": 1488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195086036987805,
+      "loss": 1.524,
+      "step": 1489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008192566601384534,
+      "loss": 1.5148,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000819004579647606,
+      "loss": 1.5097,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008187523623343566,
+      "loss": 1.4529,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008185000083068821,
+      "loss": 1.5111,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818247517673418,
+      "loss": 1.4389,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008179948905422593,
+      "loss": 1.4923,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008177421270217582,
+      "loss": 1.4887,
+      "step": 1496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008174892272203261,
+      "loss": 1.4971,
+      "step": 1497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000817236191246433,
+      "loss": 1.5223,
+      "step": 1498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008169830192086071,
+      "loss": 1.4876,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167297112154346,
+      "loss": 1.5135,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008164762673755609,
+      "loss": 1.4993,
+      "step": 1501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008162226877976886,
+      "loss": 1.4967,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008159689725905795,
+      "loss": 1.4846,
+      "step": 1503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008157151218630527,
+      "loss": 1.527,
+      "step": 1504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000815461135723986,
+      "loss": 1.4661,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152070142823153,
+      "loss": 1.5337,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008149527576470342,
+      "loss": 1.5238,
+      "step": 1507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008146983659271942,
+      "loss": 1.4479,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008144438392319055,
+      "loss": 1.5335,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008141891776703354,
+      "loss": 1.4987,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008139343813517092,
+      "loss": 1.4778,
+      "step": 1511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008136794503853102,
+      "loss": 1.4737,
+      "step": 1512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134243848804794,
+      "loss": 1.5144,
+      "step": 1513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008131691849466153,
+      "loss": 1.5394,
+      "step": 1514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008129138506931745,
+      "loss": 1.5289,
+      "step": 1515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126583822296707,
+      "loss": 1.4855,
+      "step": 1516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008124027796656757,
+      "loss": 1.5759,
+      "step": 1517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000812147043110818,
+      "loss": 1.5203,
+      "step": 1518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118911726747847,
+      "loss": 1.4889,
+      "step": 1519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000811635168467319,
+      "loss": 1.537,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113790305982227,
+      "loss": 1.4959,
+      "step": 1521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008111227591773544,
+      "loss": 1.5228,
+      "step": 1522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008108663543146298,
+      "loss": 1.4637,
+      "step": 1523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008106098161200219,
+      "loss": 1.4778,
+      "step": 1524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008103531447035613,
+      "loss": 1.4674,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008100963401753354,
+      "loss": 1.5111,
+      "step": 1526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008098394026454885,
+      "loss": 1.489,
+      "step": 1527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008095823322242224,
+      "loss": 1.5055,
+      "step": 1528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008093251290217958,
+      "loss": 1.5001,
+      "step": 1529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008090677931485239,
+      "loss": 1.5466,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008088103247147793,
+      "loss": 1.4982,
+      "step": 1531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008085527238309913,
+      "loss": 1.4761,
+      "step": 1532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808294990607646,
+      "loss": 1.5016,
+      "step": 1533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008080371251552863,
+      "loss": 1.5421,
+      "step": 1534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077791275845118,
+      "loss": 1.52,
+      "step": 1535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008075209980059786,
+      "loss": 1.5172,
+      "step": 1536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008072627365303995,
+      "loss": 1.5015,
+      "step": 1537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008070043432685441,
+      "loss": 1.5191,
+      "step": 1538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008067458183312384,
+      "loss": 1.4611,
+      "step": 1539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064871618293645,
+      "loss": 1.4975,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008062283738738619,
+      "loss": 1.5112,
+      "step": 1541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008059694545757251,
+      "loss": 1.5197,
+      "step": 1542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008057104040460061,
+      "loss": 1.4843,
+      "step": 1543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008054512223958126,
+      "loss": 1.4639,
+      "step": 1544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000805191909736309,
+      "loss": 1.527,
+      "step": 1545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008049324661787149,
+      "loss": 1.5261,
+      "step": 1546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046728918343076,
+      "loss": 1.5325,
+      "step": 1547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008044131868144191,
+      "loss": 1.497,
+      "step": 1548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000804153351230438,
+      "loss": 1.5157,
+      "step": 1549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008038933851938091,
+      "loss": 1.4751,
+      "step": 1550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008036332888160327,
+      "loss": 1.4609,
+      "step": 1551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008033730622086652,
+      "loss": 1.4858,
+      "step": 1552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803112705483319,
+      "loss": 1.4739,
+      "step": 1553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008028522187516622,
+      "loss": 1.5114,
+      "step": 1554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008025916021254187,
+      "loss": 1.4305,
+      "step": 1555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000802330855716368,
+      "loss": 1.5026,
+      "step": 1556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008020699796363452,
+      "loss": 1.5379,
+      "step": 1557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008018089739972412,
+      "loss": 1.445,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008015478389110027,
+      "loss": 1.487,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008012865744896312,
+      "loss": 1.5434,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008010251808451844,
+      "loss": 1.4516,
+      "step": 1561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008007636580897753,
+      "loss": 1.5433,
+      "step": 1562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008005020063355714,
+      "loss": 1.4959,
+      "step": 1563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000800240225694797,
+      "loss": 1.4783,
+      "step": 1564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007999783162797305,
+      "loss": 1.5285,
+      "step": 1565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007997162782027061,
+      "loss": 1.4707,
+      "step": 1566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007994541115761129,
+      "loss": 1.5161,
+      "step": 1567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007991918165123954,
+      "loss": 1.5051,
+      "step": 1568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000798929393124053,
+      "loss": 1.5001,
+      "step": 1569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007986668415236401,
+      "loss": 1.4729,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007984041618237664,
+      "loss": 1.5259,
+      "step": 1571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007981413541370961,
+      "loss": 1.4965,
+      "step": 1572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007978784185763486,
+      "loss": 1.4696,
+      "step": 1573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007976153552542982,
+      "loss": 1.4868,
+      "step": 1574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007973521642837736,
+      "loss": 1.4855,
+      "step": 1575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007970888457776588,
+      "loss": 1.4907,
+      "step": 1576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796825399848892,
+      "loss": 1.4535,
+      "step": 1577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007965618266104664,
+      "loss": 1.548,
+      "step": 1578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007962981261754295,
+      "loss": 1.5268,
+      "step": 1579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007960342986568836,
+      "loss": 1.4937,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007957703441679856,
+      "loss": 1.4521,
+      "step": 1581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007955062628219463,
+      "loss": 1.4917,
+      "step": 1582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952420547320316,
+      "loss": 1.4644,
+      "step": 1583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007949777200115615,
+      "loss": 1.4901,
+      "step": 1584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007947132587739101,
+      "loss": 1.5209,
+      "step": 1585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000794448671132506,
+      "loss": 1.4633,
+      "step": 1586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000794183957200832,
+      "loss": 1.4997,
+      "step": 1587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007939191170924249,
+      "loss": 1.4875,
+      "step": 1588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007936541509208757,
+      "loss": 1.4693,
+      "step": 1589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007933890587998297,
+      "loss": 1.4752,
+      "step": 1590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007931238408429859,
+      "loss": 1.4481,
+      "step": 1591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007928584971640975,
+      "loss": 1.5159,
+      "step": 1592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007925930278769713,
+      "loss": 1.5013,
+      "step": 1593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007923274330954683,
+      "loss": 1.4636,
+      "step": 1594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007920617129335033,
+      "loss": 1.5256,
+      "step": 1595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007917958675050448,
+      "loss": 1.4818,
+      "step": 1596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791529896924115,
+      "loss": 1.5693,
+      "step": 1597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007912638013047895,
+      "loss": 1.5324,
+      "step": 1598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007909975807611982,
+      "loss": 1.5336,
+      "step": 1599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000790731235407524,
+      "loss": 1.514,
+      "step": 1600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007904647653580036,
+      "loss": 1.4885,
+      "step": 1601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007901981707269273,
+      "loss": 1.4889,
+      "step": 1602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007899314516286385,
+      "loss": 1.5105,
+      "step": 1603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007896646081775339,
+      "loss": 1.535,
+      "step": 1604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007893976404880641,
+      "loss": 1.4738,
+      "step": 1605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891305486747327,
+      "loss": 1.4881,
+      "step": 1606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007888633328520963,
+      "loss": 1.512,
+      "step": 1607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000788595993134765,
+      "loss": 1.4818,
+      "step": 1608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007883285296374018,
+      "loss": 1.4841,
+      "step": 1609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000788060942474723,
+      "loss": 1.4916,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000787793231761498,
+      "loss": 1.5255,
+      "step": 1611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007875253976125488,
+      "loss": 1.52,
+      "step": 1612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007872574401427506,
+      "loss": 1.4688,
+      "step": 1613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007869893594670316,
+      "loss": 1.5304,
+      "step": 1614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007867211557003727,
+      "loss": 1.5066,
+      "step": 1615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007864528289578077,
+      "loss": 1.49,
+      "step": 1616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000786184379354423,
+      "loss": 1.4746,
+      "step": 1617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007859158070053577,
+      "loss": 1.4568,
+      "step": 1618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007856471120258037,
+      "loss": 1.4828,
+      "step": 1619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007853782945310053,
+      "loss": 1.4878,
+      "step": 1620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007851093546362596,
+      "loss": 1.5154,
+      "step": 1621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007848402924569158,
+      "loss": 1.4918,
+      "step": 1622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007845711081083761,
+      "loss": 1.5194,
+      "step": 1623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007843018017060946,
+      "loss": 1.4911,
+      "step": 1624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007840323733655779,
+      "loss": 1.4721,
+      "step": 1625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007837628232023852,
+      "loss": 1.5026,
+      "step": 1626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007834931513321275,
+      "loss": 1.5541,
+      "step": 1627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007832233578704682,
+      "loss": 1.5034,
+      "step": 1628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007829534429331229,
+      "loss": 1.4796,
+      "step": 1629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007826834066358592,
+      "loss": 1.5192,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007824132490944967,
+      "loss": 1.4391,
+      "step": 1631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007821429704249074,
+      "loss": 1.4891,
+      "step": 1632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818725707430148,
+      "loss": 1.4921,
+      "step": 1633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007816020501647944,
+      "loss": 1.4764,
+      "step": 1634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007813314088062738,
+      "loss": 1.4909,
+      "step": 1635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007810606467835318,
+      "loss": 1.4907,
+      "step": 1636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007807897642126998,
+      "loss": 1.5153,
+      "step": 1637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007805187612099602,
+      "loss": 1.5098,
+      "step": 1638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007802476378915475,
+      "loss": 1.5163,
+      "step": 1639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007799763943737475,
+      "loss": 1.4579,
+      "step": 1640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007797050307728979,
+      "loss": 1.4945,
+      "step": 1641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007794335472053874,
+      "loss": 1.4783,
+      "step": 1642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007791619437876566,
+      "loss": 1.4946,
+      "step": 1643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007788902206361973,
+      "loss": 1.5123,
+      "step": 1644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007786183778675526,
+      "loss": 1.527,
+      "step": 1645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007783464155983173,
+      "loss": 1.4693,
+      "step": 1646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007780743339451371,
+      "loss": 1.4465,
+      "step": 1647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007778021330247085,
+      "loss": 1.4875,
+      "step": 1648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007775298129537801,
+      "loss": 1.4802,
+      "step": 1649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000777257373849151,
+      "loss": 1.4962,
+      "step": 1650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007769848158276715,
+      "loss": 1.5125,
+      "step": 1651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007767121390062426,
+      "loss": 1.5026,
+      "step": 1652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007764393435018169,
+      "loss": 1.5332,
+      "step": 1653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007761664294313975,
+      "loss": 1.4958,
+      "step": 1654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007758933969120381,
+      "loss": 1.4971,
+      "step": 1655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007756202460608438,
+      "loss": 1.5314,
+      "step": 1656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007753469769949701,
+      "loss": 1.4888,
+      "step": 1657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000775073589831623,
+      "loss": 1.5237,
+      "step": 1658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007748000846880596,
+      "loss": 1.506,
+      "step": 1659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007745264616815875,
+      "loss": 1.4858,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007742527209295644,
+      "loss": 1.5137,
+      "step": 1661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007739788625493993,
+      "loss": 1.4907,
+      "step": 1662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000773704886658551,
+      "loss": 1.5147,
+      "step": 1663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007734307933745287,
+      "loss": 1.4825,
+      "step": 1664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007731565828148926,
+      "loss": 1.4943,
+      "step": 1665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007728822550972523,
+      "loss": 1.4955,
+      "step": 1666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007726078103392684,
+      "loss": 1.4706,
+      "step": 1667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007723332486586514,
+      "loss": 1.4733,
+      "step": 1668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000772058570173162,
+      "loss": 1.4564,
+      "step": 1669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007717837750006106,
+      "loss": 1.5797,
+      "step": 1670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007715088632588582,
+      "loss": 1.486,
+      "step": 1671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771233835065816,
+      "loss": 1.5394,
+      "step": 1672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007709586905394441,
+      "loss": 1.4978,
+      "step": 1673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706834297977537,
+      "loss": 1.4939,
+      "step": 1674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007704080529588049,
+      "loss": 1.4714,
+      "step": 1675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000770132560140708,
+      "loss": 1.4692,
+      "step": 1676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007698569514616231,
+      "loss": 1.5031,
+      "step": 1677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007695812270397601,
+      "loss": 1.5492,
+      "step": 1678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007693053869933782,
+      "loss": 1.5111,
+      "step": 1679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007690294314407862,
+      "loss": 1.4806,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768753360500343,
+      "loss": 1.4976,
+      "step": 1681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007684771742904561,
+      "loss": 1.4942,
+      "step": 1682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682008729295832,
+      "loss": 1.4468,
+      "step": 1683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007679244565362314,
+      "loss": 1.4535,
+      "step": 1684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767647925228956,
+      "loss": 1.4805,
+      "step": 1685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007673712791263634,
+      "loss": 1.5266,
+      "step": 1686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007670945183471076,
+      "loss": 1.5073,
+      "step": 1687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007668176430098929,
+      "loss": 1.504,
+      "step": 1688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007665406532334719,
+      "loss": 1.5146,
+      "step": 1689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000766263549136647,
+      "loss": 1.4973,
+      "step": 1690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007659863308382691,
+      "loss": 1.5233,
+      "step": 1691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007657089984572383,
+      "loss": 1.5074,
+      "step": 1692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007654315521125037,
+      "loss": 1.4725,
+      "step": 1693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007651539919230633,
+      "loss": 1.5147,
+      "step": 1694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007648763180079637,
+      "loss": 1.5285,
+      "step": 1695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007645985304863003,
+      "loss": 1.4979,
+      "step": 1696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007643206294772176,
+      "loss": 1.479,
+      "step": 1697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007640426150999082,
+      "loss": 1.4853,
+      "step": 1698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000763764487473614,
+      "loss": 1.4979,
+      "step": 1699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007634862467176248,
+      "loss": 1.5001,
+      "step": 1700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632078929512793,
+      "loss": 1.4943,
+      "step": 1701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007629294262939647,
+      "loss": 1.5224,
+      "step": 1702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007626508468651164,
+      "loss": 1.4845,
+      "step": 1703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007623721547842183,
+      "loss": 1.4951,
+      "step": 1704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007620933501708028,
+      "loss": 1.5068,
+      "step": 1705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00076181443314445,
+      "loss": 1.5018,
+      "step": 1706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007615354038247889,
+      "loss": 1.4869,
+      "step": 1707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007612562623314962,
+      "loss": 1.519,
+      "step": 1708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007609770087842969,
+      "loss": 1.5298,
+      "step": 1709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007606976433029639,
+      "loss": 1.4993,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007604181660073184,
+      "loss": 1.4927,
+      "step": 1711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601385770172293,
+      "loss": 1.5028,
+      "step": 1712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007598588764526136,
+      "loss": 1.4896,
+      "step": 1713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007595790644334358,
+      "loss": 1.4555,
+      "step": 1714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007592991410797086,
+      "loss": 1.4764,
+      "step": 1715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007590191065114924,
+      "loss": 1.5062,
+      "step": 1716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758738960848895,
+      "loss": 1.5279,
+      "step": 1717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007584587042120723,
+      "loss": 1.4457,
+      "step": 1718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007581783367212274,
+      "loss": 1.4952,
+      "step": 1719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007578978584966109,
+      "loss": 1.5168,
+      "step": 1720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007576172696585216,
+      "loss": 1.5132,
+      "step": 1721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007573365703273045,
+      "loss": 1.4947,
+      "step": 1722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570557606233533,
+      "loss": 1.5291,
+      "step": 1723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007567748406671084,
+      "loss": 1.4822,
+      "step": 1724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007564938105790573,
+      "loss": 1.5022,
+      "step": 1725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007562126704797352,
+      "loss": 1.5218,
+      "step": 1726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000755931420489724,
+      "loss": 1.4724,
+      "step": 1727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007556500607296534,
+      "loss": 1.5057,
+      "step": 1728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007553685913201994,
+      "loss": 1.5272,
+      "step": 1729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007550870123820857,
+      "loss": 1.4853,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007548053240360826,
+      "loss": 1.5039,
+      "step": 1731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007545235264030072,
+      "loss": 1.4535,
+      "step": 1732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000754241619603724,
+      "loss": 1.4972,
+      "step": 1733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007539596037591438,
+      "loss": 1.5054,
+      "step": 1734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007536774789902246,
+      "loss": 1.5341,
+      "step": 1735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007533952454179707,
+      "loss": 1.452,
+      "step": 1736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007531129031634332,
+      "loss": 1.497,
+      "step": 1737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007528304523477099,
+      "loss": 1.4912,
+      "step": 1738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007525478930919453,
+      "loss": 1.4897,
+      "step": 1739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007522652255173303,
+      "loss": 1.4811,
+      "step": 1740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519824497451019,
+      "loss": 1.4956,
+      "step": 1741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007516995658965442,
+      "loss": 1.5261,
+      "step": 1742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751416574092987,
+      "loss": 1.479,
+      "step": 1743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511334744558067,
+      "loss": 1.4753,
+      "step": 1744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007508502671064259,
+      "loss": 1.4946,
+      "step": 1745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007505669521663136,
+      "loss": 1.5335,
+      "step": 1746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007502835297569845,
+      "loss": 1.5294,
+      "step": 1747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075,
+      "loss": 1.4947,
+      "step": 1748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000749716363016967,
+      "loss": 1.4865,
+      "step": 1749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007494326189295384,
+      "loss": 1.5085,
+      "step": 1750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007491487678594137,
+      "loss": 1.4393,
+      "step": 1751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007488648099283373,
+      "loss": 1.474,
+      "step": 1752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007485807452581002,
+      "loss": 1.4782,
+      "step": 1753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007482965739705391,
+      "loss": 1.463,
+      "step": 1754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007480122961875359,
+      "loss": 1.508,
+      "step": 1755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007477279120310189,
+      "loss": 1.5167,
+      "step": 1756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007474434216229613,
+      "loss": 1.4875,
+      "step": 1757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007471588250853824,
+      "loss": 1.5112,
+      "step": 1758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007468741225403469,
+      "loss": 1.494,
+      "step": 1759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465893141099649,
+      "loss": 1.4985,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007463043999163918,
+      "loss": 1.4843,
+      "step": 1761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007460193800818286,
+      "loss": 1.4695,
+      "step": 1762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007457342547285217,
+      "loss": 1.5065,
+      "step": 1763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007454490239787621,
+      "loss": 1.4793,
+      "step": 1764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007451636879548867,
+      "loss": 1.5214,
+      "step": 1765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007448782467792775,
+      "loss": 1.4611,
+      "step": 1766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007445927005743611,
+      "loss": 1.4756,
+      "step": 1767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007443070494626096,
+      "loss": 1.501,
+      "step": 1768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007440212935665401,
+      "loss": 1.5409,
+      "step": 1769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007437354330087143,
+      "loss": 1.4805,
+      "step": 1770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007434494679117391,
+      "loss": 1.511,
+      "step": 1771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000743163398398266,
+      "loss": 1.5042,
+      "step": 1772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007428772245909916,
+      "loss": 1.5058,
+      "step": 1773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007425909466126568,
+      "loss": 1.5175,
+      "step": 1774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007423045645860478,
+      "loss": 1.4792,
+      "step": 1775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007420180786339947,
+      "loss": 1.4671,
+      "step": 1776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007417314888793727,
+      "loss": 1.4505,
+      "step": 1777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007414447954451013,
+      "loss": 1.4858,
+      "step": 1778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007411579984541446,
+      "loss": 1.5096,
+      "step": 1779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007408710980295108,
+      "loss": 1.5132,
+      "step": 1780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007405840942942529,
+      "loss": 1.4623,
+      "step": 1781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740296987371468,
+      "loss": 1.4544,
+      "step": 1782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007400097773842975,
+      "loss": 1.5189,
+      "step": 1783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007397224644559266,
+      "loss": 1.5372,
+      "step": 1784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007394350487095857,
+      "loss": 1.4804,
+      "step": 1785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007391475302685479,
+      "loss": 1.5121,
+      "step": 1786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007388599092561315,
+      "loss": 1.4988,
+      "step": 1787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007385721857956982,
+      "loss": 1.4449,
+      "step": 1788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000738284360010654,
+      "loss": 1.5033,
+      "step": 1789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007379964320244482,
+      "loss": 1.4744,
+      "step": 1790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007377084019605747,
+      "loss": 1.5309,
+      "step": 1791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374202699425707,
+      "loss": 1.4867,
+      "step": 1792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007371320360940171,
+      "loss": 1.4815,
+      "step": 1793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007368437005385389,
+      "loss": 1.4928,
+      "step": 1794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365552633998041,
+      "loss": 1.4467,
+      "step": 1795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007362667248015245,
+      "loss": 1.4689,
+      "step": 1796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007359780848674561,
+      "loss": 1.4865,
+      "step": 1797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356893437213971,
+      "loss": 1.5001,
+      "step": 1798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007354005014871903,
+      "loss": 1.5012,
+      "step": 1799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007351115582887211,
+      "loss": 1.5134,
+      "step": 1800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007348225142499186,
+      "loss": 1.4938,
+      "step": 1801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007345333694947547,
+      "loss": 1.5065,
+      "step": 1802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007342441241472451,
+      "loss": 1.4863,
+      "step": 1803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007339547783314482,
+      "loss": 1.5337,
+      "step": 1804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007336653321714657,
+      "loss": 1.5012,
+      "step": 1805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333757857914423,
+      "loss": 1.4671,
+      "step": 1806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007330861393155656,
+      "loss": 1.4909,
+      "step": 1807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007327963928680661,
+      "loss": 1.4903,
+      "step": 1808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007325065465732175,
+      "loss": 1.5238,
+      "step": 1809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007322166005553358,
+      "loss": 1.5246,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007319265549387802,
+      "loss": 1.4574,
+      "step": 1811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007316364098479527,
+      "loss": 1.5062,
+      "step": 1812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007313461654072974,
+      "loss": 1.4609,
+      "step": 1813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007310558217413015,
+      "loss": 1.5024,
+      "step": 1814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007307653789744947,
+      "loss": 1.4557,
+      "step": 1815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007304748372314489,
+      "loss": 1.4968,
+      "step": 1816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007301841966367788,
+      "loss": 1.4954,
+      "step": 1817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007298934573151415,
+      "loss": 1.5211,
+      "step": 1818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007296026193912362,
+      "loss": 1.5192,
+      "step": 1819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293116829898043,
+      "loss": 1.4854,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072902064823563,
+      "loss": 1.4763,
+      "step": 1821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007287295152535392,
+      "loss": 1.4848,
+      "step": 1822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007284382841684,
+      "loss": 1.485,
+      "step": 1823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007281469551051226,
+      "loss": 1.504,
+      "step": 1824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007278555281886594,
+      "loss": 1.4898,
+      "step": 1825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007275640035440044,
+      "loss": 1.4921,
+      "step": 1826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007272723812961941,
+      "loss": 1.4843,
+      "step": 1827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726980661570306,
+      "loss": 1.4734,
+      "step": 1828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007266888444914605,
+      "loss": 1.4811,
+      "step": 1829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007263969301848187,
+      "loss": 1.4823,
+      "step": 1830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726104918775584,
+      "loss": 1.5057,
+      "step": 1831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007258128103890015,
+      "loss": 1.491,
+      "step": 1832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007255206051503575,
+      "loss": 1.5058,
+      "step": 1833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007252283031849801,
+      "loss": 1.472,
+      "step": 1834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007249359046182388,
+      "loss": 1.5122,
+      "step": 1835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007246434095755449,
+      "loss": 1.456,
+      "step": 1836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007243508181823502,
+      "loss": 1.4842,
+      "step": 1837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007240581305641489,
+      "loss": 1.5164,
+      "step": 1838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007237653468464755,
+      "loss": 1.4905,
+      "step": 1839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007234724671549065,
+      "loss": 1.4885,
+      "step": 1840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000723179491615059,
+      "loss": 1.4998,
+      "step": 1841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007228864203525918,
+      "loss": 1.4867,
+      "step": 1842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007225932534932039,
+      "loss": 1.4945,
+      "step": 1843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000722299991162636,
+      "loss": 1.4973,
+      "step": 1844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007220066334866697,
+      "loss": 1.5078,
+      "step": 1845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007217131805911271,
+      "loss": 1.5158,
+      "step": 1846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007214196326018715,
+      "loss": 1.4907,
+      "step": 1847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007211259896448069,
+      "loss": 1.4645,
+      "step": 1848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007208322518458778,
+      "loss": 1.5036,
+      "step": 1849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007205384193310699,
+      "loss": 1.4887,
+      "step": 1850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007202444922264089,
+      "loss": 1.4779,
+      "step": 1851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007199504706579617,
+      "loss": 1.4999,
+      "step": 1852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007196563547518349,
+      "loss": 1.5145,
+      "step": 1853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007193621446341765,
+      "loss": 1.4803,
+      "step": 1854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007190678404311743,
+      "loss": 1.459,
+      "step": 1855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007187734422690564,
+      "loss": 1.5015,
+      "step": 1856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007184789502740917,
+      "loss": 1.5137,
+      "step": 1857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000718184364572589,
+      "loss": 1.501,
+      "step": 1858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007178896852908971,
+      "loss": 1.5064,
+      "step": 1859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007175949125554055,
+      "loss": 1.4806,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007173000464925432,
+      "loss": 1.4984,
+      "step": 1861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007170050872287797,
+      "loss": 1.4719,
+      "step": 1862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007167100348906241,
+      "loss": 1.4898,
+      "step": 1863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007164148896046256,
+      "loss": 1.49,
+      "step": 1864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007161196514973735,
+      "loss": 1.4631,
+      "step": 1865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007158243206954964,
+      "loss": 1.4922,
+      "step": 1866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007155288973256631,
+      "loss": 1.4962,
+      "step": 1867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007152333815145817,
+      "loss": 1.4535,
+      "step": 1868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007149377733890001,
+      "loss": 1.4924,
+      "step": 1869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007146420730757064,
+      "loss": 1.4749,
+      "step": 1870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000714346280701527,
+      "loss": 1.492,
+      "step": 1871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007140503963933291,
+      "loss": 1.4865,
+      "step": 1872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007137544202780185,
+      "loss": 1.4999,
+      "step": 1873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007134583524825405,
+      "loss": 1.4972,
+      "step": 1874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007131621931338798,
+      "loss": 1.4985,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007128659423590605,
+      "loss": 1.4739,
+      "step": 1876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007125696002851458,
+      "loss": 1.4675,
+      "step": 1877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000712273167039238,
+      "loss": 1.5255,
+      "step": 1878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007119766427484786,
+      "loss": 1.4556,
+      "step": 1879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007116800275400482,
+      "loss": 1.4701,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000711383321541166,
+      "loss": 1.4511,
+      "step": 1881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007110865248790909,
+      "loss": 1.5186,
+      "step": 1882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007107896376811198,
+      "loss": 1.4714,
+      "step": 1883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007104926600745891,
+      "loss": 1.5267,
+      "step": 1884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007101955921868738,
+      "loss": 1.4583,
+      "step": 1885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007098984341453873,
+      "loss": 1.4688,
+      "step": 1886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007096011860775822,
+      "loss": 1.4733,
+      "step": 1887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007093038481109494,
+      "loss": 1.5034,
+      "step": 1888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007090064203730182,
+      "loss": 1.5194,
+      "step": 1889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007087089029913567,
+      "loss": 1.5075,
+      "step": 1890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007084112960935716,
+      "loss": 1.472,
+      "step": 1891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007081135998073072,
+      "loss": 1.4459,
+      "step": 1892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007078158142602473,
+      "loss": 1.45,
+      "step": 1893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000707517939580113,
+      "loss": 1.4749,
+      "step": 1894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000707219975894664,
+      "loss": 1.4574,
+      "step": 1895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007069219233316983,
+      "loss": 1.4761,
+      "step": 1896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000706623782019052,
+      "loss": 1.5092,
+      "step": 1897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007063255520845989,
+      "loss": 1.4688,
+      "step": 1898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007060272336562512,
+      "loss": 1.4887,
+      "step": 1899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007057288268619591,
+      "loss": 1.4932,
+      "step": 1900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007054303318297101,
+      "loss": 1.4537,
+      "step": 1901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007051317486875303,
+      "loss": 1.5131,
+      "step": 1902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007048330775634833,
+      "loss": 1.4824,
+      "step": 1903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00070453431858567,
+      "loss": 1.4748,
+      "step": 1904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007042354718822297,
+      "loss": 1.4757,
+      "step": 1905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007039365375813391,
+      "loss": 1.5088,
+      "step": 1906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007036375158112121,
+      "loss": 1.4586,
+      "step": 1907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007033384067001006,
+      "loss": 1.4826,
+      "step": 1908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007030392103762937,
+      "loss": 1.4672,
+      "step": 1909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007027399269681178,
+      "loss": 1.4924,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000702440556603937,
+      "loss": 1.4828,
+      "step": 1911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007021410994121524,
+      "loss": 1.4981,
+      "step": 1912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007018415555212026,
+      "loss": 1.4877,
+      "step": 1913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007015419250595632,
+      "loss": 1.4883,
+      "step": 1914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007012422081557467,
+      "loss": 1.4388,
+      "step": 1915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007009424049383032,
+      "loss": 1.497,
+      "step": 1916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007006425155358195,
+      "loss": 1.4697,
+      "step": 1917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007003425400769192,
+      "loss": 1.4596,
+      "step": 1918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007000424786902635,
+      "loss": 1.4839,
+      "step": 1919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006997423315045496,
+      "loss": 1.5108,
+      "step": 1920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000699442098648512,
+      "loss": 1.4823,
+      "step": 1921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006991417802509219,
+      "loss": 1.5049,
+      "step": 1922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698841376440587,
+      "loss": 1.4927,
+      "step": 1923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006985408873463517,
+      "loss": 1.4569,
+      "step": 1924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006982403130970972,
+      "loss": 1.5099,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006979396538217411,
+      "loss": 1.5068,
+      "step": 1926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006976389096492372,
+      "loss": 1.5218,
+      "step": 1927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006973380807085763,
+      "loss": 1.4869,
+      "step": 1928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006970371671287847,
