diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,37242 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6202,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3475935828877e-06,
+      "loss": 5.9486,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.06951871657754e-05,
+      "loss": 5.8951,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6042780748663105e-05,
+      "loss": 5.9756,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.13903743315508e-05,
+      "loss": 5.7782,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.67379679144385e-05,
+      "loss": 5.959,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.208556149732621e-05,
+      "loss": 5.9055,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.74331550802139e-05,
+      "loss": 6.2998,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.27807486631016e-05,
+      "loss": 6.1521,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8128342245989304e-05,
+      "loss": 6.4036,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3475935828877e-05,
+      "loss": 6.501,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 6.3223,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.417112299465242e-05,
+      "loss": 6.3368,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.951871657754011e-05,
+      "loss": 6.2648,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.48663101604278e-05,
+      "loss": 6.1554,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.021390374331551e-05,
+      "loss": 5.8798,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.55614973262032e-05,
+      "loss": 5.9567,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 5.9103,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.625668449197861e-05,
+      "loss": 5.4526,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010160427807486631,
+      "loss": 5.367,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000106951871657754,
+      "loss": 5.4023,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011229946524064172,
+      "loss": 5.3629,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011764705882352942,
+      "loss": 5.0887,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012299465240641713,
+      "loss": 5.1178,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012834224598930484,
+      "loss": 4.8329,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013368983957219252,
+      "loss": 4.849,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013903743315508022,
+      "loss": 4.8322,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001443850267379679,
+      "loss": 4.9571,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001497326203208556,
+      "loss": 4.8369,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015508021390374334,
+      "loss": 4.8279,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016042780748663101,
+      "loss": 4.8095,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016577540106951872,
+      "loss": 4.8473,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001711229946524064,
+      "loss": 4.6431,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017647058823529413,
+      "loss": 4.7465,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 4.7161,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001871657754010695,
+      "loss": 4.705,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019251336898395722,
+      "loss": 4.6889,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019786096256684495,
+      "loss": 4.5606,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020320855614973263,
+      "loss": 4.6834,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020855614973262033,
+      "loss": 4.5475,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000213903743315508,
+      "loss": 4.7062,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021925133689839572,
+      "loss": 4.555,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022459893048128345,
+      "loss": 4.4943,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022994652406417113,
+      "loss": 4.6637,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023529411764705883,
+      "loss": 4.7428,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002406417112299465,
+      "loss": 4.5246,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024598930481283427,
+      "loss": 4.5559,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002513368983957219,
+      "loss": 4.3376,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002566844919786097,
+      "loss": 4.5093,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026203208556149733,
+      "loss": 4.3749,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026737967914438503,
+      "loss": 4.4069,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027272727272727274,
+      "loss": 4.5086,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027807486631016044,
+      "loss": 4.3684,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028342245989304815,
+      "loss": 4.4441,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002887700534759358,
+      "loss": 4.439,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029411764705882356,
+      "loss": 4.4707,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002994652406417112,
+      "loss": 4.2591,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003048128342245989,
+      "loss": 4.3509,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031016042780748667,
+      "loss": 4.4297,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003155080213903743,
+      "loss": 4.4056,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032085561497326203,
+      "loss": 4.398,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003262032085561498,
+      "loss": 4.2179,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00033155080213903744,
+      "loss": 4.2304,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00033689839572192514,
+      "loss": 4.3654,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003422459893048128,
+      "loss": 4.4312,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034759358288770055,
+      "loss": 4.2155,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035294117647058826,
+      "loss": 4.2967,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003582887700534759,
+      "loss": 4.2588,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036363636363636367,
+      "loss": 4.2611,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003689839572192513,
+      "loss": 4.2833,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000374331550802139,
+      "loss": 4.343,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003796791443850268,
+      "loss": 4.3456,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038502673796791443,
+      "loss": 4.141,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039037433155080214,
+      "loss": 4.4074,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003957219251336899,
+      "loss": 4.247,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040106951871657755,
+      "loss": 4.3275,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040641711229946525,
+      "loss": 4.1954,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004117647058823529,
+      "loss": 4.134,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041711229946524066,
+      "loss": 4.3486,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042245989304812837,
+      "loss": 4.3122,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000427807486631016,
+      "loss": 4.2094,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004331550802139038,
+      "loss": 4.3217,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043850267379679143,
+      "loss": 4.102,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044385026737967914,
+      "loss": 4.1534,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004491978609625669,
+      "loss": 4.2219,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045454545454545455,
+      "loss": 4.2487,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045989304812834225,
+      "loss": 4.2481,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046524064171123,
+      "loss": 4.2309,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047058823529411766,
+      "loss": 4.0941,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047593582887700537,
+      "loss": 4.0096,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000481283422459893,
+      "loss": 4.0631,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004866310160427808,
+      "loss": 4.1869,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004919786096256685,
+      "loss": 4.1159,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004973262032085562,
+      "loss": 4.1178,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005026737967914438,
+      "loss": 4.1284,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005080213903743316,
+      "loss": 4.0345,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005133689839572194,
+      "loss": 4.0867,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005187165775401069,
+      "loss": 4.057,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005240641711229947,
+      "loss": 3.937,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005294117647058824,
+      "loss": 3.9765,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005347593582887701,
+      "loss": 3.9608,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005401069518716578,
+      "loss": 4.1594,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005454545454545455,
+      "loss": 4.1302,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005508021390374331,
+      "loss": 4.1312,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005561497326203209,
+      "loss": 4.0223,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005614973262032086,
+      "loss": 4.0161,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005668449197860963,
+      "loss": 4.0451,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000572192513368984,
+      "loss": 4.1083,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005775401069518716,
+      "loss": 4.1682,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005828877005347594,
+      "loss": 4.0605,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005882352941176471,
+      "loss": 4.0765,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005935828877005348,
+      "loss": 3.8291,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005989304812834224,
+      "loss": 4.0857,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006042780748663102,
+      "loss": 4.077,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006096256684491978,
+      "loss": 4.0375,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006149732620320856,
+      "loss": 3.8772,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006203208556149733,
+      "loss": 3.9426,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006256684491978609,
+      "loss": 4.0139,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006310160427807486,
+      "loss": 4.0414,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006363636363636364,
+      "loss": 4.0246,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006417112299465241,
+      "loss": 3.9336,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006470588235294118,
+      "loss": 4.0677,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006524064171122996,
+      "loss": 4.0379,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006577540106951871,
+      "loss": 3.8605,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006631016042780749,
+      "loss": 3.8841,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006684491978609626,
+      "loss": 3.837,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006737967914438503,
+      "loss": 3.8734,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000679144385026738,
+      "loss": 3.8831,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006844919786096256,
+      "loss": 3.8443,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006898395721925133,
+      "loss": 3.8587,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006951871657754011,
+      "loss": 4.0822,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007005347593582888,
+      "loss": 3.779,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007058823529411765,
+      "loss": 3.7747,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007112299465240642,
+      "loss": 3.8958,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007165775401069518,
+      "loss": 3.8679,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007219251336898396,
+      "loss": 3.6683,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 3.8482,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000732620320855615,
+      "loss": 3.92,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007379679144385026,
+      "loss": 3.8979,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007433155080213904,
+      "loss": 3.8828,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000748663101604278,
+      "loss": 3.9528,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007540106951871658,
+      "loss": 3.9246,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007593582887700536,
+      "loss": 3.8685,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007647058823529411,
+      "loss": 3.9712,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007700534759358289,
+      "loss": 4.0532,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007754010695187166,
+      "loss": 3.9034,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007807486631016043,
+      "loss": 3.8289,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000786096256684492,
+      "loss": 3.9289,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007914438502673798,
+      "loss": 4.0015,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007967914438502673,
+      "loss": 3.8183,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008021390374331551,
+      "loss": 3.8087,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008074866310160429,
+      "loss": 3.8361,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008128342245989305,
+      "loss": 3.839,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008181818181818183,
+      "loss": 3.8806,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008235294117647058,
+      "loss": 3.8447,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008288770053475936,
+      "loss": 3.77,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008342245989304813,
+      "loss": 3.9146,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000839572192513369,
+      "loss": 3.845,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008449197860962567,
+      "loss": 3.8684,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008502673796791444,
+      "loss": 4.0146,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000855614973262032,
+      "loss": 3.7876,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008609625668449198,
+      "loss": 3.8683,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008663101604278076,
+      "loss": 3.9205,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008716577540106952,
+      "loss": 4.0075,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008770053475935829,
+      "loss": 3.7433,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008823529411764706,
+      "loss": 3.7349,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008877005347593583,
+      "loss": 3.8098,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000893048128342246,
+      "loss": 3.885,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008983957219251338,
+      "loss": 3.8511,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009037433155080213,
+      "loss": 3.6956,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009090909090909091,
+      "loss": 3.8744,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009144385026737968,
+      "loss": 3.7744,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009197860962566845,
+      "loss": 3.9944,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009251336898395723,
+      "loss": 3.8654,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00093048128342246,
+      "loss": 3.8064,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009358288770053476,
+      "loss": 3.8453,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009411764705882353,
+      "loss": 3.6996,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000946524064171123,
+      "loss": 3.9487,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009518716577540107,
+      "loss": 3.7477,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009572192513368985,
+      "loss": 3.7293,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000962566844919786,
+      "loss": 3.8472,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009679144385026738,
+      "loss": 3.7662,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009732620320855616,
+      "loss": 3.8273,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009786096256684492,
+      "loss": 3.7248,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000983957219251337,
+      "loss": 3.8215,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009893048128342245,
+      "loss": 3.8897,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009946524064171124,
+      "loss": 3.7155,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 4.0417,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999999931802496,
+      "loss": 3.7861,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999997272100022,
+      "loss": 3.8133,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999993862225744,
+      "loss": 3.623,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999998908840306,
+      "loss": 3.7479,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999982950633268,
+      "loss": 3.8731,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999975448918043,
+      "loss": 3.7765,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999966583259434,
+      "loss": 3.8732,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999995635365986,
+      "loss": 3.7426,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999944760122107,
+      "loss": 3.88,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999993180264934,
+      "loss": 3.7482,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999917481245096,
+      "loss": 3.9271,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999901795913278,
+      "loss": 3.8027,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999884746658165,
+      "loss": 3.8586,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999986633348441,
+      "loss": 3.8988,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999846556397038,
+      "loss": 3.7926,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999825415401438,
+      "loss": 3.7898,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999802910503383,
+      "loss": 3.7323,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999779041709005,
+      "loss": 3.7795,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999753809024823,
+      "loss": 3.73,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999727212457715,
+      "loss": 3.7879,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999699252014938,
+      "loss": 3.6848,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999966992770412,
+      "loss": 3.7039,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999639239533257,
+      "loss": 3.8482,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999607187510726,
+      "loss": 3.7943,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999573771645267,
+      "loss": 3.7842,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999538991945997,
+      "loss": 3.9321,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999502848422402,
+      "loss": 3.8289,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999946534108434,
+      "loss": 3.7996,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999426469942047,
+      "loss": 3.7495,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999386235006124,
+      "loss": 3.7358,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999344636287548,
+      "loss": 3.7833,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999301673797665,
+      "loss": 3.7257,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999257347548195,
+      "loss": 3.7508,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999211657551234,
+      "loss": 3.8274,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999164603819238,
+      "loss": 3.8082,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999116186365047,
+      "loss": 3.7433,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999906640520187,
+      "loss": 3.7989,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999015260343286,
+      "loss": 3.5308,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998962751803246,
+      "loss": 3.7875,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998908879596076,
+      "loss": 3.932,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998853643736468,
+      "loss": 3.8212,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999879704423949,
+      "loss": 3.789,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998739081120585,
+      "loss": 3.7505,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998679754395566,
+      "loss": 3.8381,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998619064080611,
+      "loss": 3.8199,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999855701019228,
+      "loss": 3.7177,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998493592747498,
+      "loss": 3.8603,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998428811763566,
+      "loss": 3.7391,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999836266725816,
+      "loss": 3.9236,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998295159249315,
+      "loss": 3.7047,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998226287755451,
+      "loss": 3.7526,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998156052795355,
+      "loss": 3.6538,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999808445438819,
+      "loss": 3.815,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998011492553481,
+      "loss": 3.7717,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997937167311134,
+      "loss": 3.767,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997861478681425,
+      "loss": 3.7465,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997784426685002,
+      "loss": 3.6345,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999770601134288,
+      "loss": 3.7467,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997626232676454,
+      "loss": 3.7458,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997545090707485,
+      "loss": 3.6346,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997462585458108,
+      "loss": 3.7477,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999737871695083,
+      "loss": 3.6396,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997293485208528,
+      "loss": 3.7122,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997206890254454,
+      "loss": 3.7103,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997118932112228,
+      "loss": 3.8014,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997029610805848,
+      "loss": 3.7202,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996938926359676,
+      "loss": 3.6884,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996846878798453,
+      "loss": 3.868,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996753468147286,