+      "loss": 1.4391,
+      "step": 1929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006967361690389258,
+      "loss": 1.4871,
+      "step": 1930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006964350865680991,
+      "loss": 1.4851,
+      "step": 1931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006961339198454398,
+      "loss": 1.4907,
+      "step": 1932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006958326690001196,
+      "loss": 1.4993,
+      "step": 1933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006955313341613464,
+      "loss": 1.4987,
+      "step": 1934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006952299154583639,
+      "loss": 1.5048,
+      "step": 1935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006949284130204519,
+      "loss": 1.4905,
+      "step": 1936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006946268269769261,
+      "loss": 1.5244,
+      "step": 1937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006943251574571379,
+      "loss": 1.5175,
+      "step": 1938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006940234045904746,
+      "loss": 1.4779,
+      "step": 1939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006937215685063594,
+      "loss": 1.5011,
+      "step": 1940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006934196493342511,
+      "loss": 1.4901,
+      "step": 1941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006931176472036438,
+      "loss": 1.4835,
+      "step": 1942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000692815562244068,
+      "loss": 1.5027,
+      "step": 1943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006925133945850885,
+      "loss": 1.5125,
+      "step": 1944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006922111443563069,
+      "loss": 1.457,
+      "step": 1945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006919088116873594,
+      "loss": 1.4979,
+      "step": 1946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006916063967079176,
+      "loss": 1.4892,
+      "step": 1947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006913038995476887,
+      "loss": 1.4797,
+      "step": 1948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910013203364151,
+      "loss": 1.5079,
+      "step": 1949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000690698659203874,
+      "loss": 1.5606,
+      "step": 1950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000690395916279878,
+      "loss": 1.4821,
+      "step": 1951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006900930916942753,
+      "loss": 1.4811,
+      "step": 1952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006897901855769483,
+      "loss": 1.481,
+      "step": 1953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006894871980578145,
+      "loss": 1.4721,
+      "step": 1954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000689184129266827,
+      "loss": 1.5061,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006888809793339729,
+      "loss": 1.4773,
+      "step": 1956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006885777483892746,
+      "loss": 1.5305,
+      "step": 1957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006882744365627892,
+      "loss": 1.5083,
+      "step": 1958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006879710439846083,
+      "loss": 1.5199,
+      "step": 1959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006876675707848582,
+      "loss": 1.4847,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006873640170937001,
+      "loss": 1.4868,
+      "step": 1961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006870603830413291,
+      "loss": 1.4813,
+      "step": 1962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006867566687579754,
+      "loss": 1.473,
+      "step": 1963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006864528743739033,
+      "loss": 1.4908,
+      "step": 1964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006861490000194113,
+      "loss": 1.5116,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006858450458248325,
+      "loss": 1.4904,
+      "step": 1966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006855410119205341,
+      "loss": 1.5301,
+      "step": 1967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852368984369175,
+      "loss": 1.4656,
+      "step": 1968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006849327055044182,
+      "loss": 1.4927,
+      "step": 1969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006846284332535059,
+      "loss": 1.4737,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006843240818146845,
+      "loss": 1.487,
+      "step": 1971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006840196513184912,
+      "loss": 1.5217,
+      "step": 1972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006837151418954976,
+      "loss": 1.513,
+      "step": 1973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006834105536763094,
+      "loss": 1.4833,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006831058867915654,
+      "loss": 1.4536,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006828011413719385,
+      "loss": 1.4625,
+      "step": 1976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006824963175481356,
+      "loss": 1.4988,
+      "step": 1977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006821914154508966,
+      "loss": 1.5221,
+      "step": 1978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006818864352109953,
+      "loss": 1.4837,
+      "step": 1979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006815813769592394,
+      "loss": 1.4469,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006812762408264692,
+      "loss": 1.4515,
+      "step": 1981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000680971026943559,
+      "loss": 1.5124,
+      "step": 1982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006806657354414165,
+      "loss": 1.5391,
+      "step": 1983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006803603664509822,
+      "loss": 1.4982,
+      "step": 1984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006800549201032303,
+      "loss": 1.5174,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006797493965291681,
+      "loss": 1.5013,
+      "step": 1986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006794437958598359,
+      "loss": 1.4927,
+      "step": 1987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000679138118226307,
+      "loss": 1.4737,
+      "step": 1988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000678832363759688,
+      "loss": 1.4573,
+      "step": 1989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006785265325911181,
+      "loss": 1.4752,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006782206248517695,
+      "loss": 1.4804,
+      "step": 1991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006779146406728477,
+      "loss": 1.505,
+      "step": 1992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006776085801855902,
+      "loss": 1.5015,
+      "step": 1993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773024435212678,
+      "loss": 1.4686,
+      "step": 1994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006769962308111839,
+      "loss": 1.4893,
+      "step": 1995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006766899421866741,
+      "loss": 1.532,
+      "step": 1996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006763835777791071,
+      "loss": 1.4744,
+      "step": 1997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676077137719884,
+      "loss": 1.4602,
+      "step": 1998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006757706221404378,
+      "loss": 1.4579,
+      "step": 1999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006754640311722348,
+      "loss": 1.5508,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006751573649467728,
+      "loss": 1.4605,
+      "step": 2001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006748506235955825,
+      "loss": 1.4977,
+      "step": 2002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006745438072502263,
+      "loss": 1.4606,
+      "step": 2003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742369160422993,
+      "loss": 1.4888,
+      "step": 2004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006739299501034282,
+      "loss": 1.4966,
+      "step": 2005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000673622909565272,
+      "loss": 1.5207,
+      "step": 2006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006733157945595218,
+      "loss": 1.5105,
+      "step": 2007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006730086052179003,
+      "loss": 1.4963,
+      "step": 2008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006727013416721625,
+      "loss": 1.4951,
+      "step": 2009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000672394004054095,
+      "loss": 1.5022,
+      "step": 2010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006720865924955161,
+      "loss": 1.4779,
+      "step": 2011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006717791071282759,
+      "loss": 1.5163,
+      "step": 2012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006714715480842563,
+      "loss": 1.4443,
+      "step": 2013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006711639154953706,
+      "loss": 1.4674,
+      "step": 2014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006708562094935635,
+      "loss": 1.4836,
+      "step": 2015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006705484302108118,
+      "loss": 1.4671,
+      "step": 2016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000670240577779123,
+      "loss": 1.4738,
+      "step": 2017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699326523305363,
+      "loss": 1.4834,
+      "step": 2018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006696246539971226,
+      "loss": 1.4547,
+      "step": 2019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006693165829109832,
+      "loss": 1.5,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006690084392042513,
+      "loss": 1.4933,
+      "step": 2021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006687002230090912,
+      "loss": 1.4712,
+      "step": 2022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006683919344576977,
+      "loss": 1.4961,
+      "step": 2023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006680835736822974,
+      "loss": 1.5292,
+      "step": 2024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006677751408151474,
+      "loss": 1.4902,
+      "step": 2025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006674666359885358,
+      "loss": 1.5067,
+      "step": 2026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006671580593347817,
+      "loss": 1.5111,
+      "step": 2027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666849410986235,
+      "loss": 1.5342,
+      "step": 2028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006665406910752761,
+      "loss": 1.4698,
+      "step": 2029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006662318997343162,
+      "loss": 1.4364,
+      "step": 2030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006659230370957974,
+      "loss": 1.4693,
+      "step": 2031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006656141032921922,
+      "loss": 1.4901,
+      "step": 2032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006653050984560035,
+      "loss": 1.4896,
+      "step": 2033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006649960227197648,
+      "loss": 1.4891,
+      "step": 2034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006646868762160399,
+      "loss": 1.4982,
+      "step": 2035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006643776590774231,
+      "loss": 1.4794,
+      "step": 2036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006640683714365388,
+      "loss": 1.4905,
+      "step": 2037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006637590134260422,
+      "loss": 1.488,
+      "step": 2038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006634495851786178,
+      "loss": 1.4862,
+      "step": 2039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006631400868269807,
+      "loss": 1.4917,
+      "step": 2040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006628305185038763,
+      "loss": 1.4777,
+      "step": 2041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006625208803420796,
+      "loss": 1.4873,
+      "step": 2042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006622111724743957,
+      "loss": 1.482,
+      "step": 2043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006619013950336597,
+      "loss": 1.496,
+      "step": 2044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006615915481527361,
+      "loss": 1.4806,
+      "step": 2045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00066128163196452,
+      "loss": 1.4581,
+      "step": 2046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006609716466019355,
+      "loss": 1.4598,
+      "step": 2047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006606615921979366,
+      "loss": 1.4948,
+      "step": 2048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006603514688855071,
+      "loss": 1.5093,
+      "step": 2049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00066004127679766,
+      "loss": 1.4534,
+      "step": 2050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006597310160674382,
+      "loss": 1.482,
+      "step": 2051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006594206868279134,
+      "loss": 1.5063,
+      "step": 2052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006591102892121877,
+      "loss": 1.5027,
+      "step": 2053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006587998233533916,
+      "loss": 1.5147,
+      "step": 2054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584892893846852,
+      "loss": 1.5312,
+      "step": 2055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000658178687439258,
+      "loss": 1.4755,
+      "step": 2056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006578680176503283,
+      "loss": 1.4959,
+      "step": 2057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006575572801511437,
+      "loss": 1.4532,
+      "step": 2058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006572464750749809,
+      "loss": 1.4577,
+      "step": 2059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006569356025551454,
+      "loss": 1.4957,
+      "step": 2060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006566246627249719,
+      "loss": 1.4525,
+      "step": 2061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006563136557178236,
+      "loss": 1.4904,
+      "step": 2062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006560025816670927,
+      "loss": 1.471,
+      "step": 2063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006556914407062004,
+      "loss": 1.5157,
+      "step": 2064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000655380232968596,
+      "loss": 1.5033,
+      "step": 2065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006550689585877583,
+      "loss": 1.5178,
+      "step": 2066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006547576176971937,
+      "loss": 1.4585,
+      "step": 2067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000654446210430438,
+      "loss": 1.4702,
+      "step": 2068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006541347369210548,
+      "loss": 1.5129,
+      "step": 2069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006538231973026364,
+      "loss": 1.5134,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006535115917088037,
+      "loss": 1.504,
+      "step": 2071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006531999202732055,
+      "loss": 1.5098,
+      "step": 2072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006528881831295188,
+      "loss": 1.4669,
+      "step": 2073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006525763804114494,
+      "loss": 1.4592,
+      "step": 2074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006522645122527304,
+      "loss": 1.4732,
+      "step": 2075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006519525787871234,
+      "loss": 1.4947,
+      "step": 2076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006516405801484183,
+      "loss": 1.4654,
+      "step": 2077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006513285164704323,
+      "loss": 1.4725,
+      "step": 2078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006510163878870108,
+      "loss": 1.4666,
+      "step": 2079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006507041945320273,
+      "loss": 1.481,
+      "step": 2080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006503919365393827,
+      "loss": 1.4883,
+      "step": 2081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006500796140430057,
+      "loss": 1.4548,
+      "step": 2082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000649767227176853,
+      "loss": 1.4612,
+      "step": 2083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006494547760749084,
+      "loss": 1.4941,
+      "step": 2084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006491422608711834,
+      "loss": 1.4697,
+      "step": 2085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006488296816997174,
+      "loss": 1.5044,
+      "step": 2086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006485170386945766,
+      "loss": 1.4741,
+      "step": 2087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000648204331989855,
+      "loss": 1.467,
+      "step": 2088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006478915617196739,
+      "loss": 1.5041,
+      "step": 2089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006475787280181818,
+      "loss": 1.461,
+      "step": 2090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000647265831019554,
+      "loss": 1.4918,
+      "step": 2091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000646952870857994,
+      "loss": 1.4483,
+      "step": 2092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006466398476677313,
+      "loss": 1.482,
+      "step": 2093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006463267615830228,
+      "loss": 1.4774,
+      "step": 2094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006460136127381526,
+      "loss": 1.4856,
+      "step": 2095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006457004012674315,
+      "loss": 1.4631,
+      "step": 2096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006453871273051974,
+      "loss": 1.4534,
+      "step": 2097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006450737909858146,
+      "loss": 1.4635,
+      "step": 2098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006447603924436743,
+      "loss": 1.4382,
+      "step": 2099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006444469318131947,
+      "loss": 1.4683,
+      "step": 2100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006441334092288201,
+      "loss": 1.4681,
+      "step": 2101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006438198248250219,
+      "loss": 1.4885,
+      "step": 2102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006435061787362975,
+      "loss": 1.4968,
+      "step": 2103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006431924710971714,
+      "loss": 1.4601,
+      "step": 2104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006428787020421937,
+      "loss": 1.4928,
+      "step": 2105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425648717059417,
+      "loss": 1.466,
+      "step": 2106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006422509802230181,
+      "loss": 1.5203,
+      "step": 2107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006419370277280527,
+      "loss": 1.4984,
+      "step": 2108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006416230143557007,
+      "loss": 1.4831,
+      "step": 2109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006413089402406437,
+      "loss": 1.4742,
+      "step": 2110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006409948055175898,
+      "loss": 1.4675,
+      "step": 2111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006406806103212725,
+      "loss": 1.4905,
+      "step": 2112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006403663547864513,
+      "loss": 1.4725,
+      "step": 2113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006400520390479119,
+      "loss": 1.4847,
+      "step": 2114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006397376632404657,
+      "loss": 1.4772,
+      "step": 2115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006394232274989494,
+      "loss": 1.5061,
+      "step": 2116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006391087319582263,
+      "loss": 1.4963,
+      "step": 2117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006387941767531846,
+      "loss": 1.4374,
+      "step": 2118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006384795620187385,
+      "loss": 1.4562,
+      "step": 2119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006381648878898274,
+      "loss": 1.4966,
+      "step": 2120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006378501545014164,
+      "loss": 1.4846,
+      "step": 2121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000637535361988496,
+      "loss": 1.4936,
+      "step": 2122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000637220510486082,
+      "loss": 1.4406,
+      "step": 2123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006369056001292155,
+      "loss": 1.4596,
+      "step": 2124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000636590631052963,
+      "loss": 1.4961,
+      "step": 2125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000636275603392416,
+      "loss": 1.4521,
+      "step": 2126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000635960517282691,
+      "loss": 1.4726,
+      "step": 2127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00063564537285893,
+      "loss": 1.5007,
+      "step": 2128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006353301702562999,
+      "loss": 1.4725,
+      "step": 2129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006350149096099921,
+      "loss": 1.5076,
+      "step": 2130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006346995910552233,
+      "loss": 1.4812,
+      "step": 2131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000634384214727235,
+      "loss": 1.458,
+      "step": 2132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006340687807612933,
+      "loss": 1.4894,
+      "step": 2133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006337532892926892,
+      "loss": 1.4486,
+      "step": 2134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006334377404567386,
+      "loss": 1.5386,
+      "step": 2135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006331221343887814,
+      "loss": 1.4982,
+      "step": 2136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006328064712241824,
+      "loss": 1.4786,
+      "step": 2137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000632490751098331,
+      "loss": 1.523,
+      "step": 2138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006321749741466406,
+      "loss": 1.4761,
+      "step": 2139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006318591405045495,
+      "loss": 1.4787,
+      "step": 2140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006315432503075201,
+      "loss": 1.4561,
+      "step": 2141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006312273036910389,
+      "loss": 1.4622,
+      "step": 2142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006309113007906169,
+      "loss": 1.5139,
+      "step": 2143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006305952417417889,
+      "loss": 1.5118,
+      "step": 2144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000630279126680114,
+      "loss": 1.5093,
+      "step": 2145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006299629557411752,
+      "loss": 1.4873,
+      "step": 2146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006296467290605797,
+      "loss": 1.4973,
+      "step": 2147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006293304467739584,
+      "loss": 1.4432,
+      "step": 2148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000629014109016966,
+      "loss": 1.4995,
+      "step": 2149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006286977159252812,
+      "loss": 1.4661,
+      "step": 2150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006283812676346063,
+      "loss": 1.4973,
+      "step": 2151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006280647642806673,
+      "loss": 1.4576,
+      "step": 2152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006277482059992138,
+      "loss": 1.4454,
+      "step": 2153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000627431592926019,
+      "loss": 1.5093,
+      "step": 2154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006271149251968793,
+      "loss": 1.5046,
+      "step": 2155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006267982029476152,
+      "loss": 1.4705,
+      "step": 2156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264814263140701,
+      "loss": 1.4873,
+      "step": 2157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006261645954321109,
+      "loss": 1.4665,
+      "step": 2158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006258477104376276,
+      "loss": 1.4685,
+      "step": 2159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006255307714665334,
+      "loss": 1.4477,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006252137786547647,
+      "loss": 1.4889,
+      "step": 2161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006248967321382814,
+      "loss": 1.5146,
+      "step": 2162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006245796320530659,
+      "loss": 1.4806,
+      "step": 2163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006242624785351235,
+      "loss": 1.4712,
+      "step": 2164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006239452717204831,
+      "loss": 1.516,
+      "step": 2165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006236280117451958,
+      "loss": 1.5098,
+      "step": 2166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006233106987453358,
+      "loss": 1.5054,
+      "step": 2167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006229933328569998,
+      "loss": 1.4812,
+      "step": 2168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006226759142163077,
+      "loss": 1.4645,
+      "step": 2169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006223584429594014,
+      "loss": 1.4944,
+      "step": 2170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006220409192224458,
+      "loss": 1.493,
+      "step": 2171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006217233431416279,
+      "loss": 1.4598,
+      "step": 2172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006214057148531578,
+      "loss": 1.4952,
+      "step": 2173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006210880344932674,
+      "loss": 1.4892,
+      "step": 2174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006207703021982113,
+      "loss": 1.4792,
+      "step": 2175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006204525181042659,
+      "loss": 1.466,
+      "step": 2176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006201346823477303,
+      "loss": 1.5182,
+      "step": 2177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006198167950649258,
+      "loss": 1.4658,
+      "step": 2178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006194988563921952,
+      "loss": 1.4998,
+      "step": 2179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000619180866465904,
+      "loss": 1.537,
+      "step": 2180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006188628254224394,
+      "loss": 1.5147,
+      "step": 2181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006185447333982103,
+      "loss": 1.4781,
+      "step": 2182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006182265905296479,
+      "loss": 1.4629,
+      "step": 2183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006179083969532051,
+      "loss": 1.4629,
+      "step": 2184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006175901528053564,
+      "loss": 1.4557,
+      "step": 2185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006172718582225978,
+      "loss": 1.5023,
+      "step": 2186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006169535133414473,
+      "loss": 1.4893,
+      "step": 2187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006166351182984446,
+      "loss": 1.4367,
+      "step": 2188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006163166732301506,
+      "loss": 1.5071,
+      "step": 2189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006159981782731474,
+      "loss": 1.4573,
+      "step": 2190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156796335640391,
+      "loss": 1.5025,
+      "step": 2191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006153610392394508,
+      "loss": 1.4738,
+      "step": 2192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006150423954360292,
+      "loss": 1.4851,
+      "step": 2193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006147237022904415,
+      "loss": 1.5088,
+      "step": 2194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006144049599393766,
+      "loss": 1.4568,
+      "step": 2195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006140861685195449,
+      "loss": 1.491,
+      "step": 2196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006137673281676768,
+      "loss": 1.4776,
+      "step": 2197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006134484390205246,
+      "loss": 1.4586,
+      "step": 2198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006131295012148612,
+      "loss": 1.4581,
+      "step": 2199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006128105148874801,
+      "loss": 1.5126,
+      "step": 2200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006124914801751961,
+      "loss": 1.5297,
+      "step": 2201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121723972148444,
+      "loss": 1.4962,
+      "step": 2202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006118532661432812,
+      "loss": 1.4809,
+      "step": 2203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006115340870973829,
+      "loss": 1.4883,
+      "step": 2204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000611214860214047,
+      "loss": 1.4802,
+      "step": 2205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006108955856301911,
+      "loss": 1.5032,
+      "step": 2206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006105762634827535,
+      "loss": 1.5322,
+      "step": 2207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006102568939086926,
+      "loss": 1.4784,
+      "step": 2208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099374770449876,
+      "loss": 1.505,
+      "step": 2209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006096180130286376,
+      "loss": 1.5235,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000609298501996662,
+      "loss": 1.4753,
+      "step": 2211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006089789440861006,
+      "loss": 1.4751,
+      "step": 2212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000608659339434013,
+      "loss": 1.4906,
+      "step": 2213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006083396881774789,
+      "loss": 1.4797,
+      "step": 2214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080199904535984,
+      "loss": 1.4604,
+      "step": 2215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006077002463994907,
+      "loss": 1.4976,
+      "step": 2216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006073804561522958,
+      "loss": 1.482,
+      "step": 2217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006070606198491727,
+      "loss": 1.4752,
+      "step": 2218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006067407376273009,
+      "loss": 1.4541,
+      "step": 2219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006064208096238791,
+      "loss": 1.4693,
+      "step": 2220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061008359761256,
+      "loss": 1.4517,
+      "step": 2221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006057808168212787,
+      "loss": 1.502,
+      "step": 2222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006054607522965958,
+      "loss": 1.4661,
+      "step": 2223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000605140642539354,
+      "loss": 1.4449,
+      "step": 2224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006048204876868495,
+      "loss": 1.4816,
+      "step": 2225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006045002878763983,
+      "loss": 1.5061,
+      "step": 2226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006041800432453353,
+      "loss": 1.4824,
+      "step": 2227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006038597539310148,
+      "loss": 1.4917,
+      "step": 2228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006035394200708104,
+      "loss": 1.4881,
+      "step": 2229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006032190418021145,
+      "loss": 1.4418,
+      "step": 2230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006028986192623386,
+      "loss": 1.4587,
+      "step": 2231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006025781525889133,
+      "loss": 1.482,
+      "step": 2232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006022576419192882,
+      "loss": 1.4499,
+      "step": 2233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006019370873909315,
+      "loss": 1.4934,
+      "step": 2234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006016164891413305,
+      "loss": 1.4862,
+      "step": 2235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006012958473079914,
+      "loss": 1.4586,
+      "step": 2236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006009751620284383,
+      "loss": 1.4931,
+      "step": 2237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006006544334402148,
+      "loss": 1.4458,
+      "step": 2238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006003336616808827,
+      "loss": 1.4916,
+      "step": 2239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006000128468880223,
+      "loss": 1.5023,
+      "step": 2240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005996919891992323,
+      "loss": 1.4431,
+      "step": 2241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005993710887521302,
+      "loss": 1.4909,
+      "step": 2242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005990501456843513,
+      "loss": 1.5081,
+      "step": 2243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005987291601335494,
+      "loss": 1.5,
+      "step": 2244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005984081322373968,
+      "loss": 1.5022,
+      "step": 2245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005980870621335835,
+      "loss": 1.4659,
+      "step": 2246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005977659499598178,
+      "loss": 1.4731,
+      "step": 2247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005974447958538262,
+      "loss": 1.5209,
+      "step": 2248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005971235999533531,
+      "loss": 1.4772,
+      "step": 2249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005968023623961605,
+      "loss": 1.4759,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005964810833200287,
+      "loss": 1.4641,
+      "step": 2251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005961597628627557,
+      "loss": 1.4788,
+      "step": 2252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000595838401162157,
+      "loss": 1.4798,
+      "step": 2253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005955169983560663,
+      "loss": 1.4716,
+      "step": 2254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005951955545823342,
+      "loss": 1.5126,
+      "step": 2255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005948740699788295,
+      "loss": 1.508,
+      "step": 2256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005945525446834386,
+      "loss": 1.4785,
+      "step": 2257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005942309788340644,
+      "loss": 1.4711,
+      "step": 2258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005939093725686282,
+      "loss": 1.4653,
+      "step": 2259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005935877260250684,
+      "loss": 1.4995,
+      "step": 2260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005932660393413403,
+      "loss": 1.4584,
+      "step": 2261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005929443126554168,
+      "loss": 1.499,
+      "step": 2262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926225461052877,
+      "loss": 1.4612,
+      "step": 2263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005923007398289603,
+      "loss": 1.4702,
+      "step": 2264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005919788939644583,
+      "loss": 1.4907,
+      "step": 2265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000591657008649823,
+      "loss": 1.4908,
+      "step": 2266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005913350840231124,
+      "loss": 1.4772,
+      "step": 2267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910131202224011,
+      "loss": 1.4913,
+      "step": 2268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005906911173857809,
+      "loss": 1.4795,
+      "step": 2269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00059036907565136,
+      "loss": 1.4646,
+      "step": 2270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005900469951572636,
+      "loss": 1.5158,
+      "step": 2271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005897248760416334,
+      "loss": 1.4538,
+      "step": 2272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005894027184426274,
+      "loss": 1.4706,
+      "step": 2273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890805224984205,
+      "loss": 1.4907,
+      "step": 2274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005887582883472041,
+      "loss": 1.492,
+      "step": 2275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005884360161271854,
+      "loss": 1.4883,
+      "step": 2276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005881137059765885,
+      "loss": 1.4851,
+      "step": 2277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005877913580336536,
+      "loss": 1.4964,
+      "step": 2278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587468972436637,
+      "loss": 1.4884,
+      "step": 2279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005871465493238112,
+      "loss": 1.4639,
+      "step": 2280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005868240888334653,
+      "loss": 1.414,
+      "step": 2281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005865015911039033,
+      "loss": 1.467,
+      "step": 2282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005861790562734459,
+      "loss": 1.473,
+      "step": 2283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005858564844804302,
+      "loss": 1.4665,
+      "step": 2284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000585533875863208,
+      "loss": 1.4412,
+      "step": 2285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005852112305601477,
+      "loss": 1.4683,
+      "step": 2286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005848885487096334,
+      "loss": 1.4898,
+      "step": 2287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005845658304500644,
+      "loss": 1.4672,
+      "step": 2288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000584243075919856,
+      "loss": 1.5013,
+      "step": 2289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005839202852574392,
+      "loss": 1.461,
+      "step": 2290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00058359745860126,
+      "loss": 1.4751,
+      "step": 2291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005832745960897802,
+      "loss": 1.4827,
+      "step": 2292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005829516978614769,
+      "loss": 1.4388,
+      "step": 2293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005826287640548425,
+      "loss": 1.4603,
+      "step": 2294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005823057948083847,
+      "loss": 1.4631,
+      "step": 2295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005819827902606261,
+      "loss": 1.4738,
+      "step": 2296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005816597505501052,
+      "loss": 1.452,
+      "step": 2297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005813366758153746,
+      "loss": 1.4729,
+      "step": 2298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005810135661950028,
+      "loss": 1.4497,
+      "step": 2299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005806904218275727,
+      "loss": 1.4466,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000580367242851682,
+      "loss": 1.507,
+      "step": 2301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000580044029405944,
+      "loss": 1.4821,
+      "step": 2302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005797207816289861,
+      "loss": 1.4658,
+      "step": 2303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005793974996594506,
+      "loss": 1.4578,
+      "step": 2304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005790741836359944,
+      "loss": 1.4916,
+      "step": 2305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005787508336972893,
+      "loss": 1.4797,
+      "step": 2306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005784274499820213,
+      "loss": 1.4515,
+      "step": 2307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005781040326288911,
+      "loss": 1.4918,
+      "step": 2308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005777805817766137,
+      "loss": 1.5152,
+      "step": 2309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005774570975639186,
+      "loss": 1.4825,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005771335801295495,
+      "loss": 1.4703,
+      "step": 2311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005768100296122644,
+      "loss": 1.4708,
+      "step": 2312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005764864461508354,
+      "loss": 1.4849,
+      "step": 2313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000576162829884049,
+      "loss": 1.4758,
+      "step": 2314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005758391809507055,
+      "loss": 1.4801,
+      "step": 2315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000575515499489619,
+      "loss": 1.4771,
+      "step": 2316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005751917856396181,
+      "loss": 1.4812,
+      "step": 2317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005748680395395451,
+      "loss": 1.4749,
+      "step": 2318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005745442613282559,
+      "loss": 1.4489,
+      "step": 2319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005742204511446203,
+      "loss": 1.4428,
+      "step": 2320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000573896609127522,
+      "loss": 1.4609,
+      "step": 2321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000573572735415858,
+      "loss": 1.4881,
+      "step": 2322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005732488301485395,
+      "loss": 1.4685,
+      "step": 2323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005729248934644903,
+      "loss": 1.444,
+      "step": 2324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005726009255026484,
+      "loss": 1.4503,
+      "step": 2325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005722769264019652,
+      "loss": 1.472,
+      "step": 2326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005719528963014052,
+      "loss": 1.4877,
+      "step": 2327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005716288353399461,
+      "loss": 1.4765,
+      "step": 2328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005713047436565792,
+      "loss": 1.5134,
+      "step": 2329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005709806213903086,
+      "loss": 1.4694,
+      "step": 2330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005706564686801519,
+      "loss": 1.4911,
+      "step": 2331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005703322856651393,
+      "loss": 1.4608,
+      "step": 2332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005700080724843147,
+      "loss": 1.4565,
+      "step": 2333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000569683829276734,
+      "loss": 1.4675,
+      "step": 2334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005693595561814665,
+      "loss": 1.4686,
+      "step": 2335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005690352533375948,
+      "loss": 1.4446,
+      "step": 2336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005687109208842131,
+      "loss": 1.4986,
+      "step": 2337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683865589604294,
+      "loss": 1.4799,
+      "step": 2338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005680621677053637,
+      "loss": 1.5109,
+      "step": 2339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005677377472581488,
+      "loss": 1.4754,
+      "step": 2340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005674132977579302,
+      "loss": 1.4828,
+      "step": 2341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005670888193438653,
+      "loss": 1.478,
+      "step": 2342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005667643121551244,
+      "loss": 1.4238,
+      "step": 2343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005664397763308902,
+      "loss": 1.5151,
+      "step": 2344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005661152120103573,
+      "loss": 1.4779,
+      "step": 2345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005657906193327325,
+      "loss": 1.4812,
+      "step": 2346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005654659984372351,
+      "loss": 1.4193,
+      "step": 2347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005651413494630966,
+      "loss": 1.4835,
+      "step": 2348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00056481667254956,
+      "loss": 1.4801,
+      "step": 2349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005644919678358806,
+      "loss": 1.4704,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000564167235461326,
+      "loss": 1.4643,
+      "step": 2351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005638424755651748,
+      "loss": 1.4782,
+      "step": 2352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000563517688286718,
+      "loss": 1.4826,
+      "step": 2353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005631928737652584,
+      "loss": 1.4616,
+      "step": 2354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005628680321401101,
+      "loss": 1.4819,
+      "step": 2355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005625431635505993,
+      "loss": 1.4482,
+      "step": 2356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005622182681360633,
+      "loss": 1.4435,
+      "step": 2357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005618933460358513,
+      "loss": 1.4667,
+      "step": 2358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005615683973893234,
+      "loss": 1.4311,
+      "step": 2359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005612434223358518,
+      "loss": 1.4698,
+      "step": 2360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005609184210148195,
+      "loss": 1.4366,
+      "step": 2361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005605933935656208,
+      "loss": 1.4962,
+      "step": 2362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005602683401276614,
+      "loss": 1.4849,
+      "step": 2363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005599432608403586,
+      "loss": 1.4696,
+      "step": 2364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005596181558431394,
+      "loss": 1.4673,
+      "step": 2365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005592930252754432,
+      "loss": 1.4628,
+      "step": 2366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005589678692767199,
+      "loss": 1.4496,
+      "step": 2367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005586426879864301,
+      "loss": 1.5075,
+      "step": 2368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005583174815440454,
+      "loss": 1.462,
+      "step": 2369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005579922500890483,
+      "loss": 1.4617,
+      "step": 2370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000557666993760932,
+      "loss": 1.4323,
+      "step": 2371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005573417126992003,
+      "loss": 1.4968,
+      "step": 2372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005570164070433675,
+      "loss": 1.4511,
+      "step": 2373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005566910769329584,
+      "loss": 1.4786,
+      "step": 2374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005563657225075088,
+      "loss": 1.4995,
+      "step": 2375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005560403439065643,
+      "loss": 1.484,
+      "step": 2376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005557149412696814,
+      "loss": 1.4749,
+      "step": 2377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005553895147364264,
+      "loss": 1.439,
+      "step": 2378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005550640644463765,
+      "loss": 1.4483,
+      "step": 2379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005547385905391181,
+      "loss": 1.4641,
+      "step": 2380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005544130931542488,
+      "loss": 1.4645,
+      "step": 2381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005540875724313756,
+      "loss": 1.4622,
+      "step": 2382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005537620285101158,
+      "loss": 1.4736,
+      "step": 2383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005534364615300964,
+      "loss": 1.527,
+      "step": 2384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005531108716309548,
+      "loss": 1.4752,
+      "step": 2385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005527852589523372,
+      "loss": 1.4828,
+      "step": 2386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005524596236339009,
+      "loss": 1.4671,
+      "step": 2387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552133965815312,
+      "loss": 1.4569,
+      "step": 2388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005518082856362464,
+      "loss": 1.4841,
+      "step": 2389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005514825832363899,
+      "loss": 1.4556,
+      "step": 2390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005511568587554375,
+      "loss": 1.4513,
+      "step": 2391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005508311123330939,
+      "loss": 1.4598,
+      "step": 2392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005505053441090729,
+      "loss": 1.4941,
+      "step": 2393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005501795542230982,
+      "loss": 1.4966,
+      "step": 2394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000549853742814902,
+      "loss": 1.4872,
+      "step": 2395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005495279100242266,
+      "loss": 1.4361,
+      "step": 2396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005492020559908229,
+      "loss": 1.4818,
+      "step": 2397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005488761808544509,
+      "loss": 1.4588,
+      "step": 2398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00054855028475488,
+      "loss": 1.4529,
+      "step": 2399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005482243678318885,
+      "loss": 1.4212,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005478984302252632,
+      "loss": 1.5158,
+      "step": 2401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005475724720748002,
+      "loss": 1.4631,
+      "step": 2402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005472464935203045,
+      "loss": 1.451,
+      "step": 2403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005469204947015897,
+      "loss": 1.4778,
+      "step": 2404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005465944757584777,
+      "loss": 1.4522,
+      "step": 2405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005462684368308,
+      "loss": 1.4756,
+      "step": 2406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005459423780583955,
+      "loss": 1.5033,
+      "step": 2407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005456162995811123,
+      "loss": 1.4226,
+      "step": 2408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005452902015388069,
+      "loss": 1.4733,
+      "step": 2409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005449640840713443,
+      "loss": 1.4768,
+      "step": 2410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005446379473185972,
+      "loss": 1.4935,
+      "step": 2411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005443117914204475,
+      "loss": 1.4805,
+      "step": 2412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005439856165167844,
+      "loss": 1.4409,
+      "step": 2413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005436594227475061,
+      "loss": 1.5164,
+      "step": 2414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005433332102525182,
+      "loss": 1.4887,
+      "step": 2415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005430069791717346,
+      "loss": 1.4557,
+      "step": 2416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005426807296450776,
+      "loss": 1.4333,
+      "step": 2417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423544618124763,
+      "loss": 1.4762,
+      "step": 2418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005420281758138687,
+      "loss": 1.4845,
+      "step": 2419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005417018717892004,
+      "loss": 1.4615,
+      "step": 2420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005413755498784243,
+      "loss": 1.4506,
+      "step": 2421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005410492102215015,
+      "loss": 1.4753,
+      "step": 2422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005407228529584,
+      "loss": 1.4603,
+      "step": 2423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005403964782290963,
+      "loss": 1.5056,
+      "step": 2424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005400700861735735,
+      "loss": 1.4237,
+      "step": 2425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005397436769318229,
+      "loss": 1.4349,
+      "step": 2426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005394172506438423,
+      "loss": 1.4695,
+      "step": 2427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005390908074496377,
+      "loss": 1.4691,
+      "step": 2428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000538764347489222,
+      "loss": 1.4679,
+      "step": 2429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005384378709026149,
+      "loss": 1.4747,
+      "step": 2430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005381113778298439,
+      "loss": 1.5051,
+      "step": 2431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000537784868410943,
+      "loss": 1.4788,
+      "step": 2432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005374583427859535,
+      "loss": 1.4794,
+      "step": 2433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005371318010949237,
+      "loss": 1.5035,
+      "step": 2434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005368052434779088,
+      "loss": 1.4964,
+      "step": 2435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005364786700749705,
+      "loss": 1.4597,
+      "step": 2436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005361520810261779,
+      "loss": 1.4464,
+      "step": 2437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005358254764716059,
+      "loss": 1.4867,
+      "step": 2438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005354988565513367,
+      "loss": 1.5204,
+      "step": 2439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005351722214054592,
+      "loss": 1.4233,
+      "step": 2440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005348455711740684,
+      "loss": 1.49,
+      "step": 2441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005345189059972659,
+      "loss": 1.452,
+      "step": 2442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005341922260151599,
+      "loss": 1.4779,
+      "step": 2443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005338655313678649,
+      "loss": 1.4778,
+      "step": 2444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005335388221955012,
+      "loss": 1.4743,
+      "step": 2445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005332120986381962,
+      "loss": 1.4969,
+      "step": 2446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005328853608360828,
+      "loss": 1.443,
+      "step": 2447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005325586089293,
+      "loss": 1.4811,
+      "step": 2448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005322318430579934,
+      "loss": 1.4557,
+      "step": 2449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005319050633623141,
+      "loss": 1.452,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005315782699824194,
+      "loss": 1.4442,
+      "step": 2451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005312514630584724,
+      "loss": 1.5058,
+      "step": 2452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005309246427306417,
+      "loss": 1.5127,
+      "step": 2453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005305978091391019,
+      "loss": 1.4687,
+      "step": 2454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005302709624240336,
+      "loss": 1.4746,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005299441027256226,
+      "loss": 1.4854,
+      "step": 2456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005296172301840602,
+      "loss": 1.485,
+      "step": 2457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005292903449395437,
+      "loss": 1.4605,
+      "step": 2458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005289634471322752,
+      "loss": 1.4587,
+      "step": 2459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005286365369024629,
+      "loss": 1.4702,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005283096143903199,
+      "loss": 1.4413,
+      "step": 2461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005279826797360644,
+      "loss": 1.4886,
+      "step": 2462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005276557330799204,
+      "loss": 1.4494,
+      "step": 2463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005273287745621162,
+      "loss": 1.5099,
+      "step": 2464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005270018043228861,
+      "loss": 1.4713,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005266748225024689,
+      "loss": 1.4667,
+      "step": 2466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005263478292411085,
+      "loss": 1.4451,
+      "step": 2467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005260208246790537,
+      "loss": 1.5005,
+      "step": 2468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005256938089565579,
+      "loss": 1.4925,
+      "step": 2469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005253667822138799,
+      "loss": 1.4965,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005250397445912824,
+      "loss": 1.4843,
+      "step": 2471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005247126962290335,
+      "loss": 1.4751,
+      "step": 2472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005243856372674057,
+      "loss": 1.5007,
+      "step": 2473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005240585678466755,
+      "loss": 1.492,
+      "step": 2474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005237314881071248,
+      "loss": 1.4818,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005234043981890394,
+      "loss": 1.5116,
+      "step": 2476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005230772982327093,
+      "loss": 1.4386,
+      "step": 2477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005227501883784294,
+      "loss": 1.4553,
+      "step": 2478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005224230687664981,
+      "loss": 1.4506,
+      "step": 2479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005220959395372185,
+      "loss": 1.4589,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005217688008308979,
+      "loss": 1.476,
+      "step": 2481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005214416527878475,
+      "loss": 1.461,
+      "step": 2482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005211144955483821,
+      "loss": 1.5251,
+      "step": 2483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005207873292528211,
+      "loss": 1.5105,
+      "step": 2484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005204601540414872,
+      "loss": 1.467,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005201329700547076,
+      "loss": 1.4481,
+      "step": 2486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005198057774328127,
+      "loss": 1.4628,
+      "step": 2487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005194785763161367,
+      "loss": 1.4858,
+      "step": 2488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005191513668450177,
+      "loss": 1.4943,
+      "step": 2489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005188241491597971,
+      "loss": 1.4713,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00051849692340082,
+      "loss": 1.4867,
+      "step": 2491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005181696897084351,
+      "loss": 1.5134,
+      "step": 2492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005178424482229938,
+      "loss": 1.4525,
+      "step": 2493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005175151990848515,
+      "loss": 1.4404,
+      "step": 2494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005171879424343671,
+      "loss": 1.4776,
+      "step": 2495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000516860678411902,
+      "loss": 1.4772,
+      "step": 2496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005165334071578212,
+      "loss": 1.4767,
+      "step": 2497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005162061288124929,
+      "loss": 1.4569,
+      "step": 2498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005158788435162878,
+      "loss": 1.4425,
+      "step": 2499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005155515514095802,
+      "loss": 1.4719,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005152242526327471,
+      "loss": 1.4808,
+      "step": 2501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000514896947326168,
+      "loss": 1.4595,
+      "step": 2502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005145696356302257,
+      "loss": 1.4963,
+      "step": 2503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142423176853058,
+      "loss": 1.4459,
+      "step": 2504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000513914993631796,
+      "loss": 1.4748,
+      "step": 2505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005135876636100869,
+      "loss": 1.497,
+      "step": 2506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005132603277605722,
+      "loss": 1.4909,
+      "step": 2507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000512932986223647,
+      "loss": 1.454,
+      "step": 2508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005126056391397099,
+      "loss": 1.4566,
+      "step": 2509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005122782866491613,
+      "loss": 1.4888,
+      "step": 2510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005119509288924041,
+      "loss": 1.4769,
+      "step": 2511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005116235660098433,
+      "loss": 1.4448,
+      "step": 2512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005112961981418864,
+      "loss": 1.4931,
+      "step": 2513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005109688254289427,
+      "loss": 1.5049,
+      "step": 2514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005106414480114238,
+      "loss": 1.4509,
+      "step": 2515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005103140660297432,
+      "loss": 1.4548,
+      "step": 2516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005099866796243163,
+      "loss": 1.4572,
+      "step": 2517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005096592889355608,
+      "loss": 1.4758,
+      "step": 2518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005093318941038957,
+      "loss": 1.4585,
+      "step": 2519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005090044952697421,
+      "loss": 1.4739,
+      "step": 2520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005086770925735225,
+      "loss": 1.4769,
+      "step": 2521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005083496861556619,
+      "loss": 1.4633,
+      "step": 2522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005080222761565856,
+      "loss": 1.4518,
+      "step": 2523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005076948627167215,
+      "loss": 1.477,
+      "step": 2524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005073674459764985,
+      "loss": 1.4812,
+      "step": 2525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005070400260763469,
+      "loss": 1.4489,
+      "step": 2526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005067126031566987,
+      "loss": 1.498,
+      "step": 2527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000506385177357987,
+      "loss": 1.4996,
+      "step": 2528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005060577488206457,
+      "loss": 1.5003,
+      "step": 2529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005057303176851107,
+      "loss": 1.4947,
+      "step": 2530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005054028840918184,
+      "loss": 1.4468,
+      "step": 2531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005050754481812064,
+      "loss": 1.43,
+      "step": 2532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005047480100937133,
+      "loss": 1.462,
+      "step": 2533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000504420569969779,
+      "loss": 1.4341,
+      "step": 2534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005040931279498435,
+      "loss": 1.4488,
+      "step": 2535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005037656841743482,
+      "loss": 1.4819,
+      "step": 2536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005034382387837354,
+      "loss": 1.5048,
+      "step": 2537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005031107919184475,
+      "loss": 1.4659,
+      "step": 2538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005027833437189278,
+      "loss": 1.4751,
+      "step": 2539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005024558943256202,
+      "loss": 1.4883,
+      "step": 2540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005021284438789694,
+      "loss": 1.427,
+      "step": 2541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00050180099251942,
+      "loss": 1.4299,
+      "step": 2542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005014735403874174,
+      "loss": 1.4899,
+      "step": 2543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005011460876234072,
+      "loss": 1.5004,
+      "step": 2544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005008186343678352,
+      "loss": 1.4819,
+      "step": 2545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005004911807611475,
+      "loss": 1.5011,
+      "step": 2546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005001637269437903,
+      "loss": 1.471,
+      "step": 2547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004998362730562099,
+      "loss": 1.4206,
+      "step": 2548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004995088192388527,
+      "loss": 1.4835,
+      "step": 2549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499181365632165,
+      "loss": 1.4421,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004988539123765928,
+      "loss": 1.4679,
+      "step": 2551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004985264596125826,
+      "loss": 1.446,
+      "step": 2552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00049819900748058,
+      "loss": 1.4369,
+      "step": 2553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004978715561210307,
+      "loss": 1.4614,
+      "step": 2554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004975441056743799,
+      "loss": 1.4282,
+      "step": 2555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004972166562810723,
+      "loss": 1.4676,
+      "step": 2556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004968892080815527,
+      "loss": 1.4628,
+      "step": 2557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004965617612162647,
+      "loss": 1.4489,
+      "step": 2558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004962343158256517,
+      "loss": 1.4651,
+      "step": 2559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004959068720501568,
+      "loss": 1.4784,
+      "step": 2560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004955794300302213,
+      "loss": 1.4583,
+      "step": 2561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004952519899062868,
+      "loss": 1.469,
+      "step": 2562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004949245518187938,
+      "loss": 1.4629,
+      "step": 2563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004945971159081816,
+      "loss": 1.4817,
+      "step": 2564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004942696823148894,
+      "loss": 1.4877,
+      "step": 2565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004939422511793545,
+      "loss": 1.4634,
+      "step": 2566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004936148226420133,
+      "loss": 1.4698,
+      "step": 2567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004932873968433014,
+      "loss": 1.473,
+      "step": 2568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004929599739236532,
+      "loss": 1.4701,
+      "step": 2569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004926325540235017,
+      "loss": 1.4925,
+      "step": 2570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004923051372832786,
+      "loss": 1.5094,
+      "step": 2571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004919777238434146,
+      "loss": 1.4749,
+      "step": 2572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004916503138443384,
+      "loss": 1.452,
+      "step": 2573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004913229074264775,
+      "loss": 1.4815,
+      "step": 2574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004909955047302581,
+      "loss": 1.5039,
+      "step": 2575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004906681058961044,
+      "loss": 1.4883,
+      "step": 2576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004903407110644392,
+      "loss": 1.4902,
+      "step": 2577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004900133203756839,
+      "loss": 1.5079,
+      "step": 2578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000489685933970257,
+      "loss": 1.4465,
+      "step": 2579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004893585519885763,
+      "loss": 1.4776,
+      "step": 2580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004890311745710574,
+      "loss": 1.4803,
+      "step": 2581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004887038018581136,
+      "loss": 1.4642,
+      "step": 2582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004883764339901566,
+      "loss": 1.4698,
+      "step": 2583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004880490711075959,
+      "loss": 1.4888,
+      "step": 2584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004877217133508388,
+      "loss": 1.4536,
+      "step": 2585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004873943608602901,
+      "loss": 1.4745,
+      "step": 2586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000487067013776353,
+      "loss": 1.4906,
+      "step": 2587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048673967223942794,
+      "loss": 1.4708,
+      "step": 2588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000486412336389913,
+      "loss": 1.5103,
+      "step": 2589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000486085006368204,
+      "loss": 1.4746,
+      "step": 2590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004857576823146944,
+      "loss": 1.4665,
+      "step": 2591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048543036436977434,
+      "loss": 1.5329,
+      "step": 2592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004851030526738321,
+      "loss": 1.4483,
+      "step": 2593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000484775747367253,
+      "loss": 1.4604,
+      "step": 2594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048444844859041976,
+      "loss": 1.5035,
+      "step": 2595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004841211564837121,
+      "loss": 1.4805,
+      "step": 2596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004837938711875073,
+      "loss": 1.4722,
+      "step": 2597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004834665928421789,
+      "loss": 1.5175,
+      "step": 2598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004831393215880981,
+      "loss": 1.4742,
+      "step": 2599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048281205756563303,
+      "loss": 1.4737,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048248480091514853,
+      "loss": 1.4965,
+      "step": 2601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004821575517770064,
+      "loss": 1.458,
+      "step": 2602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004818303102915652,
+      "loss": 1.4559,
+      "step": 2603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048150307659918006,
+      "loss": 1.4451,
+      "step": 2604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048117585084020295,
+      "loss": 1.5044,
+      "step": 2605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048084863315498236,
+      "loss": 1.4676,
+      "step": 2606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048052142368386325,
+      "loss": 1.483,
+      "step": 2607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004801942225671873,
+      "loss": 1.444,
+      "step": 2608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004798670299452926,
+      "loss": 1.4732,
+      "step": 2609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004795398459585129,
+      "loss": 1.4906,
+      "step": 2610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004792126707471791,
+      "loss": 1.447,
+      "step": 2611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000478885504451618,
+      "loss": 1.4594,
+      "step": 2612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047855834721215255,
+      "loss": 1.462,
+      "step": 2613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000478231199169102,
+      "loss": 1.4426,
+      "step": 2614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047790406046278145,
+      "loss": 1.4237,
+      "step": 2615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000477576931233502,
+      "loss": 1.4704,
+      "step": 2616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004772498116215708,
+      "loss": 1.475,
+      "step": 2617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047692270176729067,
+      "loss": 1.4476,
+      "step": 2618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047659560181096067,
+      "loss": 1.4813,
+      "step": 2619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047626851189287516,
+      "loss": 1.4933,
+      "step": 2620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047594143215332464,
+      "loss": 1.4534,
+      "step": 2621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004756143627325946,
+      "loss": 1.4633,
+      "step": 2622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004752873037709666,
+      "loss": 1.4472,
+      "step": 2623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004749602554087177,
+      "loss": 1.4077,
+      "step": 2624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004746332177861202,
+      "loss": 1.4511,
+      "step": 2625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004743061910434421,
+      "loss": 1.4795,
+      "step": 2626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047397917532094653,
+      "loss": 1.4378,
+      "step": 2627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004736521707588916,
+      "loss": 1.4814,
+      "step": 2628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004733251774975312,
+      "loss": 1.4297,
+      "step": 2629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004729981956771139,
+      "loss": 1.5143,
+      "step": 2630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004726712254378839,
+      "loss": 1.4763,
+      "step": 2631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004723442669200798,
+      "loss": 1.4874,
+      "step": 2632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047201732026393574,
+      "loss": 1.4459,
+      "step": 2633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004716903856096803,
+      "loss": 1.4695,
+      "step": 2634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004713634630975372,
+      "loss": 1.4686,
+      "step": 2635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047103655286772483,
+      "loss": 1.4578,
+      "step": 2636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047070965506045644,
+      "loss": 1.4652,
+      "step": 2637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047038276981593987,
+      "loss": 1.4869,
+      "step": 2638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004700558972743777,
+      "loss": 1.4859,
+      "step": 2639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004697290375759666,
+      "loss": 1.5187,
+      "step": 2640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004694021908608982,
+      "loss": 1.4752,
+      "step": 2641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046907535726935847,
+      "loss": 1.4757,
+      "step": 2642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004687485369415277,
+      "loss": 1.4654,
+      "step": 2643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004684217300175805,
+      "loss": 1.4805,
+      "step": 2644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046809493663768577,
+      "loss": 1.4479,
+      "step": 2645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004677681569420066,
+      "loss": 1.4402,
+      "step": 2646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046744139107070005,
+      "loss": 1.472,
+      "step": 2647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046711463916391735,
+      "loss": 1.4782,
+      "step": 2648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004667879013618038,
+      "loss": 1.422,
+      "step": 2649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046646117780449876,
+      "loss": 1.5073,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046613446863213515,
+      "loss": 1.5035,
+      "step": 2651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004658077739848402,
+      "loss": 1.4838,
+      "step": 2652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004654810940027342,
+      "loss": 1.465,
+      "step": 2653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046515442882593175,
+      "loss": 1.4587,
+      "step": 2654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046482777859454085,
+      "loss": 1.4625,
+      "step": 2655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004645011434486633,
+      "loss": 1.4708,
+      "step": 2656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046417452352839417,
+      "loss": 1.4596,
+      "step": 2657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046384791897382237,
+      "loss": 1.4842,
+      "step": 2658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046352132992502957,
+      "loss": 1.4734,
+      "step": 2659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004631947565220913,
+      "loss": 1.4723,
+      "step": 2660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004628681989050763,
+      "loss": 1.4522,
+      "step": 2661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046254165721404655,
+      "loss": 1.4648,
+      "step": 2662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004622151315890571,
+      "loss": 1.4332,
+      "step": 2663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004618886221701563,
+      "loss": 1.4698,
+      "step": 2664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046156212909738523,
+      "loss": 1.4793,
+      "step": 2665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046123565251077815,
+      "loss": 1.4935,
+      "step": 2666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004609091925503623,
+      "loss": 1.466,
+      "step": 2667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004605827493561577,
+      "loss": 1.49,
+      "step": 2668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004602563230681772,
+      "loss": 1.4346,
+      "step": 2669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004599299138264267,
+      "loss": 1.4332,
+      "step": 2670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045960352177090393,
+      "loss": 1.4413,
+      "step": 2671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045927714704160005,
+      "loss": 1.4575,
+      "step": 2672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004589507897784987,
+      "loss": 1.4526,
+      "step": 2673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045862445012157573,
+      "loss": 1.4964,
+      "step": 2674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004582981282107996,
+      "loss": 1.4498,
+      "step": 2675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004579718241861314,
+      "loss": 1.4625,
+      "step": 2676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004576455381875238,
+      "loss": 1.4512,
+      "step": 2677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004573192703549226,
+      "loss": 1.4388,
+      "step": 2678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004569930208282653,
+      "loss": 1.473,
+      "step": 2679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045666678974748176,
+      "loss": 1.4887,
+      "step": 2680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004563405772524939,
+      "loss": 1.4767,
+      "step": 2681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004560143834832157,
+      "loss": 1.5001,
+      "step": 2682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004556882085795527,
+      "loss": 1.4793,
+      "step": 2683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004553620526814029,
+      "loss": 1.4638,
+      "step": 2684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004550359159286559,
+      "loss": 1.4492,
+      "step": 2685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004547097984611931,
+      "loss": 1.4662,
+      "step": 2686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045438370041888775,
+      "loss": 1.4367,
+      "step": 2687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004540576219416048,
+      "loss": 1.4429,
+      "step": 2688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004537315631692002,
+      "loss": 1.4673,
+      "step": 2689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004534055242415223,
+      "loss": 1.464,
+      "step": 2690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045307950529841043,
+      "loss": 1.4048,
+      "step": 2691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004527535064796955,
+      "loss": 1.5079,
+      "step": 2692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004524275279251998,
+      "loss": 1.4229,
+      "step": 2693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045210156977473705,
+      "loss": 1.4634,
+      "step": 2694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004517756321681117,
+      "loss": 1.4987,
+      "step": 2695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045144971524512006,
+      "loss": 1.4392,
+      "step": 2696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045112381914554913,
+      "loss": 1.4387,
+      "step": 2697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045079794400917716,
+      "loss": 1.4741,
+      "step": 2698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004504720899757734,
+      "loss": 1.438,
+      "step": 2699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004501462571850981,
+      "loss": 1.4504,
+      "step": 2700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000449820445776902,
+      "loss": 1.4355,
+      "step": 2701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004494946558909271,
+      "loss": 1.4748,
+      "step": 2702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044916888766690624,
+      "loss": 1.4729,
+      "step": 2703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004488431412445625,
+      "loss": 1.4335,
+      "step": 2704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004485174167636101,
+      "loss": 1.4585,
+      "step": 2705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004481917143637537,
+      "loss": 1.495,
+      "step": 2706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004478660341846881,
+      "loss": 1.4534,
+      "step": 2707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044754037636609916,
+      "loss": 1.4181,
+      "step": 2708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044721474104766277,
+      "loss": 1.4884,
+      "step": 2709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004468891283690454,
+      "loss": 1.5079,
+      "step": 2710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004465635384699037,
+      "loss": 1.4571,
+      "step": 2711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044623797148988424,
+      "loss": 1.5198,
+      "step": 2712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044591242756862446,
+      "loss": 1.4659,
+      "step": 2713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004455869068457513,
+      "loss": 1.4505,
+      "step": 2714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044526140946088194,
+      "loss": 1.4235,
+      "step": 2715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044493593555362365,
+      "loss": 1.4642,
+      "step": 2716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004446104852635735,
+      "loss": 1.4693,
+      "step": 2717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044428505873031856,
+      "loss": 1.5188,
+      "step": 2718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044395965609343583,
+      "loss": 1.4328,
+      "step": 2719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004436342774924914,
+      "loss": 1.4454,
+      "step": 2720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044330892306704175,
+      "loss": 1.4564,
+      "step": 2721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004429835929566327,
+      "loss": 1.4798,
+      "step": 2722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004426582873007998,
+      "loss": 1.4819,
+      "step": 2723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044233300623906796,
+      "loss": 1.4564,
+      "step": 2724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004420077499109517,
+      "loss": 1.4591,
+      "step": 2725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044168251845595464,
+      "loss": 1.4884,
+      "step": 2726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044135731201357,
+      "loss": 1.481,
+      "step": 2727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044103213072328013,
+      "loss": 1.4555,
+      "step": 2728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004407069747245568,
+      "loss": 1.4679,
+      "step": 2729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044038184415686057,
+      "loss": 1.4424,
+      "step": 2730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044005673915964167,
+      "loss": 1.4693,
+      "step": 2731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043973165987233853,
+      "loss": 1.4794,
+      "step": 2732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004394066064343793,
+      "loss": 1.4491,
+      "step": 2733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004390815789851806,
+      "loss": 1.4789,
+      "step": 2734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043875657766414823,
+      "loss": 1.487,
+      "step": 2735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043843160261067656,
+      "loss": 1.4742,
+      "step": 2736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004381066539641488,
+      "loss": 1.4939,
+      "step": 2737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004377817318639368,
+      "loss": 1.4682,
+      "step": 2738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004374568364494007,
+      "loss": 1.4578,
+      "step": 2739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004371319678598899,
+      "loss": 1.4386,
+      "step": 2740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004368071262347417,
+      "loss": 1.4526,
+      "step": 2741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043648231171328206,
+      "loss": 1.4506,
+      "step": 2742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043615752443482543,
+      "loss": 1.5004,
+      "step": 2743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043583276453867427,
+      "loss": 1.4676,
+      "step": 2744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043550803216411944,
+      "loss": 1.4537,
+      "step": 2745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043518332745044015,
+      "loss": 1.4359,
+      "step": 2746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004348586505369035,
+      "loss": 1.4542,
+      "step": 2747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043453400156276493,
+      "loss": 1.502,
+      "step": 2748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004342093806672678,
+      "loss": 1.4625,
+      "step": 2749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000433884787989643,
+      "loss": 1.4773,
+      "step": 2750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004335602236691099,
+      "loss": 1.4527,
+      "step": 2751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043323568784487553,
+      "loss": 1.459,
+      "step": 2752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004329111806561347,
+      "loss": 1.4442,
+      "step": 2753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043258670224206983,
+      "loss": 1.47,
+      "step": 2754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043226225274185115,
+      "loss": 1.4373,
+      "step": 2755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043193783229463634,
+      "loss": 1.4546,
+      "step": 2756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043161344103957066,
+      "loss": 1.4343,
+      "step": 2757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043128907911578696,
+      "loss": 1.4672,
+      "step": 2758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004309647466624054,
+      "loss": 1.4657,
+      "step": 2759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043064044381853345,
+      "loss": 1.4594,
+      "step": 2760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043031617072326634,
+      "loss": 1.451,
+      "step": 2761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042999192751568563,
+      "loss": 1.4173,
+      "step": 2762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042966771433486074,
+      "loss": 1.4419,
+      "step": 2763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004293435313198482,
+      "loss": 1.4703,
+      "step": 2764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042901937860969144,
+      "loss": 1.4805,
+      "step": 2765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004286952563434208,
+      "loss": 1.4793,
+      "step": 2766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042837116466005413,
+      "loss": 1.4161,
+      "step": 2767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000428047103698595,
+      "loss": 1.471,
+      "step": 2768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042772307359803486,
+      "loss": 1.4459,
+      "step": 2769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042739907449735167,
+      "loss": 1.485,
+      "step": 2770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042707510653550987,
+      "loss": 1.4649,
+      "step": 2771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042675116985146065,
+      "loss": 1.4932,
+      "step": 2772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004264272645841419,
+      "loss": 1.4557,
+      "step": 2773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004261033908724782,
+      "loss": 1.476,
+      "step": 2774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042577954885537986,
+      "loss": 1.4463,
+      "step": 2775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042545573867174425,
+      "loss": 1.4538,
+      "step": 2776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000425131960460455,
+      "loss": 1.4574,
+      "step": 2777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042480821436038186,
+      "loss": 1.5038,
+      "step": 2778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000424484500510381,
+      "loss": 1.482,
+      "step": 2779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004241608190492948,
+      "loss": 1.4749,
+      "step": 2780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004238371701159511,
+      "loss": 1.4771,
+      "step": 2781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004235135538491646,
+      "loss": 1.497,
+      "step": 2782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004231899703877356,
+      "loss": 1.3925,
+      "step": 2783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004228664198704505,
+      "loss": 1.4672,
+      "step": 2784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042254290243608136,
+      "loss": 1.4397,
+      "step": 2785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004222194182233863,
+      "loss": 1.4709,
+      "step": 2786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000421895967371109,
+      "loss": 1.4759,
+      "step": 2787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042157255001797875,
+      "loss": 1.4844,
+      "step": 2788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004212491663027107,
+      "loss": 1.4496,
+      "step": 2789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004209258163640056,
+      "loss": 1.4381,
+      "step": 2790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004206025003405494,
+      "loss": 1.4415,
+      "step": 2791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000420279218371014,
+      "loss": 1.4883,
+      "step": 2792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041995597059405604,
+      "loss": 1.4435,
+      "step": 2793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041963275714831797,
+      "loss": 1.4428,
+      "step": 2794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041930957817242745,
+      "loss": 1.4383,
+      "step": 2795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041898643380499723,
+      "loss": 1.449,
+      "step": 2796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041866332418462527,
+      "loss": 1.4859,
+      "step": 2797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000418340249449895,
+      "loss": 1.4688,
+      "step": 2798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041801720973937395,
+      "loss": 1.4591,
+      "step": 2799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004176942051916155,
+      "loss": 1.4966,
+      "step": 2800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004173712359451576,