+      "loss": 3.6932,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999665869443166,
+      "loss": 3.7192,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996562557677425,
+      "loss": 3.7406,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996465057910805,
+      "loss": 3.778,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996366195158402,
+      "loss": 3.7166,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999626596944718,
+      "loss": 3.7759,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996164380804483,
+      "loss": 3.774,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996061429258022,
+      "loss": 3.7561,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995957114835878,
+      "loss": 3.837,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995851437566514,
+      "loss": 3.6542,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999574439747875,
+      "loss": 3.6925,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995635994601792,
+      "loss": 3.6578,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995526228965205,
+      "loss": 3.7383,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999541510059894,
+      "loss": 3.7585,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995302609533303,
+      "loss": 3.8008,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995188755798985,
+      "loss": 3.6539,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995073539427047,
+      "loss": 3.8845,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994956960448912,
+      "loss": 3.5674,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994839018896387,
+      "loss": 3.6741,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994719714801641,
+      "loss": 3.7959,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994599048197223,
+      "loss": 3.7257,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994477019116049,
+      "loss": 3.6391,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994353627591403,
+      "loss": 3.6798,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999422887365695,
+      "loss": 3.6971,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994102757346721,
+      "loss": 3.6488,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993975278695117,
+      "loss": 3.7403,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993846437736913,
+      "loss": 3.6619,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999371623450726,
+      "loss": 3.6132,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999358466904167,
+      "loss": 3.7227,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993451741376034,
+      "loss": 3.8561,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993317451546617,
+      "loss": 3.6644,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999318179959005,
+      "loss": 3.7534,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993044785543337,
+      "loss": 3.7825,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992906409443854,
+      "loss": 3.7098,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999276667132935,
+      "loss": 3.684,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999262557123794,
+      "loss": 3.6203,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992483109208122,
+      "loss": 3.5992,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999233928527875,
+      "loss": 3.6556,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992194099489064,
+      "loss": 3.7577,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992047551878667,
+      "loss": 3.6756,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991899642487535,
+      "loss": 3.8233,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991750371356016,
+      "loss": 3.7358,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991599738524831,
+      "loss": 3.6257,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991447744035072,
+      "loss": 3.6915,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099912943879282,
+      "loss": 3.731,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991139670246047,
+      "loss": 3.7566,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990983591030825,
+      "loss": 3.6893,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990826150325103,
+      "loss": 3.6207,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990667348171833,
+      "loss": 3.6676,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990507184614336,
+      "loss": 3.6716,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099903456596963,
+      "loss": 3.86,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990182773461792,
+      "loss": 3.8431,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999001852595524,
+      "loss": 3.6791,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989852917221451,
+      "loss": 3.717,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989685947305602,
+      "loss": 3.6599,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989517616253243,
+      "loss": 3.6526,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998934792411029,
+      "loss": 3.6757,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989176870923034,
+      "loss": 3.6942,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989004456738137,
+      "loss": 3.7508,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988830681602631,
+      "loss": 3.601,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998865554556392,
+      "loss": 3.5761,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988479048669782,
+      "loss": 3.6358,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988301190968362,
+      "loss": 3.7138,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988121972508176,
+      "loss": 3.6576,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987941393338115,
+      "loss": 3.572,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987759453507438,
+      "loss": 3.7492,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998757615306578,
+      "loss": 3.6342,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987391492063138,
+      "loss": 3.5609,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998720547054989,
+      "loss": 3.5772,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987018088576777,
+      "loss": 3.751,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986829346194917,
+      "loss": 3.6777,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099866392434558,
+      "loss": 3.7342,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986447780411278,
+      "loss": 3.6687,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986254957113586,
+      "loss": 3.5875,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986060773615319,
+      "loss": 3.6872,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985865229969453,
+      "loss": 3.5409,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985668326229328,
+      "loss": 3.597,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985470062448658,
+      "loss": 3.6946,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985270438681525,
+      "loss": 3.7482,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985069454982389,
+      "loss": 3.7012,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984867111406076,
+      "loss": 3.5765,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984663408007777,
+      "loss": 3.5477,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984458344843067,
+      "loss": 3.8007,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998425192196788,
+      "loss": 3.6531,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998404413943853,
+      "loss": 3.6274,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983834997311697,
+      "loss": 3.7646,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983624495644432,
+      "loss": 3.7375,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983412634494157,
+      "loss": 3.6213,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998319941391867,
+      "loss": 3.646,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998298483397613,
+      "loss": 3.7726,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982768894725076,
+      "loss": 3.6982,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982551596224411,
+      "loss": 3.5818,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982332938533417,
+      "loss": 3.7393,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982112921711735,
+      "loss": 3.6548,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981891545819388,
+      "loss": 3.6664,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981668810916764,
+      "loss": 3.5646,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981444717064622,
+      "loss": 3.5343,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981219264324093,
+      "loss": 3.8004,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998099245275668,
+      "loss": 3.6435,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980764282424251,
+      "loss": 3.5236,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998053475338905,
+      "loss": 3.7884,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980303865713693,
+      "loss": 3.6316,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980071619461162,
+      "loss": 3.6923,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979838014694813,
+      "loss": 3.6906,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979603051478367,
+      "loss": 3.7143,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979366729875924,
+      "loss": 3.6242,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979129049951948,
+      "loss": 3.6223,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978890011771275,
+      "loss": 3.7303,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978649615399113,
+      "loss": 3.7574,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978407860901043,
+      "loss": 3.5757,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978164748343007,
+      "loss": 3.5999,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997792027779133,
+      "loss": 3.6576,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977674449312692,
+      "loss": 3.6334,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977427262974163,
+      "loss": 3.5762,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977178718843169,
+      "loss": 3.6162,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997692881698751,
+      "loss": 3.6593,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976677557475353,
+      "loss": 3.623,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976424940375246,
+      "loss": 3.7449,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976170965756098,
+      "loss": 3.5766,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975915633687188,
+      "loss": 3.4851,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975658944238169,
+      "loss": 3.5713,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975400897479065,
+      "loss": 3.6258,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997514149348027,
+      "loss": 3.6143,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974880732312541,
+      "loss": 3.8064,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974618614047017,
+      "loss": 3.7325,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974355138755197,
+      "loss": 3.5524,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974090306508958,
+      "loss": 3.674,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973824117380542,
+      "loss": 3.5691,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973556571442564,
+      "loss": 3.5528,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973287668768004,
+      "loss": 3.7901,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997301740943022,
+      "loss": 3.4705,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972745793502933,
+      "loss": 3.6123,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972472821060238,
+      "loss": 3.6343,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972198492176603,
+      "loss": 3.5437,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971922806926857,
+      "loss": 3.6402,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971645765386207,
+      "loss": 3.7182,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971367367630226,
+      "loss": 3.5644,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971087613734858,
+      "loss": 3.653,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997080650377642,
+      "loss": 3.6866,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997052403783159,
+      "loss": 3.6885,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970240215977427,
+      "loss": 3.5455,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969955038291352,
+      "loss": 3.6225,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996966850485116,
+      "loss": 3.6166,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969380615735017,
+      "loss": 3.6286,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969091371021452,
+      "loss": 3.5394,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996880077078937,
+      "loss": 3.6536,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968508815118042,
+      "loss": 3.5967,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968215504087114,
+      "loss": 3.5914,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967920837776595,
+      "loss": 3.6483,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967624816266869,
+      "loss": 3.7576,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967327439638688,
+      "loss": 3.6956,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996702870797317,
+      "loss": 3.6324,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996672862135181,
+      "loss": 3.5627,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966427179856467,
+      "loss": 3.8247,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966124383569371,
+      "loss": 3.5794,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965820232573125,
+      "loss": 3.7232,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996551472695069,
+      "loss": 3.6576,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965207866785414,
+      "loss": 3.5436,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964899652161003,
+      "loss": 3.51,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964590083161532,
+      "loss": 3.6684,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964279159871449,
+      "loss": 3.5939,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963966882375575,
+      "loss": 3.7163,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996365325075909,
+      "loss": 3.6201,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963338265107553,
+      "loss": 3.6707,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996302192550689,
+      "loss": 3.5358,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962704232043392,
+      "loss": 3.6123,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962385184803726,
+      "loss": 3.6365,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962064783874923,
+      "loss": 3.6887,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961743029344382,
+      "loss": 3.4337,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961419921299883,
+      "loss": 3.4868,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996109545982956,
+      "loss": 3.6961,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960769645021923,
+      "loss": 3.785,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960442476965852,
+      "loss": 3.4476,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099601139557506,
+      "loss": 3.6732,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959784081465775,
+      "loss": 3.5934,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995945285420137,
+      "loss": 3.5392,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959120274047738,
+      "loss": 3.4776,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958786341095604,
+      "loss": 3.7242,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958451055436064,
+      "loss": 3.5389,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958114417160577,
+      "loss": 3.5481,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957776426360975,
+      "loss": 3.5544,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957437083129464,
+      "loss": 3.6186,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957096387558605,
+      "loss": 3.4458,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995675433974134,
+      "loss": 3.5658,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956410939770978,
+      "loss": 3.5481,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956066187741194,
+      "loss": 3.4711,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955720083746034,
+      "loss": 3.5488,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995537262787991,
+      "loss": 3.5963,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955023820237602,
+      "loss": 3.6598,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954673660914266,
+      "loss": 3.6826,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995432215000542,
+      "loss": 3.5356,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995396928760695,
+      "loss": 3.6369,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995361507381512,
+      "loss": 3.7795,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995325950872655,
+      "loss": 3.5622,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952902592438237,
+      "loss": 3.6422,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952544325047543,
+      "loss": 3.6076,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099521847066522,
+      "loss": 3.7241,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951823737350313,
+      "loss": 3.5244,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951461417240341,
+      "loss": 3.5843,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951097746421131,
+      "loss": 3.6061,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950732724991885,
+      "loss": 3.6687,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950366353052174,
+      "loss": 3.6813,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949998630701946,
+      "loss": 3.4376,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994962955804151,
+      "loss": 3.6877,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949259135171543,
+      "loss": 3.5971,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948887362193095,
+      "loss": 3.5688,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994851423920758,
+      "loss": 3.5673,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948139766316785,
+      "loss": 3.5814,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947763943622865,
+      "loss": 3.6455,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947386771228333,
+      "loss": 3.4994,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009947008249236082,
+      "loss": 3.5484,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994662837774937,
+      "loss": 3.4904,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994624715687182,
+      "loss": 3.6338,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945864586707427,
+      "loss": 3.5473,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945480667360549,
+      "loss": 3.6766,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994509539893592,
+      "loss": 3.5347,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944708781538635,
+      "loss": 3.5936,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994432081527416,
+      "loss": 3.6759,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994393150024833,
+      "loss": 3.4666,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994354083656734,
+      "loss": 3.6506,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943148824337765,
+      "loss": 3.5375,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942755463666542,
+      "loss": 3.4339,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994236075466097,
+      "loss": 3.5631,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994196469742873,
+      "loss": 3.5491,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941567292077857,
+      "loss": 3.5338,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994116853871676,
+      "loss": 3.6489,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940768437454218,
+      "loss": 3.6265,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940366988399369,
+      "loss": 3.487,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993996419166173,
+      "loss": 3.5466,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939560047351175,
+      "loss": 3.6031,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939154555577954,
+      "loss": 3.5594,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938747716452681,
+      "loss": 3.6552,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938339530086336,
+      "loss": 3.571,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937929996590266,
+      "loss": 3.5924,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937519116076194,
+      "loss": 3.7261,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099371068886562,
+      "loss": 3.5448,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936693314442737,
+      "loss": 3.5326,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993627839354862,
+      "loss": 3.4735,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935862126087042,
+      "loss": 3.5275,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993544451217155,
+      "loss": 3.7592,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935025551916067,
+      "loss": 3.5127,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934605245434886,
+      "loss": 3.6119,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934183592842656,
+      "loss": 3.3895,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933760594254403,
+      "loss": 3.493,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933336249785514,
+      "loss": 3.6469,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932910559551748,
+      "loss": 3.5954,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993248352366923,
+      "loss": 3.5558,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993205514225445,
+      "loss": 3.4001,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931625415424265,
+      "loss": 3.6699,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099311943432959,
+      "loss": 3.4861,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930761925986953,
+      "loss": 3.5784,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930328163615374,
+      "loss": 3.6346,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929893056299496,
+      "loss": 3.5611,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992945660415801,
+      "loss": 3.5395,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929018807309975,
+      "loss": 3.6389,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928579665874817,
+      "loss": 3.4937,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992813917997233,
+      "loss": 3.4859,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927697349722676,
+      "loss": 3.4833,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992725417524638,
+      "loss": 3.5376,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926809656664334,
+      "loss": 3.5175,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926363794097804,
+      "loss": 3.6341,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992591658766841,