+      "loss": 1.4742,
+      "step": 2801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004170483021385232,
+      "loss": 1.4549,
+      "step": 2802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041672540391021984,
+      "loss": 1.4188,
+      "step": 2803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041640254139874015,
+      "loss": 1.4819,
+      "step": 2804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004160797147425609,
+      "loss": 1.4558,
+      "step": 2805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000415756924080144,
+      "loss": 1.4554,
+      "step": 2806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004154341695499357,
+      "loss": 1.478,
+      "step": 2807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004151114512903667,
+      "loss": 1.4086,
+      "step": 2808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004147887694398523,
+      "loss": 1.4573,
+      "step": 2809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004144661241367922,
+      "loss": 1.4779,
+      "step": 2810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041414351551957,
+      "loss": 1.4742,
+      "step": 2811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004138209437265541,
+      "loss": 1.4296,
+      "step": 2812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041349840889609693,
+      "loss": 1.4733,
+      "step": 2813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041317591116653486,
+      "loss": 1.4364,
+      "step": 2814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041285345067618867,
+      "loss": 1.4801,
+      "step": 2815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004125310275633631,
+      "loss": 1.4698,
+      "step": 2816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004122086419663465,
+      "loss": 1.4873,
+      "step": 2817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041188629402341155,
+      "loss": 1.5089,
+      "step": 2818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004115639838728147,
+      "loss": 1.4308,
+      "step": 2819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000411241711652796,
+      "loss": 1.4832,
+      "step": 2820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041091947750157944,
+      "loss": 1.5058,
+      "step": 2821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041059728155737286,
+      "loss": 1.4616,
+      "step": 2822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004102751239583669,
+      "loss": 1.4633,
+      "step": 2823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004099530048427366,
+      "loss": 1.4391,
+      "step": 2824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040963092434864013,
+      "loss": 1.4863,
+      "step": 2825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004093088826142192,
+      "loss": 1.4958,
+      "step": 2826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040898687977759893,
+      "loss": 1.468,
+      "step": 2827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004086649159768878,
+      "loss": 1.437,
+      "step": 2828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004083429913501771,
+      "loss": 1.4269,
+      "step": 2829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004080211060355418,
+      "loss": 1.457,
+      "step": 2830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040769926017103985,
+      "loss": 1.422,
+      "step": 2831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040737745389471234,
+      "loss": 1.4336,
+      "step": 2832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004070556873445833,
+      "loss": 1.4824,
+      "step": 2833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004067339606586599,
+      "loss": 1.4835,
+      "step": 2834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004064122739749318,
+      "loss": 1.46,
+      "step": 2835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004060906274313719,
+      "loss": 1.4541,
+      "step": 2836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004057690211659357,
+      "loss": 1.424,
+      "step": 2837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040544745531656155,
+      "loss": 1.4395,
+      "step": 2838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004051259300211704,
+      "loss": 1.4081,
+      "step": 2839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040480444541766573,
+      "loss": 1.4719,
+      "step": 2840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040448300164393393,
+      "loss": 1.4848,
+      "step": 2841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004041615988378431,
+      "loss": 1.4598,
+      "step": 2842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004038402371372444,
+      "loss": 1.4631,
+      "step": 2843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004035189166799713,
+      "loss": 1.4442,
+      "step": 2844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004031976376038395,
+      "loss": 1.4716,
+      "step": 2845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040287640004664694,
+      "loss": 1.4647,
+      "step": 2846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004025552041461738,
+      "loss": 1.4514,
+      "step": 2847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004022340500401822,
+      "loss": 1.4695,
+      "step": 2848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040191293786641656,
+      "loss": 1.4502,
+      "step": 2849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004015918677626033,
+      "loss": 1.4327,
+      "step": 2850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004012708398664506,
+      "loss": 1.464,
+      "step": 2851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004009498543156487,
+      "loss": 1.4463,
+      "step": 2852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040062891124787,
+      "loss": 1.4584,
+      "step": 2853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040030801080076777,
+      "loss": 1.4804,
+      "step": 2854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003999871531119779,
+      "loss": 1.4891,
+      "step": 2855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039966633831911737,
+      "loss": 1.4705,
+      "step": 2856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003993455665597852,
+      "loss": 1.4584,
+      "step": 2857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003990248379715617,
+      "loss": 1.4343,
+      "step": 2858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003987041526920089,
+      "loss": 1.4622,
+      "step": 2859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039838351085866953,
+      "loss": 1.4763,
+      "step": 2860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039806291260906864,
+      "loss": 1.4553,
+      "step": 2861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000397742358080712,
+      "loss": 1.3943,
+      "step": 2862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039742184741108685,
+      "loss": 1.5135,
+      "step": 2863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003971013807376616,
+      "loss": 1.4438,
+      "step": 2864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039678095819788573,
+      "loss": 1.467,
+      "step": 2865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003964605799291897,
+      "loss": 1.5005,
+      "step": 2866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003961402460689852,
+      "loss": 1.4495,
+      "step": 2867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039581995675466475,
+      "loss": 1.4519,
+      "step": 2868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039549971212360173,
+      "loss": 1.4553,
+      "step": 2869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003951795123131505,
+      "loss": 1.4933,
+      "step": 2870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039485935746064627,
+      "loss": 1.485,
+      "step": 2871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039453924770340435,
+      "loss": 1.4854,
+      "step": 2872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003942191831787214,
+      "loss": 1.4445,
+      "step": 2873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039389916402387436,
+      "loss": 1.4596,
+      "step": 2874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003935791903761209,
+      "loss": 1.4864,
+      "step": 2875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039325926237269906,
+      "loss": 1.4235,
+      "step": 2876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003929393801508273,
+      "loss": 1.4989,
+      "step": 2877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039261954384770437,
+      "loss": 1.4498,
+      "step": 2878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039229975360050934,
+      "loss": 1.4385,
+      "step": 2879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039198000954640175,
+      "loss": 1.4764,
+      "step": 2880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039166031182252104,
+      "loss": 1.4513,
+      "step": 2881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039134066056598706,
+      "loss": 1.4341,
+      "step": 2882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039102105591389956,
+      "loss": 1.4566,
+      "step": 2883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003907014980033381,
+      "loss": 1.4446,
+      "step": 2884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003903819869713625,
+      "loss": 1.4153,
+      "step": 2885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003900625229550125,
+      "loss": 1.4675,
+      "step": 2886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003897431060913074,
+      "loss": 1.4471,
+      "step": 2887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038942373651724657,
+      "loss": 1.4671,
+      "step": 2888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038910441436980906,
+      "loss": 1.4648,
+      "step": 2889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038878513978595316,
+      "loss": 1.4828,
+      "step": 2890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038846591290261716,
+      "loss": 1.4456,
+      "step": 2891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038814673385671896,
+      "loss": 1.4744,
+      "step": 2892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038782760278515565,
+      "loss": 1.4592,
+      "step": 2893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038750851982480403,
+      "loss": 1.4461,
+      "step": 2894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003871894851125201,
+      "loss": 1.4561,
+      "step": 2895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038687049878513905,
+      "loss": 1.4325,
+      "step": 2896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038655156097947555,
+      "loss": 1.4685,
+      "step": 2897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038623267183232325,
+      "loss": 1.4623,
+      "step": 2898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003859138314804552,
+      "loss": 1.4805,
+      "step": 2899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003855950400606233,
+      "loss": 1.4781,
+      "step": 2900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038527629770955855,
+      "loss": 1.4591,
+      "step": 2901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000384957604563971,
+      "loss": 1.4885,
+      "step": 2902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003846389607605492,
+      "loss": 1.4338,
+      "step": 2903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003843203664359609,
+      "loss": 1.4131,
+      "step": 2904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038400182172685266,
+      "loss": 1.4518,
+      "step": 2905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003836833267698495,
+      "loss": 1.418,
+      "step": 2906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003833648817015554,
+      "loss": 1.4758,
+      "step": 2907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003830464866585527,
+      "loss": 1.4717,
+      "step": 2908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038272814177740236,
+      "loss": 1.4451,
+      "step": 2909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038240984719464376,
+      "loss": 1.4726,
+      "step": 2910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003820916030467949,
+      "loss": 1.4691,
+      "step": 2911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038177340947035207,
+      "loss": 1.4652,
+      "step": 2912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038145526660178963,
+      "loss": 1.489,
+      "step": 2913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038113717457756083,
+      "loss": 1.4793,
+      "step": 2914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003808191335340961,
+      "loss": 1.5374,
+      "step": 2915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038050114360780484,
+      "loss": 1.461,
+      "step": 2916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003801832049350743,
+      "loss": 1.4753,
+      "step": 2917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003798653176522696,
+      "loss": 1.4726,
+      "step": 2918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003795474818957341,
+      "loss": 1.4282,
+      "step": 2919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000379229697801789,
+      "loss": 1.4606,
+      "step": 2920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003789119655067327,
+      "loss": 1.4397,
+      "step": 2921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037859428514684223,
+      "loss": 1.4812,
+      "step": 2922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003782766568583721,
+      "loss": 1.4969,
+      "step": 2923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003779590807775544,
+      "loss": 1.4543,
+      "step": 2924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037764155704059866,
+      "loss": 1.4797,
+      "step": 2925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003773240857836925,
+      "loss": 1.458,
+      "step": 2926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037700666714300023,
+      "loss": 1.4402,
+      "step": 2927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037668930125466427,
+      "loss": 1.4929,
+      "step": 2928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037637198825480424,
+      "loss": 1.4483,
+      "step": 2929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037605472827951684,
+      "loss": 1.4358,
+      "step": 2930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003757375214648764,
+      "loss": 1.4877,
+      "step": 2931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003754203679469344,
+      "loss": 1.4441,
+      "step": 2932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003751032678617187,
+      "loss": 1.4222,
+      "step": 2933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003747862213452353,
+      "loss": 1.4845,
+      "step": 2934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003744692285334668,
+      "loss": 1.4437,
+      "step": 2935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003741522895623725,
+      "loss": 1.498,
+      "step": 2936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003738354045678891,
+      "loss": 1.4389,
+      "step": 2937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003735185736859299,
+      "loss": 1.486,
+      "step": 2938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003732017970523848,
+      "loss": 1.4795,
+      "step": 2939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003728850748031207,
+      "loss": 1.4805,
+      "step": 2940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037256840707398123,
+      "loss": 1.4049,
+      "step": 2941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003722517940007863,
+      "loss": 1.4315,
+      "step": 2942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003719352357193327,
+      "loss": 1.4638,
+      "step": 2943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003716187323653939,
+      "loss": 1.455,
+      "step": 2944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003713022840747189,
+      "loss": 1.4291,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003709858909830341,
+      "loss": 1.4713,
+      "step": 2946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003706695532260417,
+      "loss": 1.4346,
+      "step": 2947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003703532709394203,
+      "loss": 1.4862,
+      "step": 2948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003700370442588248,
+      "loss": 1.4495,
+      "step": 2949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003697208733198862,
+      "loss": 1.4676,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003694047582582113,
+      "loss": 1.4695,
+      "step": 2951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036908869920938326,
+      "loss": 1.4427,
+      "step": 2952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036877269630896113,
+      "loss": 1.4498,
+      "step": 2953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036845674969247994,
+      "loss": 1.4435,
+      "step": 2954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003681408594954505,
+      "loss": 1.4267,
+      "step": 2955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036782502585335956,
+      "loss": 1.4615,
+      "step": 2956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036750924890166926,
+      "loss": 1.4686,
+      "step": 2957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003671935287758177,
+      "loss": 1.479,
+      "step": 2958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003668778656112187,
+      "loss": 1.4655,
+      "step": 2959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036656225954326147,
+      "loss": 1.4442,
+      "step": 2960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003662467107073107,
+      "loss": 1.4799,
+      "step": 2961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003659312192387069,
+      "loss": 1.4577,
+      "step": 2962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003656157852727652,
+      "loss": 1.4428,
+      "step": 2963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036530040894477685,
+      "loss": 1.4944,
+      "step": 2964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000364985090390008,
+      "loss": 1.4748,
+      "step": 2965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003646698297437001,
+      "loss": 1.4851,
+      "step": 2966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036435462714106987,
+      "loss": 1.4469,
+      "step": 2967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003640394827173089,
+      "loss": 1.4371,
+      "step": 2968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003637243966075842,
+      "loss": 1.4357,
+      "step": 2969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003634093689470371,
+      "loss": 1.4293,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036309439987078455,
+      "loss": 1.4711,
+      "step": 2971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003627794895139181,
+      "loss": 1.414,
+      "step": 2972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003624646380115041,
+      "loss": 1.4393,
+      "step": 2973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036214984549858367,
+      "loss": 1.4671,
+      "step": 2974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036183511211017284,
+      "loss": 1.4536,
+      "step": 2975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036152043798126166,
+      "loss": 1.4648,
+      "step": 2976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003612058232468154,
+      "loss": 1.4456,
+      "step": 2977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003608912680417737,
+      "loss": 1.4149,
+      "step": 2978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003605767725010505,
+      "loss": 1.4637,
+      "step": 2979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036026233675953435,
+      "loss": 1.4926,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035994796095208825,
+      "loss": 1.4857,
+      "step": 2981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003596336452135488,
+      "loss": 1.4546,
+      "step": 2982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003593193896787277,
+      "loss": 1.4797,
+      "step": 2983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035900519448241023,
+      "loss": 1.4481,
+      "step": 2984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003586910597593563,
+      "loss": 1.423,
+      "step": 2985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035837698564429945,
+      "loss": 1.4983,
+      "step": 2986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003580629722719475,
+      "loss": 1.5208,
+      "step": 2987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035774901977698195,
+      "loss": 1.4182,
+      "step": 2988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003574351282940584,
+      "loss": 1.4947,
+      "step": 2989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003571212979578062,
+      "loss": 1.4532,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035680752890282865,
+      "loss": 1.4463,
+      "step": 2991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003564938212637023,
+      "loss": 1.4426,
+      "step": 2992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035618017517497825,
+      "loss": 1.459,
+      "step": 2993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035586659077118,
+      "loss": 1.4209,
+      "step": 2994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003555530681868054,
+      "loss": 1.4675,
+      "step": 2995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003552396075563257,
+      "loss": 1.4382,
+      "step": 2996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003549262090141855,
+      "loss": 1.4495,
+      "step": 2997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035461287269480257,
+      "loss": 1.4424,
+      "step": 2998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035429959873256844,
+      "loss": 1.4648,
+      "step": 2999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003539863872618474,
+      "loss": 1.4383,
+      "step": 3000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035367323841697723,
+      "loss": 1.4621,
+      "step": 3001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003533601523322688,
+      "loss": 1.4341,
+      "step": 3002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000353047129142006,
+      "loss": 1.481,
+      "step": 3003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035273416898044585,
+      "loss": 1.4522,
+      "step": 3004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035242127198181853,
+      "loss": 1.4414,
+      "step": 3005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003521084382803262,
+      "loss": 1.4216,
+      "step": 3006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035179566801014515,
+      "loss": 1.4291,
+      "step": 3007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003514829613054236,
+      "loss": 1.4307,
+      "step": 3008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003511703183002827,
+      "loss": 1.5083,
+      "step": 3009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003508577391288166,
+      "loss": 1.4675,
+      "step": 3010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003505452239250918,
+      "loss": 1.4601,
+      "step": 3011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035023277282314713,
+      "loss": 1.4257,
+      "step": 3012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003499203859569943,
+      "loss": 1.3892,
+      "step": 3013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034960806346061735,
+      "loss": 1.4654,
+      "step": 3014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034929580546797277,
+      "loss": 1.443,
+      "step": 3015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034898361211298923,
+      "loss": 1.4636,
+      "step": 3016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003486714835295679,
+      "loss": 1.434,
+      "step": 3017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003483594198515818,
+      "loss": 1.4353,
+      "step": 3018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034804742121287657,
+      "loss": 1.4648,
+      "step": 3019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003477354877472697,
+      "loss": 1.4575,
+      "step": 3020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003474236195885507,
+      "loss": 1.4663,
+      "step": 3021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003471118168704811,
+      "loss": 1.4847,
+      "step": 3022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034680007972679475,
+      "loss": 1.4784,
+      "step": 3023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003464884082911964,
+      "loss": 1.4708,
+      "step": 3024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034617680269736367,
+      "loss": 1.4552,
+      "step": 3025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034586526307894534,
+      "loss": 1.4683,
+      "step": 3026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034555378956956215,
+      "loss": 1.4886,
+      "step": 3027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034524238230280633,
+      "loss": 1.4848,
+      "step": 3028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003449310414122417,
+      "loss": 1.4636,
+      "step": 3029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034461976703140394,
+      "loss": 1.4739,
+      "step": 3030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003443085592937997,
+      "loss": 1.446,
+      "step": 3031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034399741833290734,
+      "loss": 1.4528,
+      "step": 3032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003436863442821765,
+      "loss": 1.4325,
+      "step": 3033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003433753372750281,
+      "loss": 1.4762,
+      "step": 3034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003430643974448545,
+      "loss": 1.4404,
+      "step": 3035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003427535249250192,
+      "loss": 1.4459,
+      "step": 3036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003424427198488564,
+      "loss": 1.473,
+      "step": 3037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003421319823496718,
+      "loss": 1.4589,
+      "step": 3038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000341821312560742,
+      "loss": 1.4584,
+      "step": 3039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034151071061531475,
+      "loss": 1.4633,
+      "step": 3040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034120017664660836,
+      "loss": 1.4726,
+      "step": 3041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003408897107878124,
+      "loss": 1.4279,
+      "step": 3042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003405793131720866,
+      "loss": 1.4327,
+      "step": 3043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000340268983932562,
+      "loss": 1.494,
+      "step": 3044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003399587232023401,
+      "loss": 1.4651,
+      "step": 3045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000339648531114493,
+      "loss": 1.4729,
+      "step": 3046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003393384078020634,
+      "loss": 1.4677,
+      "step": 3047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003390283533980646,
+      "loss": 1.4587,
+      "step": 3048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003387183680354801,
+      "loss": 1.4394,
+      "step": 3049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033840845184726387,
+      "loss": 1.4648,
+      "step": 3050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003380986049663405,
+      "loss": 1.4591,
+      "step": 3051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033778882752560436,
+      "loss": 1.463,
+      "step": 3052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003374791196579204,
+      "loss": 1.458,
+      "step": 3053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003371694814961238,
+      "loss": 1.4379,
+      "step": 3054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033685991317301936,
+      "loss": 1.4732,
+      "step": 3055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033655041482138234,
+      "loss": 1.4451,
+      "step": 3056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033624098657395786,
+      "loss": 1.4435,
+      "step": 3057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003359316285634611,
+      "loss": 1.4361,
+      "step": 3058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033562234092257695,
+      "loss": 1.4386,
+      "step": 3059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033531312378396025,
+      "loss": 1.4321,
+      "step": 3060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033500397728023537,
+      "loss": 1.4622,
+      "step": 3061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003346949015439966,
+      "loss": 1.4607,
+      "step": 3062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003343858967078079,
+      "loss": 1.4264,
+      "step": 3063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033407696290420265,
+      "loss": 1.4426,
+      "step": 3064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033376810026568385,
+      "loss": 1.4741,
+      "step": 3065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003334593089247241,
+      "loss": 1.4675,
+      "step": 3066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033315058901376516,
+      "loss": 1.4716,
+      "step": 3067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033284194066521836,
+      "loss": 1.4562,
+      "step": 3068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033253336401146424,
+      "loss": 1.4234,
+      "step": 3069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033222485918485265,
+      "loss": 1.4483,
+      "step": 3070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003319164263177026,
+      "loss": 1.4471,
+      "step": 3071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033160806554230243,
+      "loss": 1.425,
+      "step": 3072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033129977699090896,
+      "loss": 1.4575,
+      "step": 3073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003309915607957487,
+      "loss": 1.4336,
+      "step": 3074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003306834170890168,
+      "loss": 1.4512,
+      "step": 3075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033037534600287744,
+      "loss": 1.4597,
+      "step": 3076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033006734766946355,
+      "loss": 1.4542,
+      "step": 3077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032975942222087713,
+      "loss": 1.5007,
+      "step": 3078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032945156978918835,
+      "loss": 1.481,
+      "step": 3079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003291437905064365,
+      "loss": 1.5101,
+      "step": 3080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003288360845046295,
+      "loss": 1.4624,
+      "step": 3081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003285284519157438,
+      "loss": 1.4124,
+      "step": 3082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003282208928717241,
+      "loss": 1.4976,
+      "step": 3083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032791340750448414,
+      "loss": 1.4444,
+      "step": 3084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032760599594590525,
+      "loss": 1.5009,
+      "step": 3085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003272986583278376,
+      "loss": 1.4708,
+      "step": 3086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003269913947820998,
+      "loss": 1.489,
+      "step": 3087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003266842054404783,
+      "loss": 1.4522,
+      "step": 3088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032637709043472805,
+      "loss": 1.4506,
+      "step": 3089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003260700498965721,
+      "loss": 1.4564,
+      "step": 3090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003257630839577009,
+      "loss": 1.3944,
+      "step": 3091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003254561927497738,
+      "loss": 1.461,
+      "step": 3092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003251493764044176,
+      "loss": 1.4673,
+      "step": 3093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032484263505322717,
+      "loss": 1.4424,
+      "step": 3094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003245359688277653,
+      "loss": 1.4256,
+      "step": 3095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003242293778595622,
+      "loss": 1.4842,
+      "step": 3096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032392286228011627,
+      "loss": 1.4316,
+      "step": 3097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032361642222089295,
+      "loss": 1.4808,
+      "step": 3098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000323310057813326,
+      "loss": 1.4435,
+      "step": 3099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032300376918881625,
+      "loss": 1.4487,
+      "step": 3100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032269755647873217,
+      "loss": 1.427,
+      "step": 3101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003223914198144098,
+      "loss": 1.4921,
+      "step": 3102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032208535932715245,
+      "loss": 1.455,
+      "step": 3103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003217793751482305,
+      "loss": 1.4258,
+      "step": 3104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032147346740888207,
+      "loss": 1.4731,
+      "step": 3105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003211676362403121,
+      "loss": 1.47,
+      "step": 3106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032086188177369305,
+      "loss": 1.4638,
+      "step": 3107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032055620414016414,
+      "loss": 1.4684,
+      "step": 3108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032025060347083196,
+      "loss": 1.4236,
+      "step": 3109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003199450798967697,
+      "loss": 1.4584,
+      "step": 3110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031963963354901786,
+      "loss": 1.4533,
+      "step": 3111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003193342645585836,
+      "loss": 1.4335,
+      "step": 3112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000319028973056441,
+      "loss": 1.4756,
+      "step": 3113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003187237591735308,
+      "loss": 1.4782,
+      "step": 3114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003184186230407608,
+      "loss": 1.4841,
+      "step": 3115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003181135647890047,
+      "loss": 1.502,
+      "step": 3116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003178085845491035,
+      "loss": 1.4719,
+      "step": 3117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031750368245186446,
+      "loss": 1.4458,
+      "step": 3118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003171988586280614,
+      "loss": 1.4611,
+      "step": 3119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031689411320843456,
+      "loss": 1.4645,
+      "step": 3120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003165894463236907,
+      "loss": 1.4212,
+      "step": 3121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031628485810450234,
+      "loss": 1.4457,
+      "step": 3122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003159803486815089,
+      "loss": 1.4223,
+      "step": 3123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003156759181853156,
+      "loss": 1.4044,
+      "step": 3124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031537156674649403,
+      "loss": 1.4679,
+      "step": 3125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003150672944955818,
+      "loss": 1.4517,
+      "step": 3126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031476310156308273,
+      "loss": 1.4481,
+      "step": 3127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031445898807946616,
+      "loss": 1.4671,
+      "step": 3128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003141549541751677,
+      "loss": 1.4588,
+      "step": 3129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031385099998058885,
+      "loss": 1.4381,
+      "step": 3130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003135471256260968,
+      "loss": 1.4612,
+      "step": 3131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031324333124202453,
+      "loss": 1.4554,
+      "step": 3132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031293961695867103,
+      "loss": 1.4859,
+      "step": 3133
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003126359829063001,
+      "loss": 1.4881,
+      "step": 3134
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003123324292151418,
+      "loss": 1.4874,
+      "step": 3135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031202895601539185,
+      "loss": 1.4629,
+      "step": 3136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003117255634372109,
+      "loss": 1.4721,
+      "step": 3137
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003114222516107255,
+      "loss": 1.4548,
+      "step": 3138
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031111902066602727,
+      "loss": 1.4839,
+      "step": 3139
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031081587073317316,
+      "loss": 1.4028,
+      "step": 3140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003105128019421855,
+      "loss": 1.4472,
+      "step": 3141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003102098144230519,
+      "loss": 1.452,
+      "step": 3142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003099069083057247,
+      "loss": 1.4547,
+      "step": 3143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030960408372012187,
+      "loss": 1.4549,
+      "step": 3144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003093013407961263,
+      "loss": 1.4603,
+      "step": 3145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003089986796635851,
+      "loss": 1.4665,
+      "step": 3146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003086961004523114,
+      "loss": 1.4379,
+      "step": 3147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030839360329208243,
+      "loss": 1.4675,
+      "step": 3148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003080911883126407,
+      "loss": 1.4774,
+      "step": 3149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003077888556436931,
+      "loss": 1.4615,
+      "step": 3150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003074866054149116,
+      "loss": 1.4516,
+      "step": 3151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003071844377559323,
+      "loss": 1.4496,
+      "step": 3152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003068823527963562,
+      "loss": 1.451,
+      "step": 3153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003065803506657491,
+      "loss": 1.44,
+      "step": 3154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003062784314936407,
+      "loss": 1.4505,
+      "step": 3155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030597659540952546,
+      "loss": 1.4738,
+      "step": 3156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030567484254286217,
+      "loss": 1.4453,
+      "step": 3157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030537317302307406,
+      "loss": 1.4066,
+      "step": 3158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030507158697954816,
+      "loss": 1.4521,
+      "step": 3159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003047700845416361,
+      "loss": 1.4327,
+      "step": 3160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003044686658386537,
+      "loss": 1.4396,
+      "step": 3161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003041673309998805,
+      "loss": 1.4461,
+      "step": 3162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003038660801545603,
+      "loss": 1.4488,
+      "step": 3163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030356491343190117,
+      "loss": 1.472,
+      "step": 3164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003032638309610742,
+      "loss": 1.4441,
+      "step": 3165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003029628328712154,
+      "loss": 1.4767,
+      "step": 3166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003026619192914238,
+      "loss": 1.4836,
+      "step": 3167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030236109035076275,
+      "loss": 1.4684,
+      "step": 3168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003020603461782589,
+      "loss": 1.4903,
+      "step": 3169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003017596869029028,
+      "loss": 1.4533,
+      "step": 3170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030145911265364836,
+      "loss": 1.4576,
+      "step": 3171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030115862355941314,
+      "loss": 1.4401,
+      "step": 3172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003008582197490782,
+      "loss": 1.4832,
+      "step": 3173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000300557901351488,
+      "loss": 1.456,
+      "step": 3174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003002576684954504,
+      "loss": 1.4483,
+      "step": 3175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029995752130973665,
+      "loss": 1.4729,
+      "step": 3176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002996574599230808,
+      "loss": 1.4439,
+      "step": 3177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002993574844641807,
+      "loss": 1.4466,
+      "step": 3178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029905759506169684,
+      "loss": 1.4384,
+      "step": 3179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029875779184425333,
+      "loss": 1.4816,
+      "step": 3180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029845807494043687,
+      "loss": 1.4099,
+      "step": 3181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002981584444787975,
+      "loss": 1.4341,
+      "step": 3182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002978589005878476,
+      "loss": 1.4306,
+      "step": 3183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029755944339606306,
+      "loss": 1.455,
+      "step": 3184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029726007303188223,
+      "loss": 1.4579,
+      "step": 3185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969607896237064,
+      "loss": 1.4347,
+      "step": 3186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002966615932998994,
+      "loss": 1.4843,
+      "step": 3187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029636248418878796,
+      "loss": 1.441,
+      "step": 3188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000296063462418661,
+      "loss": 1.4391,
+      "step": 3189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029576452811777034,
+      "loss": 1.4399,
+      "step": 3190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029546568141433004,
+      "loss": 1.5057,
+      "step": 3191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002951669224365169,
+      "loss": 1.4633,
+      "step": 3192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002948682513124697,
+      "loss": 1.4525,
+      "step": 3193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002945696681702901,
+      "loss": 1.4369,
+      "step": 3194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002942711731380411,
+      "loss": 1.4269,
+      "step": 3195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002939727663437488,
+      "loss": 1.4731,
+      "step": 3196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002936744479154011,
+      "loss": 1.4827,
+      "step": 3197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029337621798094805,
+      "loss": 1.4523,
+      "step": 3198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029307807666830165,
+      "loss": 1.4576,
+      "step": 3199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000292780024105336,
+      "loss": 1.44,
+      "step": 3200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029248206041988713,
+      "loss": 1.4691,
+      "step": 3201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002921841857397528,
+      "loss": 1.4535,
+      "step": 3202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002918864001926929,
+      "loss": 1.466,
+      "step": 3203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029158870390642866,
+      "loss": 1.4336,
+      "step": 3204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029129109700864336,
+      "loss": 1.4496,
+      "step": 3205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002909935796269819,
+      "loss": 1.4346,
+      "step": 3206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002906961518890507,
+      "loss": 1.4685,
+      "step": 3207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002903988139224177,
+      "loss": 1.4899,
+      "step": 3208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002901015658546126,
+      "loss": 1.459,
+      "step": 3209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028980440781312616,
+      "loss": 1.4478,
+      "step": 3210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002895073399254108,
+      "loss": 1.4977,
+      "step": 3211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892103623188803,
+      "loss": 1.4447,
+      "step": 3212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002889134751209092,
+      "loss": 1.4465,
+      "step": 3213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000288616678458834,
+      "loss": 1.4785,
+      "step": 3214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028831997245995185,
+      "loss": 1.4747,
+      "step": 3215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002880233572515213,
+      "loss": 1.5056,
+      "step": 3216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028772683296076197,
+      "loss": 1.4518,
+      "step": 3217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002874303997148543,
+      "loss": 1.4243,
+      "step": 3218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002871340576409396,
+      "loss": 1.4342,
+      "step": 3219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028683780686612027,
+      "loss": 1.4319,
+      "step": 3220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002865416475174596,
+      "loss": 1.4771,
+      "step": 3221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028624557972198154,
+      "loss": 1.421,
+      "step": 3222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028594960360667085,
+      "loss": 1.4532,
+      "step": 3223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028565371929847286,
+      "loss": 1.4659,
+      "step": 3224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028535792692429386,
+      "loss": 1.469,
+      "step": 3225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002850622266109999,
+      "loss": 1.4213,
+      "step": 3226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002847666184854185,
+      "loss": 1.4201,
+      "step": 3227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002844711026743371,
+      "loss": 1.3898,
+      "step": 3228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002841756793045036,
+      "loss": 1.4381,
+      "step": 3229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028388034850262647,
+      "loss": 1.4588,
+      "step": 3230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028358511039537447,
+      "loss": 1.4317,
+      "step": 3231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028328996510937603,
+      "loss": 1.4777,
+      "step": 3232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002829949127712205,
+      "loss": 1.4566,
+      "step": 3233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028269995350745694,
+      "loss": 1.4203,
+      "step": 3234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028240508744459465,
+      "loss": 1.485,
+      "step": 3235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000282110314709103,
+      "loss": 1.4334,
+      "step": 3236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002818156354274111,
+      "loss": 1.4285,
+      "step": 3237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002815210497259083,
+      "loss": 1.4413,
+      "step": 3238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028122655773094377,
+      "loss": 1.4364,
+      "step": 3239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002809321595688259,
+      "loss": 1.4782,
+      "step": 3240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028063785536582356,
+      "loss": 1.4395,
+      "step": 3241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002803436452481651,
+      "loss": 1.4693,
+      "step": 3242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028004952934203834,
+      "loss": 1.4611,
+      "step": 3243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027975550777359104,
+      "loss": 1.432,
+      "step": 3244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027946158066893004,
+      "loss": 1.4354,
+      "step": 3245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000279167748154122,
+      "loss": 1.474,
+      "step": 3246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027887401035519313,
+      "loss": 1.4873,
+      "step": 3247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002785803673981284,
+      "loss": 1.4422,
+      "step": 3248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027828681940887304,
+      "loss": 1.422,
+      "step": 3249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027799336651333046,
+      "loss": 1.3836,
+      "step": 3250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002777000088373641,
+      "loss": 1.4327,
+      "step": 3251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002774067465067962,
+      "loss": 1.4602,
+      "step": 3252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027711357964740836,
+      "loss": 1.4562,
+      "step": 3253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027682050838494086,
+      "loss": 1.4699,
+      "step": 3254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027652753284509367,
+      "loss": 1.4598,
+      "step": 3255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002762346531535246,
+      "loss": 1.4157,
+      "step": 3256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027594186943585126,
+      "loss": 1.4759,
+      "step": 3257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002756491818176498,
+      "loss": 1.4592,
+      "step": 3258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002753565904244552,
+      "loss": 1.4746,
+      "step": 3259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027506409538176113,
+      "loss": 1.4866,
+      "step": 3260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002747716968150201,
+      "loss": 1.4393,
+      "step": 3261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027447939484964267,
+      "loss": 1.4781,
+      "step": 3262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002741871896109986,
+      "loss": 1.451,
+      "step": 3263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002738950812244161,
+      "loss": 1.4238,
+      "step": 3264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027360306981518144,
+      "loss": 1.4552,
+      "step": 3265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002733111555085397,
+      "loss": 1.473,
+      "step": 3266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027301933842969415,
+      "loss": 1.439,
+      "step": 3267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027272761870380624,
+      "loss": 1.4364,
+      "step": 3268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027243599645599575,
+      "loss": 1.4539,
+      "step": 3269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027214447181134085,
+      "loss": 1.4804,
+      "step": 3270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002718530448948775,
+      "loss": 1.4502,
+      "step": 3271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002715617158316002,
+      "loss": 1.432,
+      "step": 3272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002712704847464609,
+      "loss": 1.4719,
+      "step": 3273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027097935176437,
+      "loss": 1.4244,
+      "step": 3274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002706883170101957,
+      "loss": 1.4562,
+      "step": 3275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002703973806087638,
+      "loss": 1.4443,
+      "step": 3276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027010654268485847,
+      "loss": 1.4457,
+      "step": 3277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026981580336322107,
+      "loss": 1.4375,
+      "step": 3278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002695251627685512,
+      "loss": 1.4681,
+      "step": 3279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002692346210255055,
+      "loss": 1.4649,
+      "step": 3280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026894417825869857,
+      "loss": 1.4606,
+      "step": 3281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026865383459270266,
+      "loss": 1.4423,
+      "step": 3282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002683635901520474,
+      "loss": 1.4286,
+      "step": 3283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002680734450612197,
+      "loss": 1.4499,
+      "step": 3284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002677833994446642,
+      "loss": 1.4456,
+      "step": 3285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026749345342678266,
+      "loss": 1.4689,
+      "step": 3286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026720360713193396,
+      "loss": 1.4548,
+      "step": 3287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002669138606844345,
+      "loss": 1.4571,
+      "step": 3288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026662421420855777,
+      "loss": 1.3784,
+      "step": 3289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026633466782853435,
+      "loss": 1.4261,
+      "step": 3290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026604522166855175,
+      "loss": 1.5217,
+      "step": 3291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002657558758527551,
+      "loss": 1.4627,
+      "step": 3292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026546663050524546,
+      "loss": 1.432,
+      "step": 3293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002651774857500816,
+      "loss": 1.4819,
+      "step": 3294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000264888441711279,
+      "loss": 1.4592,
+      "step": 3295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002645994985128098,
+      "loss": 1.4289,
+      "step": 3296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643106562786029,
+      "loss": 1.4329,
+      "step": 3297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002640219151325442,
+      "loss": 1.4367,
+      "step": 3298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002637332751984756,
+      "loss": 1.4537,
+      "step": 3299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002634447366001962,
+      "loss": 1.4377,
+      "step": 3300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002631562994614613,
+      "loss": 1.4444,
+      "step": 3301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002628679639059829,
+      "loss": 1.4325,
+      "step": 3302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026257973005742937,
+      "loss": 1.4916,
+      "step": 3303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026229159803942527,
+      "loss": 1.4686,
+      "step": 3304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026200356797555175,
+      "loss": 1.4777,
+      "step": 3305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026171563998934606,
+      "loss": 1.4238,
+      "step": 3306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026142781420430175,
+      "loss": 1.4482,
+      "step": 3307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026114009074386846,
+      "loss": 1.4561,
+      "step": 3308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026085246973145203,
+      "loss": 1.4228,
+      "step": 3309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026056495129041457,
+      "loss": 1.4062,
+      "step": 3310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002602775355440734,
+      "loss": 1.4344,
+      "step": 3311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002599902226157027,
+      "loss": 1.4236,
+      "step": 3312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000259703012628532,
+      "loss": 1.4265,
+      "step": 3313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025941590570574714,
+      "loss": 1.4328,
+      "step": 3314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002591289019704892,
+      "loss": 1.4544,
+      "step": 3315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025884200154585563,
+      "loss": 1.443,
+      "step": 3316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025855520455489884,
+      "loss": 1.4557,
+      "step": 3317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025826851112062746,
+      "loss": 1.4543,
+      "step": 3318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002579819213660054,
+      "loss": 1.4553,
+      "step": 3319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025769543541395224,
+      "loss": 1.4585,
+      "step": 3320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002574090533873431,
+      "loss": 1.4446,
+      "step": 3321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025712277540900863,
+      "loss": 1.4193,
+      "step": 3322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002568366016017342,
+      "loss": 1.4586,
+      "step": 3323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025655053208826107,
+      "loss": 1.4282,
+      "step": 3324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025626456699128587,
+      "loss": 1.4599,
+      "step": 3325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025597870643346,
+      "loss": 1.4661,
+      "step": 3326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002556929505373904,
+      "loss": 1.4658,
+      "step": 3327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002554072994256391,
+      "loss": 1.446,
+      "step": 3328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025512175322072274,
+      "loss": 1.4602,
+      "step": 3329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002548363120451134,
+      "loss": 1.4222,
+      "step": 3330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002545509760212381,
+      "loss": 1.4577,
+      "step": 3331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002542657452714785,
+      "loss": 1.42,
+      "step": 3332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025398061991817143,
+      "loss": 1.4592,
+      "step": 3333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025369560008360825,
+      "loss": 1.4961,
+      "step": 3334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002534106858900351,
+      "loss": 1.445,
+      "step": 3335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002531258774596531,
+      "loss": 1.4337,
+      "step": 3336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002528411749146176,
+      "loss": 1.4098,
+      "step": 3337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002525565783770387,
+      "loss": 1.4437,
+      "step": 3338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002522720879689811,
+      "loss": 1.496,
+      "step": 3339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025198770381246416,
+      "loss": 1.4451,
+      "step": 3340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000251703426029461,
+      "loss": 1.4614,
+      "step": 3341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025141925474189973,
+      "loss": 1.4743,
+      "step": 3342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025113519007166277,
+      "loss": 1.4431,
+      "step": 3343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002508512321405864,
+      "loss": 1.4262,
+      "step": 3344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002505673810704615,
+      "loss": 1.4416,
+      "step": 3345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025028363698303323,
+      "loss": 1.4573,
+      "step": 3346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 1.484,
+      "step": 3347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024971647024301546,
+      "loss": 1.4398,
+      "step": 3348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024943304783368647,
+      "loss": 1.4591,
+      "step": 3349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002491497328935741,
+      "loss": 1.4561,
+      "step": 3350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002488665255441934,
+      "loss": 1.4375,
+      "step": 3351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024858342590701303,
+      "loss": 1.4051,
+      "step": 3352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000248300434103456,
+      "loss": 1.4414,
+      "step": 3353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024801755025489813,
+      "loss": 1.4396,
+      "step": 3354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024773477448266983,
+      "loss": 1.4566,
+      "step": 3355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024745210690805473,
+      "loss": 1.4743,
+      "step": 3356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024716954765229015,
+      "loss": 1.4517,
+      "step": 3357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000246887096836567,
+      "loss": 1.4749,
+      "step": 3358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002466047545820297,
+      "loss": 1.426,
+      "step": 3359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002463225210097756,
+      "loss": 1.4571,
+      "step": 3360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002460403962408563,
+      "loss": 1.4777,
+      "step": 3361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002457583803962761,
+      "loss": 1.4142,
+      "step": 3362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002454764735969929,
+      "loss": 1.4539,
+      "step": 3363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024519467596391756,
+      "loss": 1.4534,
+      "step": 3364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002449129876179144,
+      "loss": 1.4384,
+      "step": 3365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024463140867980054,
+      "loss": 1.4267,
+      "step": 3366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024434993927034666,
+      "loss": 1.4826,
+      "step": 3367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024406857951027596,
+      "loss": 1.4699,
+      "step": 3368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024378732952026484,
+      "loss": 1.4342,
+      "step": 3369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024350618942094266,
+      "loss": 1.4328,
+      "step": 3370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024322515933289173,
+      "loss": 1.4528,
+      "step": 3371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024294423937664672,
+      "loss": 1.45,
+      "step": 3372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002426634296726955,
+      "loss": 1.4746,
+      "step": 3373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024238273034147862,
+      "loss": 1.4717,
+      "step": 3374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024210214150338904,
+      "loss": 1.4103,
+      "step": 3375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024182166327877265,
+      "loss": 1.4706,
+      "step": 3376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024154129578792783,
+      "loss": 1.4277,
+      "step": 3377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024126103915110504,
+      "loss": 1.438,
+      "step": 3378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024098089348850767,
+      "loss": 1.4722,
+      "step": 3379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024070085892029141,
+      "loss": 1.4324,
+      "step": 3380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024042093556656425,
+      "loss": 1.4406,
+      "step": 3381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024014112354738654,
+      "loss": 1.4615,
+      "step": 3382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023986142298277092,
+      "loss": 1.4605,
+      "step": 3383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002395818339926818,
+      "loss": 1.4705,
+      "step": 3384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023930235669703627,
+      "loss": 1.4756,
+      "step": 3385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023902299121570332,
+      "loss": 1.4056,
+      "step": 3386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000238743737668504,
+      "loss": 1.4455,
+      "step": 3387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023846459617521128,
+      "loss": 1.4062,
+      "step": 3388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023818556685555026,
+      "loss": 1.4491,
+      "step": 3389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023790664982919753,
+      "loss": 1.4281,
+      "step": 3390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023762784521578185,
+      "loss": 1.4612,
+      "step": 3391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023734915313488377,
+      "loss": 1.4316,
+      "step": 3392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023707057370603546,
+      "loss": 1.445,
+      "step": 3393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002367921070487208,
+      "loss": 1.4259,
+      "step": 3394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002365137532823753,
+      "loss": 1.4728,
+      "step": 3395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023623551252638608,
+      "loss": 1.4588,
+      "step": 3396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023595738490009177,
+      "loss": 1.4582,
+      "step": 3397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023567937052278244,
+      "loss": 1.4307,
+      "step": 3398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023540146951369967,
+      "loss": 1.4877,
+      "step": 3399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002351236819920363,
+      "loss": 1.438,
+      "step": 3400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023484600807693686,
+      "loss": 1.4706,
+      "step": 3401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023456844788749637,
+      "loss": 1.453,
+      "step": 3402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002342910015427618,
+      "loss": 1.4528,
+      "step": 3403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000234013669161731,
+      "loss": 1.5048,
+      "step": 3404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002337364508633531,
+      "loss": 1.446,
+      "step": 3405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023345934676652808,
+      "loss": 1.4577,
+      "step": 3406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002331823569901073,
+      "loss": 1.4598,
+      "step": 3407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023290548165289245,
+      "loss": 1.4464,
+      "step": 3408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023262872087363669,
+      "loss": 1.4919,
+      "step": 3409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023235207477104392,
+      "loss": 1.4548,
+      "step": 3410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023207554346376874,
+      "loss": 1.4315,
+      "step": 3411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023179912707041667,
+      "loss": 1.4712,
+      "step": 3412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002315228257095438,
+      "loss": 1.4305,
+      "step": 3413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002312466394996572,
+      "loss": 1.4278,
+      "step": 3414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023097056855921383,
+      "loss": 1.4415,
+      "step": 3415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023069461300662193,
+      "loss": 1.4184,
+      "step": 3416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023041877296023995,
+      "loss": 1.4704,
+      "step": 3417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023014304853837686,
+      "loss": 1.4591,
+      "step": 3418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022986743985929208,
+      "loss": 1.4872,
+      "step": 3419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002295919470411954,
+      "loss": 1.403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022931657020224656,
+      "loss": 1.4544,
+      "step": 3421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022904130946055595,
+      "loss": 1.4499,
+      "step": 3422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022876616493418416,
+      "loss": 1.4524,
+      "step": 3423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022849113674114176,
+      "loss": 1.4344,
+      "step": 3424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022821622499938948,
+      "loss": 1.4805,
+      "step": 3425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002279414298268382,
+      "loss": 1.4557,
+      "step": 3426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022766675134134863,
+      "loss": 1.4411,
+      "step": 3427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022739218966073154,
+      "loss": 1.4467,
+      "step": 3428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022711774490274768,
+      "loss": 1.425,
+      "step": 3429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022684341718510743,
+      "loss": 1.4077,
+      "step": 3430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022656920662547121,
+      "loss": 1.4856,
+      "step": 3431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002262951133414492,
+      "loss": 1.4669,
+      "step": 3432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022602113745060076,
+      "loss": 1.459,
+      "step": 3433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022574727907043556,
+      "loss": 1.4255,
+      "step": 3434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022547353831841262,
+      "loss": 1.4147,
+      "step": 3435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002251999153119404,
+      "loss": 1.4276,
+      "step": 3436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000224926410168377,
+      "loss": 1.4306,
+      "step": 3437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002246530230050301,
+      "loss": 1.4869,
+      "step": 3438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022437975393915628,
+      "loss": 1.4487,
+      "step": 3439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022410660308796193,
+      "loss": 1.4473,
+      "step": 3440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002238335705686026,
+      "loss": 1.4608,
+      "step": 3441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022356065649818308,
+      "loss": 1.4432,
+      "step": 3442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022328786099375737,
+      "loss": 1.4669,
+      "step": 3443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022301518417232875,
+      "loss": 1.4762,
+      "step": 3444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022274262615084916,
+      "loss": 1.4134,
+      "step": 3445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022247018704622001,
+      "loss": 1.4403,
+      "step": 3446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002221978669752916,
+      "loss": 1.4534,
+      "step": 3447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022192566605486313,
+      "loss": 1.424,
+      "step": 3448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002216535844016827,
+      "loss": 1.4717,
+      "step": 3449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022138162213244751,
+      "loss": 1.4626,
+      "step": 3450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002211097793638029,
+      "loss": 1.4846,
+      "step": 3451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002208380562123436,
+      "loss": 1.4523,
+      "step": 3452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022056645279461273,
+      "loss": 1.4262,
+      "step": 3453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022029496922710222,
+      "loss": 1.4349,
+      "step": 3454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022002360562625256,
+      "loss": 1.4449,
+      "step": 3455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021975236210845258,
+      "loss": 1.4347,
+      "step": 3456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021948123879003985,
+      "loss": 1.4183,
+      "step": 3457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021921023578730026,
+      "loss": 1.4395,
+      "step": 3458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021893935321646825,
+      "loss": 1.4682,
+      "step": 3459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021866859119372634,
+      "loss": 1.4787,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021839794983520555,
+      "loss": 1.4408,
+      "step": 3461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002181274292569853,
+      "loss": 1.4263,
+      "step": 3462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021785702957509268,
+      "loss": 1.4629,
+      "step": 3463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021758675090550328,
+      "loss": 1.4594,
+      "step": 3464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002173165933641409,
+      "loss": 1.4509,
+      "step": 3465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002170465570668772,
+      "loss": 1.451,
+      "step": 3466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021677664212953185,
+      "loss": 1.4845,
+      "step": 3467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021650684866787273,
+      "loss": 1.4688,
+      "step": 3468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021623717679761494,
+      "loss": 1.4793,
+      "step": 3469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021596762663442215,
+      "loss": 1.4703,
+      "step": 3470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021569819829390553,
+      "loss": 1.4673,
+      "step": 3471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021542889189162402,
+      "loss": 1.443,
+      "step": 3472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021515970754308424,
+      "loss": 1.4531,
+      "step": 3473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021489064536374064,
+      "loss": 1.4051,
+      "step": 3474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002146217054689949,
+      "loss": 1.4599,
+      "step": 3475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021435288797419643,
+      "loss": 1.482,
+      "step": 3476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021408419299464242,
+      "loss": 1.4415,
+      "step": 3477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021381562064557707,
+      "loss": 1.4464,
+      "step": 3478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021354717104219234,
+      "loss": 1.4276,
+      "step": 3479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002132788442996273,
+      "loss": 1.4557,
+      "step": 3480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002130106405329686,
+      "loss": 1.4707,
+      "step": 3481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002127425598572496,
+      "loss": 1.4424,
+      "step": 3482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021247460238745148,
+      "loss": 1.4431,
+      "step": 3483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021220676823850227,
+      "loss": 1.4333,
+      "step": 3484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002119390575252771,
+      "loss": 1.418,
+      "step": 3485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021167147036259832,
+      "loss": 1.4381,
+      "step": 3486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021140400686523508,
+      "loss": 1.4367,
+      "step": 3487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021113666714790374,
+      "loss": 1.4248,
+      "step": 3488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002108694513252673,
+      "loss": 1.4089,
+      "step": 3489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002106023595119358,
+      "loss": 1.4954,
+      "step": 3490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021033539182246602,
+      "loss": 1.4416,
+      "step": 3491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021006854837136153,
+      "loss": 1.4076,
+      "step": 3492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020980182927307278,
+      "loss": 1.4601,
+      "step": 3493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020953523464199643,
+      "loss": 1.4299,
+      "step": 3494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020926876459247606,
+      "loss": 1.428,
+      "step": 3495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020900241923880187,
+      "loss": 1.4507,
+      "step": 3496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020873619869521054,
+      "loss": 1.4631,
+      "step": 3497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020847010307588516,
+      "loss": 1.4242,
+      "step": 3498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020820413249495536,
+      "loss": 1.4462,
+      "step": 3499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020793828706649675,
+      "loss": 1.4443,
+      "step": 3500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020767256690453173,
+      "loss": 1.4205,
+      "step": 3501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002074069721230288,
+      "loss": 1.4249,
+      "step": 3502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002071415028359026,
+      "loss": 1.4359,
+      "step": 3503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020687615915701408,
+      "loss": 1.4735,
+      "step": 3504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020661094120017043,
+      "loss": 1.4302,
+      "step": 3505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002063458490791244,
+      "loss": 1.4127,
+      "step": 3506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020608088290757526,
+      "loss": 1.4763,
+      "step": 3507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020581604279916815,
+      "loss": 1.441,
+      "step": 3508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020555132886749405,
+      "loss": 1.4694,
+      "step": 3509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020528674122608997,
+      "loss": 1.4675,
+      "step": 3510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002050222799884387,
+      "loss": 1.4337,
+      "step": 3511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020475794526796855,
+      "loss": 1.4536,
+      "step": 3512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020449373717805385,
+      "loss": 1.4976,
+      "step": 3513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002042296558320147,
+      "loss": 1.3969,
+      "step": 3514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020396570134311655,
+      "loss": 1.4343,
+      "step": 3515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020370187382457068,
+      "loss": 1.4665,
+      "step": 3516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020343817338953376,
+      "loss": 1.4501,
+      "step": 3517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020317460015110807,
+      "loss": 1.4438,
+      "step": 3518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020291115422234123,
+      "loss": 1.4227,
+      "step": 3519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002026478357162263,
+      "loss": 1.4607,
+      "step": 3520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002023846447457018,
+      "loss": 1.427,
+      "step": 3521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002021215814236513,
+      "loss": 1.4495,
+      "step": 3522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020185864586290398,
+      "loss": 1.4279,
+      "step": 3523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020159583817623366,
+      "loss": 1.4357,
+      "step": 3524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002013331584763599,
+      "loss": 1.4589,
+      "step": 3525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002010706068759471,
+      "loss": 1.4274,
+      "step": 3526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002008081834876046,
+      "loss": 1.4445,
+      "step": 3527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020054588842388705,
+      "loss": 1.4386,
+      "step": 3528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020028372179729405,
+      "loss": 1.4612,
+      "step": 3529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020002168372026957,
+      "loss": 1.4306,
+      "step": 3530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019975977430520308,
+      "loss": 1.4351,
+      "step": 3531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019949799366442855,
+      "loss": 1.485,
+      "step": 3532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019923634191022484,
+      "loss": 1.3974,
+      "step": 3533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019897481915481547,
+      "loss": 1.4717,
+      "step": 3534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019871342551036885,
+      "loss": 1.4445,
+      "step": 3535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019845216108899745,
+      "loss": 1.4828,
+      "step": 3536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001981910260027588,
+      "loss": 1.4448,
+      "step": 3537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019793002036365493,
+      "loss": 1.4431,
+      "step": 3538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019766914428363214,
+      "loss": 1.438,
+      "step": 3539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019740839787458136,
+      "loss": 1.4538,
+      "step": 3540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019714778124833775,
+      "loss": 1.4566,
+      "step": 3541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019688729451668114,
+      "loss": 1.4557,
+      "step": 3542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019662693779133494,
+      "loss": 1.4632,
+      "step": 3543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019636671118396755,
+      "loss": 1.453,
+      "step": 3544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019610661480619107,
+      "loss": 1.465,
+      "step": 3545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019584664876956203,
+      "loss": 1.4582,
+      "step": 3546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019558681318558097,
+      "loss": 1.4259,
+      "step": 3547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019532710816569238,
+      "loss": 1.4782,
+      "step": 3548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019506753382128494,
+      "loss": 1.4369,
+      "step": 3549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019480809026369112,
+      "loss": 1.4302,
+      "step": 3550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019454877760418732,
+      "loss": 1.4707,
+      "step": 3551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019428959595399386,
+      "loss": 1.4812,
+      "step": 3552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019403054542427485,
+      "loss": 1.4574,
+      "step": 3553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001937716261261383,
+      "loss": 1.4164,
+      "step": 3554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019351283817063548,
+      "loss": 1.4937,
+      "step": 3555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019325418166876168,
+      "loss": 1.4582,
+      "step": 3556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019299565673145592,
+      "loss": 1.4099,
+      "step": 3557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019273726346960053,
+      "loss": 1.4382,
+      "step": 3558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019247900199402147,
+      "loss": 1.4147,
+      "step": 3559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019222087241548835,
+      "loss": 1.4832,
+      "step": 3560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019196287484471376,
+      "loss": 1.4559,
+      "step": 3561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019170500939235397,
+      "loss": 1.4172,
+      "step": 3562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001914472761690087,
+      "loss": 1.438,
+      "step": 3563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019118967528522067,
+      "loss": 1.4201,
+      "step": 3564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001909322068514761,
+      "loss": 1.4484,
+      "step": 3565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001906748709782044,
+      "loss": 1.4358,
+      "step": 3566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019041766777577767,
+      "loss": 1.4677,
+      "step": 3567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019016059735451157,
+      "loss": 1.4579,
+      "step": 3568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018990365982466474,
+      "loss": 1.4506,
+      "step": 3569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018964685529643878,
+      "loss": 1.444,
+      "step": 3570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018939018387997814,
+      "loss": 1.4382,
+      "step": 3571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001891336456853705,
+      "loss": 1.4284,
+      "step": 3572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018887724082264584,
+      "loss": 1.4101,
+      "step": 3573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018862096940177743,
+      "loss": 1.4324,
+      "step": 3574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018836483153268115,
+      "loss": 1.42,
+      "step": 3575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018810882732521561,
+      "loss": 1.4847,
+      "step": 3576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018785295688918208,
+      "loss": 1.4468,
+      "step": 3577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018759722033432448,
+      "loss": 1.4579,
+      "step": 3578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018734161777032933,
+      "loss": 1.4096,
+      "step": 3579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018708614930682555,
+      "loss": 1.4524,
+      "step": 3580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018683081505338467,
+      "loss": 1.4226,
+      "step": 3581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018657561511952064,
+      "loss": 1.4489,
+      "step": 3582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001863205496146898,
+      "loss": 1.4715,
+      "step": 3583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018606561864829098,
+      "loss": 1.4617,
+      "step": 3584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018581082232966474,
+      "loss": 1.4587,
+      "step": 3585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001855561607680945,
+      "loss": 1.4331,
+      "step": 3586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853016340728057,
+      "loss": 1.4628,
+      "step": 3587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018504724235296588,
+      "loss": 1.3974,
+      "step": 3588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018479298571768472,
+      "loss": 1.4373,
+      "step": 3589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018453886427601408,
+      "loss": 1.4636,
+      "step": 3590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018428487813694743,
+      "loss": 1.4482,
+      "step": 3591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018403102740942068,
+      "loss": 1.4541,
+      "step": 3592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001837773122023114,
+      "loss": 1.4748,
+      "step": 3593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018352373262443917,
+      "loss": 1.4289,
+      "step": 3594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001832702887845653,
+      "loss": 1.419,
+      "step": 3595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018301698079139313,
+      "loss": 1.4323,
+      "step": 3596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018276380875356709,
+      "loss": 1.4202,
+      "step": 3597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018251077277967398,
+      "loss": 1.4405,
+      "step": 3598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018225787297824193,
+      "loss": 1.4284,
+      "step": 3599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018200510945774078,
+      "loss": 1.4416,
+      "step": 3600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018175248232658186,
+      "loss": 1.4138,
+      "step": 3601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018149999169311815,
+      "loss": 1.4123,
+      "step": 3602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018124763766564362,
+      "loss": 1.3733,
+      "step": 3603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018099542035239407,
+      "loss": 1.4183,
+      "step": 3604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018074333986154673,
+      "loss": 1.4351,
+      "step": 3605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001804913963012198,
+      "loss": 1.4333,
+      "step": 3606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018023958977947301,
+      "loss": 1.4623,
+      "step": 3607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017998792040430724,
+      "loss": 1.4291,
+      "step": 3608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017973638828366455,
+      "loss": 1.4063,
+      "step": 3609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017948499352542807,
+      "loss": 1.4452,
+      "step": 3610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017923373623742213,
+      "loss": 1.4872,
+      "step": 3611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017898261652741194,
+      "loss": 1.4736,
+      "step": 3612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017873163450310393,
+      "loss": 1.4282,
+      "step": 3613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001784807902721452,
+      "loss": 1.4301,
+      "step": 3614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001782300839421242,
+      "loss": 1.4694,
+      "step": 3615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017797951562056947,
+      "loss": 1.4482,
+      "step": 3616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017772908541495102,
+      "loss": 1.395,
+      "step": 3617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017747879343267949,
+      "loss": 1.4591,
+      "step": 3618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017722863978110608,
+      "loss": 1.435,
+      "step": 3619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017697862456752273,
+      "loss": 1.4429,
+      "step": 3620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017672874789916227,
+      "loss": 1.4523,
+      "step": 3621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017647900988319737,
+      "loss": 1.4296,
+      "step": 3622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017622941062674202,
+      "loss": 1.4488,
+      "step": 3623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017597995023685038,
+      "loss": 1.4494,
+      "step": 3624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017573062882051692,
+      "loss": 1.4455,
+      "step": 3625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001754814464846768,
+      "loss": 1.4749,
+      "step": 3626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017523240333620544,
+      "loss": 1.4164,
+      "step": 3627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017498349948191818,
+      "loss": 1.4227,
+      "step": 3628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017473473502857111,
+      "loss": 1.4499,
+      "step": 3629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017448611008286037,
+      "loss": 1.4288,
+      "step": 3630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001742376247514222,
+      "loss": 1.4709,
+      "step": 3631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017398927914083297,
+      "loss": 1.4242,
+      "step": 3632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017374107335760936,
+      "loss": 1.4824,
+      "step": 3633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017349300750820756,
+      "loss": 1.4561,
+      "step": 3634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017324508169902419,
+      "loss": 1.4535,
+      "step": 3635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017299729603639568,
+      "loss": 1.4214,
+      "step": 3636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017274965062659837,
+      "loss": 1.4072,
+      "step": 3637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017250214557584836,
+      "loss": 1.4509,
+      "step": 3638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001722547809903016,
+      "loss": 1.4647,
+      "step": 3639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017200755697605386,
+      "loss": 1.4583,
+      "step": 3640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017176047363914054,
+      "loss": 1.4262,
+      "step": 3641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001715135310855367,
+      "loss": 1.4481,
+      "step": 3642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017126672942115696,
+      "loss": 1.4163,
+      "step": 3643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017102006875185572,
+      "loss": 1.4692,
+      "step": 3644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001707735491834269,
+      "loss": 1.3827,
+      "step": 3645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017052717082160346,
+      "loss": 1.4434,
+      "step": 3646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017028093377205821,
+      "loss": 1.4066,
+      "step": 3647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017003483814040328,
+      "loss": 1.476,
+      "step": 3648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016978888403219018,
+      "loss": 1.4548,
+      "step": 3649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001695430715529096,
+      "loss": 1.4309,
+      "step": 3650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016929740080799167,
+      "loss": 1.4781,
+      "step": 3651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001690518719028054,
+      "loss": 1.4428,
+      "step": 3652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001688064849426592,
+      "loss": 1.4098,
+      "step": 3653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016856124003280064,
+      "loss": 1.4708,
+      "step": 3654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016831613727841626,
+      "loss": 1.4213,
+      "step": 3655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016807117678463174,
+      "loss": 1.4711,
+      "step": 3656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016782635865651168,
+      "loss": 1.4564,
+      "step": 3657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016758168299905942,
+      "loss": 1.5122,
+      "step": 3658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016733714991721738,
+      "loss": 1.4456,
+      "step": 3659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000167092759515867,
+      "loss": 1.4325,
+      "step": 3660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016684851189982826,
+      "loss": 1.4399,
+      "step": 3661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016660440717385994,
+      "loss": 1.4089,
+      "step": 3662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016636044544265984,
+      "loss": 1.4237,
+      "step": 3663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016611662681086374,
+      "loss": 1.4389,
+      "step": 3664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016587295138304677,
+      "loss": 1.4791,
+      "step": 3665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016562941926372227,
+      "loss": 1.4555,
+      "step": 3666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016538603055734214,
+      "loss": 1.451,
+      "step": 3667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016514278536829686,
+      "loss": 1.4071,