+      "loss": 3.5758,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992546803749815,
+      "loss": 3.5091,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992501814370938,
+      "loss": 3.4318,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924566906424832,
+      "loss": 3.4936,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924114325767595,
+      "loss": 3.5361,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992366040186113,
+      "loss": 3.6106,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992320513482926,
+      "loss": 3.4672,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922748524796184,
+      "loss": 3.4872,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922290571886452,
+      "loss": 3.6643,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921831276224996,
+      "loss": 3.5878,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921370637937105,
+      "loss": 3.5023,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920908657148435,
+      "loss": 3.5572,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000992044533398501,
+      "loss": 3.4962,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919980668573222,
+      "loss": 3.5311,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919514661039827,
+      "loss": 3.5491,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919047311511945,
+      "loss": 3.5165,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918578620117067,
+      "loss": 3.4012,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918108586983044,
+      "loss": 3.3695,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099176372122381,
+      "loss": 3.5244,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917164496010818,
+      "loss": 3.6515,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991669043843015,
+      "loss": 3.5018,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991621503962542,
+      "loss": 3.4892,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915738299726305,
+      "loss": 3.6485,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991526021886286,
+      "loss": 3.5323,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914780797165496,
+      "loss": 3.4927,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914300034764999,
+      "loss": 3.4527,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913817931792512,
+      "loss": 3.6255,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913334488379551,
+      "loss": 3.605,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912849704657996,
+      "loss": 3.6616,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912363580760085,
+      "loss": 3.428,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911876116818433,
+      "loss": 3.599,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911387312966013,
+      "loss": 3.444,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910897169336168,
+      "loss": 3.5013,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910405686062603,
+      "loss": 3.5512,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909912863279389,
+      "loss": 3.5392,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909418701120965,
+      "loss": 3.4871,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908923199722132,
+      "loss": 3.4046,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908426359218055,
+      "loss": 3.5671,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907928179744272,
+      "loss": 3.6767,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907428661436678,
+      "loss": 3.5237,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906927804431542,
+      "loss": 3.5909,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906425608865487,
+      "loss": 3.585,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905922074875508,
+      "loss": 3.4669,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905417202598968,
+      "loss": 3.6374,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904910992173585,
+      "loss": 3.5054,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904403443737452,
+      "loss": 3.5301,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903894557429023,
+      "loss": 3.4528,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903384333387116,
+      "loss": 3.4843,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902872771750915,
+      "loss": 3.4409,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902359872659972,
+      "loss": 3.4456,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901845636254196,
+      "loss": 3.7117,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901330062673868,
+      "loss": 3.5793,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990081315205963,
+      "loss": 3.5781,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900294904552492,
+      "loss": 3.5186,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899775320293827,
+      "loss": 3.5342,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989925439942537,
+      "loss": 3.5156,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898732142089224,
+      "loss": 3.5947,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898208548427855,
+      "loss": 3.625,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897683618584098,
+      "loss": 3.5469,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897157352701143,
+      "loss": 3.502,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896629750922555,
+      "loss": 3.5049,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896100813392254,
+      "loss": 3.5742,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895570540254532,
+      "loss": 3.5773,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895038931654041,
+      "loss": 3.5103,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00098945059877358,
+      "loss": 3.6207,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893971708645189,
+      "loss": 3.5775,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893436094527953,
+      "loss": 3.5351,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892899145530205,
+      "loss": 3.5495,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892360861798418,
+      "loss": 3.5029,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989182124347943,
+      "loss": 3.5773,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891280290720446,
+      "loss": 3.6013,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890738003669028,
+      "loss": 3.6577,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989019438247311,
+      "loss": 3.5552,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889649427280987,
+      "loss": 3.4023,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889103138241314,
+      "loss": 3.7719,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888555515503116,
+      "loss": 3.6188,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888006559215777,
+      "loss": 3.5023,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887456269529048,
+      "loss": 3.4684,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988690464659304,
+      "loss": 3.5242,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886351690558237,
+      "loss": 3.588,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885797401575473,
+      "loss": 3.5065,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885241779795954,
+      "loss": 3.6716,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988468482537125,
+      "loss": 3.5966,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884126538453294,
+      "loss": 3.4471,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883566919194375,
+      "loss": 3.5159,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883005967747157,
+      "loss": 3.651,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988244368426466,
+      "loss": 3.5517,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988188006890027,
+      "loss": 3.4959,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881315121807734,
+      "loss": 3.5669,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880748843141168,
+      "loss": 3.6149,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988018123305504,
+      "loss": 3.5215,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879612291704197,
+      "loss": 3.5427,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879042019243834,
+      "loss": 3.4381,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878470415829517,
+      "loss": 3.6386,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877897481617175,
+      "loss": 3.4823,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877323216763098,
+      "loss": 3.5773,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876747621423941,
+      "loss": 3.4238,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987617069575672,
+      "loss": 3.5172,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875592439918812,
+      "loss": 3.5516,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875012854067965,
+      "loss": 3.4397,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874431938362276,
+      "loss": 3.4421,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873849692960221,
+      "loss": 3.7115,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873266118020627,
+      "loss": 3.5163,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872681213702688,
+      "loss": 3.4918,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987209498016596,
+      "loss": 3.4362,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871507417570362,
+      "loss": 3.45,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870918526076176,
+      "loss": 3.5738,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870328305844043,
+      "loss": 3.4739,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869736757034972,
+      "loss": 3.4735,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986914387981033,
+      "loss": 3.4475,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868549674331847,
+      "loss": 3.4443,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986795414076162,
+      "loss": 3.4398,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098673572792621,
+      "loss": 3.5642,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866759089996108,
+      "loss": 3.4707,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866159573126824,
+      "loss": 3.4429,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865558728817789,
+      "loss": 3.5282,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864956557232908,
+      "loss": 3.5258,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864353058536446,
+      "loss": 3.4504,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863748232893034,
+      "loss": 3.532,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863142080467663,
+      "loss": 3.5252,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986253460142568,
+      "loss": 3.3678,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861925795932804,
+      "loss": 3.5199,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986131566415511,
+      "loss": 3.558,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860704206259034,
+      "loss": 3.4935,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860091422411379,
+      "loss": 3.5435,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859477312779302,
+      "loss": 3.5333,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985886187753033,
+      "loss": 3.5841,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858245116832346,
+      "loss": 3.5702,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857627030853595,
+      "loss": 3.5011,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857007619762687,
+      "loss": 3.5814,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985638688372859,
+      "loss": 3.4576,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855764822920634,
+      "loss": 3.5555,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855141437508513,
+      "loss": 3.4445,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854516727662277,
+      "loss": 3.591,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853890693552344,
+      "loss": 3.5724,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009853263335349487,
+      "loss": 3.4615,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852634653224846,
+      "loss": 3.4789,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852004647349917,
+      "loss": 3.4826,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985137331789656,
+      "loss": 3.5983,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850740665036993,
+      "loss": 3.5091,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850106688943803,
+      "loss": 3.5704,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849471389789927,
+      "loss": 3.458,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848834767748672,
+      "loss": 3.5689,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848196822993698,
+      "loss": 3.4767,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847557555699034,
+      "loss": 3.5393,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846916966039063,
+      "loss": 3.5799,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846275054188533,
+      "loss": 3.5126,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984563182032255,
+      "loss": 3.4894,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844987264616585,
+      "loss": 3.5242,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844341387246463,
+      "loss": 3.5015,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843694188388374,
+      "loss": 3.4522,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843045668218866,
+      "loss": 3.4047,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984239582691485,
+      "loss": 3.3793,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841744664653596,
+      "loss": 3.6124,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841092181612737,
+      "loss": 3.4762,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984043837797026,
+      "loss": 3.5768,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839783253904517,
+      "loss": 3.4672,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983912680959422,
+      "loss": 3.5286,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838469045218442,
+      "loss": 3.5863,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983780996095661,
+      "loss": 3.5107,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837149556988524,
+      "loss": 3.6364,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836487833494328,
+      "loss": 3.5758,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835824790654535,
+      "loss": 3.4237,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835160428650017,
+      "loss": 3.3763,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834494747662007,
+      "loss": 3.6611,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009833827747872093,
+      "loss": 3.3836,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009833159429462229,
+      "loss": 3.3775,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832489792614722,
+      "loss": 3.5003,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831818837512248,
+      "loss": 3.5504,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983114656433783,
+      "loss": 3.4241,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830472973274861,
+      "loss": 3.5479,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982979806450709,
+      "loss": 3.4593,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982912183821863,
+      "loss": 3.5261,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828444294593937,
+      "loss": 3.4643,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827765433817848,
+      "loss": 3.6277,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827085256075545,
+      "loss": 3.6039,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826403761552579,
+      "loss": 3.4462,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825720950434847,
+      "loss": 3.6047,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825036822908616,
+      "loss": 3.3855,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824351379160514,
+      "loss": 3.5885,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823664619377516,
+      "loss": 3.5475,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982297654374697,
+      "loss": 3.5979,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822287152456571,
+      "loss": 3.5535,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982159644569438,
+      "loss": 3.5769,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820904423648816,
+      "loss": 3.5396,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820211086508652,
+      "loss": 3.427,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819516434463028,
+      "loss": 3.4448,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818820467701434,
+      "loss": 3.4626,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818123186413725,
+      "loss": 3.3334,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817424590790114,
+      "loss": 3.4315,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816724681021168,
+      "loss": 3.5492,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816023457297819,
+      "loss": 3.5175,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815320919811348,
+      "loss": 3.4223,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814617068753405,
+      "loss": 3.5122,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813911904315992,
+      "loss": 3.5805,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981320542669147,
+      "loss": 3.5595,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812497636072562,
+      "loss": 3.4128,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811788532652342,
+      "loss": 3.4413,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000981107811662425,
+      "loss": 3.54,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810366388182078,
+      "loss": 3.6898,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980965334751998,
+      "loss": 3.5101,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808938994832464,
+      "loss": 3.5096,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808223330314402,
+      "loss": 3.4713,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807506354161017,
+      "loss": 3.3966,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806788066567896,
+      "loss": 3.4621,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806068467730976,
+      "loss": 3.5022,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980534755784656,
+      "loss": 3.392,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804625337111305,
+      "loss": 3.3015,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803901805722224,
+      "loss": 3.5334,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803176963876692,
+      "loss": 3.3377,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802450811772433,
+      "loss": 3.5432,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801723349607539,
+      "loss": 3.3393,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800994577580453,
+      "loss": 3.4604,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800264495889977,
+      "loss": 3.4806,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979953310473527,
+      "loss": 3.5224,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798800404315848,
+      "loss": 3.347,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798066394831584,
+      "loss": 3.4114,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979733107648271,
+      "loss": 3.4063,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796594449469809,
+      "loss": 3.4683,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979585651399383,
+      "loss": 3.4805,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795117270256074,
+      "loss": 3.3007,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794376718458197,
+      "loss": 3.3117,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979363485880222,
+      "loss": 3.3804,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792891691490506,
+      "loss": 3.4172,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792147216725789,
+      "loss": 3.5357,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791401434711153,
+      "loss": 3.4155,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979065434565004,
+      "loss": 3.483,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789905949746252,
+      "loss": 3.4455,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789156247203938,
+      "loss": 3.4546,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788405238227611,
+      "loss": 3.5603,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978765292302214,
+      "loss": 3.435,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786899301792752,
+      "loss": 3.4021,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786144374745022,
+      "loss": 3.3085,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978538814208489,
+      "loss": 3.7027,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784630604018647,
+      "loss": 3.4932,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978387176075294,
+      "loss": 3.5908,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978311161249478,
+      "loss": 3.3883,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782350159451521,
+      "loss": 3.5029,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009781587401830885,
+      "loss": 3.4665,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978082333984094,
+      "loss": 3.483,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978005797369012,
+      "loss": 3.476,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779291303587203,
+      "loss": 3.431,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778523329741335,
+      "loss": 3.4099,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777754052362007,
+      "loss": 3.534,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776983471659073,
+      "loss": 3.4808,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776211587842738,
+      "loss": 3.5587,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775438401123566,
+      "loss": 3.4687,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977466391171247,
+      "loss": 3.4258,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977388811982073,
+      "loss": 3.4678,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773111025659972,
+      "loss": 3.4692,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772332629442175,
+      "loss": 3.4615,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977155293137968,
+      "loss": 3.4073,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770771931685185,
+      "loss": 3.613,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769989630571737,
+      "loss": 3.6737,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769206028252735,
+      "loss": 3.5279,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768421124941944,
+      "loss": 3.5014,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767634920853474,
+      "loss": 3.4013,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766847416201796,
+      "loss": 3.4528,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766058611201732,
+      "loss": 3.395,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765268506068459,
+      "loss": 3.4454,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764477101017514,
+      "loss": 3.4507,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000976368439626478,
+      "loss": 3.4283,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762890392026502,
+      "loss": 3.4319,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762095088519274,
+      "loss": 3.3616,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761298485960048,
+      "loss": 3.3573,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000976050058456613,
+      "loss": 3.3781,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759701384555177,
+      "loss": 3.537,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758900886145205,
+      "loss": 3.5197,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758099089554584,
+      "loss": 3.5167,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757295995002031,
+      "loss": 3.5496,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756491602706625,
+      "loss": 3.4181,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755685912887798,
+      "loss": 3.3405,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754878925765329,
+      "loss": 3.3718,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754070641559362,
+      "loss": 3.5102,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753261060490384,
+      "loss": 3.4478,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752450182779243,
+      "loss": 3.4663,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751638008647139,
+      "loss": 3.4649,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750824538315623,
+      "loss": 3.4826,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750009772006603,
+      "loss": 3.6714,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749193709942339,
+      "loss": 3.4362,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748376352345443,
+      "loss": 3.5046,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747557699438884,
+      "loss": 3.4076,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974673775144598,
+      "loss": 3.608,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745916508590408,
+      "loss": 3.4245,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745093971096193,
+      "loss": 3.362,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744270139187716,
+      "loss": 3.5268,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743445013089707,
+      "loss": 3.5477,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742618593027255,
+      "loss": 3.5585,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097417908792258,
+      "loss": 3.5237,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974096187191113,
+      "loss": 3.4769,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740131571309394,
+      "loss": 3.5084,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973929997764709,
+      "loss": 3.361,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738467091151063,
+      "loss": 3.4713,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737632912048522,
+      "loss": 3.3813,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973679744056702,
+      "loss": 3.5314,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735960676934465,
+      "loss": 3.5343,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973512262137912,
+      "loss": 3.5633,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734283274129595,
+      "loss": 3.4784,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733442635414858,
+      "loss": 3.5808,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732600705464228,
+      "loss": 3.4311,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731757484507371,
+      "loss": 3.4009,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730912972774311,
+      "loss": 3.4182,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730067170495425,
+      "loss": 3.3137,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729220077901435,
+      "loss": 3.4959,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728371695223424,
+      "loss": 3.4695,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727522022692817,
+      "loss": 3.5425,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009726671060541401,
+      "loss": 3.5181,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725818809001308,
+      "loss": 3.4539,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724965268305024,
+      "loss": 3.4355,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724110438685386,
+      "loss": 3.4231,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723254320375583,
+      "loss": 3.3532,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722396913609157,
+      "loss": 3.5464,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721538218619997,
+      "loss": 3.4366,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972067823564235,
+      "loss": 3.3732,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719816964910809,
+      "loss": 3.3369,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718954406660317,
+      "loss": 3.4348,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718090561126176,
+      "loss": 3.4791,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717225428544034,
+      "loss": 3.3915,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716359009149889,
+      "loss": 3.5801,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971549130318009,
+      "loss": 3.2591,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009714622310871342,
+      "loss": 3.4877,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713752032460694,
+      "loss": 3.5312,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712880468185552,
+      "loss": 3.3617,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000971200761828367,
+      "loss": 3.5226,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711133482993151,
+      "loss": 3.3586,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710258062552451,
+      "loss": 3.579,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709381357200376,
+      "loss": 3.489,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708503367176083,
+      "loss": 3.4277,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707624092719078,
+      "loss": 3.3935,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970674353406922,
+      "loss": 3.5132,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705861691466716,
+      "loss": 3.3905,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970497856515212,
+      "loss": 3.4973,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704094155366345,
+      "loss": 3.4043,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703208462350649,
+      "loss": 3.4351,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702321486346637,
+      "loss": 3.5083,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970143322759627,
+      "loss": 3.4914,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700543686341854,
+      "loss": 3.5055,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699652862826049,
+      "loss": 3.3615,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969876075729186,
+      "loss": 3.4662,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697867369982647,
+      "loss": 3.4596,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696972701142118,
+      "loss": 3.4821,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696076751014327,
+      "loss": 3.4982,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969517951984368,
+      "loss": 3.4358,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694281007874935,
+      "loss": 3.4106,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693381215353196,
+      "loss": 3.4424,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692480142523918,
+      "loss": 3.5152,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691577789632904,
+      "loss": 3.4181,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690674156926308,
+      "loss": 3.4328,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689769244650631,
+      "loss": 3.4373,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688863053052722,
+      "loss": 3.4083,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687955582379785,
+      "loss": 3.5364,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687046832879367,
+      "loss": 3.4056,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686136804799365,
+      "loss": 3.5085,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685225498388026,
+      "loss": 3.4764,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684312913893947,
+      "loss": 3.3951,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968339905156607,
+      "loss": 3.5044,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968248391165369,
+      "loss": 3.3641,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681567494406445,
+      "loss": 3.4339,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680649800074323,
+      "loss": 3.5734,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679730828907669,
+      "loss": 3.5639,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678810581157162,
+      "loss": 3.4469,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967788905707384,
+      "loss": 3.559,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676966256909085,
+      "loss": 3.5335,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676042180914626,
+      "loss": 3.5049,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675116829342543,
+      "loss": 3.4148,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674190202445264,
+      "loss": 3.2746,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673262300475562,
+      "loss": 3.4149,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967233312368656,
+      "loss": 3.4076,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671402672331727,
+      "loss": 3.3937,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009670470946664883,
+      "loss": 3.4546,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966953794694019,
+      "loss": 3.5755,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668603673412164,
+      "loss": 3.3033,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667668126335665,
+      "loss": 3.4493,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666731305965901,
+      "loss": 3.4355,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665793212558427,
+      "loss": 3.4193,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664853846369143,
+      "loss": 3.4251,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663913207654303,
+      "loss": 3.4735,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662971296670502,
+      "loss": 3.5888,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662028113674683,
+      "loss": 3.2766,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661083658924139,
+      "loss": 3.4096,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660137932676505,
+      "loss": 3.4307,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009659190935189768,
+      "loss": 3.3639,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658242666722258,
+      "loss": 3.4123,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009657293127532654,
+      "loss": 3.3955,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965634231787998,
+      "loss": 3.3262,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655390238023609,
+      "loss": 3.3778,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965443688822326,
+      "loss": 3.5448,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653482268738991,
+      "loss": 3.5188,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965252637983122,
+      "loss": 3.3605,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651569221760699,
+      "loss": 3.4631,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650610794788533,
+      "loss": 3.3609,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009649651099176172,
+      "loss": 3.2997,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964869013518541,
+      "loss": 3.5119,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964772790307839,
+      "loss": 3.443,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646764403117598,
+      "loss": 3.4665,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645799635565869,
+      "loss": 3.4322,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644833600686378,
+      "loss": 3.3975,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643866298742654,
+      "loss": 3.3586,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642897729998563,
+      "loss": 3.4003,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641927894718325,
+      "loss": 3.4892,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640956793166501,
+      "loss": 3.5207,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639984425607995,
+      "loss": 3.4757,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963901079230806,
+      "loss": 3.5099,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638035893532297,
+      "loss": 3.4433,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637059729546645,
+      "loss": 3.4504,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963608230061739,
+      "loss": 3.4371,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963510360701117,
+      "loss": 3.4395,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634123648994959,
+      "loss": 3.4227,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633142426836081,
+      "loss": 3.4331,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632159940802205,
+      "loss": 3.4253,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631176191161341,
+      "loss": 3.5397,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009630191178181848,
+      "loss": 3.4202,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629204902132425,
+      "loss": 3.4039,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628217363282123,
+      "loss": 3.3886,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627228561900328,
+      "loss": 3.4761,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626238498256776,
+      "loss": 3.4573,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625247172621549,
+      "loss": 3.4066,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624254585265068,
+      "loss": 3.4431,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623260736458104,
+      "loss": 3.3837,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622265626471766,
+      "loss": 3.3437,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621269255577512,
+      "loss": 3.462,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009620271624047141,
+      "loss": 3.589,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619272732152796,
+      "loss": 3.4145,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618272580166967,
+      "loss": 3.4257,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617271168362485,
+      "loss": 3.5116,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616268497012523,
+      "loss": 3.4112,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615264566390602,
+      "loss": 3.2934,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614259376770584,
+      "loss": 3.4571,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613252928426674,
+      "loss": 3.3936,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000961224522163342,
+      "loss": 3.4374,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611236256665718,
+      "loss": 3.5991,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610226033798799,
+      "loss": 3.5437,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609214553308246,
+      "loss": 3.305,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608201815469977,
+      "loss": 3.2431,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607187820560258,
+      "loss": 3.3745,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009606172568855698,
+      "loss": 3.52,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605156060633246,
+      "loss": 3.3794,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009604138296170197,
+      "loss": 3.5553,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603119275744188,
+      "loss": 3.3365,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602098999633192,
+      "loss": 3.2153,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601077468115535,
+      "loss": 3.3288,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960005468146988,
+      "loss": 3.4794,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000959903063997523,
+      "loss": 3.4947,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598005343910938,
+      "loss": 3.4697,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009596978793556693,
+      "loss": 3.5168,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009595950989192524,
+      "loss": 3.3613,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000959492193109881,
+      "loss": 3.476,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593891619556265,
+      "loss": 3.4309,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000959286005484595,
+      "loss": 3.1764,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591827237249264,
+      "loss": 3.3749,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590793167047951,
+      "loss": 3.4264,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589757844524094,
+      "loss": 3.5059,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588721269960118,
+      "loss": 3.3735,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958768344363879,
+      "loss": 3.3721,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586644365843221,
+      "loss": 3.4304,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585604036856859,
+      "loss": 3.2022,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584562456963496,
+      "loss": 3.3411,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583519626447265,
+      "loss": 3.5984,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582475545592637,
+      "loss": 3.4496,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009581430214684431,
+      "loss": 3.4124,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580383634007801,
+      "loss": 3.3729,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579335803848244,
+      "loss": 3.4016,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578286724491595,
+      "loss": 3.4017,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577236396224036,
+      "loss": 3.3784,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576184819332085,
+      "loss": 3.5411,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575131994102601,
+      "loss": 3.3887,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574077920822784,
+      "loss": 3.2414,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573022599780177,
+      "loss": 3.4576,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571966031262657,
+      "loss": 3.4938,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570908215558449,
+      "loss": 3.483,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569849152956114,
+      "loss": 3.3474,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568788843744552,
+      "loss": 3.3386,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567727288213005,
+      "loss": 3.4209,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009566664486651055,
+      "loss": 3.6608,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565600439348625,
+      "loss": 3.5509,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564535146595976,
+      "loss": 3.2398,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563468608683709,
+      "loss": 3.3412,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562400825902764,
+      "loss": 3.4999,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561331798544422,
+      "loss": 3.5068,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560261526900303,
+      "loss": 3.4056,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559190011262369,
+      "loss": 3.4596,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558117251922913,
+      "loss": 3.283,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557043249174578,
+      "loss": 3.4376,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955596800331034,
+      "loss": 3.3436,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554891514623514,
+      "loss": 3.362,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553813783407756,
+      "loss": 3.5383,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552734809957062,
+      "loss": 3.4304,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551654594565763,
+      "loss": 3.378,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550573137528532,
+      "loss": 3.2956,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954949043914038,
+      "loss": 3.4008,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548406499696653,
+      "loss": 3.1183,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009547321319493045,
+      "loss": 3.4021,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546234898825578,
+      "loss": 3.4357,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545147237990617,
+      "loss": 3.4669,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544058337284867,
+      "loss": 3.4969,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542968197005366,
+      "loss": 3.3618,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541876817449498,
+      "loss": 3.481,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540784198914976,
+      "loss": 3.3194,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539690341699857,
+      "loss": 3.5061,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538595246102535,
+      "loss": 3.3064,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537498912421741,
+      "loss": 3.4014,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536401340956542,
+      "loss": 3.5274,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535302532006348,
+      "loss": 3.4335,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00095342024858709,
+      "loss": 3.3156,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533101202850282,
+      "loss": 3.3732,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009531998683244911,
+      "loss": 3.3095,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009530894927355545,
+      "loss": 3.4049,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529789935483275,
+      "loss": 3.533,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528683707929535,
+      "loss": 3.4015,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952757624499609,
+      "loss": 3.561,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526467546985048,
+      "loss": 3.392,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525357614198848,
+      "loss": 3.357,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952424644694027,
+      "loss": 3.5334,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523134045512429,
+      "loss": 3.3527,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522020410218775,
+      "loss": 3.3857,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520905541363099,
+      "loss": 3.3311,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519789439249527,
+      "loss": 3.4424,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009518672104182517,
+      "loss": 3.4414,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517553536466872,
+      "loss": 3.5108,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951643373640772,
+      "loss": 3.3373,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515312704310535,
+      "loss": 3.4415,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514190440481123,
+      "loss": 3.397,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513066945225626,