+      "step": 3668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016489968380091537,
+      "loss": 1.44,
+      "step": 3669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016465672595946486,
+      "loss": 1.4198,
+      "step": 3670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016441391194815097,
+      "loss": 1.452,
+      "step": 3671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016417124187111776,
+      "loss": 1.4541,
+      "step": 3672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016392871583244728,
+      "loss": 1.4707,
+      "step": 3673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016368633393616013,
+      "loss": 1.4514,
+      "step": 3674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016344409628621482,
+      "loss": 1.4653,
+      "step": 3675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016320200298650822,
+      "loss": 1.4452,
+      "step": 3676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000162960054140875,
+      "loss": 1.3919,
+      "step": 3677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016271824985308802,
+      "loss": 1.4389,
+      "step": 3678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016247659022685824,
+      "loss": 1.4514,
+      "step": 3679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016223507536583447,
+      "loss": 1.4062,
+      "step": 3680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016199370537360347,
+      "loss": 1.454,
+      "step": 3681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016175248035368994,
+      "loss": 1.4399,
+      "step": 3682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016151140040955608,
+      "loss": 1.471,
+      "step": 3683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016127046564460217,
+      "loss": 1.4228,
+      "step": 3684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001610296761621662,
+      "loss": 1.4232,
+      "step": 3685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001607890320655237,
+      "loss": 1.4508,
+      "step": 3686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000160548533457888,
+      "loss": 1.4553,
+      "step": 3687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016030818044241008,
+      "loss": 1.4821,
+      "step": 3688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016006797312217815,
+      "loss": 1.4139,
+      "step": 3689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015982791160021814,
+      "loss": 1.4407,
+      "step": 3690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015958799597949353,
+      "loss": 1.4094,
+      "step": 3691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015934822636290514,
+      "loss": 1.4529,
+      "step": 3692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015910860285329109,
+      "loss": 1.4945,
+      "step": 3693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001588691255534272,
+      "loss": 1.52,
+      "step": 3694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015862979456602582,
+      "loss": 1.4661,
+      "step": 3695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015839060999373728,
+      "loss": 1.4619,
+      "step": 3696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001581515719391488,
+      "loss": 1.4611,
+      "step": 3697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015791268050478486,
+      "loss": 1.4387,
+      "step": 3698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001576739357931069,
+      "loss": 1.439,
+      "step": 3699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001574353379065136,
+      "loss": 1.4705,
+      "step": 3700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015719688694734057,
+      "loss": 1.4231,
+      "step": 3701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015695858301786048,
+      "loss": 1.4396,
+      "step": 3702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015672042622028276,
+      "loss": 1.4434,
+      "step": 3703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000156482416656754,
+      "loss": 1.4372,
+      "step": 3704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015624455442935742,
+      "loss": 1.448,
+      "step": 3705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001560068396401133,
+      "loss": 1.4285,
+      "step": 3706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001557692723909782,
+      "loss": 1.446,
+      "step": 3707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015553185278384586,
+      "loss": 1.4467,
+      "step": 3708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015529458092054655,
+      "loss": 1.438,
+      "step": 3709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001550574569028471,
+      "loss": 1.4493,
+      "step": 3710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015482048083245115,
+      "loss": 1.4562,
+      "step": 3711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015458365281099877,
+      "loss": 1.4278,
+      "step": 3712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015434697294006622,
+      "loss": 1.4729,
+      "step": 3713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015411044132116665,
+      "loss": 1.5083,
+      "step": 3714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001538740580557494,
+      "loss": 1.4163,
+      "step": 3715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015363782324520031,
+      "loss": 1.4224,
+      "step": 3716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001534017369908415,
+      "loss": 1.4502,
+      "step": 3717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001531657993939314,
+      "loss": 1.429,
+      "step": 3718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001529300105556644,
+      "loss": 1.4625,
+      "step": 3719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001526943705771715,
+      "loss": 1.4627,
+      "step": 3720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015245887955951966,
+      "loss": 1.4647,
+      "step": 3721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015222353760371195,
+      "loss": 1.4137,
+      "step": 3722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015198834481068753,
+      "loss": 1.4341,
+      "step": 3723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001517533012813217,
+      "loss": 1.4577,
+      "step": 3724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015151840711642535,
+      "loss": 1.4569,
+      "step": 3725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001512836624167457,
+      "loss": 1.4236,
+      "step": 3726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015104906728296568,
+      "loss": 1.4171,
+      "step": 3727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015081462181570427,
+      "loss": 1.4743,
+      "step": 3728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015058032611551603,
+      "loss": 1.4507,
+      "step": 3729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015034618028289139,
+      "loss": 1.4114,
+      "step": 3730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015011218441825642,
+      "loss": 1.4229,
+      "step": 3731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014987833862197298,
+      "loss": 1.4927,
+      "step": 3732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001496446429943385,
+      "loss": 1.4819,
+      "step": 3733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014941109763558601,
+      "loss": 1.429,
+      "step": 3734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014917770264588414,
+      "loss": 1.4261,
+      "step": 3735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001489444581253369,
+      "loss": 1.4347,
+      "step": 3736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014871136417398407,
+      "loss": 1.4472,
+      "step": 3737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014847842089180024,
+      "loss": 1.4118,
+      "step": 3738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014824562837869603,
+      "loss": 1.4591,
+      "step": 3739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014801298673451703,
+      "loss": 1.4498,
+      "step": 3740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001477804960590442,
+      "loss": 1.4663,
+      "step": 3741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001475481564519938,
+      "loss": 1.4262,
+      "step": 3742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001473159680130175,
+      "loss": 1.4445,
+      "step": 3743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014708393084170146,
+      "loss": 1.4598,
+      "step": 3744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001468520450375676,
+      "loss": 1.4499,
+      "step": 3745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014662031070007271,
+      "loss": 1.4527,
+      "step": 3746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014638872792860862,
+      "loss": 1.45,
+      "step": 3747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014615729682250218,
+      "loss": 1.3683,
+      "step": 3748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001459260174810153,
+      "loss": 1.4164,
+      "step": 3749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014569489000334436,
+      "loss": 1.4695,
+      "step": 3750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014546391448862095,
+      "loss": 1.45,
+      "step": 3751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001452330910359116,
+      "loss": 1.4341,
+      "step": 3752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014500241974421736,
+      "loss": 1.4252,
+      "step": 3753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014477190071247414,
+      "loss": 1.416,
+      "step": 3754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014454153403955266,
+      "loss": 1.4262,
+      "step": 3755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014431131982425795,
+      "loss": 1.448,
+      "step": 3756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001440812581653298,
+      "loss": 1.4066,
+      "step": 3757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014385134916144278,
+      "loss": 1.5143,
+      "step": 3758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014362159291120575,
+      "loss": 1.4451,
+      "step": 3759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014339198951316217,
+      "loss": 1.4724,
+      "step": 3760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014316253906578995,
+      "loss": 1.472,
+      "step": 3761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001429332416675012,
+      "loss": 1.4334,
+      "step": 3762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014270409741664269,
+      "loss": 1.4668,
+      "step": 3763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014247510641149524,
+      "loss": 1.4397,
+      "step": 3764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014224626875027413,
+      "loss": 1.4221,
+      "step": 3765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014201758453112877,
+      "loss": 1.4191,
+      "step": 3766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014178905385214293,
+      "loss": 1.4106,
+      "step": 3767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000141560676811334,
+      "loss": 1.5047,
+      "step": 3768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014133245350665414,
+      "loss": 1.4231,
+      "step": 3769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014110438403598912,
+      "loss": 1.4231,
+      "step": 3770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014087646849715907,
+      "loss": 1.4026,
+      "step": 3771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001406487069879178,
+      "loss": 1.4433,
+      "step": 3772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001404210996059534,
+      "loss": 1.4191,
+      "step": 3773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014019364644888722,
+      "loss": 1.465,
+      "step": 3774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013996634761427511,
+      "loss": 1.4554,
+      "step": 3775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013973920319960653,
+      "loss": 1.4099,
+      "step": 3776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013951221330230456,
+      "loss": 1.4657,
+      "step": 3777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013928537801972618,
+      "loss": 1.4604,
+      "step": 3778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001390586974491622,
+      "loss": 1.4727,
+      "step": 3779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013883217168783645,
+      "loss": 1.4775,
+      "step": 3780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013860580083290702,
+      "loss": 1.4796,
+      "step": 3781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001383795849814653,
+      "loss": 1.4674,
+      "step": 3782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013815352423053617,
+      "loss": 1.4671,
+      "step": 3783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013792761867707808,
+      "loss": 1.4166,
+      "step": 3784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013770186841798303,
+      "loss": 1.4483,
+      "step": 3785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001374762735500759,
+      "loss": 1.4228,
+      "step": 3786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013725083417011547,
+      "loss": 1.4542,
+      "step": 3787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013702555037479363,
+      "loss": 1.4159,
+      "step": 3788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013680042226073554,
+      "loss": 1.4429,
+      "step": 3789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001365754499244996,
+      "loss": 1.4234,
+      "step": 3790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013635063346257732,
+      "loss": 1.4319,
+      "step": 3791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013612597297139357,
+      "loss": 1.4232,
+      "step": 3792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000135901468547306,
+      "loss": 1.4743,
+      "step": 3793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001356771202866056,
+      "loss": 1.4529,
+      "step": 3794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013545292828551632,
+      "loss": 1.484,
+      "step": 3795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013522889264019496,
+      "loss": 1.4597,
+      "step": 3796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013500501344673138,
+      "loss": 1.4704,
+      "step": 3797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013478129080114848,
+      "loss": 1.4101,
+      "step": 3798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013455772479940153,
+      "loss": 1.4386,
+      "step": 3799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013433431553737903,
+      "loss": 1.4011,
+      "step": 3800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001341110631109022,
+      "loss": 1.4478,
+      "step": 3801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013388796761572492,
+      "loss": 1.4144,
+      "step": 3802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013366502914753382,
+      "loss": 1.4166,
+      "step": 3803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013344224780194826,
+      "loss": 1.4094,
+      "step": 3804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013321962367451978,
+      "loss": 1.463,
+      "step": 3805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013299715686073294,
+      "loss": 1.4275,
+      "step": 3806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013277484745600476,
+      "loss": 1.3891,
+      "step": 3807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013255269555568466,
+      "loss": 1.4461,
+      "step": 3808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001323307012550545,
+      "loss": 1.445,
+      "step": 3809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013210886464932865,
+      "loss": 1.4519,
+      "step": 3810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013188718583365355,
+      "loss": 1.4554,
+      "step": 3811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013166566490310826,
+      "loss": 1.4305,
+      "step": 3812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013144430195270417,
+      "loss": 1.4142,
+      "step": 3813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013122309707738462,
+      "loss": 1.4231,
+      "step": 3814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001310020503720254,
+      "loss": 1.4612,
+      "step": 3815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013078116193143447,
+      "loss": 1.434,
+      "step": 3816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013056043185035154,
+      "loss": 1.4552,
+      "step": 3817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001303398602234488,
+      "loss": 1.4652,
+      "step": 3818
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013011944714533036,
+      "loss": 1.4524,
+      "step": 3819
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001298991927105323,
+      "loss": 1.4124,
+      "step": 3820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012967909701352265,
+      "loss": 1.4206,
+      "step": 3821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001294591601487014,
+      "loss": 1.4313,
+      "step": 3822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012923938221040044,
+      "loss": 1.436,
+      "step": 3823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001290197632928834,
+      "loss": 1.4654,
+      "step": 3824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012880030349034577,
+      "loss": 1.4195,
+      "step": 3825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012858100289691476,
+      "loss": 1.4565,
+      "step": 3826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012836186160664932,
+      "loss": 1.4704,
+      "step": 3827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012814287971354022,
+      "loss": 1.4266,
+      "step": 3828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012792405731150942,
+      "loss": 1.4512,
+      "step": 3829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012770539449441092,
+      "loss": 1.4224,
+      "step": 3830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012748689135603014,
+      "loss": 1.4506,
+      "step": 3831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001272685479900839,
+      "loss": 1.4439,
+      "step": 3832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012705036449022072,
+      "loss": 1.4269,
+      "step": 3833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012683234095002045,
+      "loss": 1.4204,
+      "step": 3834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001266144774629941,
+      "loss": 1.4361,
+      "step": 3835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012639677412258426,
+      "loss": 1.4669,
+      "step": 3836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012617923102216494,
+      "loss": 1.4751,
+      "step": 3837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001259618482550412,
+      "loss": 1.4521,
+      "step": 3838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001257446259144494,
+      "loss": 1.4419,
+      "step": 3839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012552756409355738,
+      "loss": 1.4797,
+      "step": 3840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001253106628854635,
+      "loss": 1.4481,
+      "step": 3841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012509392238319766,
+      "loss": 1.439,
+      "step": 3842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012487734267972096,
+      "loss": 1.4581,
+      "step": 3843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012466092386792516,
+      "loss": 1.4219,
+      "step": 3844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012444466604063332,
+      "loss": 1.4116,
+      "step": 3845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012422856929059938,
+      "loss": 1.4117,
+      "step": 3846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012401263371050793,
+      "loss": 1.4587,
+      "step": 3847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012379685939297474,
+      "loss": 1.4494,
+      "step": 3848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012358124643054635,
+      "loss": 1.4517,
+      "step": 3849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012336579491570006,
+      "loss": 1.4386,
+      "step": 3850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001231505049408439,
+      "loss": 1.457,
+      "step": 3851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012293537659831673,
+      "loss": 1.4184,
+      "step": 3852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012272040998038787,
+      "loss": 1.4783,
+      "step": 3853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012250560517925745,
+      "loss": 1.3978,
+      "step": 3854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012229096228705621,
+      "loss": 1.4044,
+      "step": 3855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012207648139584537,
+      "loss": 1.4566,
+      "step": 3856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012186216259761662,
+      "loss": 1.4594,
+      "step": 3857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012164800598429238,
+      "loss": 1.4544,
+      "step": 3858
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000121434011647725,
+      "loss": 1.4014,
+      "step": 3859
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012122017967969772,
+      "loss": 1.3871,
+      "step": 3860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000121006510171924,
+      "loss": 1.4203,
+      "step": 3861
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012079300321604753,
+      "loss": 1.4934,
+      "step": 3862
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012057965890364237,
+      "loss": 1.431,
+      "step": 3863
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012036647732621275,
+      "loss": 1.4457,
+      "step": 3864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012015345857519339,
+      "loss": 1.4329,
+      "step": 3865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011994060274194851,
+      "loss": 1.4544,
+      "step": 3866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001197279099177731,
+      "loss": 1.4258,
+      "step": 3867
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011951538019389196,
+      "loss": 1.4579,
+      "step": 3868
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011930301366145996,
+      "loss": 1.4508,
+      "step": 3869
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011909081041156206,
+      "loss": 1.4407,
+      "step": 3870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011887877053521328,
+      "loss": 1.4338,
+      "step": 3871
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011866689412335801,
+      "loss": 1.4182,
+      "step": 3872
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011845518126687121,
+      "loss": 1.4465,
+      "step": 3873
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001182436320565573,
+      "loss": 1.4264,
+      "step": 3874
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001180322465831507,
+      "loss": 1.4265,
+      "step": 3875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001178210249373155,
+      "loss": 1.4311,
+      "step": 3876
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011760996720964573,
+      "loss": 1.4248,
+      "step": 3877
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011739907349066453,
+      "loss": 1.4266,
+      "step": 3878
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011718834387082539,
+      "loss": 1.4838,
+      "step": 3879
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 1.4698,
+      "step": 3880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011676737729003389,
+      "loss": 1.3791,
+      "step": 3881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001165571405096359,
+      "loss": 1.4049,
+      "step": 3882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011634706818948848,
+      "loss": 1.4484,
+      "step": 3883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011613716041969257,
+      "loss": 1.4241,
+      "step": 3884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011592741729027845,
+      "loss": 1.3987,
+      "step": 3885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011571783889120585,
+      "loss": 1.4284,
+      "step": 3886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011550842531236377,
+      "loss": 1.4023,
+      "step": 3887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011529917664357064,
+      "loss": 1.404,
+      "step": 3888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011509009297457423,
+      "loss": 1.4384,
+      "step": 3889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011488117439505108,
+      "loss": 1.3943,
+      "step": 3890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011467242099460745,
+      "loss": 1.4453,
+      "step": 3891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011446383286277845,
+      "loss": 1.4336,
+      "step": 3892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011425541008902851,
+      "loss": 1.4359,
+      "step": 3893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011404715276275096,
+      "loss": 1.459,
+      "step": 3894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001138390609732684,
+      "loss": 1.4389,
+      "step": 3895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011363113480983201,
+      "loss": 1.4262,
+      "step": 3896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011342337436162226,
+      "loss": 1.438,
+      "step": 3897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001132157797177486,
+      "loss": 1.4195,
+      "step": 3898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011300835096724915,
+      "loss": 1.4554,
+      "step": 3899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011280108819909102,
+      "loss": 1.4206,
+      "step": 3900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001125939915021702,
+      "loss": 1.4285,
+      "step": 3901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011238706096531104,
+      "loss": 1.4456,
+      "step": 3902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011218029667726704,
+      "loss": 1.4369,
+      "step": 3903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011197369872672037,
+      "loss": 1.4178,
+      "step": 3904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011176726720228158,
+      "loss": 1.4358,
+      "step": 3905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011156100219249022,
+      "loss": 1.4434,
+      "step": 3906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001113549037858142,
+      "loss": 1.4267,
+      "step": 3907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011114897207064983,
+      "loss": 1.4457,
+      "step": 3908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011094320713532213,
+      "loss": 1.4486,
+      "step": 3909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011073760906808455,
+      "loss": 1.4209,
+      "step": 3910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011053217795711906,
+      "loss": 1.4354,
+      "step": 3911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011032691389053578,
+      "loss": 1.4505,
+      "step": 3912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011012181695637347,
+      "loss": 1.4524,
+      "step": 3913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010991688724259902,
+      "loss": 1.409,
+      "step": 3914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010971212483710758,
+      "loss": 1.41,
+      "step": 3915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010950752982772266,
+      "loss": 1.4406,
+      "step": 3916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010930310230219603,
+      "loss": 1.4115,
+      "step": 3917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010909884234820728,
+      "loss": 1.457,
+      "step": 3918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010889475005336446,
+      "loss": 1.4485,
+      "step": 3919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010869082550520365,
+      "loss": 1.4489,
+      "step": 3920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010848706879118891,
+      "loss": 1.4102,
+      "step": 3921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010828347999871235,
+      "loss": 1.4922,
+      "step": 3922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010808005921509406,
+      "loss": 1.4252,
+      "step": 3923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010787680652758209,
+      "loss": 1.4503,
+      "step": 3924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010767372202335229,
+      "loss": 1.4557,
+      "step": 3925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010747080578950874,
+      "loss": 1.4514,
+      "step": 3926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010726805791308264,
+      "loss": 1.4412,
+      "step": 3927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010706547848103365,
+      "loss": 1.4352,
+      "step": 3928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010686306758024889,
+      "loss": 1.4757,
+      "step": 3929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010666082529754317,
+      "loss": 1.4154,
+      "step": 3930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010645875171965925,
+      "loss": 1.4137,
+      "step": 3931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010625684693326726,
+      "loss": 1.4486,
+      "step": 3932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010605511102496484,
+      "loss": 1.4472,
+      "step": 3933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010585354408127757,
+      "loss": 1.4567,
+      "step": 3934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010565214618865821,
+      "loss": 1.4012,
+      "step": 3935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010545091743348723,
+      "loss": 1.4822,
+      "step": 3936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010524985790207254,
+      "loss": 1.429,
+      "step": 3937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010504896768064942,
+      "loss": 1.4129,
+      "step": 3938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010484824685538041,
+      "loss": 1.4552,
+      "step": 3939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010464769551235553,
+      "loss": 1.462,
+      "step": 3940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010444731373759219,
+      "loss": 1.4248,
+      "step": 3941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010424710161703488,
+      "loss": 1.4584,
+      "step": 3942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001040470592365555,
+      "loss": 1.4505,
+      "step": 3943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010384718668195314,
+      "loss": 1.46,
+      "step": 3944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010364748403895369,
+      "loss": 1.4355,
+      "step": 3945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010344795139321067,
+      "loss": 1.4654,
+      "step": 3946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010324858883030442,
+      "loss": 1.4519,
+      "step": 3947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010304939643574234,
+      "loss": 1.4578,
+      "step": 3948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010285037429495897,
+      "loss": 1.4347,
+      "step": 3949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010265152249331572,
+      "loss": 1.4517,
+      "step": 3950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010245284111610087,
+      "loss": 1.4105,
+      "step": 3951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010225433024852982,
+      "loss": 1.4534,
+      "step": 3952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010205598997574461,
+      "loss": 1.4562,
+      "step": 3953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010185782038281433,
+      "loss": 1.4221,
+      "step": 3954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010165982155473463,
+      "loss": 1.4205,
+      "step": 3955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010146199357642827,
+      "loss": 1.4498,
+      "step": 3956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010126433653274413,
+      "loss": 1.3868,
+      "step": 3957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010106685050845838,
+      "loss": 1.4244,
+      "step": 3958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010086953558827349,
+      "loss": 1.4181,
+      "step": 3959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010067239185681882,
+      "loss": 1.4423,
+      "step": 3960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010047541939864995,
+      "loss": 1.4386,
+      "step": 3961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010027861829824952,
+      "loss": 1.4239,
+      "step": 3962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010008198864002593,
+      "loss": 1.4178,
+      "step": 3963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.988553050831467e-05,
+      "loss": 1.4242,
+      "step": 3964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.968924398737745e-05,
+      "loss": 1.3811,
+      "step": 3965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.949312916140235e-05,
+      "loss": 1.4424,
+      "step": 3966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.929718611450389e-05,
+      "loss": 1.4231,
+      "step": 3967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.9101414930723e-05,
+      "loss": 1.3982,
+      "step": 3968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.89058156940264e-05,
+      "loss": 1.4195,
+      "step": 3969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.871038848830765e-05,
+      "loss": 1.473,
+      "step": 3970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.851513339738626e-05,
+      "loss": 1.4474,
+      "step": 3971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.832005050500803e-05,
+      "loss": 1.4137,
+      "step": 3972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.812513989484473e-05,
+      "loss": 1.4496,
+      "step": 3973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.79304016504945e-05,
+      "loss": 1.4635,
+      "step": 3974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.773583585548113e-05,
+      "loss": 1.4521,
+      "step": 3975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.754144259325487e-05,
+      "loss": 1.4208,
+      "step": 3976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.734722194719176e-05,
+      "loss": 1.4391,
+      "step": 3977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.715317400059386e-05,
+      "loss": 1.49,
+      "step": 3978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.695929883668919e-05,
+      "loss": 1.4323,
+      "step": 3979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.67655965386316e-05,
+      "loss": 1.397,
+      "step": 3980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.657206718950079e-05,
+      "loss": 1.4316,
+      "step": 3981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.637871087230227e-05,
+      "loss": 1.4346,
+      "step": 3982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.618552766996752e-05,
+      "loss": 1.4677,
+      "step": 3983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.599251766535344e-05,
+      "loss": 1.4465,
+      "step": 3984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.579968094124303e-05,
+      "loss": 1.4412,
+      "step": 3985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.560701758034474e-05,
+      "loss": 1.4042,
+      "step": 3986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.541452766529246e-05,
+      "loss": 1.4076,
+      "step": 3987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.522221127864611e-05,
+      "loss": 1.4327,
+      "step": 3988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.503006850289097e-05,
+      "loss": 1.4228,
+      "step": 3989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.483809942043792e-05,
+      "loss": 1.3941,
+      "step": 3990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.464630411362324e-05,
+      "loss": 1.4329,
+      "step": 3991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.445468266470875e-05,
+      "loss": 1.4566,
+      "step": 3992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.426323515588182e-05,
+      "loss": 1.3901,
+      "step": 3993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.407196166925486e-05,
+      "loss": 1.4352,
+      "step": 3994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.388086228686604e-05,
+      "loss": 1.4884,
+      "step": 3995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.368993709067858e-05,
+      "loss": 1.4191,
+      "step": 3996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.349918616258113e-05,
+      "loss": 1.4145,
+      "step": 3997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.330860958438764e-05,
+      "loss": 1.4414,
+      "step": 3998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.311820743783728e-05,
+      "loss": 1.4167,
+      "step": 3999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.292797980459405e-05,
+      "loss": 1.4901,
+      "step": 4000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.273792676624749e-05,
+      "loss": 1.4507,
+      "step": 4001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.254804840431225e-05,
+      "loss": 1.4239,
+      "step": 4002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.235834480022786e-05,
+      "loss": 1.417,
+      "step": 4003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.216881603535899e-05,
+      "loss": 1.4233,
+      "step": 4004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.197946219099551e-05,
+      "loss": 1.4255,
+      "step": 4005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.179028334835182e-05,
+      "loss": 1.3937,
+      "step": 4006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.160127958856757e-05,
+      "loss": 1.4372,
+      "step": 4007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.141245099270734e-05,
+      "loss": 1.4637,
+      "step": 4008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.122379764176048e-05,
+      "loss": 1.438,
+      "step": 4009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.103531961664119e-05,
+      "loss": 1.404,
+      "step": 4010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.084701699818853e-05,
+      "loss": 1.4208,
+      "step": 4011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.065888986716625e-05,
+      "loss": 1.471,
+      "step": 4012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.047093830426284e-05,
+      "loss": 1.4568,
+      "step": 4013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.028316239009154e-05,
+      "loss": 1.4627,
+      "step": 4014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.009556220519028e-05,
+      "loss": 1.424,
+      "step": 4015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.990813783002149e-05,
+      "loss": 1.4118,
+      "step": 4016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.972088934497241e-05,
+      "loss": 1.4154,
+      "step": 4017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.953381683035444e-05,
+      "loss": 1.4384,
+      "step": 4018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.934692036640396e-05,
+      "loss": 1.4516,
+      "step": 4019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.916020003328157e-05,
+      "loss": 1.4753,
+      "step": 4020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.897365591107242e-05,
+      "loss": 1.4286,
+      "step": 4021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.878728807978614e-05,
+      "loss": 1.4325,
+      "step": 4022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.860109661935672e-05,
+      "loss": 1.4352,
+      "step": 4023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.841508160964218e-05,
+      "loss": 1.431,
+      "step": 4024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.822924313042535e-05,
+      "loss": 1.4051,
+      "step": 4025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.80435812614131e-05,
+      "loss": 1.3911,
+      "step": 4026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.785809608223655e-05,
+      "loss": 1.4109,
+      "step": 4027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.767278767245113e-05,
+      "loss": 1.4603,
+      "step": 4028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.748765611153648e-05,
+      "loss": 1.4649,
+      "step": 4029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.730270147889602e-05,
+      "loss": 1.5298,
+      "step": 4030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.711792385385781e-05,
+      "loss": 1.4604,
+      "step": 4031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.693332331567361e-05,
+      "loss": 1.4295,
+      "step": 4032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.674889994351942e-05,
+      "loss": 1.4284,
+      "step": 4033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.656465381649525e-05,
+      "loss": 1.4414,
+      "step": 4034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.638058501362512e-05,
+      "loss": 1.464,
+      "step": 4035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.619669361385663e-05,
+      "loss": 1.4303,
+      "step": 4036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.60129796960617e-05,
+      "loss": 1.4259,
+      "step": 4037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.582944333903609e-05,
+      "loss": 1.4115,
+      "step": 4038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.564608462149919e-05,
+      "loss": 1.4456,
+      "step": 4039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.546290362209442e-05,
+      "loss": 1.4472,
+      "step": 4040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.527990041938882e-05,
+      "loss": 1.4408,
+      "step": 4041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.509707509187331e-05,
+      "loss": 1.4547,
+      "step": 4042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.491442771796232e-05,
+      "loss": 1.4222,
+      "step": 4043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.473195837599418e-05,
+      "loss": 1.4477,
+      "step": 4044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.454966714423068e-05,
+      "loss": 1.4024,
+      "step": 4045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.436755410085733e-05,
+      "loss": 1.4631,
+      "step": 4046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.418561932398327e-05,
+      "loss": 1.4172,
+      "step": 4047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.400386289164091e-05,
+      "loss": 1.4272,
+      "step": 4048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.38222848817864e-05,
+      "loss": 1.4441,
+      "step": 4049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.364088537229935e-05,
+      "loss": 1.4405,
+      "step": 4050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.34596644409828e-05,
+      "loss": 1.4446,
+      "step": 4051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.327862216556309e-05,
+      "loss": 1.4308,
+      "step": 4052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.309775862369007e-05,
+      "loss": 1.4363,
+      "step": 4053
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.291707389293695e-05,
+      "loss": 1.424,
+      "step": 4054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.273656805079994e-05,
+      "loss": 1.4325,
+      "step": 4055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25562411746989e-05,
+      "loss": 1.4294,
+      "step": 4056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.237609334197677e-05,
+      "loss": 1.4379,
+      "step": 4057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.219612462989967e-05,
+      "loss": 1.3998,
+      "step": 4058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.201633511565703e-05,
+      "loss": 1.404,
+      "step": 4059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.183672487636135e-05,
+      "loss": 1.413,
+      "step": 4060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.165729398904803e-05,
+      "loss": 1.4183,
+      "step": 4061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.147804253067581e-05,
+      "loss": 1.4406,
+      "step": 4062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.129897057812641e-05,
+      "loss": 1.4234,
+      "step": 4063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.112007820820461e-05,
+      "loss": 1.4525,
+      "step": 4064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.094136549763797e-05,
+      "loss": 1.4527,
+      "step": 4065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.076283252307732e-05,
+      "loss": 1.4622,
+      "step": 4066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.058447936109597e-05,
+      "loss": 1.4654,
+      "step": 4067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.040630608819039e-05,
+      "loss": 1.4634,
+      "step": 4068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.022831278077996e-05,
+      "loss": 1.4376,
+      "step": 4069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.005049951520666e-05,
+      "loss": 1.4488,
+      "step": 4070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.987286636773533e-05,
+      "loss": 1.394,
+      "step": 4071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.969541341455366e-05,
+      "loss": 1.4781,
+      "step": 4072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.951814073177188e-05,
+      "loss": 1.4564,
+      "step": 4073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.934104839542306e-05,
+      "loss": 1.4335,
+      "step": 4074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.916413648146281e-05,
+      "loss": 1.4186,
+      "step": 4075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.898740506576935e-05,
+      "loss": 1.4472,
+      "step": 4076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.881085422414364e-05,
+      "loss": 1.4796,
+      "step": 4077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.863448403230905e-05,
+      "loss": 1.4125,
+      "step": 4078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.845829456591142e-05,
+      "loss": 1.4499,
+      "step": 4079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.828228590051911e-05,
+      "loss": 1.4397,
+      "step": 4080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.810645811162309e-05,
+      "loss": 1.4332,
+      "step": 4081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.793081127463658e-05,
+      "loss": 1.451,
+      "step": 4082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.775534546489526e-05,
+      "loss": 1.4427,
+      "step": 4083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.758006075765722e-05,
+      "loss": 1.4597,
+      "step": 4084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.74049572281027e-05,
+      "loss": 1.4392,
+      "step": 4085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.72300349513343e-05,
+      "loss": 1.3895,
+      "step": 4086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.705529400237704e-05,
+      "loss": 1.4314,
+      "step": 4087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.688073445617799e-05,
+      "loss": 1.4516,
+      "step": 4088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.670635638760649e-05,
+      "loss": 1.4711,
+      "step": 4089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.653215987145412e-05,
+      "loss": 1.4559,
+      "step": 4090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.635814498243426e-05,
+      "loss": 1.4688,
+      "step": 4091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.618431179518276e-05,
+      "loss": 1.4366,
+      "step": 4092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.601066038425741e-05,
+      "loss": 1.4373,
+      "step": 4093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.583719082413798e-05,
+      "loss": 1.4407,
+      "step": 4094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.56639031892264e-05,
+      "loss": 1.4349,
+      "step": 4095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.549079755384646e-05,
+      "loss": 1.4372,
+      "step": 4096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.531787399224371e-05,
+      "loss": 1.4249,
+      "step": 4097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.514513257858597e-05,
+      "loss": 1.4452,
+      "step": 4098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.497257338696261e-05,