+      "loss": 3.3632,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511942218850523,
+      "loss": 3.4813,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510816261662627,
+      "loss": 3.3691,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009509689073969088,
+      "loss": 3.296,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508560656077392,
+      "loss": 3.394,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950743100829536,
+      "loss": 3.426,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506300130931149,
+      "loss": 3.5267,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505168024293249,
+      "loss": 3.4061,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950403468869049,
+      "loss": 3.4177,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502900124432035,
+      "loss": 3.383,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501764331827378,
+      "loss": 3.3181,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009500627311186356,
+      "loss": 3.3809,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499489062819136,
+      "loss": 3.4403,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498349587036219,
+      "loss": 3.4243,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497208884148444,
+      "loss": 3.3877,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496066954466983,
+      "loss": 3.4378,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494923798303343,
+      "loss": 3.4132,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493779415969365,
+      "loss": 3.4078,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492633807777227,
+      "loss": 3.5369,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491486974039438,
+      "loss": 3.4259,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490338915068841,
+      "loss": 3.5392,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009489189631178618,
+      "loss": 3.4965,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488039122682283,
+      "loss": 3.4576,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486887389893679,
+      "loss": 3.3752,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485734433126991,
+      "loss": 3.4839,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484580252696733,
+      "loss": 3.4843,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483424848917753,
+      "loss": 3.3902,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482268222105234,
+      "loss": 3.3114,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009481110372574693,
+      "loss": 3.4135,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947995130064198,
+      "loss": 3.4888,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478791006623276,
+      "loss": 3.4122,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00094776294908351,
+      "loss": 3.3093,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476466753594302,
+      "loss": 3.3862,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475302795218061,
+      "loss": 3.423,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474137616023899,
+      "loss": 3.4152,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009472971216329662,
+      "loss": 3.3902,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471803596453535,
+      "loss": 3.2852,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470634756714027,
+      "loss": 3.3719,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469464697429992,
+      "loss": 3.3613,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468293418920608,
+      "loss": 3.4722,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009467120921505388,
+      "loss": 3.3449,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465947205504178,
+      "loss": 3.2864,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464772271237155,
+      "loss": 3.4195,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463596119024831,
+      "loss": 3.3884,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462418749188048,
+      "loss": 3.5345,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000946124016204798,
+      "loss": 3.4601,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460060357926134,
+      "loss": 3.3562,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945887933714435,
+      "loss": 3.2827,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457697100024798,
+      "loss": 3.5575,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945651364688998,
+      "loss": 3.3943,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945532897806273,
+      "loss": 3.4887,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454143093866216,
+      "loss": 3.4859,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452955994623932,
+      "loss": 3.3214,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451767680659709,
+      "loss": 3.287,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450578152297706,
+      "loss": 3.4778,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009449387409862415,
+      "loss": 3.3501,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448195453678661,
+      "loss": 3.3924,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447002284071593,
+      "loss": 3.4862,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445807901366699,
+      "loss": 3.4371,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444612305889795,
+      "loss": 3.5001,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443415497967026,
+      "loss": 3.3799,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000944221747792487,
+      "loss": 3.428,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441018246090134,
+      "loss": 3.4268,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439817802789957,
+      "loss": 3.4337,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438616148351809,
+      "loss": 3.3007,
+      "step": 1103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009437413283103486,
+      "loss": 3.3873,
+      "step": 1104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436209207373123,
+      "loss": 3.4679,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435003921489176,
+      "loss": 3.4424,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433797425780435,
+      "loss": 3.4173,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009432589720576021,
+      "loss": 3.433,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009431380806205385,
+      "loss": 3.4495,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430170682998305,
+      "loss": 3.5405,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000942895935128489,
+      "loss": 3.3693,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427746811395581,
+      "loss": 3.3385,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426533063661147,
+      "loss": 3.3986,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425318108412684,
+      "loss": 3.4789,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000942410194598162,
+      "loss": 3.4064,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009422884576699715,
+      "loss": 3.4204,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421666000899052,
+      "loss": 3.3817,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009420446218912047,
+      "loss": 3.3647,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419225231071446,
+      "loss": 3.2934,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418003037710321,
+      "loss": 3.4512,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416779639162072,
+      "loss": 3.2144,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009415555035760434,
+      "loss": 3.4466,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009414329227839464,
+      "loss": 3.5094,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413102215733553,
+      "loss": 3.4368,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411873999777414,
+      "loss": 3.352,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410644580306092,
+      "loss": 3.2927,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009409413957654964,
+      "loss": 3.3452,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408182132159728,
+      "loss": 3.4336,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406949104156417,
+      "loss": 3.3229,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009405714873981386,
+      "loss": 3.4584,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404479441971321,
+      "loss": 3.5503,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009403242808463236,
+      "loss": 3.3838,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009402004973794474,
+      "loss": 3.3587,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094007659383027,
+      "loss": 3.3125,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399525702325915,
+      "loss": 3.3735,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398284266202439,
+      "loss": 3.514,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397041630270926,
+      "loss": 3.4996,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009395797794870354,
+      "loss": 3.328,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394552760340029,
+      "loss": 3.4831,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393306527019584,
+      "loss": 3.3892,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009392059095248978,
+      "loss": 3.3798,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390810465368499,
+      "loss": 3.3717,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000938956063771876,
+      "loss": 3.3771,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388309612640703,
+      "loss": 3.476,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387057390475593,
+      "loss": 3.3482,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385803971565025,
+      "loss": 3.343,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000938454935625092,
+      "loss": 3.3992,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009383293544875523,
+      "loss": 3.2775,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382036537781408,
+      "loss": 3.4508,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380778335311472,
+      "loss": 3.3439,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379518937808941,
+      "loss": 3.452,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378258345617368,
+      "loss": 3.3057,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009376996559080628,
+      "loss": 3.6207,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375733578542925,
+      "loss": 3.4744,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374469404348784,
+      "loss": 3.3499,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373204036843064,
+      "loss": 3.2892,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371937476370942,
+      "loss": 3.4037,
+      "step": 1157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009370669723277922,
+      "loss": 3.1885,
+      "step": 1158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369400777909838,
+      "loss": 3.2903,
+      "step": 1159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368130640612842,
+      "loss": 3.4753,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366859311733418,
+      "loss": 3.3379,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365586791618368,
+      "loss": 3.2838,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364313080614826,
+      "loss": 3.4207,
+      "step": 1163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363038179070245,
+      "loss": 3.4738,
+      "step": 1164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361762087332408,
+      "loss": 3.3545,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360484805749418,
+      "loss": 3.269,
+      "step": 1166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009359206334669707,
+      "loss": 3.2842,
+      "step": 1167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009357926674442027,
+      "loss": 3.2509,
+      "step": 1168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356645825415459,
+      "loss": 3.4129,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355363787939404,
+      "loss": 3.5417,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009354080562363588,
+      "loss": 3.4689,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352796149038063,
+      "loss": 3.3206,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351510548313204,
+      "loss": 3.3774,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350223760539713,
+      "loss": 3.4419,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009348935786068608,
+      "loss": 3.453,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347646625251238,
+      "loss": 3.3074,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346356278439274,
+      "loss": 3.4048,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009345064745984709,
+      "loss": 3.3332,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343772028239858,
+      "loss": 3.4451,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342478125557366,
+      "loss": 3.3869,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009341183038290193,
+      "loss": 3.3183,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339886766791628,
+      "loss": 3.4331,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933858931141528,
+      "loss": 3.3845,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337290672515081,
+      "loss": 3.299,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335990850445288,
+      "loss": 3.5026,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334689845560479,
+      "loss": 3.434,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333387658215555,
+      "loss": 3.3191,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933208428876574,
+      "loss": 3.3561,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330779737566581,
+      "loss": 3.2574,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009329474004973945,
+      "loss": 3.2602,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009328167091344024,
+      "loss": 3.2468,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326858997033329,
+      "loss": 3.4466,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009325549722398698,
+      "loss": 3.3555,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324239267797287,
+      "loss": 3.2234,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322927633586575,
+      "loss": 3.3295,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932161482012436,
+      "loss": 3.272,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009320300827768769,
+      "loss": 3.4187,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009318985656878243,
+      "loss": 3.4064,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317669307811547,
+      "loss": 3.3096,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931635178092777,
+      "loss": 3.393,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315033076586319,
+      "loss": 3.2821,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009313713195146923,
+      "loss": 3.3809,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312392136969634,
+      "loss": 3.3547,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311069902414821,
+      "loss": 3.5618,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309746491843181,
+      "loss": 3.3222,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308421905615722,
+      "loss": 3.3861,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000930709614409378,
+      "loss": 3.5047,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000930576920763901,
+      "loss": 3.4651,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304441096613386,
+      "loss": 3.4263,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009303111811379206,
+      "loss": 3.3901,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301781352299081,
+      "loss": 3.3603,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000930044971973595,
+      "loss": 3.3481,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299116914053071,
+      "loss": 3.2465,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297782935614017,
+      "loss": 3.4738,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296447784782684,
+      "loss": 3.3136,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009295111461923289,
+      "loss": 3.2506,
+      "step": 1216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009293773967400368,
+      "loss": 3.2414,
+      "step": 1217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009292435301578773,
+      "loss": 3.3469,
+      "step": 1218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009291095464823683,
+      "loss": 3.3221,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928975445750059,
+      "loss": 3.357,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009288412279975307,
+      "loss": 3.3918,
+      "step": 1221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287068932613967,
+      "loss": 3.4859,
+      "step": 1222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009285724415783023,
+      "loss": 3.4007,
+      "step": 1223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284378729849243,
+      "loss": 3.2551,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928303187517972,
+      "loss": 3.3322,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928168385214186,
+      "loss": 3.2775,
+      "step": 1226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280334661103391,
+      "loss": 3.4409,
+      "step": 1227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278984302432358,
+      "loss": 3.3964,
+      "step": 1228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277632776497129,
+      "loss": 3.3164,
+      "step": 1229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009276280083666382,
+      "loss": 3.3247,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009274926224309121,
+      "loss": 3.2771,
+      "step": 1231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273571198794663,
+      "loss": 3.3696,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000927221500749265,
+      "loss": 3.2238,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270857650773032,
+      "loss": 3.2579,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269499129006085,
+      "loss": 3.4019,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00092681394425624,
+      "loss": 3.4263,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009266778591812885,
+      "loss": 3.4831,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009265416577128769,
+      "loss": 3.5093,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009264053398881593,
+      "loss": 3.4835,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926268905744322,
+      "loss": 3.473,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926132355318583,
+      "loss": 3.3827,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259956886481916,
+      "loss": 3.4095,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258589057704291,
+      "loss": 3.2897,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009257220067226088,
+      "loss": 3.4457,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009255849915420754,
+      "loss": 3.3444,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009254478602662049,
+      "loss": 3.345,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253106129324056,
+      "loss": 3.2801,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251732495781171,
+      "loss": 3.4944,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925035770240811,
+      "loss": 3.3267,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009248981749579899,
+      "loss": 3.4009,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247604637671887,
+      "loss": 3.4202,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246226367059736,
+      "loss": 3.2489,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244846938119422,
+      "loss": 3.4814,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243466351227243,
+      "loss": 3.416,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009242084606759809,
+      "loss": 3.3531,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009240701705094044,
+      "loss": 3.3162,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000923931764660719,
+      "loss": 3.2706,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237932431676806,
+      "loss": 3.3022,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236546060680761,
+      "loss": 3.3965,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235158533997249,
+      "loss": 3.4596,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009233769852004769,
+      "loss": 3.3808,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232380015082143,
+      "loss": 3.3786,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00092309890236085,
+      "loss": 3.4302,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009229596877963292,
+      "loss": 3.3764,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009228203578526281,
+      "loss": 3.4333,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009226809125677548,
+      "loss": 3.5228,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225413519797482,
+      "loss": 3.2941,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224016761266793,
+      "loss": 3.3698,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00092226188504665,
+      "loss": 3.3965,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009221219787777942,
+      "loss": 3.3558,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009219819573582768,
+      "loss": 3.5112,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009218418208262944,
+      "loss": 3.3175,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217015692200745,
+      "loss": 3.3449,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215612025778767,
+      "loss": 3.3651,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009214207209379914,
+      "loss": 3.4455,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009212801243387406,
+      "loss": 3.4602,
+      "step": 1276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211394128184778,
+      "loss": 3.5385,
+      "step": 1277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209985864155877,
+      "loss": 3.2499,
+      "step": 1278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000920857645168486,
+      "loss": 3.4178,
+      "step": 1279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009207165891156205,
+      "loss": 3.452,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009205754182954696,
+      "loss": 3.3734,
+      "step": 1281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009204341327465434,
+      "loss": 3.4437,
+      "step": 1282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202927325073832,
+      "loss": 3.4694,
+      "step": 1283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201512176165615,
+      "loss": 3.4712,
+      "step": 1284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009200095881126822,
+      "loss": 3.3471,
+      "step": 1285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198678440343804,
+      "loss": 3.4713,
+      "step": 1286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009197259854203226,
+      "loss": 3.3782,
+      "step": 1287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919584012309206,
+      "loss": 3.4582,
+      "step": 1288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00091944192473976,
+      "loss": 3.4302,
+      "step": 1289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009192997227507443,