+      "loss": 1.4329,
+      "step": 4099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.480019649138514e-05,
+      "loss": 1.4413,
+      "step": 4100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.462800196578661e-05,
+      "loss": 1.3914,
+      "step": 4101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.445598988402207e-05,
+      "loss": 1.4027,
+      "step": 4102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.428416031986818e-05,
+      "loss": 1.4151,
+      "step": 4103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.411251334702335e-05,
+      "loss": 1.4262,
+      "step": 4104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.39410490391077e-05,
+      "loss": 1.4473,
+      "step": 4105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.376976746966302e-05,
+      "loss": 1.4476,
+      "step": 4106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.359866871215265e-05,
+      "loss": 1.4322,
+      "step": 4107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.342775283996173e-05,
+      "loss": 1.4617,
+      "step": 4108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.325701992639649e-05,
+      "loss": 1.4074,
+      "step": 4109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.308647004468517e-05,
+      "loss": 1.3892,
+      "step": 4110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.291610326797732e-05,
+      "loss": 1.4438,
+      "step": 4111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.27459196693439e-05,
+      "loss": 1.4247,
+      "step": 4112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.257591932177748e-05,
+      "loss": 1.4179,
+      "step": 4113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.240610229819194e-05,
+      "loss": 1.4446,
+      "step": 4114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.223646867142237e-05,
+      "loss": 1.3994,
+      "step": 4115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.206701851422536e-05,
+      "loss": 1.4472,
+      "step": 4116
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.189775189927889e-05,
+      "loss": 1.4091,
+      "step": 4117
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.172866889918206e-05,
+      "loss": 1.4517,
+      "step": 4118
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.155976958645532e-05,
+      "loss": 1.4272,
+      "step": 4119
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.139105403354024e-05,
+      "loss": 1.4057,
+      "step": 4120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.122252231279974e-05,
+      "loss": 1.4656,
+      "step": 4121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.105417449651763e-05,
+      "loss": 1.4115,
+      "step": 4122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.08860106568991e-05,
+      "loss": 1.4172,
+      "step": 4123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.071803086607021e-05,
+      "loss": 1.4433,
+      "step": 4124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.055023519607828e-05,
+      "loss": 1.3936,
+      "step": 4125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.038262371889159e-05,
+      "loss": 1.4433,
+      "step": 4126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.021519650639951e-05,
+      "loss": 1.4059,
+      "step": 4127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.004795363041205e-05,
+      "loss": 1.4148,
+      "step": 4128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.98808951626605e-05,
+      "loss": 1.4351,
+      "step": 4129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.971402117479686e-05,
+      "loss": 1.4022,
+      "step": 4130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.954733173839416e-05,
+      "loss": 1.431,
+      "step": 4131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.938082692494618e-05,
+      "loss": 1.4488,
+      "step": 4132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.921450680586756e-05,
+      "loss": 1.4458,
+      "step": 4133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.904837145249365e-05,
+      "loss": 1.4351,
+      "step": 4134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.888242093608066e-05,
+      "loss": 1.4122,
+      "step": 4135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.871665532780535e-05,
+      "loss": 1.4111,
+      "step": 4136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.855107469876543e-05,
+      "loss": 1.427,
+      "step": 4137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8385679119979e-05,
+      "loss": 1.4186,
+      "step": 4138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.822046866238513e-05,
+      "loss": 1.4428,
+      "step": 4139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.805544339684295e-05,
+      "loss": 1.4905,
+      "step": 4140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.789060339413272e-05,
+      "loss": 1.4049,
+      "step": 4141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.772594872495491e-05,
+      "loss": 1.4238,
+      "step": 4142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.756147945993063e-05,
+      "loss": 1.4159,
+      "step": 4143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.739719566960146e-05,
+      "loss": 1.4348,
+      "step": 4144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.723309742442951e-05,
+      "loss": 1.4437,
+      "step": 4145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.706918479479696e-05,
+      "loss": 1.4559,
+      "step": 4146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.690545785100671e-05,
+      "loss": 1.432,
+      "step": 4147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6741916663282e-05,
+      "loss": 1.4856,
+      "step": 4148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.657856130176632e-05,
+      "loss": 1.377,
+      "step": 4149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.641539183652345e-05,
+      "loss": 1.3983,
+      "step": 4150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.625240833753755e-05,
+      "loss": 1.4344,
+      "step": 4151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.608961087471276e-05,
+      "loss": 1.4498,
+      "step": 4152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.592699951787362e-05,
+      "loss": 1.4274,
+      "step": 4153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.576457433676492e-05,
+      "loss": 1.4574,
+      "step": 4154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.560233540105143e-05,
+      "loss": 1.4231,
+      "step": 4155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.544028278031816e-05,
+      "loss": 1.4167,
+      "step": 4156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.527841654407024e-05,
+      "loss": 1.4387,
+      "step": 4157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.511673676173252e-05,
+      "loss": 1.4534,
+      "step": 4158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.495524350265037e-05,
+      "loss": 1.3765,
+      "step": 4159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.479393683608875e-05,
+      "loss": 1.44,
+      "step": 4160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.46328168312329e-05,
+      "loss": 1.4189,
+      "step": 4161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.447188355718786e-05,
+      "loss": 1.427,
+      "step": 4162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.431113708297853e-05,
+      "loss": 1.469,
+      "step": 4163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.415057747754976e-05,
+      "loss": 1.4044,
+      "step": 4164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.399020480976625e-05,
+      "loss": 1.4441,
+      "step": 4165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.383001914841253e-05,
+      "loss": 1.4221,
+      "step": 4166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.367002056219284e-05,
+      "loss": 1.4718,
+      "step": 4167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.351020911973132e-05,
+      "loss": 1.4238,
+      "step": 4168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.335058488957179e-05,
+      "loss": 1.4192,
+      "step": 4169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.31911479401776e-05,
+      "loss": 1.4311,
+      "step": 4170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.303189833993207e-05,
+      "loss": 1.4486,
+      "step": 4171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.287283615713796e-05,
+      "loss": 1.4684,
+      "step": 4172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.271396146001767e-05,
+      "loss": 1.4675,
+      "step": 4173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.255527431671332e-05,
+      "loss": 1.4408,
+      "step": 4174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.23967747952865e-05,
+      "loss": 1.4424,
+      "step": 4175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.223846296371821e-05,
+      "loss": 1.397,
+      "step": 4176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.208033888990911e-05,
+      "loss": 1.4497,
+      "step": 4177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.192240264167926e-05,
+      "loss": 1.465,
+      "step": 4178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.17646542867682e-05,
+      "loss": 1.4151,
+      "step": 4179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.160709389283481e-05,
+      "loss": 1.4274,
+      "step": 4180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.144972152745753e-05,
+      "loss": 1.4181,
+      "step": 4181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.129253725813405e-05,
+      "loss": 1.4033,
+      "step": 4182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.113554115228115e-05,
+      "loss": 1.4392,
+      "step": 4183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.097873327723519e-05,
+      "loss": 1.4377,
+      "step": 4184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.082211370025181e-05,
+      "loss": 1.4309,
+      "step": 4185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0665682488505735e-05,
+      "loss": 1.4229,
+      "step": 4186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.050943970909101e-05,
+      "loss": 1.4049,
+      "step": 4187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.035338542902091e-05,
+      "loss": 1.4862,
+      "step": 4188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.019751971522758e-05,
+      "loss": 1.4421,
+      "step": 4189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0041842634562525e-05,
+      "loss": 1.4788,
+      "step": 4190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.988635425379635e-05,
+      "loss": 1.4433,
+      "step": 4191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.973105463961864e-05,
+      "loss": 1.3871,
+      "step": 4192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.957594385863807e-05,
+      "loss": 1.4409,
+      "step": 4193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.942102197738231e-05,
+      "loss": 1.4481,
+      "step": 4194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.926628906229803e-05,
+      "loss": 1.424,
+      "step": 4195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.911174517975077e-05,
+      "loss": 1.4099,
+      "step": 4196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.895739039602505e-05,
+      "loss": 1.3983,
+      "step": 4197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.88032247773243e-05,
+      "loss": 1.4146,
+      "step": 4198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.864924838977087e-05,
+      "loss": 1.4942,
+      "step": 4199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.849546129940586e-05,
+      "loss": 1.4103,
+      "step": 4200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.834186357218907e-05,
+      "loss": 1.4477,
+      "step": 4201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.81884552739993e-05,
+      "loss": 1.4285,
+      "step": 4202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8035236470634024e-05,
+      "loss": 1.4301,
+      "step": 4203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.788220722780935e-05,
+      "loss": 1.4128,
+      "step": 4204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7729367611160265e-05,
+      "loss": 1.4339,
+      "step": 4205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.75767176862404e-05,
+      "loss": 1.4593,
+      "step": 4206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.74242575185217e-05,
+      "loss": 1.4074,
+      "step": 4207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.72719871733951e-05,
+      "loss": 1.3807,
+      "step": 4208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.711990671616995e-05,
+      "loss": 1.4795,
+      "step": 4209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6968016212074313e-05,
+      "loss": 1.4508,
+      "step": 4210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.681631572625451e-05,
+      "loss": 1.4269,
+      "step": 4211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6664805323775715e-05,
+      "loss": 1.4353,
+      "step": 4212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6513485069621174e-05,
+      "loss": 1.4758,
+      "step": 4213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.636235502869286e-05,
+      "loss": 1.4338,
+      "step": 4214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.621141526581103e-05,
+      "loss": 1.3977,
+      "step": 4215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.606066584571446e-05,
+      "loss": 1.4438,
+      "step": 4216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.591010683306014e-05,
+      "loss": 1.4052,
+      "step": 4217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.575973829242364e-05,
+      "loss": 1.4153,
+      "step": 4218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.560956028829839e-05,
+      "loss": 1.4354,
+      "step": 4219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.545957288509651e-05,
+      "loss": 1.45,
+      "step": 4220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.530977614714816e-05,
+      "loss": 1.3694,
+      "step": 4221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5160170138701895e-05,
+      "loss": 1.4587,
+      "step": 4222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5010754923924267e-05,
+      "loss": 1.4294,
+      "step": 4223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.486153056690013e-05,
+      "loss": 1.42,
+      "step": 4224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4712497131632396e-05,
+      "loss": 1.4631,
+      "step": 4225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.456365468204222e-05,
+      "loss": 1.4358,
+      "step": 4226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.441500328196869e-05,
+      "loss": 1.403,
+      "step": 4227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4266542995168975e-05,
+      "loss": 1.4032,
+      "step": 4228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.411827388531837e-05,
+      "loss": 1.4692,
+      "step": 4229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3970196016010274e-05,
+      "loss": 1.472,
+      "step": 4230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3822309450755555e-05,
+      "loss": 1.4856,
+      "step": 4231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.367461425298359e-05,
+      "loss": 1.445,
+      "step": 4232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3527110486041476e-05,
+      "loss": 1.4685,
+      "step": 4233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.337979821319416e-05,
+      "loss": 1.4105,
+      "step": 4234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3232677497624504e-05,
+      "loss": 1.3658,
+      "step": 4235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3085748402433244e-05,
+      "loss": 1.4188,
+      "step": 4236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.29390109906388e-05,
+      "loss": 1.428,
+      "step": 4237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.279246532517745e-05,
+      "loss": 1.4565,
+      "step": 4238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.264611146890336e-05,
+      "loss": 1.4047,
+      "step": 4239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2499949484588196e-05,
+      "loss": 1.4379,
+      "step": 4240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235397943492154e-05,
+      "loss": 1.4063,
+      "step": 4241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.220820138251053e-05,
+      "loss": 1.4067,
+      "step": 4242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.206261538987994e-05,
+      "loss": 1.4418,
+      "step": 4243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191722151947226e-05,
+      "loss": 1.4284,
+      "step": 4244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1772019833647486e-05,
+      "loss": 1.4804,
+      "step": 4245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.162701039468326e-05,
+      "loss": 1.4453,
+      "step": 4246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1482193264774714e-05,
+      "loss": 1.4568,
+      "step": 4247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.133756850603461e-05,
+      "loss": 1.431,
+      "step": 4248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.119313618049309e-05,
+      "loss": 1.4443,
+      "step": 4249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.104889635009763e-05,
+      "loss": 1.4206,
+      "step": 4250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0904849076713445e-05,
+      "loss": 1.4903,
+      "step": 4251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.076099442212301e-05,
+      "loss": 1.4774,
+      "step": 4252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.061733244802608e-05,
+      "loss": 1.4589,
+      "step": 4253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0473863216040026e-05,
+      "loss": 1.4303,
+      "step": 4254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.03305867876993e-05,
+      "loss": 1.4458,
+      "step": 4255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.018750322445581e-05,
+      "loss": 1.4651,
+      "step": 4256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.004461258767873e-05,
+      "loss": 1.4585,
+      "step": 4257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.990191493865437e-05,
+      "loss": 1.3859,
+      "step": 4258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.975941033858639e-05,
+      "loss": 1.4212,
+      "step": 4259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.961709884859561e-05,
+      "loss": 1.4265,
+      "step": 4260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.94749805297201e-05,
+      "loss": 1.3955,
+      "step": 4261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.933305544291483e-05,
+      "loss": 1.4303,
+      "step": 4262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.919132364905221e-05,
+      "loss": 1.4169,
+      "step": 4263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.90497852089215e-05,
+      "loss": 1.3815,
+      "step": 4264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.890844018322921e-05,
+      "loss": 1.4391,
+      "step": 4265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8767288632598774e-05,
+      "loss": 1.4043,
+      "step": 4266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.862633061757082e-05,
+      "loss": 1.4403,
+      "step": 4267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.848556619860262e-05,
+      "loss": 1.4077,
+      "step": 4268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.834499543606869e-05,
+      "loss": 1.4534,
+      "step": 4269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.820461839026047e-05,
+      "loss": 1.4243,
+      "step": 4270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8064435121386194e-05,
+      "loss": 1.4496,
+      "step": 4271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.79244456895711e-05,
+      "loss": 1.4715,
+      "step": 4272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7784650154857335e-05,
+      "loss": 1.4033,
+      "step": 4273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.764504857720364e-05,
+      "loss": 1.4372,
+      "step": 4274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.750564101648575e-05,
+      "loss": 1.4532,
+      "step": 4275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.736642753249615e-05,
+      "loss": 1.4294,
+      "step": 4276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7227408184944145e-05,
+      "loss": 1.4516,
+      "step": 4277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.70885830334557e-05,
+      "loss": 1.4176,
+      "step": 4278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6949952137573524e-05,
+      "loss": 1.4381,
+      "step": 4279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.681151555675689e-05,
+      "loss": 1.4514,
+      "step": 4280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6673273350381906e-05,
+      "loss": 1.4307,
+      "step": 4281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.65352255777412e-05,
+      "loss": 1.4392,
+      "step": 4282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.639737229804403e-05,
+      "loss": 1.4111,
+      "step": 4283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.625971357041631e-05,
+      "loss": 1.4147,
+      "step": 4284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6122249453900424e-05,
+      "loss": 1.3992,
+      "step": 4285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5984980007455265e-05,
+      "loss": 1.4063,
+      "step": 4286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5847905289956304e-05,
+      "loss": 1.4473,
+      "step": 4287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57110253601955e-05,
+      "loss": 1.4067,
+      "step": 4288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.557434027688118e-05,
+      "loss": 1.4579,
+      "step": 4289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.54378500986381e-05,
+      "loss": 1.4519,
+      "step": 4290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5301554884007725e-05,
+      "loss": 1.4109,
+      "step": 4291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.516545469144728e-05,
+      "loss": 1.4052,
+      "step": 4292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.502954957933092e-05,
+      "loss": 1.4168,
+      "step": 4293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.489383960594884e-05,
+      "loss": 1.4302,
+      "step": 4294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4758324829507646e-05,
+      "loss": 1.4252,
+      "step": 4295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.462300530813024e-05,
+      "loss": 1.427,
+      "step": 4296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.44878810998558e-05,
+      "loss": 1.4268,
+      "step": 4297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4352952262639414e-05,
+      "loss": 1.4355,
+      "step": 4298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.421821885435279e-05,
+      "loss": 1.4393,
+      "step": 4299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.408368093278364e-05,
+      "loss": 1.4366,
+      "step": 4300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.394933855563582e-05,
+      "loss": 1.4571,
+      "step": 4301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.38151917805294e-05,
+      "loss": 1.442,
+      "step": 4302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.368124066500051e-05,
+      "loss": 1.4264,
+      "step": 4303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.354748526650115e-05,
+      "loss": 1.4271,
+      "step": 4304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3413925642399765e-05,
+      "loss": 1.413,
+      "step": 4305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.328056184998058e-05,
+      "loss": 1.4638,
+      "step": 4306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3147393946443845e-05,
+      "loss": 1.4306,
+      "step": 4307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3014421988905874e-05,
+      "loss": 1.43,
+      "step": 4308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.288164603439892e-05,
+      "loss": 1.4142,
+      "step": 4309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.274906613987123e-05,
+      "loss": 1.3948,
+      "step": 4310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.261668236218663e-05,
+      "loss": 1.3979,
+      "step": 4311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.248449475812521e-05,
+      "loss": 1.4106,
+      "step": 4312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2352503384382846e-05,
+      "loss": 1.4313,
+      "step": 4313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.222070829757107e-05,
+      "loss": 1.4281,
+      "step": 4314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.208910955421747e-05,
+      "loss": 1.3852,
+      "step": 4315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.195770721076525e-05,
+      "loss": 1.4586,
+      "step": 4316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.182650132357341e-05,
+      "loss": 1.4189,
+      "step": 4317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1695491948916695e-05,
+      "loss": 1.4353,
+      "step": 4318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.156467914298562e-05,
+      "loss": 1.4551,
+      "step": 4319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1434062961886285e-05,
+      "loss": 1.4616,
+      "step": 4320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.130364346164056e-05,
+      "loss": 1.4239,
+      "step": 4321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1173420698186025e-05,
+      "loss": 1.4561,
+      "step": 4322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.104339472737556e-05,
+      "loss": 1.4426,
+      "step": 4323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.091356560497794e-05,
+      "loss": 1.4972,
+      "step": 4324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0783933386677464e-05,
+      "loss": 1.4447,
+      "step": 4325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.065449812807387e-05,
+      "loss": 1.4427,
+      "step": 4326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0525259884682475e-05,
+      "loss": 1.4502,
+      "step": 4327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0396218711934276e-05,
+      "loss": 1.444,
+      "step": 4328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.026737466517533e-05,
+      "loss": 1.4053,
+      "step": 4329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0138727799667484e-05,
+      "loss": 1.3965,
+      "step": 4330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.001027817058789e-05,
+      "loss": 1.436,
+      "step": 4331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.988202583302925e-05,
+      "loss": 1.4782,
+      "step": 4332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.975397084199939e-05,
+      "loss": 1.4381,
+      "step": 4333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.962611325242177e-05,
+      "loss": 1.4413,
+      "step": 4334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.949845311913491e-05,
+      "loss": 1.3994,
+      "step": 4335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9370990496892856e-05,
+      "loss": 1.4294,
+      "step": 4336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9243725440364806e-05,
+      "loss": 1.4042,
+      "step": 4337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.911665800413533e-05,
+      "loss": 1.415,
+      "step": 4338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.898978824270416e-05,
+      "loss": 1.4347,
+      "step": 4339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.886311621048638e-05,
+      "loss": 1.4532,
+      "step": 4340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8736641961811925e-05,
+      "loss": 1.3926,
+      "step": 4341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8610365550926285e-05,
+      "loss": 1.4512,
+      "step": 4342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8484287031989864e-05,
+      "loss": 1.4433,
+      "step": 4343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.83584064590784e-05,
+      "loss": 1.4103,
+      "step": 4344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.823272388618243e-05,
+      "loss": 1.4546,
+      "step": 4345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.810723936720789e-05,
+      "loss": 1.4334,
+      "step": 4346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7981952955975505e-05,
+      "loss": 1.4381,
+      "step": 4347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.785686470622118e-05,
+      "loss": 1.4238,
+      "step": 4348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7731974671595835e-05,
+      "loss": 1.4107,
+      "step": 4349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.760728290566523e-05,
+      "loss": 1.4288,
+      "step": 4350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7482789461910305e-05,
+      "loss": 1.3985,
+      "step": 4351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.735849439372685e-05,
+      "loss": 1.4649,
+      "step": 4352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.723439775442533e-05,
+      "loss": 1.4094,
+      "step": 4353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.711049959723145e-05,
+      "loss": 1.4202,
+      "step": 4354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.698679997528559e-05,
+      "loss": 1.435,
+      "step": 4355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6863298941643144e-05,
+      "loss": 1.4142,
+      "step": 4356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.673999654927413e-05,
+      "loss": 1.4255,
+      "step": 4357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.661689285106362e-05,
+      "loss": 1.4257,
+      "step": 4358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6493987899811066e-05,
+      "loss": 1.4358,
+      "step": 4359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.637128174823101e-05,
+      "loss": 1.4011,
+      "step": 4360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.624877444895269e-05,
+      "loss": 1.4342,
+      "step": 4361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.612646605451997e-05,
+      "loss": 1.4372,
+      "step": 4362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6004356617391375e-05,
+      "loss": 1.4276,
+      "step": 4363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5882446189940355e-05,
+      "loss": 1.4483,
+      "step": 4364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.576073482445452e-05,
+      "loss": 1.4263,
+      "step": 4365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5639222573136566e-05,
+      "loss": 1.4113,
+      "step": 4366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5517909488103514e-05,
+      "loss": 1.4949,
+      "step": 4367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5396795621387144e-05,
+      "loss": 1.4565,
+      "step": 4368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5275881024933675e-05,
+      "loss": 1.4057,
+      "step": 4369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.51551657506039e-05,
+      "loss": 1.4478,
+      "step": 4370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5034649850173074e-05,
+      "loss": 1.4603,
+      "step": 4371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.49143333753309e-05,
+      "loss": 1.4174,
+      "step": 4372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.479421637768176e-05,
+      "loss": 1.3977,
+      "step": 4373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.467429890874424e-05,
+      "loss": 1.394,
+      "step": 4374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.455458101995146e-05,
+      "loss": 1.4374,
+      "step": 4375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.443506276265096e-05,
+      "loss": 1.452,
+      "step": 4376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4315744188104626e-05,
+      "loss": 1.4342,
+      "step": 4377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.419662534748863e-05,
+      "loss": 1.4756,
+      "step": 4378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.407770629189361e-05,
+      "loss": 1.4209,
+      "step": 4379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.395898707232442e-05,
+      "loss": 1.428,
+      "step": 4380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.384046773970023e-05,
+      "loss": 1.4041,
+      "step": 4381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.372214834485449e-05,
+      "loss": 1.4345,
+      "step": 4382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.360402893853498e-05,
+      "loss": 1.4634,
+      "step": 4383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3486109571403414e-05,
+      "loss": 1.4496,
+      "step": 4384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.336839029403599e-05,
+      "loss": 1.3995,
+      "step": 4385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.325087115692299e-05,
+      "loss": 1.4381,
+      "step": 4386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3133552210468876e-05,
+      "loss": 1.4463,
+      "step": 4387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.301643350499228e-05,
+      "loss": 1.4154,
+      "step": 4388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2899515090725916e-05,
+      "loss": 1.4526,
+      "step": 4389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2782797017816504e-05,
+      "loss": 1.3987,
+      "step": 4390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.266627933632488e-05,
+      "loss": 1.3876,
+      "step": 4391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.254996209622607e-05,
+      "loss": 1.4174,
+      "step": 4392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2433845347408955e-05,
+      "loss": 1.4418,
+      "step": 4393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.231792913967657e-05,
+      "loss": 1.4329,
+      "step": 4394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2202213522745825e-05,
+      "loss": 1.4199,
+      "step": 4395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.208669854624763e-05,
+      "loss": 1.4002,
+      "step": 4396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.197138425972679e-05,
+      "loss": 1.4221,
+      "step": 4397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1856270712642185e-05,
+      "loss": 1.3894,
+      "step": 4398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.174135795436639e-05,
+      "loss": 1.4278,
+      "step": 4399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.162664603418608e-05,
+      "loss": 1.4502,
+      "step": 4400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.151213500130168e-05,
+      "loss": 1.4273,
+      "step": 4401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1397824904827345e-05,
+      "loss": 1.4289,
+      "step": 4402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.128371579379119e-05,
+      "loss": 1.4351,
+      "step": 4403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1169807717135126e-05,
+      "loss": 1.4569,
+      "step": 4404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.105610072371473e-05,
+      "loss": 1.4412,
+      "step": 4405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.094259486229956e-05,
+      "loss": 1.4231,
+      "step": 4406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.08292901815726e-05,
+      "loss": 1.4445,
+      "step": 4407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.071618673013077e-05,
+      "loss": 1.4285,
+      "step": 4408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.060328455648464e-05,
+      "loss": 1.4287,
+      "step": 4409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0490583709058384e-05,
+      "loss": 1.4029,
+      "step": 4410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.037808423618993e-05,
+      "loss": 1.4246,
+      "step": 4411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0265786186130627e-05,
+      "loss": 1.4112,
+      "step": 4412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0153689607045842e-05,
+      "loss": 1.4556,
+      "step": 4413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0041794547014e-05,
+      "loss": 1.4332,
+      "step": 4414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9930101054027437e-05,
+      "loss": 1.4284,
+      "step": 4415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.981860917599205e-05,
+      "loss": 1.4217,
+      "step": 4416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9707318960727038e-05,
+      "loss": 1.428,
+      "step": 4417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.959623045596538e-05,
+      "loss": 1.4133,
+      "step": 4418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.948534370935335e-05,
+      "loss": 1.4541,
+      "step": 4419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.937465876845069e-05,
+      "loss": 1.4515,
+      "step": 4420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9264175680730698e-05,
+      "loss": 1.4563,
+      "step": 4421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.915389449357997e-05,
+      "loss": 1.4824,
+      "step": 4422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9043815254298612e-05,
+      "loss": 1.4509,
+      "step": 4423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8933938010100026e-05,
+      "loss": 1.4154,
+      "step": 4424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8824262808111177e-05,
+      "loss": 1.4609,
+      "step": 4425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8714789695372055e-05,
+      "loss": 1.4009,
+      "step": 4426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8605518718836156e-05,
+      "loss": 1.4342,
+      "step": 4427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849644992537026e-05,
+      "loss": 1.44,
+      "step": 4428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838758336175451e-05,
+      "loss": 1.4196,
+      "step": 4429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8278919074682163e-05,
+      "loss": 1.433,
+      "step": 4430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8170457110759885e-05,
+      "loss": 1.4495,
+      "step": 4431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8062197516507348e-05,
+      "loss": 1.4258,
+      "step": 4432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7954140338357526e-05,
+      "loss": 1.4115,
+      "step": 4433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7846285622656732e-05,
+      "loss": 1.3877,
+      "step": 4434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7738633415664183e-05,
+      "loss": 1.4368,
+      "step": 4435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7631183763552393e-05,
+      "loss": 1.3988,
+      "step": 4436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7523936712406995e-05,
+      "loss": 1.4177,
+      "step": 4437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7416892308226694e-05,
+      "loss": 1.4197,
+      "step": 4438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.731005059692332e-05,
+      "loss": 1.4028,
+      "step": 4439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.720341162432166e-05,
+      "loss": 1.4147,
+      "step": 4440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7096975436159633e-05,
+      "loss": 1.4172,
+      "step": 4441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.699074207808816e-05,
+      "loss": 1.4409,
+      "step": 4442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.688471159567124e-05,
+      "loss": 1.4301,
+      "step": 4443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.677888403438583e-05,
+      "loss": 1.4659,
+      "step": 4444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.667325943962162e-05,
+      "loss": 1.4557,
+      "step": 4445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.656783785668154e-05,
+      "loss": 1.4314,
+      "step": 4446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6462619330781424e-05,
+      "loss": 1.459,
+      "step": 4447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.635760390704983e-05,
+      "loss": 1.4366,
+      "step": 4448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.625279163052835e-05,
+      "loss": 1.4191,
+      "step": 4449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.614818254617152e-05,
+      "loss": 1.4164,
+      "step": 4450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6043776698846445e-05,
+      "loss": 1.4007,
+      "step": 4451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.593957413333331e-05,
+      "loss": 1.4725,
+      "step": 4452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5835574894324975e-05,
+      "loss": 1.4025,
+      "step": 4453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.573177902642726e-05,
+      "loss": 1.414,
+      "step": 4454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.562818657415855e-05,
+      "loss": 1.4174,
+      "step": 4455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5524797581950143e-05,
+      "loss": 1.4341,
+      "step": 4456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5421612094146008e-05,
+      "loss": 1.4019,
+      "step": 4457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.531863015500274e-05,
+      "loss": 1.4126,
+      "step": 4458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5215851808689795e-05,
+      "loss": 1.4257,
+      "step": 4459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.511327709928929e-05,
+      "loss": 1.4147,
+      "step": 4460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501090607079587e-05,
+      "loss": 1.4224,
+      "step": 4461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.490873876711697e-05,
+      "loss": 1.4167,
+      "step": 4462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4806775232072532e-05,
+      "loss": 1.3974,
+      "step": 4463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470501550939508e-05,
+      "loss": 1.4425,
+      "step": 4464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4603459642729864e-05,
+      "loss": 1.4074,
+      "step": 4465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.45021076756346e-05,
+      "loss": 1.4096,
+      "step": 4466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.440095965157957e-05,
+      "loss": 1.4436,
+      "step": 4467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4300015613947568e-05,
+      "loss": 1.3753,
+      "step": 4468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4199275606033966e-05,
+      "loss": 1.4018,
+      "step": 4469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.409873967104659e-05,
+      "loss": 1.4055,
+      "step": 4470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3998407852105607e-05,
+      "loss": 1.3872,
+      "step": 4471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.389828019224388e-05,
+      "loss": 1.4131,
+      "step": 4472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3798356734406502e-05,
+      "loss": 1.4614,
+      "step": 4473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3698637521451184e-05,
+      "loss": 1.3908,
+      "step": 4474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.359912259614777e-05,
+      "loss": 1.4069,
+      "step": 4475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3499812001178678e-05,
+      "loss": 1.4529,
+      "step": 4476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.340070577913861e-05,
+      "loss": 1.4355,
+      "step": 4477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.330180397253473e-05,
+      "loss": 1.4595,
+      "step": 4478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.320310662378633e-05,
+      "loss": 1.4071,
+      "step": 4479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.310461377522527e-05,
+      "loss": 1.412,
+      "step": 4480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.300632546909537e-05,
+      "loss": 1.4457,
+      "step": 4481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.290824174755296e-05,
+      "loss": 1.4364,
+      "step": 4482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.281036265266656e-05,
+      "loss": 1.4042,
+      "step": 4483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2712688226416976e-05,
+      "loss": 1.384,
+      "step": 4484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2615218510697088e-05,
+      "loss": 1.4303,
+      "step": 4485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.251795354731223e-05,
+      "loss": 1.4563,
+      "step": 4486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2420893377979533e-05,
+      "loss": 1.5022,
+      "step": 4487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.232403804432864e-05,
+      "loss": 1.4051,
+      "step": 4488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2227387587901216e-05,
+      "loss": 1.4082,
+      "step": 4489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.213094205015098e-05,
+      "loss": 1.4317,
+      "step": 4490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.203470147244385e-05,
+      "loss": 1.4312,
+      "step": 4491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1938665896057918e-05,
+      "loss": 1.4507,
+      "step": 4492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1842835362183068e-05,
+      "loss": 1.409,
+      "step": 4493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1747209911921528e-05,
+      "loss": 1.4384,
+      "step": 4494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.165178958628744e-05,
+      "loss": 1.4372,
+      "step": 4495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1556574426206954e-05,
+      "loss": 1.4338,
+      "step": 4496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.146156447251829e-05,
+      "loss": 1.4098,
+      "step": 4497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1366759765971577e-05,
+      "loss": 1.3828,
+      "step": 4498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1272160347229007e-05,
+      "loss": 1.4455,
+      "step": 4499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1177766256864628e-05,
+      "loss": 1.4202,
+      "step": 4500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.108357753536444e-05,
+      "loss": 1.4457,
+      "step": 4501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.098959422312641e-05,
+      "loss": 1.4251,
+      "step": 4502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0895816360460395e-05,
+      "loss": 1.4138,
+      "step": 4503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0802243987588066e-05,
+      "loss": 1.4663,
+      "step": 4504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.070887714464309e-05,
+      "loss": 1.4336,
+      "step": 4505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.061571587167077e-05,
+      "loss": 1.4241,
+      "step": 4506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0522760208628423e-05,
+      "loss": 1.416,
+      "step": 4507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.043001019538515e-05,
+      "loss": 1.405,
+      "step": 4508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.033746587172186e-05,
+      "loss": 1.3909,
+      "step": 4509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0245127277331076e-05,
+      "loss": 1.4535,
+      "step": 4510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.015299445181734e-05,
+      "loss": 1.4721,
+      "step": 4511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0061067434696712e-05,
+      "loss": 1.4123,
+      "step": 4512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9969346265397094e-05,
+      "loss": 1.4163,
+      "step": 4513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9877830983258126e-05,
+      "loss": 1.4716,
+      "step": 4514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.978652162753103e-05,
+      "loss": 1.4407,
+      "step": 4515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969541823737886e-05,
+      "loss": 1.442,
+      "step": 4516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9604520851876197e-05,
+      "loss": 1.4459,
+      "step": 4517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.951382951000924e-05,
+      "loss": 1.4537,
+      "step": 4518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.942334425067599e-05,
+      "loss": 1.4084,
+      "step": 4519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9333065112685845e-05,
+      "loss": 1.43,
+      "step": 4520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9242992134760052e-05,
+      "loss": 1.4296,
+      "step": 4521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9153125355531153e-05,
+      "loss": 1.4317,
+      "step": 4522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.906346481354354e-05,
+      "loss": 1.4368,
+      "step": 4523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.897401054725284e-05,
+      "loss": 1.4023,