+      "loss": 3.3561,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191574063809504,
+      "loss": 3.41,
+      "step": 1291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009190149756692006,
+      "loss": 3.3291,
+      "step": 1292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188724306543486,
+      "loss": 3.4251,
+      "step": 1293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009187297713752794,
+      "loss": 3.4574,
+      "step": 1294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185869978709087,
+      "loss": 3.2576,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009184441101801841,
+      "loss": 3.5057,
+      "step": 1296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183011083420838,
+      "loss": 3.2183,
+      "step": 1297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181579923956171,
+      "loss": 3.336,
+      "step": 1298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009180147623798249,
+      "loss": 3.3807,
+      "step": 1299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178714183337787,
+      "loss": 3.3969,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009177279602965813,
+      "loss": 3.4602,
+      "step": 1301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175843883073667,
+      "loss": 3.2714,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009174407024053,
+      "loss": 3.3338,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009172969026295769,
+      "loss": 3.175,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009171529890194252,
+      "loss": 3.3754,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009170089616141026,
+      "loss": 3.4411,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168648204528983,
+      "loss": 3.258,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009167205655751329,
+      "loss": 3.2787,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009165761970201574,
+      "loss": 3.401,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009164317148273543,
+      "loss": 3.4217,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009162871190361369,
+      "loss": 3.3705,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009161424096859492,
+      "loss": 3.4489,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159975868162668,
+      "loss": 3.3931,
+      "step": 1313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009158526504665957,
+      "loss": 3.2986,
+      "step": 1314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009157076006764733,
+      "loss": 3.3457,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009155624374854676,
+      "loss": 3.4797,
+      "step": 1316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009154171609331778,
+      "loss": 3.3956,
+      "step": 1317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009152717710592336,
+      "loss": 3.3376,
+      "step": 1318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009151262679032963,
+      "loss": 3.3388,
+      "step": 1319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009149806515050573,
+      "loss": 3.4063,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148349219042396,
+      "loss": 3.5079,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009146890791405966,
+      "loss": 3.2323,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914543123253913,
+      "loss": 3.3223,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143970542840039,
+      "loss": 3.3247,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009142508722707154,
+      "loss": 3.4344,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141045772539247,
+      "loss": 3.4103,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139581692735395,
+      "loss": 3.2872,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009138116483694984,
+      "loss": 3.3249,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136650145817708,
+      "loss": 3.3037,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009135182679503573,
+      "loss": 3.2295,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133714085152884,
+      "loss": 3.2756,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009132244363166262,
+      "loss": 3.2556,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009130773513944632,
+      "loss": 3.4014,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009129301537889227,
+      "loss": 3.2324,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009127828435401587,
+      "loss": 3.2639,
+      "step": 1335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000912635420688356,
+      "loss": 3.5575,
+      "step": 1336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00091248788527373,
+      "loss": 3.5006,
+      "step": 1337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009123402373365269,
+      "loss": 3.4526,
+      "step": 1338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121924769170237,
+      "loss": 3.3776,
+      "step": 1339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009120446040555279,
+      "loss": 3.4802,
+      "step": 1340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118966187923777,
+      "loss": 3.4072,
+      "step": 1341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911748521167942,
+      "loss": 3.5645,
+      "step": 1342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116003112226205,
+      "loss": 3.3619,
+      "step": 1343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009114519889968434,
+      "loss": 3.3224,
+      "step": 1344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911303554531071,
+      "loss": 3.249,
+      "step": 1345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009111550078657956,
+      "loss": 3.1902,
+      "step": 1346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110063490415385,
+      "loss": 3.4587,
+      "step": 1347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009108575780988527,
+      "loss": 3.3273,
+      "step": 1348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009107086950783215,
+      "loss": 3.3583,
+      "step": 1349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105597000205585,
+      "loss": 3.4856,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104105929662081,
+      "loss": 3.3383,
+      "step": 1351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009102613739559452,
+      "loss": 3.4071,
+      "step": 1352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009101120430304755,
+      "loss": 3.3558,
+      "step": 1353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009099626002305346,
+      "loss": 3.4049,
+      "step": 1354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009098130455968892,
+      "loss": 3.3784,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009096633791703364,
+      "loss": 3.2711,
+      "step": 1356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095136009917036,
+      "loss": 3.3998,
+      "step": 1357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009093637111018487,
+      "loss": 3.3993,
+      "step": 1358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009092137095416604,
+      "loss": 3.5481,
+      "step": 1359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009090635963520575,
+      "loss": 3.4415,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009089133715739893,
+      "loss": 3.3139,
+      "step": 1361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087630352484357,
+      "loss": 3.2524,
+      "step": 1362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908612587416407,
+      "loss": 3.2122,
+      "step": 1363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009084620281189438,
+      "loss": 3.4643,
+      "step": 1364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009083113573971171,
+      "loss": 3.4022,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009081605752920285,
+      "loss": 3.3301,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00090800968184481,
+      "loss": 3.4332,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009078586770966236,
+      "loss": 3.269,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077075610886616,
+      "loss": 3.3823,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009075563338621476,
+      "loss": 3.2805,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074049954583344,
+      "loss": 3.2919,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009072535459185056,
+      "loss": 3.2892,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009071019852839755,
+      "loss": 3.3164,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000906950313596088,
+      "loss": 3.4243,
+      "step": 1374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009067985308962176,
+      "loss": 3.5045,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009066466372257694,
+      "loss": 3.3844,
+      "step": 1376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009064946326261784,
+      "loss": 3.3357,
+      "step": 1377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009063425171389097,
+      "loss": 3.3924,
+      "step": 1378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061902908054589,
+      "loss": 3.3363,
+      "step": 1379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009060379536673521,
+      "loss": 3.1473,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009058855057661452,
+      "loss": 3.3926,
+      "step": 1381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057329471434244,
+      "loss": 3.3823,
+      "step": 1382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055802778408063,
+      "loss": 3.379,
+      "step": 1383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009054274978999373,
+      "loss": 3.3205,
+      "step": 1384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052746073624947,
+      "loss": 3.4581,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000905121606270185,
+      "loss": 3.4159,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009049684946647458,
+      "loss": 3.4125,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048152725879442,
+      "loss": 3.3618,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046619400815777,
+      "loss": 3.2907,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009045084971874737,
+      "loss": 3.341,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009043549439474903,
+      "loss": 3.3855,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000904201280403515,
+      "loss": 3.4016,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040475065974656,
+      "loss": 3.3087,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009038936225712901,
+      "loss": 3.3458,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009037396283669667,
+      "loss": 3.3492,
+      "step": 1395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009035855240265037,
+      "loss": 3.2055,
+      "step": 1396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034313095919386,
+      "loss": 3.2368,
+      "step": 1397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009032769851053399,
+      "loss": 3.5413,
+      "step": 1398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009031225506088057,
+      "loss": 3.3277,
+      "step": 1399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029680061444645,
+      "loss": 3.4812,
+      "step": 1400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009028133517544741,
+      "loss": 3.4095,
+      "step": 1401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009026585874810227,
+      "loss": 3.3131,
+      "step": 1402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025037133663287,
+      "loss": 3.3154,
+      "step": 1403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00090234872945264,
+      "loss": 3.3992,
+      "step": 1404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021936357822347,
+      "loss": 3.3944,
+      "step": 1405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009020384323974209,
+      "loss": 3.4425,
+      "step": 1406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018831193405365,
+      "loss": 3.2933,
+      "step": 1407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009017276966539491,
+      "loss": 3.3987,
+      "step": 1408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000901572164380057,
+      "loss": 3.4202,
+      "step": 1409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009014165225612874,
+      "loss": 3.3525,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000901260771240098,
+      "loss": 3.3778,
+      "step": 1411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009011049104589759,
+      "loss": 3.3594,
+      "step": 1412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009489402604389,
+      "loss": 3.2807,
+      "step": 1413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009007928606870339,
+      "loss": 3.4698,
+      "step": 1414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009006366717813377,
+      "loss": 3.321,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009004803735859571,
+      "loss": 3.4248,
+      "step": 1416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003239661435289,
+      "loss": 3.4691,
+      "step": 1417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001674494967191,
+      "loss": 3.2513,
+      "step": 1418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009000108236882245,
+      "loss": 3.3327,
+      "step": 1419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008998540887607705,
+      "loss": 3.4059,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899697244757113,
+      "loss": 3.3752,
+      "step": 1421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008995402917200373,
+      "loss": 3.3514,
+      "step": 1422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993832296923591,
+      "loss": 3.3647,
+      "step": 1423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899226058716923,
+      "loss": 3.2287,
+      "step": 1424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990687788366037,
+      "loss": 3.3707,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008989113900943056,
+      "loss": 3.3156,
+      "step": 1426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987538925329628,
+      "loss": 3.4474,
+      "step": 1427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898596286195539,
+      "loss": 3.3224,
+      "step": 1428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008984385711250277,
+      "loss": 3.4184,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008982807473644521,
+      "loss": 3.3285,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008981228149568647,
+      "loss": 3.348,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897964773945348,
+      "loss": 3.3918,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897806624373014,
+      "loss": 3.3156,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008976483662830046,
+      "loss": 3.4797,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008974899997184906,
+      "loss": 3.3313,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897331524722673,
+      "loss": 3.2881,
+      "step": 1436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971729413387822,
+      "loss": 3.3171,
+      "step": 1437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008970142496100782,
+      "loss": 3.3463,
+      "step": 1438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008968554495798505,
+      "loss": 3.1659,
+      "step": 1439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896696541291418,
+      "loss": 3.2594,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008965375247881296,
+      "loss": 3.3362,
+      "step": 1441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896378400113363,
+      "loss": 3.3815,
+      "step": 1442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008962191673105263,
+      "loss": 3.3254,
+      "step": 1443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008960598264230563,
+      "loss": 3.306,
+      "step": 1444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959003774944198,
+      "loss": 3.3453,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008957408205681128,
+      "loss": 3.4091,
+      "step": 1446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955811556876605,
+      "loss": 3.3559,
+      "step": 1447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008954213828966185,
+      "loss": 3.4489,
+      "step": 1448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008952615022385709,
+      "loss": 3.5211,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008951015137571314,
+      "loss": 3.338,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949414174959434,
+      "loss": 3.4531,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947812134986797,
+      "loss": 3.3413,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008946209018090422,
+      "loss": 3.2718,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008944604824707623,
+      "loss": 3.2736,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000894299955527601,
+      "loss": 3.3943,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008941393210233482,
+      "loss": 3.1605,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939785790018235,
+      "loss": 3.3159,
+      "step": 1457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008938177295068757,
+      "loss": 3.2299,
+      "step": 1458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000893656772582383,
+      "loss": 3.3505,
+      "step": 1459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008934957082722528,
+      "loss": 3.1961,
+      "step": 1460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933345366204218,
+      "loss": 3.5531,
+      "step": 1461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000893173257670856,
+      "loss": 3.4106,
+      "step": 1462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008930118714675508,
+      "loss": 3.5009,
+      "step": 1463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008928503780545307,
+      "loss": 3.3692,
+      "step": 1464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008926887774758493,
+      "loss": 3.3384,
+      "step": 1465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00089252706977559,
+      "loss": 3.2305,
+      "step": 1466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923652549978647,
+      "loss": 3.3736,
+      "step": 1467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008922033331868149,
+      "loss": 3.3896,
+      "step": 1468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008920413043866116,
+      "loss": 3.3662,
+      "step": 1469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008918791686414543,
+      "loss": 3.4063,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891716925995572,
+      "loss": 3.2955,
+      "step": 1471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891554576493223,
+      "loss": 3.2443,
+      "step": 1472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913921201786947,
+      "loss": 3.3375,
+      "step": 1473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008912295570963033,
+      "loss": 3.3762,
+      "step": 1474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910668872903946,
+      "loss": 3.4337,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008909041108053433,
+      "loss": 3.2725,
+      "step": 1476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907412276855532,
+      "loss": 3.3808,
+      "step": 1477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000890578237975457,
+      "loss": 3.2695,
+      "step": 1478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008904151417195167,
+      "loss": 3.2181,
+      "step": 1479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008902519389622237,
+      "loss": 3.2118,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008900886297480977,
+      "loss": 3.4683,
+      "step": 1481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008899252141216879,
+      "loss": 3.3104,
+      "step": 1482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008897616921275724,
+      "loss": 3.1985,
+      "step": 1483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895980638103587,
+      "loss": 3.3502,
+      "step": 1484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008894343292146825,
+      "loss": 3.246,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892704883852092,
+      "loss": 3.3688,
+      "step": 1486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889106541366633,
+      "loss": 3.2304,
+      "step": 1487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008889424882036769,
+      "loss": 3.3946,
+      "step": 1488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887783289410931,
+      "loss": 3.3653,
+      "step": 1489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008886140636236624,
+      "loss": 3.4106,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008884496922961949,
+      "loss": 3.2781,
+      "step": 1491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882852150035295,
+      "loss": 3.3197,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008881206317905337,
+      "loss": 3.3429,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008879559427021044,
+      "loss": 3.3133,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008877911477831671,
+      "loss": 3.3206,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887626247078676,
+      "loss": 3.2705,
+      "step": 1496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874612406336147,
+      "loss": 3.2695,
+      "step": 1497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008872961284929952,
+      "loss": 3.329,
+      "step": 1498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008871309107018584,
+      "loss": 3.2673,
+      "step": 1499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008869655873052738,
+      "loss": 3.315,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008868001583483405,
+      "loss": 3.3725,
+      "step": 1501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008866346238761855,
+      "loss": 3.2297,
+      "step": 1502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864689839339652,
+      "loss": 3.3058,
+      "step": 1503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008863032385668641,
+      "loss": 3.4595,
+      "step": 1504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861373878200964,
+      "loss": 3.3501,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859714317389041,
+      "loss": 3.3853,
+      "step": 1506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008858053703685587,
+      "loss": 3.4266,
+      "step": 1507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008856392037543599,
+      "loss": 3.2929,
+      "step": 1508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008854729319416363,
+      "loss": 3.3024,
+      "step": 1509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853065549757451,
+      "loss": 3.302,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851400729020726,
+      "loss": 3.2701,
+      "step": 1511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849734857660331,
+      "loss": 3.2209,
+      "step": 1512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008848067936130701,
+      "loss": 3.2898,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008846399964886555,
+      "loss": 3.4593,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844730944382899,
+      "loss": 3.1545,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008843060875075024,
+      "loss": 3.3224,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000884138975741851,