+      "step": 4524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.888476259502636e-05,
+      "loss": 1.4083,
+      "step": 4525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.879572099514304e-05,
+      "loss": 1.4871,
+      "step": 4526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.870688578579316e-05,
+      "loss": 1.4368,
+      "step": 4527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8618257005078464e-05,
+      "loss": 1.4271,
+      "step": 4528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8529834691012216e-05,
+      "loss": 1.421,
+      "step": 4529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8441618881519185e-05,
+      "loss": 1.4173,
+      "step": 4530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.835360961443544e-05,
+      "loss": 1.4558,
+      "step": 4531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.826580692750851e-05,
+      "loss": 1.4294,
+      "step": 4532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.817821085839738e-05,
+      "loss": 1.4542,
+      "step": 4533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.809082144467239e-05,
+      "loss": 1.4172,
+      "step": 4534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.800363872381522e-05,
+      "loss": 1.4495,
+      "step": 4535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7916662733218848e-05,
+      "loss": 1.4506,
+      "step": 4536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.78298935101876e-05,
+      "loss": 1.4172,
+      "step": 4537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.774333109193732e-05,
+      "loss": 1.4351,
+      "step": 4538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7656975515594863e-05,
+      "loss": 1.442,
+      "step": 4539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7570826818198495e-05,
+      "loss": 1.457,
+      "step": 4540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7484885036697928e-05,
+      "loss": 1.4435,
+      "step": 4541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.739915020795374e-05,
+      "loss": 1.3879,
+      "step": 4542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7313622368738013e-05,
+      "loss": 1.4396,
+      "step": 4543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7228301555734016e-05,
+      "loss": 1.414,
+      "step": 4544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7143187805536254e-05,
+      "loss": 1.4542,
+      "step": 4545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.70582811546503e-05,
+      "loss": 1.4001,
+      "step": 4546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6973581639493085e-05,
+      "loss": 1.4422,
+      "step": 4547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6889089296392435e-05,
+      "loss": 1.4456,
+      "step": 4548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6804804161587526e-05,
+      "loss": 1.4334,
+      "step": 4549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6720726271228615e-05,
+      "loss": 1.436,
+      "step": 4550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6636855661377013e-05,
+      "loss": 1.3965,
+      "step": 4551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6553192368005286e-05,
+      "loss": 1.4377,
+      "step": 4552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6469736426997008e-05,
+      "loss": 1.4164,
+      "step": 4553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6386487874146537e-05,
+      "loss": 1.4713,
+      "step": 4554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.630344674515971e-05,
+      "loss": 1.4471,
+      "step": 4555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220613075653202e-05,
+      "loss": 1.4224,
+      "step": 4556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61379869011547e-05,
+      "loss": 1.4537,
+      "step": 4557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6055568257102914e-05,
+      "loss": 1.485,
+      "step": 4558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5973357178847515e-05,
+      "loss": 1.4253,
+      "step": 4559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5891353701649235e-05,
+      "loss": 1.4349,
+      "step": 4560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5809557860679723e-05,
+      "loss": 1.3967,
+      "step": 4561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.572796969102147e-05,
+      "loss": 1.4008,
+      "step": 4562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.564658922766804e-05,
+      "loss": 1.4228,
+      "step": 4563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.55654165055239e-05,
+      "loss": 1.4165,
+      "step": 4564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5484451559404312e-05,
+      "loss": 1.4109,
+      "step": 4565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5403694424035498e-05,
+      "loss": 1.4863,
+      "step": 4566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5323145134054526e-05,
+      "loss": 1.4297,
+      "step": 4567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.524280372400927e-05,
+      "loss": 1.4324,
+      "step": 4568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.51626702283586e-05,
+      "loss": 1.4561,
+      "step": 4569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5082744681472027e-05,
+      "loss": 1.3898,
+      "step": 4570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5003027117630019e-05,
+      "loss": 1.4379,
+      "step": 4571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4923517571023782e-05,
+      "loss": 1.4216,
+      "step": 4572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4844216075755201e-05,
+      "loss": 1.3956,
+      "step": 4573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4765122665837073e-05,
+      "loss": 1.3857,
+      "step": 4574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4686237375192868e-05,
+      "loss": 1.4588,
+      "step": 4575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4607560237656858e-05,
+      "loss": 1.4143,
+      "step": 4576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4529091286973995e-05,
+      "loss": 1.4263,
+      "step": 4577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4450830556799965e-05,
+      "loss": 1.4413,
+      "step": 4578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4372778080701032e-05,
+      "loss": 1.4021,
+      "step": 4579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.429493389215425e-05,
+      "loss": 1.3752,
+      "step": 4580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.421729802454741e-05,
+      "loss": 1.4441,
+      "step": 4581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4139870511178766e-05,
+      "loss": 1.4418,
+      "step": 4582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4062651385257364e-05,
+      "loss": 1.4282,
+      "step": 4583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3985640679902877e-05,
+      "loss": 1.3885,
+      "step": 4584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3908838428145331e-05,
+      "loss": 1.4437,
+      "step": 4585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3832244662925598e-05,
+      "loss": 1.3809,
+      "step": 4586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3755859417095174e-05,
+      "loss": 1.4228,
+      "step": 4587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.367968272341591e-05,
+      "loss": 1.3936,
+      "step": 4588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3603714614560337e-05,
+      "loss": 1.402,
+      "step": 4589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3527955123111447e-05,
+      "loss": 1.4276,
+      "step": 4590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3452404281562913e-05,
+      "loss": 1.4804,
+      "step": 4591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3377062122318705e-05,
+      "loss": 1.4032,
+      "step": 4592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3301928677693475e-05,
+      "loss": 1.442,
+      "step": 4593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3227003979912278e-05,
+      "loss": 1.3879,
+      "step": 4594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3152288061110518e-05,
+      "loss": 1.4129,
+      "step": 4595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.307778095333423e-05,
+      "loss": 1.4077,
+      "step": 4596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3003482688539792e-05,
+      "loss": 1.4464,
+      "step": 4597
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2929393298594106e-05,
+      "loss": 1.41,
+      "step": 4598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2855512815274418e-05,
+      "loss": 1.4033,
+      "step": 4599
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2781841270268324e-05,
+      "loss": 1.3863,
+      "step": 4600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2708378695173884e-05,
+      "loss": 1.4384,
+      "step": 4601
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2635125121499558e-05,
+      "loss": 1.4181,
+      "step": 4602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2562080580664048e-05,
+      "loss": 1.4051,
+      "step": 4603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2489245103996405e-05,
+      "loss": 1.4455,
+      "step": 4604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2416618722736195e-05,
+      "loss": 1.4474,
+      "step": 4605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2344201468033112e-05,
+      "loss": 1.4238,
+      "step": 4606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.22719933709472e-05,
+      "loss": 1.4304,
+      "step": 4607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2199994462448905e-05,
+      "loss": 1.4208,
+      "step": 4608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2128204773418806e-05,
+      "loss": 1.416,
+      "step": 4609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2056624334647715e-05,
+      "loss": 1.4112,
+      "step": 4610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1985253176836908e-05,
+      "loss": 1.4386,
+      "step": 4611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1914091330597732e-05,
+      "loss": 1.4208,
+      "step": 4612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1843138826451826e-05,
+      "loss": 1.4597,
+      "step": 4613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1772395694831017e-05,
+      "loss": 1.4481,
+      "step": 4614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1701861966077254e-05,
+      "loss": 1.4237,
+      "step": 4615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1631537670442781e-05,
+      "loss": 1.4158,
+      "step": 4616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.156142283809003e-05,
+      "loss": 1.3714,
+      "step": 4617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1491517499091497e-05,
+      "loss": 1.4081,
+      "step": 4618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1421821683429923e-05,
+      "loss": 1.4165,
+      "step": 4619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.135233542099806e-05,
+      "loss": 1.4339,
+      "step": 4620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1283058741598962e-05,
+      "loss": 1.442,
+      "step": 4621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1213991674945634e-05,
+      "loss": 1.4003,
+      "step": 4622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1145134250661216e-05,
+      "loss": 1.4048,
+      "step": 4623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1076486498278915e-05,
+      "loss": 1.3874,
+      "step": 4624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1008048447242126e-05,
+      "loss": 1.4318,
+      "step": 4625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0939820126904143e-05,
+      "loss": 1.4166,
+      "step": 4626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.087180156652845e-05,
+      "loss": 1.4011,
+      "step": 4627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0803992795288431e-05,
+      "loss": 1.4144,
+      "step": 4628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0736393842267545e-05,
+      "loss": 1.4013,
+      "step": 4629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0669004736459376e-05,
+      "loss": 1.4039,
+      "step": 4630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0601825506767248e-05,
+      "loss": 1.4197,
+      "step": 4631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.053485618200467e-05,
+      "loss": 1.4383,
+      "step": 4632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0468096790895166e-05,
+      "loss": 1.4411,
+      "step": 4633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0401547362071938e-05,
+      "loss": 1.4564,
+      "step": 4634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0335207924078439e-05,
+      "loss": 1.4204,
+      "step": 4635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0269078505367901e-05,
+      "loss": 1.4138,
+      "step": 4636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0203159134303474e-05,
+      "loss": 1.3986,
+      "step": 4637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0137449839158319e-05,
+      "loss": 1.4551,
+      "step": 4638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.007195064811539e-05,
+      "loss": 1.4141,
+      "step": 4639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0006661589267552e-05,
+      "loss": 1.4192,
+      "step": 4640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.941582690617623e-06,
+      "loss": 1.4127,
+      "step": 4641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.876713980078112e-06,
+      "loss": 1.4302,
+      "step": 4642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812055485471539e-06,
+      "loss": 1.4461,
+      "step": 4643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.74760723453022e-06,
+      "loss": 1.4678,
+      "step": 4644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.683369254896268e-06,
+      "loss": 1.4133,
+      "step": 4645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.619341574121642e-06,
+      "loss": 1.3979,
+      "step": 4646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.555524219667989e-06,
+      "loss": 1.4394,
+      "step": 4647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.49191721890691e-06,
+      "loss": 1.4733,
+      "step": 4648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.42852059911975e-06,
+      "loss": 1.4259,
+      "step": 4649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.365334387497481e-06,
+      "loss": 1.4266,
+      "step": 4650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.302358611141093e-06,
+      "loss": 1.3942,
+      "step": 4651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.239593297061032e-06,
+      "loss": 1.4112,
+      "step": 4652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.177038472177601e-06,
+      "loss": 1.4146,
+      "step": 4653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.114694163320891e-06,
+      "loss": 1.3924,
+      "step": 4654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.052560397230625e-06,
+      "loss": 1.4145,
+      "step": 4655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.990637200556151e-06,
+      "loss": 1.4058,
+      "step": 4656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.928924599856725e-06,
+      "loss": 1.4519,
+      "step": 4657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.867422621601062e-06,
+      "loss": 1.4193,
+      "step": 4658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.806131292167618e-06,
+      "loss": 1.3996,
+      "step": 4659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.745050637844532e-06,
+      "loss": 1.4233,
+      "step": 4660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.684180684829512e-06,
+      "loss": 1.4021,
+      "step": 4661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.623521459229955e-06,
+      "loss": 1.4364,
+      "step": 4662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.563072987062882e-06,
+      "loss": 1.4584,
+      "step": 4663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.502835294254885e-06,
+      "loss": 1.4293,
+      "step": 4664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.442808406642132e-06,
+      "loss": 1.4364,
+      "step": 4665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.38299234997042e-06,
+      "loss": 1.3754,
+      "step": 4666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.323387149895113e-06,
+      "loss": 1.4094,
+      "step": 4667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.263992831981149e-06,
+      "loss": 1.4066,
+      "step": 4668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.204809421702986e-06,
+      "loss": 1.3941,
+      "step": 4669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.145836944444651e-06,
+      "loss": 1.4594,
+      "step": 4670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.087075425499634e-06,
+      "loss": 1.4135,
+      "step": 4671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.028524890071054e-06,
+      "loss": 1.4525,
+      "step": 4672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.970185363271432e-06,
+      "loss": 1.3848,
+      "step": 4673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.912056870122863e-06,
+      "loss": 1.4221,
+      "step": 4674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.854139435556962e-06,
+      "loss": 1.4439,
+      "step": 4675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.79643308441469e-06,
+      "loss": 1.4423,
+      "step": 4676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.73893784144647e-06,
+      "loss": 1.4314,
+      "step": 4677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.68165373131241e-06,
+      "loss": 1.449,
+      "step": 4678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.6245807785818025e-06,
+      "loss": 1.4097,
+      "step": 4679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.567719007733509e-06,
+      "loss": 1.3803,
+      "step": 4680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5110684431558e-06,
+      "loss": 1.4098,
+      "step": 4681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.454629109146294e-06,
+      "loss": 1.4222,
+      "step": 4682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.398401029912127e-06,
+      "loss": 1.3881,
+      "step": 4683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.342384229569621e-06,
+      "loss": 1.4146,
+      "step": 4684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.286578732144778e-06,
+      "loss": 1.4858,
+      "step": 4685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.230984561572729e-06,
+      "loss": 1.4162,
+      "step": 4686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.1756017416980126e-06,
+      "loss": 1.4183,
+      "step": 4687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.120430296274683e-06,
+      "loss": 1.3945,
+      "step": 4688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.065470248965867e-06,
+      "loss": 1.4165,
+      "step": 4689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.010721623344207e-06,
+      "loss": 1.4428,
+      "step": 4690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.956184442891589e-06,
+      "loss": 1.3848,
+      "step": 4691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.901858730999355e-06,
+      "loss": 1.4074,
+      "step": 4692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8477445109678685e-06,
+      "loss": 1.4276,
+      "step": 4693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.793841806007117e-06,
+      "loss": 1.4017,
+      "step": 4694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.740150639236053e-06,
+      "loss": 1.4318,
+      "step": 4695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.686671033683089e-06,
+      "loss": 1.4546,
+      "step": 4696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.633403012285877e-06,
+      "loss": 1.4467,
+      "step": 4697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.580346597891251e-06,
+      "loss": 1.4594,
+      "step": 4698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.5275018132553435e-06,
+      "loss": 1.4376,
+      "step": 4699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.474868681043577e-06,
+      "loss": 1.4309,
+      "step": 4700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.422447223830452e-06,
+      "loss": 1.4203,
+      "step": 4701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.370237464099704e-06,
+      "loss": 1.4347,
+      "step": 4702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3182394242444205e-06,
+      "loss": 1.4744,
+      "step": 4703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.266453126566707e-06,
+      "loss": 1.4244,
+      "step": 4704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.214878593277962e-06,
+      "loss": 1.4029,
+      "step": 4705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.163515846498713e-06,
+      "loss": 1.4706,
+      "step": 4706
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.112364908258616e-06,
+      "loss": 1.4409,
+      "step": 4707
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.061425800496567e-06,
+      "loss": 1.4233,
+      "step": 4708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.010698545060589e-06,
+      "loss": 1.3977,
+      "step": 4709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9601831637077775e-06,
+      "loss": 1.4179,
+      "step": 4710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.909879678104357e-06,
+      "loss": 1.4033,
+      "step": 4711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.859788109825792e-06,
+      "loss": 1.4206,
+      "step": 4712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.809908480356452e-06,
+      "loss": 1.4125,
+      "step": 4713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.760240811090001e-06,
+      "loss": 1.4136,
+      "step": 4714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7107851233291205e-06,
+      "loss": 1.4038,
+      "step": 4715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.661541438285512e-06,
+      "loss": 1.4677,
+      "step": 4716
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.612509777079999e-06,
+      "loss": 1.4142,
+      "step": 4717
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.563690160742485e-06,
+      "loss": 1.413,
+      "step": 4718
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.515082610211941e-06,
+      "loss": 1.4414,
+      "step": 4719
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.466687146336302e-06,
+      "loss": 1.4191,
+      "step": 4720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.418503789872575e-06,
+      "loss": 1.4442,
+      "step": 4721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.370532561486896e-06,
+      "loss": 1.4204,
+      "step": 4722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.322773481754195e-06,
+      "loss": 1.4215,
+      "step": 4723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.275226571158697e-06,
+      "loss": 1.4592,
+      "step": 4724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.227891850093314e-06,
+      "loss": 1.421,
+      "step": 4725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1807693388601916e-06,
+      "loss": 1.4586,
+      "step": 4726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.133859057670332e-06,
+      "loss": 1.4429,
+      "step": 4727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.087161026643749e-06,
+      "loss": 1.3782,
+      "step": 4728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.040675265809536e-06,
+      "loss": 1.4179,
+      "step": 4729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.994401795105519e-06,
+      "loss": 1.4117,
+      "step": 4730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.948340634378545e-06,
+      "loss": 1.423,
+      "step": 4731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.90249180338459e-06,
+      "loss": 1.399,
+      "step": 4732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.856855321788256e-06,
+      "loss": 1.4139,
+      "step": 4733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.81143120916333e-06,
+      "loss": 1.4542,
+      "step": 4734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.76621948499234e-06,
+      "loss": 1.3971,
+      "step": 4735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7212201686668285e-06,
+      "loss": 1.4284,
+      "step": 4736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.676433279487135e-06,
+      "loss": 1.4357,
+      "step": 4737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.631858836662562e-06,
+      "loss": 1.4296,
+      "step": 4738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.587496859311313e-06,
+      "loss": 1.3867,
+      "step": 4739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.543347366460448e-06,
+      "loss": 1.447,
+      "step": 4740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.499410377045765e-06,
+      "loss": 1.4055,
+      "step": 4741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4556859099121325e-06,
+      "loss": 1.4141,
+      "step": 4742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.412173983813106e-06,
+      "loss": 1.3907,
+      "step": 4743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.368874617411089e-06,
+      "loss": 1.4037,
+      "step": 4744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.32578782927745e-06,
+      "loss": 1.4151,
+      "step": 4745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.282913637892239e-06,
+      "loss": 1.4009,
+      "step": 4746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.240252061644412e-06,
+      "loss": 1.4361,
+      "step": 4747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.197803118831611e-06,
+      "loss": 1.4722,
+      "step": 4748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.155566827660495e-06,
+      "loss": 1.3899,
+      "step": 4749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.113543206246295e-06,
+      "loss": 1.4396,
+      "step": 4750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.071732272613149e-06,
+      "loss": 1.4371,
+      "step": 4751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0301340446939334e-06,
+      "loss": 1.4099,
+      "step": 4752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.988748540330267e-06,
+      "loss": 1.4071,
+      "step": 4753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9475757772726716e-06,
+      "loss": 1.4126,
+      "step": 4754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.906615773180244e-06,
+      "loss": 1.4182,
+      "step": 4755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.865868545620876e-06,
+      "loss": 1.4293,
+      "step": 4756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8253341120713105e-06,
+      "loss": 1.4018,
+      "step": 4757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7850124899168616e-06,
+      "loss": 1.4424,
+      "step": 4758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7449036964516403e-06,
+      "loss": 1.4204,
+      "step": 4759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.705007748878497e-06,
+      "loss": 1.4271,
+      "step": 4760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.665324664309022e-06,
+      "loss": 1.4496,
+      "step": 4761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.625854459763378e-06,
+      "loss": 1.4221,
+      "step": 4762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5865971521705233e-06,
+      "loss": 1.4489,
+      "step": 4763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5475527583681e-06,
+      "loss": 1.4106,
+      "step": 4764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.508721295102435e-06,
+      "loss": 1.4006,
+      "step": 4765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4701027790284277e-06,
+      "loss": 1.4292,
+      "step": 4766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4316972267097734e-06,
+      "loss": 1.4506,
+      "step": 4767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.393504654618795e-06,
+      "loss": 1.3996,
+      "step": 4768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.35552507913639e-06,
+      "loss": 1.4103,
+      "step": 4769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3177585165522495e-06,
+      "loss": 1.426,
+      "step": 4770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.280204983064472e-06,
+      "loss": 1.441,
+      "step": 4771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2428644947800624e-06,
+      "loss": 1.4474,
+      "step": 4772
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2057370677144316e-06,
+      "loss": 1.4317,
+      "step": 4773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.168822717791675e-06,
+      "loss": 1.4338,
+      "step": 4774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.132121460844628e-06,
+      "loss": 1.4132,
+      "step": 4775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0956333126144764e-06,
+      "loss": 1.4222,
+      "step": 4776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0593582887512015e-06,
+      "loss": 1.4524,
+      "step": 4777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.023296404813303e-06,
+      "loss": 1.4418,
+      "step": 4778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.987447676267907e-06,
+      "loss": 1.429,
+      "step": 4779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9518121184906044e-06,
+      "loss": 1.4383,
+      "step": 4780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.916389746765724e-06,
+      "loss": 1.4384,
+      "step": 4781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8811805762860577e-06,
+      "loss": 1.4148,
+      "step": 4782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8461846221529143e-06,
+      "loss": 1.3984,
+      "step": 4783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8114018993762314e-06,
+      "loss": 1.4404,
+      "step": 4784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7768324228744644e-06,
+      "loss": 1.424,
+      "step": 4785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.742476207474587e-06,
+      "loss": 1.3933,
+      "step": 4786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7083332679122e-06,
+      "loss": 1.3999,
+      "step": 4787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.674403618831256e-06,
+      "loss": 1.4507,
+      "step": 4788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6406872747843925e-06,
+      "loss": 1.4329,
+      "step": 4789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6071842502326525e-06,
+      "loss": 1.4244,
+      "step": 4790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.573894559545653e-06,
+      "loss": 1.4498,
+      "step": 4791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5408182170014173e-06,
+      "loss": 1.4088,
+      "step": 4792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5079552367865965e-06,
+      "loss": 1.3958,
+      "step": 4793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4753056329962496e-06,
+      "loss": 1.4222,
+      "step": 4794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.442869419633953e-06,
+      "loss": 1.4127,
+      "step": 4795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4106466106116333e-06,
+      "loss": 1.4496,
+      "step": 4796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.378637219749902e-06,
+      "loss": 1.4104,
+      "step": 4797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3468412607776655e-06,
+      "loss": 1.4013,
+      "step": 4798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3152587473323495e-06,
+      "loss": 1.4101,
+      "step": 4799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2838896929598397e-06,
+      "loss": 1.4389,
+      "step": 4800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2527341111144293e-06,
+      "loss": 1.457,
+      "step": 4801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2217920151588746e-06,
+      "loss": 1.4498,
+      "step": 4802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1910634183644475e-06,
+      "loss": 1.4156,
+      "step": 4803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1605483339106614e-06,
+      "loss": 1.4477,
+      "step": 4804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.130246774885658e-06,
+      "loss": 1.429,
+      "step": 4805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1001587542858193e-06,
+      "loss": 1.4392,
+      "step": 4806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0702842850160995e-06,
+      "loss": 1.4584,
+      "step": 4807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0406233798896944e-06,
+      "loss": 1.4189,
+      "step": 4808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0111760516284273e-06,
+      "loss": 1.4305,
+      "step": 4809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9819423128622505e-06,
+      "loss": 1.4104,
+      "step": 4810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.95292217612969e-06,
+      "loss": 1.429,
+      "step": 4811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.924115653877567e-06,
+      "loss": 1.4146,
+      "step": 4812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8955227584612212e-06,
+      "loss": 1.4255,
+      "step": 4813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.867143502144175e-06,
+      "loss": 1.4132,
+      "step": 4814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.838977897098415e-06,
+      "loss": 1.4603,
+      "step": 4815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.811025955404333e-06,
+      "loss": 1.4067,
+      "step": 4816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7832876890505612e-06,
+      "loss": 1.4497,
+      "step": 4817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7557631099342497e-06,
+      "loss": 1.4497,
+      "step": 4818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.728452229860733e-06,
+      "loss": 1.4452,
+      "step": 4819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7013550605438078e-06,
+      "loss": 1.4582,
+      "step": 4820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6744716136055105e-06,
+      "loss": 1.4264,
+      "step": 4821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.64780190057634e-06,
+      "loss": 1.4283,
+      "step": 4822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6213459328950354e-06,
+      "loss": 1.4317,
+      "step": 4823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.595103721908575e-06,
+      "loss": 1.4069,
+      "step": 4824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569075278872456e-06,
+      "loss": 1.4253,
+      "step": 4825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5432606149503036e-06,
+      "loss": 1.4268,
+      "step": 4826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5176597412142613e-06,
+      "loss": 1.4187,
+      "step": 4827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4922726686445453e-06,
+      "loss": 1.4291,
+      "step": 4828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4670994081297795e-06,
+      "loss": 1.4383,
+      "step": 4829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.442139970466938e-06,
+      "loss": 1.4354,
+      "step": 4830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.41739436636118e-06,
+      "loss": 1.3987,
+      "step": 4831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3928626064260153e-06,
+      "loss": 1.4314,
+      "step": 4832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3685447011831941e-06,
+      "loss": 1.425,
+      "step": 4833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3444406610628734e-06,
+      "loss": 1.4327,
+      "step": 4834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3205504964032278e-06,
+      "loss": 1.3955,
+      "step": 4835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2968742174509495e-06,
+      "loss": 1.4128,
+      "step": 4836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2734118343608603e-06,
+      "loss": 1.4417,
+      "step": 4837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.250163357196077e-06,
+      "loss": 1.4428,
+      "step": 4838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2271287959279564e-06,
+      "loss": 1.4417,
+      "step": 4839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2043081604360962e-06,
+      "loss": 1.4473,
+      "step": 4840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1817014605084443e-06,
+      "loss": 1.4113,
+      "step": 4841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1593087058410779e-06,
+      "loss": 1.4667,
+      "step": 4842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1371299060383144e-06,
+      "loss": 1.4111,
+      "step": 4843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1151650706127115e-06,
+      "loss": 1.4194,
+      "step": 4844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0934142089851773e-06,
+      "loss": 1.4085,
+      "step": 4845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.071877330484694e-06,
+      "loss": 1.3991,
+      "step": 4846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0505544443484283e-06,
+      "loss": 1.4581,
+      "step": 4847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0294455597220088e-06,
+      "loss": 1.4218,
+      "step": 4848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0085506856590265e-06,
+      "loss": 1.4329,
+      "step": 4849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.878698311214241e-07,
+      "loss": 1.4216,
+      "step": 4850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.674030049792725e-07,
+      "loss": 1.4112,
+      "step": 4851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.471502160108836e-07,
+      "loss": 1.4426,
+      "step": 4852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.271114729028085e-07,
+      "loss": 1.444,
+      "step": 4853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.072867842497279e-07,
+      "loss": 1.4593,
+      "step": 4854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.876761585545067e-07,
+      "loss": 1.418,
+      "step": 4855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.682796042282503e-07,
+      "loss": 1.3876,
+      "step": 4856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.490971295901928e-07,
+      "loss": 1.425,
+      "step": 4857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.301287428678084e-07,
+      "loss": 1.4074,
+      "step": 4858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.113744521967004e-07,
+      "loss": 1.4237,
+      "step": 4859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.928342656206567e-07,
+      "loss": 1.407,
+      "step": 4860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.745081910917051e-07,
+      "loss": 1.4381,
+      "step": 4861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.563962364698917e-07,
+      "loss": 1.3935,
+      "step": 4862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.384984095235025e-07,
+      "loss": 1.3979,
+      "step": 4863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.208147179291191e-07,
+      "loss": 1.4364,
+      "step": 4864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.033451692711745e-07,
+      "loss": 1.4019,
+      "step": 4865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.86089771042564e-07,
+      "loss": 1.4361,
+      "step": 4866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.690485306442007e-07,
+      "loss": 1.4009,
+      "step": 4867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.522214553850158e-07,
+      "loss": 1.4205,
+      "step": 4868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.356085524823474e-07,
+      "loss": 1.4079,
+      "step": 4869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.192098290614956e-07,
+      "loss": 1.4277,
+      "step": 4870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.030252921558899e-07,
+      "loss": 1.4454,
+      "step": 4871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.870549487071997e-07,
+      "loss": 1.3917,
+      "step": 4872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.71298805565168e-07,
+      "loss": 1.384,
+      "step": 4873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.557568694877224e-07,
+      "loss": 1.4088,
+      "step": 4874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.40429147140753e-07,
+      "loss": 1.4354,
+      "step": 4875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.253156450984453e-07,
+      "loss": 1.4256,
+      "step": 4876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.104163698430587e-07,
+      "loss": 1.4613,
+      "step": 4877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.957313277648701e-07,
+      "loss": 1.4483,
+      "step": 4878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.812605251624525e-07,
+      "loss": 1.4387,
+      "step": 4879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6700396824239656e-07,
+      "loss": 1.3917,
+      "step": 4880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.529616631193112e-07,
+      "loss": 1.3988,
+      "step": 4881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3913361581610076e-07,
+      "loss": 1.3693,
+      "step": 4882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.255198322636322e-07,
+      "loss": 1.424,
+      "step": 4883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1212031830084594e-07,
+      "loss": 1.3911,
+      "step": 4884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.989350796749225e-07,
+      "loss": 1.4248,
+      "step": 4885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8596412204106037e-07,
+      "loss": 1.4373,
+      "step": 4886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7320745096258715e-07,
+      "loss": 1.4286,
+      "step": 4887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.606650719108484e-07,
+      "loss": 1.4817,
+      "step": 4888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.483369902653188e-07,
+      "loss": 1.4315,
+      "step": 4889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.362232113135466e-07,
+      "loss": 1.4588,
+      "step": 4890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.243237402512089e-07,
+      "loss": 1.4318,
+      "step": 4891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1263858218205657e-07,
+      "loss": 1.4693,
+      "step": 4892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0116774211791377e-07,
+      "loss": 1.4334,
+      "step": 4893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.899112249786229e-07,
+      "loss": 1.4569,
+      "step": 4894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7886903559221076e-07,
+      "loss": 1.4263,
+      "step": 4895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.680411786946668e-07,
+      "loss": 1.4504,
+      "step": 4896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5742765893010943e-07,
+      "loss": 1.4167,
+      "step": 4897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.470284808507306e-07,
+      "loss": 1.4107,
+      "step": 4898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3684364891674026e-07,
+      "loss": 1.4076,
+      "step": 4899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.268731674965885e-07,
+      "loss": 1.4557,
+      "step": 4900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1711704086646577e-07,
+      "loss": 1.4022,
+      "step": 4901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0757527321096924e-07,
+      "loss": 1.4192,
+      "step": 4902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9824786862260303e-07,
+      "loss": 1.4049,
+      "step": 4903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.891348311017782e-07,
+      "loss": 1.417,
+      "step": 4904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8023616455731251e-07,
+      "loss": 1.4104,
+      "step": 4905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.715518728057086e-07,
+      "loss": 1.4285,
+      "step": 4906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6308195957182026e-07,
+      "loss": 1.4434,
+      "step": 4907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5482642848835272e-07,
+      "loss": 1.4368,
+      "step": 4908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4678528309614025e-07,
+      "loss": 1.4111,
+      "step": 4909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3895852684409072e-07,
+      "loss": 1.4393,
+      "step": 4910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3134616308918545e-07,
+      "loss": 1.4638,
+      "step": 4911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2394819509631282e-07,
+      "loss": 1.4174,
+      "step": 4912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1676462603849026e-07,
+      "loss": 1.4329,
+      "step": 4913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0979545899686416e-07,
+      "loss": 1.4226,
+      "step": 4914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.030406969604325e-07,
+      "loss": 1.4092,
+      "step": 4915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.650034282643327e-08,
+      "loss": 1.4086,
+      "step": 4916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.017439940001148e-08,
+      "loss": 1.4212,
+      "step": 4917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.406286939438568e-08,
+      "loss": 1.4316,
+      "step": 4918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.816575543084791e-08,
+      "loss": 1.4248,
+      "step": 4919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.24830600386528e-08,
+      "loss": 1.4279,
+      "step": 4920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.701478565518392e-08,
+      "loss": 1.4242,
+      "step": 4921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.176093462573196e-08,
+      "loss": 1.4229,
+      "step": 4922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.672150920377206e-08,
+      "loss": 1.433,
+      "step": 4923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.189651155068642e-08,
+      "loss": 1.4382,
+      "step": 4924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.728594373593076e-08,
+      "loss": 1.4153,
+      "step": 4925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.288980773703433e-08,
+      "loss": 1.4213,
+      "step": 4926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.870810543948888e-08,
+      "loss": 1.4217,
+      "step": 4927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.474083863691524e-08,
+      "loss": 1.4396,
+      "step": 4928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.098800903078569e-08,
+      "loss": 1.4367,
+      "step": 4929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7449618230757088e-08,
+      "loss": 1.4426,
+      "step": 4930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.41256677544488e-08,
+      "loss": 1.431,
+      "step": 4931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1016159027553716e-08,
+      "loss": 1.4204,
+      "step": 4932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.812109338367174e-08,
+      "loss": 1.3903,
+      "step": 4933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5440472064587317e-08,
+      "loss": 1.4237,
+      "step": 4934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2974296220047421e-08,
+      "loss": 1.3888,
+      "step": 4935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.072256690770601e-08,
+      "loss": 1.4361,
+      "step": 4936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.685285093401606e-09,
+      "loss": 1.4189,
+      "step": 4937
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.8624516509352415e-09,
+      "loss": 1.3926,
+      "step": 4938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.254067362125969e-09,
+      "loss": 1.4115,
+      "step": 4939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8601329167553544e-09,
+      "loss": 1.4031,
+      "step": 4940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6806489127895183e-09,
+      "loss": 1.4785,
+      "step": 4941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.715615856046071e-09,
+      "loss": 1.4038,
+      "step": 4942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.650341604716673e-10,
+      "loss": 1.4029,
+      "step": 4943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2890414797547294e-10,
+      "loss": 1.4361,
+      "step": 4944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0722604848467655e-10,
+      "loss": 1.4318,
+      "step": 4945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.4198,
+      "step": 4946
+    },
+    {
+      "epoch": 1.0,
+      "step": 4946,
+      "total_flos": 1.5823970447172567e+19,
+      "train_loss": 1.513081687112786,
+      "train_runtime": 54263.8475,
+      "train_samples_per_second": 23.335,
+      "train_steps_per_second": 0.091
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4946,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 24000,
+  "total_flos": 1.5823970447172567e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}