+      "loss": 3.3049,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008839717591869221,
+      "loss": 3.3861,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008838044378883305,
+      "loss": 3.5483,
+      "step": 1519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008836370118917201,
+      "loss": 3.2823,
+      "step": 1520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834694812427629,
+      "loss": 3.432,
+      "step": 1521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833018459871593,
+      "loss": 3.4038,
+      "step": 1522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883134106170639,
+      "loss": 3.3181,
+      "step": 1523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008829662618389595,
+      "loss": 3.3551,
+      "step": 1524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827983130379072,
+      "loss": 3.233,
+      "step": 1525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008826302598132964,
+      "loss": 3.4419,
+      "step": 1526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882462102210971,
+      "loss": 3.3194,
+      "step": 1527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008822938402768024,
+      "loss": 3.2747,
+      "step": 1528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821254740566906,
+      "loss": 3.2828,
+      "step": 1529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819570035965645,
+      "loss": 3.3737,
+      "step": 1530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817884289423812,
+      "loss": 3.2664,
+      "step": 1531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000881619750140126,
+      "loss": 3.267,
+      "step": 1532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008814509672358126,
+      "loss": 3.5011,
+      "step": 1533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008812820802754837,
+      "loss": 3.4976,
+      "step": 1534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008811130893052099,
+      "loss": 3.4919,
+      "step": 1535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00088094399437109,
+      "loss": 3.2727,
+      "step": 1536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008807747955192517,
+      "loss": 3.2272,
+      "step": 1537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008806054927958506,
+      "loss": 3.2577,
+      "step": 1538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008804360862470709,
+      "loss": 3.4307,
+      "step": 1539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802665759191247,
+      "loss": 3.4068,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000880096961858253,
+      "loss": 3.3839,
+      "step": 1541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008799272441107249,
+      "loss": 3.428,
+      "step": 1542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008797574227228375,
+      "loss": 3.3235,
+      "step": 1543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008795874977409167,
+      "loss": 3.4361,
+      "step": 1544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008794174692113158,
+      "loss": 3.3737,
+      "step": 1545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008792473371804174,
+      "loss": 3.2859,
+      "step": 1546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790771016946315,
+      "loss": 3.3732,
+      "step": 1547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008789067628003968,
+      "loss": 3.4734,
+      "step": 1548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008787363205441799,
+      "loss": 3.2824,
+      "step": 1549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000878565774972476,
+      "loss": 3.471,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008783951261318079,
+      "loss": 3.3632,
+      "step": 1551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008782243740687272,
+      "loss": 3.2654,
+      "step": 1552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008780535188298134,
+      "loss": 3.2931,
+      "step": 1553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008778825604616737,
+      "loss": 3.3474,
+      "step": 1554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008777114990109442,
+      "loss": 3.3624,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008775403345242885,
+      "loss": 3.2995,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008773690670483988,
+      "loss": 3.4203,
+      "step": 1557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000877197696629995,
+      "loss": 3.3728,
+      "step": 1558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008770262233158253,
+      "loss": 3.3482,
+      "step": 1559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008768546471526659,
+      "loss": 3.1999,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766829681873212,
+      "loss": 3.3516,
+      "step": 1561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008765111864666231,
+      "loss": 3.3693,
+      "step": 1562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008763393020374324,
+      "loss": 3.3647,
+      "step": 1563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008761673149466373,
+      "loss": 3.376,
+      "step": 1564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875995225241154,
+      "loss": 3.2895,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008758230329679271,
+      "loss": 3.2676,
+      "step": 1566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008756507381739287,
+      "loss": 3.3312,
+      "step": 1567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008754783409061593,
+      "loss": 3.2945,
+      "step": 1568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875305841211647,
+      "loss": 3.3013,
+      "step": 1569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008751332391374482,
+      "loss": 3.322,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008749605347306467,
+      "loss": 3.3417,
+      "step": 1571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008747877280383548,
+      "loss": 3.3849,
+      "step": 1572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746148191077123,
+      "loss": 3.2775,
+      "step": 1573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744418079858873,
+      "loss": 3.2642,
+      "step": 1574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008742686947200751,
+      "loss": 3.4024,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008740954793574995,
+      "loss": 3.2527,
+      "step": 1576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008739221619454121,
+      "loss": 3.3368,
+      "step": 1577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008737487425310917,
+      "loss": 3.2765,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735752211618456,
+      "loss": 3.3055,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873401597885009,
+      "loss": 3.3211,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008732278727479441,
+      "loss": 3.3662,
+      "step": 1581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008730540457980417,
+      "loss": 3.2515,
+      "step": 1582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00087288011708272,
+      "loss": 3.4034,
+      "step": 1583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008727060866494249,
+      "loss": 3.3333,
+      "step": 1584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008725319545456303,
+      "loss": 3.4605,
+      "step": 1585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008723577208188377,
+      "loss": 3.3403,
+      "step": 1586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008721833855165761,
+      "loss": 3.2783,
+      "step": 1587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720089486864029,
+      "loss": 3.1317,
+      "step": 1588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008718344103759021,
+      "loss": 3.2456,
+      "step": 1589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008716597706326867,
+      "loss": 3.3759,
+      "step": 1590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008714850295043961,
+      "loss": 3.2944,
+      "step": 1591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008713101870386981,
+      "loss": 3.2833,
+      "step": 1592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711352432832881,
+      "loss": 3.2855,
+      "step": 1593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008709601982858891,
+      "loss": 3.3516,
+      "step": 1594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707850520942512,
+      "loss": 3.3091,
+      "step": 1595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008706098047561529,
+      "loss": 3.4651,
+      "step": 1596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008704344563193998,
+      "loss": 3.3412,
+      "step": 1597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702590068318252,
+      "loss": 3.3547,
+      "step": 1598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008700834563412902,
+      "loss": 3.2924,
+      "step": 1599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008699078048956828,
+      "loss": 3.363,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008697320525429194,
+      "loss": 3.4448,
+      "step": 1601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008695561993309431,
+      "loss": 3.5214,
+      "step": 1602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008693802453077251,
+      "loss": 3.4056,
+      "step": 1603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692041905212637,
+      "loss": 3.5342,
+      "step": 1604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008690280350195853,
+      "loss": 3.4405,
+      "step": 1605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000868851778850743,
+      "loss": 3.3173,
+      "step": 1606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008686754220628179,
+      "loss": 3.3101,
+      "step": 1607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008684989647039183,
+      "loss": 3.2712,
+      "step": 1608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00086832240682218,
+      "loss": 3.3881,
+      "step": 1609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008681457484657662,
+      "loss": 3.4118,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008679689896828677,
+      "loss": 3.3789,
+      "step": 1611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008677921305217022,
+      "loss": 3.252,
+      "step": 1612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008676151710305156,
+      "loss": 3.1559,
+      "step": 1613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674381112575802,
+      "loss": 3.4185,
+      "step": 1614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008672609512511964,
+      "loss": 3.265,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008670836910596917,
+      "loss": 3.2801,
+      "step": 1616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008669063307314207,
+      "loss": 3.2141,
+      "step": 1617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008667288703147658,
+      "loss": 3.3782,
+      "step": 1618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665513098581363,
+      "loss": 3.307,
+      "step": 1619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008663736494099688,
+      "loss": 3.2769,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661958890187276,
+      "loss": 3.3125,
+      "step": 1621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008660180287329036,
+      "loss": 3.2936,
+      "step": 1622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008658400686010155,
+      "loss": 3.3425,
+      "step": 1623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865662008671609,
+      "loss": 3.2615,
+      "step": 1624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008654838489932573,
+      "loss": 3.3733,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008653055896145602,
+      "loss": 3.4423,
+      "step": 1626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008651272305841454,
+      "loss": 3.3877,
+      "step": 1627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649487719506671,
+      "loss": 3.3195,
+      "step": 1628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008647702137628074,
+      "loss": 3.322,
+      "step": 1629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864591556069275,
+      "loss": 3.2975,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864412798918806,
+      "loss": 3.2645,
+      "step": 1631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008642339423601636,
+      "loss": 3.4851,
+      "step": 1632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008640549864421381,
+      "loss": 3.164,
+      "step": 1633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008638759312135467,
+      "loss": 3.114,
+      "step": 1634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863696776723234,
+      "loss": 3.3485,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008635175230200715,
+      "loss": 3.4216,
+      "step": 1636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863338170152958,
+      "loss": 3.2377,
+      "step": 1637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863158718170819,
+      "loss": 3.4166,
+      "step": 1638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008629791671226072,
+      "loss": 3.1966,
+      "step": 1639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008627995170573025,
+      "loss": 3.4511,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008626197680239115,
+      "loss": 3.3347,
+      "step": 1641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862439920071468,
+      "loss": 3.1548,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008622599732490326,
+      "loss": 3.2822,
+      "step": 1643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008620799276056932,
+      "loss": 3.4061,
+      "step": 1644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008618997831905644,
+      "loss": 3.2086,
+      "step": 1645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617195400527877,
+      "loss": 3.2865,
+      "step": 1646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008615391982415316,
+      "loss": 3.2712,
+      "step": 1647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008613587578059916,
+      "loss": 3.3824,
+      "step": 1648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008611782187953903,
+      "loss": 3.3385,
+      "step": 1649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008609975812589766,
+      "loss": 3.3252,
+      "step": 1650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008608168452460265,
+      "loss": 3.312,
+      "step": 1651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008606360108058435,
+      "loss": 3.2779,
+      "step": 1652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008604550779877571,
+      "loss": 3.3992,
+      "step": 1653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860274046841124,
+      "loss": 3.3449,
+      "step": 1654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008600929174153275,
+      "loss": 3.2202,
+      "step": 1655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008599116897597784,
+      "loss": 3.3143,
+      "step": 1656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008597303639239133,
+      "loss": 3.2843,
+      "step": 1657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008595489399571964,
+      "loss": 3.4379,
+      "step": 1658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593674179091182,
+      "loss": 3.5006,
+      "step": 1659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000859185797829196,
+      "loss": 3.2389,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008590040797669741,
+      "loss": 3.3283,
+      "step": 1661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588222637720233,
+      "loss": 3.2288,
+      "step": 1662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008586403498939414,
+      "loss": 3.2369,
+      "step": 1663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008584583381823523,
+      "loss": 3.3218,
+      "step": 1664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008582762286869073,
+      "loss": 3.2937,
+      "step": 1665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008580940214572841,
+      "loss": 3.2713,
+      "step": 1666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579117165431867,
+      "loss": 3.3341,
+      "step": 1667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857729313994346,
+      "loss": 3.3384,
+      "step": 1668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008575468138605203,
+      "loss": 3.3807,
+      "step": 1669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857364216191493,
+      "loss": 3.4699,
+      "step": 1670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008571815210370753,
+      "loss": 3.3673,
+      "step": 1671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008569987284471046,
+      "loss": 3.3418,
+      "step": 1672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008568158384714448,
+      "loss": 3.2977,
+      "step": 1673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008566328511599864,
+      "loss": 3.3425,
+      "step": 1674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008564497665626468,
+      "loss": 3.293,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008562665847293695,
+      "loss": 3.2654,
+      "step": 1676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008560833057101246,
+      "loss": 3.3123,
+      "step": 1677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008558999295549088,
+      "loss": 3.3273,
+      "step": 1678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008557164563137454,
+      "loss": 3.3569,
+      "step": 1679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000855532886036684,
+      "loss": 3.2548,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553492187738005,
+      "loss": 3.1176,
+      "step": 1681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008551654545751981,
+      "loss": 3.3414,
+      "step": 1682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549815934910052,
+      "loss": 3.2832,
+      "step": 1683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008547976355713776,
+      "loss": 3.3259,
+      "step": 1684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008546135808664972,
+      "loss": 3.3804,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008544294294265722,
+      "loss": 3.4757,
+      "step": 1686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008542451813018372,
+      "loss": 3.2337,
+      "step": 1687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008540608365425532,
+      "loss": 3.3022,
+      "step": 1688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000853876395199008,
+      "loss": 3.3891,
+      "step": 1689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000853691857321515,
+      "loss": 3.4109,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008535072229604144,
+      "loss": 3.2647,
+      "step": 1691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008533224921660725,
+      "loss": 3.3116,
+      "step": 1692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008531376649888822,
+      "loss": 3.3762,
+      "step": 1693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529527414792623,
+      "loss": 3.2327,
+      "step": 1694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008527677216876584,
+      "loss": 3.2331,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008525826056645418,
+      "loss": 3.405,
+      "step": 1696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008523973934604104,
+      "loss": 3.2509,
+      "step": 1697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008522120851257881,
+      "loss": 3.354,
+      "step": 1698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008520266807112252,
+      "loss": 3.3016,
+      "step": 1699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008518411802672984,
+      "loss": 3.3542,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008516555838446101,
+      "loss": 3.3469,
+      "step": 1701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008514698914937894,
+      "loss": 3.3224,
+      "step": 1702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008512841032654911,
+      "loss": 3.3695,
+      "step": 1703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510982192103965,
+      "loss": 3.353,
+      "step": 1704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008509122393792129,
+      "loss": 3.3689,
+      "step": 1705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008507261638226735,
+      "loss": 3.3009,
+      "step": 1706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008505399925915382,
+      "loss": 3.2994,
+      "step": 1707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008503537257365925,
+      "loss": 3.2778,
+      "step": 1708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008501673633086482,
+      "loss": 3.4669,
+      "step": 1709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849980905358543,
+      "loss": 3.4403,
+      "step": 1710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849794351937141,
+      "loss": 3.3289,
+      "step": 1711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008496077030953318,
+      "loss": 3.3796,
+      "step": 1712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008494209588840313,
+      "loss": 3.3422,
+      "step": 1713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849234119354182,
+      "loss": 3.337,
+      "step": 1714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008490471845567513,
+      "loss": 3.2886,
+      "step": 1715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008488601545427333,
+      "loss": 3.2523,
+      "step": 1716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008486730293631482,
+      "loss": 3.2376,
+      "step": 1717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484858090690415,
+      "loss": 3.3598,
+      "step": 1718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000848298493711485,
+      "loss": 3.3352,
+      "step": 1719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008481110833415769,
+      "loss": 3.316,
+      "step": 1720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008479235780104405,
+      "loss": 3.3215,
+      "step": 1721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008477359777692255,
+      "loss": 3.2302,
+      "step": 1722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008475482826691075,
+      "loss": 3.299,
+      "step": 1723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008473604927612874,
+      "loss": 3.1687,
+      "step": 1724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000847172608096993,
+      "loss": 3.2639,
+      "step": 1725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846984628727477,
+      "loss": 3.4036,
+      "step": 1726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008467965547040184,
+      "loss": 3.2366,
+      "step": 1727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008466083860779219,
+      "loss": 3.2039,
+      "step": 1728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008464201229005182,
+      "loss": 3.3338,
+      "step": 1729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008462317652231631,
+      "loss": 3.3872,
+      "step": 1730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008460433130972392,
+      "loss": 3.2931,
+      "step": 1731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008458547665741542,
+      "loss": 3.4087,
+      "step": 1732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008456661257053418,
+      "loss": 3.3458,
+      "step": 1733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000845477390542261,
+      "loss": 3.3708,
+      "step": 1734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008452885611363974,
+      "loss": 3.3202,
+      "step": 1735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008450996375392613,
+      "loss": 3.3844,
+      "step": 1736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008449106198023892,
+      "loss": 3.1221,
+      "step": 1737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008447215079773438,
+      "loss": 3.1685,
+      "step": 1738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008445323021157123,
+      "loss": 3.2708,
+      "step": 1739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008443430022691085,
+      "loss": 3.3997,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008441536084891714,
+      "loss": 3.3485,
+      "step": 1741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008439641208275657,
+      "loss": 3.5026,
+      "step": 1742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008437745393359817,
+      "loss": 3.4389,
+      "step": 1743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008435848640661356,
+      "loss": 3.2722,
+      "step": 1744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008433950950697686,
+      "loss": 3.2661,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000843205232398648,
+      "loss": 3.4215,
+      "step": 1746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008430152761045664,
+      "loss": 3.345,
+      "step": 1747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008428252262393419,
+      "loss": 3.3567,
+      "step": 1748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008426350828548182,
+      "loss": 3.3495,
+      "step": 1749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008424448460028647,
+      "loss": 3.244,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008422545157353759,
+      "loss": 3.373,
+      "step": 1751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008420640921042721,
+      "loss": 3.2599,
+      "step": 1752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841873575161499,
+      "loss": 3.3489,
+      "step": 1753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416829649590277,
+      "loss": 3.5301,
+      "step": 1754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008414922615488546,
+      "loss": 3.2543,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841301464983002,
+      "loss": 3.4799,
+      "step": 1756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008411105753135169,
+      "loss": 3.2386,
+      "step": 1757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409195925924725,
+      "loss": 3.3996,
+      "step": 1758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008407285168719667,
+      "loss": 3.2367,
+      "step": 1759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008405373482041232,
+      "loss": 3.2706,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008403460866410908,
+      "loss": 3.2546,
+      "step": 1761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008401547322350438,
+      "loss": 3.3229,
+      "step": 1762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008399632850381817,
+      "loss": 3.3638,
+      "step": 1763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008397717451027296,
+      "loss": 3.3542,
+      "step": 1764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008395801124809374,
+      "loss": 3.2764,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008393883872250807,
+      "loss": 3.3415,
+      "step": 1766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008391965693874602,
+      "loss": 3.2087,
+      "step": 1767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000839004659020402,
+      "loss": 3.3824,
+      "step": 1768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008388126561762572,
+      "loss": 3.2754,
+      "step": 1769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008386205609074024,
+      "loss": 3.3948,
+      "step": 1770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008384283732662391,
+      "loss": 3.3137,
+      "step": 1771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008382360933051943,
+      "loss": 3.2394,
+      "step": 1772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083804372107672,
+      "loss": 3.3858,
+      "step": 1773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008378512566332934,
+      "loss": 3.2368,
+      "step": 1774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008376587000274169,
+      "loss": 3.2317,
+      "step": 1775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008374660513116181,
+      "loss": 3.37,
+      "step": 1776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008372733105384496,
+      "loss": 3.3567,
+      "step": 1777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837080477760489,
+      "loss": 3.4351,
+      "step": 1778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008368875530303394,
+      "loss": 3.2047,
+      "step": 1779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008366945364006287,
+      "loss": 3.3192,
+      "step": 1780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008365014279240096,
+      "loss": 3.1908,
+      "step": 1781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008363082276531607,
+      "loss": 3.3268,
+      "step": 1782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008361149356407845,
+      "loss": 3.3165,
+      "step": 1783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008359215519396096,
+      "loss": 3.337,
+      "step": 1784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000835728076602389,
+      "loss": 3.2683,
+      "step": 1785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355345096819008,
+      "loss": 3.1733,
+      "step": 1786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008353408512309482,
+      "loss": 3.3719,
+      "step": 1787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008351471013023591,
+      "loss": 3.4096,
+      "step": 1788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008349532599489869,
+      "loss": 3.3843,
+      "step": 1789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008347593272237092,
+      "loss": 3.3216,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008345653031794292,
+      "loss": 3.4051,
+      "step": 1791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008343711878690746,
+      "loss": 3.4066,
+      "step": 1792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341769813455979,
+      "loss": 3.3974,
+      "step": 1793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008339826836619771,
+      "loss": 3.1576,
+      "step": 1794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008337882948712146,
+      "loss": 3.2871,
+      "step": 1795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008335938150263373,
+      "loss": 3.3098,
+      "step": 1796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000833399244180398,
+      "loss": 3.1953,
+      "step": 1797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008332045823864733,
+      "loss": 3.2946,
+      "step": 1798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000833009829697665,
+      "loss": 3.3054,
+      "step": 1799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008328149861670997,
+      "loss": 3.3388,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000832620051847929,
+      "loss": 3.4077,
+      "step": 1801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008324250267933285,
+      "loss": 3.3008,
+      "step": 1802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008322299110564997,
+      "loss": 3.2172,
+      "step": 1803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000832034704690668,
+      "loss": 3.2018,
+      "step": 1804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008318394077490835,
+      "loss": 3.4325,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008316440202850216,
+      "loss": 3.23,
+      "step": 1806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314485423517819,
+      "loss": 3.4174,
+      "step": 1807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008312529740026888,
+      "loss": 3.4263,
+      "step": 1808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008310573152910916,
+      "loss": 3.4368,
+      "step": 1809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008308615662703638,
+      "loss": 3.2422,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008306657269939037,
+      "loss": 3.243,
+      "step": 1811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008304697975151347,
+      "loss": 3.4144,
+      "step": 1812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008302737778875041,
+      "loss": 3.2929,
+      "step": 1813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008300776681644841,
+      "loss": 3.2921,
+      "step": 1814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008298814683995715,
+      "loss": 3.2962,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008296851786462877,
+      "loss": 3.3011,
+      "step": 1816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294887989581786,
+      "loss": 3.3418,
+      "step": 1817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008292923293888145,
+      "loss": 3.3993,
+      "step": 1818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008290957699917905,
+      "loss": 3.3173,
+      "step": 1819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000828899120820726,
+      "loss": 3.3227,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008287023819292648,
+      "loss": 3.5335,
+      "step": 1821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008285055533710752,
+      "loss": 3.2822,
+      "step": 1822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008283086351998505,
+      "loss": 3.3104,
+      "step": 1823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008281116274693078,
+      "loss": 3.3159,
+      "step": 1824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008279145302331888,
+      "loss": 3.2211,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008277173435452597,
+      "loss": 3.2887,
+      "step": 1826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008275200674593108,
+      "loss": 3.2633,
+      "step": 1827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008273227020291574,
+      "loss": 3.3126,
+      "step": 1828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271252473086388,
+      "loss": 3.2822,
+      "step": 1829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008269277033516184,
+      "loss": 3.2569,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008267300702119845,
+      "loss": 3.2969,
+      "step": 1831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008265323479436493,
+      "loss": 3.4729,
+      "step": 1832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008263345366005494,
+      "loss": 3.4062,
+      "step": 1833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826136636236646,
+      "loss": 3.3166,
+      "step": 1834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008259386469059242,
+      "loss": 3.3158,
+      "step": 1835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008257405686623936,
+      "loss": 3.2941,
+      "step": 1836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008255424015600877,
+      "loss": 3.4495,
+      "step": 1837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008253441456530647,
+      "loss": 3.2866,
+      "step": 1838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008251458009954069,
+      "loss": 3.4028,
+      "step": 1839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008249473676412207,
+      "loss": 3.3452,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008247488456446366,
+      "loss": 3.3798,
+      "step": 1841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008245502350598096,
+      "loss": 3.4339,
+      "step": 1842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008243515359409185,
+      "loss": 3.2506,
+      "step": 1843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008241527483421665,
+      "loss": 3.3814,
+      "step": 1844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000823953872317781,
+      "loss": 3.3459,
+      "step": 1845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237549079220135,
+      "loss": 3.4041,
+      "step": 1846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008235558552091391,
+      "loss": 3.358,
+      "step": 1847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008233567142334576,
+      "loss": 3.382,
+      "step": 1848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008231574850492926,
+      "loss": 3.2565,
+      "step": 1849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008229581677109921,
+      "loss": 3.2941,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008227587622729275,
+      "loss": 3.4547,
+      "step": 1851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008225592687894949,
+      "loss": 3.4747,
+      "step": 1852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008223596873151142,
+      "loss": 3.2097,
+      "step": 1853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008221600179042288,
+      "loss": 3.1926,
+      "step": 1854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219602606113068,
+      "loss": 3.3898,
+      "step": 1855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00082176041549084,
+      "loss": 3.2946,
+      "step": 1856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215604825973442,
+      "loss": 3.2356,
+      "step": 1857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008213604619853591,
+      "loss": 3.2952,
+      "step": 1858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000821160353709448,
+      "loss": 3.2517,
+      "step": 1859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008209601578241989,
+      "loss": 3.4935,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000820759874384223,
+      "loss": 3.296,
+      "step": 1861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205595034441556,
+      "loss": 3.2877,
+      "step": 1862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000820359045058656,
+      "loss": 3.3523,
+      "step": 1863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008201584992824072,
+      "loss": 3.242,
+      "step": 1864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008199578661701161,
+      "loss": 3.3313,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008197571457765134,
+      "loss": 3.1903,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195563381563535,
+      "loss": 3.276,
+      "step": 1867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008193554433644149,
+      "loss": 3.3687,
+      "step": 1868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008191544614554996,
+      "loss": 3.3583,
+      "step": 1869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008189533924844335,
+      "loss": 3.2527,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818752236506066,
+      "loss": 3.1176,
+      "step": 1871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008185509935752708,
+      "loss": 3.3722,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008183496637469449,
+      "loss": 3.3747,
+      "step": 1873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008181482470760086,
+      "loss": 3.3236,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008179467436174071,
+      "loss": 3.3958,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000817745153426108,
+      "loss": 3.4579,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175434765571031,
+      "loss": 3.2303,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008173417130654081,
+      "loss": 3.3073,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000817139863006062,
+      "loss": 3.2312,
+      "step": 1879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008169379264341273,
+      "loss": 3.4417,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167359034046906,
+      "loss": 3.3127,
+      "step": 1881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008165337939728615,
+      "loss": 3.282,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008163315981937734,
+      "loss": 3.3903,
+      "step": 1883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008161293161225834,
+      "loss": 3.299,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000815926947814472,
+      "loss": 3.3398,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008157244933246433,
+      "loss": 3.2728,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008155219527083248,
+      "loss": 3.2947,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008153193260207677,
+      "loss": 3.4153,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008151166133172464,
+      "loss": 3.2443,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008149138146530588,
+      "loss": 3.134,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008147109300835268,
+      "loss": 3.3258,
+      "step": 1891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008145079596639947,
+      "loss": 3.373,
+      "step": 1892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008143049034498311,
+      "loss": 3.5346,
+      "step": 1893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008141017614964278,
+      "loss": 3.4896,
+      "step": 1894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008138985338591998,
+      "loss": 3.3701,
+      "step": 1895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008136952205935854,
+      "loss": 3.1809,
+      "step": 1896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134918217550467,
+      "loss": 3.1437,
+      "step": 1897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008132883373990688,
+      "loss": 3.2098,
+      "step": 1898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008130847675811599,
+      "loss": 3.2158,
+      "step": 1899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008128811123568522,
+      "loss": 3.3247,
+      "step": 1900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126773717817007,
+      "loss": 3.3004,
+      "step": 1901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008124735459112837,
+      "loss": 3.4167,
+      "step": 1902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000812269634801203,
+      "loss": 3.3174,
+      "step": 1903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008120656385070837,
+      "loss": 3.2891,
+      "step": 1904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118615570845734,
+      "loss": 3.3537,
+      "step": 1905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008116573905893439,
+      "loss": 3.2605,
+      "step": 1906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008114531390770896,
+      "loss": 3.2651,
+      "step": 1907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008112488026035284,
+      "loss": 3.2966,
+      "step": 1908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000811044381224401,
+      "loss": 3.3765,
+      "step": 1909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000810839874995472,
+      "loss": 3.2007,
+      "step": 1910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008106352839725282,
+      "loss": 3.3245,
+      "step": 1911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008104306082113801,
+      "loss": 3.3985,
+      "step": 1912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008102258477678613,
+      "loss": 3.2951,
+      "step": 1913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008100210026978283,
+      "loss": 3.3456,
+      "step": 1914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008098160730571609,
+      "loss": 3.0625,
+      "step": 1915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008096110589017617,
+      "loss": 3.3031,
+      "step": 1916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008094059602875567,
+      "loss": 3.3961,
+      "step": 1917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008092007772704948,
+      "loss": 3.2407,
+      "step": 1918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008089955099065476,
+      "loss": 3.2703,
+      "step": 1919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008087901582517101,
+      "loss": 3.2769,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008085847223620002,
+      "loss": 3.1523,
+      "step": 1921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008083792022934589,
+      "loss": 3.3082,
+      "step": 1922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008081735981021499,
+      "loss": 3.2886,
+      "step": 1923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008079679098441599,
+      "loss": 3.1875,
+      "step": 1924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077621375755987,
+      "loss": 3.2664,
+      "step": 1925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807556281352599,
+      "loss": 3.1982,
+      "step": 1926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008073503412313159,
+      "loss": 3.2889,
+      "step": 1927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008071443172679285,
+      "loss": 3.3018,
+      "step": 1928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069382095186375,
+      "loss": 3.2903,
+      "step": 1929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008067320180396672,
+      "loss": 3.3355,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008065257428872647,
+      "loss": 3.2915,
+      "step": 1931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008063193841176999,
+      "loss": 3.4211,
+      "step": 1932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000806112941787265,
+      "loss": 3.3267,
+      "step": 1933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008059064159522757,
+      "loss": 3.261,
+      "step": 1934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056998066690702,
+      "loss": 3.2882,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008054931139940092,
+      "loss": 3.3267,
+      "step": 1936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008052863379834767,
+      "loss": 3.2724,
+      "step": 1937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008050794786938791,
+      "loss": 3.354,
+      "step": 1938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008048725361816454,
+      "loss": 3.1559,
+      "step": 1939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046655105032273,
+      "loss": 3.379,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008044584017150997,
+      "loss": 3.1259,
+      "step": 1941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008042512098737597,
+      "loss": 3.333,
+      "step": 1942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008040439350357272,
+      "loss": 3.3844,
+      "step": 1943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008038365772575444,
+      "loss": 3.2899,
+      "step": 1944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008036291365957768,
+      "loss": 3.2145,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803421613107012,
+      "loss": 3.2227,
+      "step": 1946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008032140068478603,
+      "loss": 3.1662,
+      "step": 1947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008030063178749548,
+      "loss": 3.3444,
+      "step": 1948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008027985462449505,
+      "loss": 3.1388,
+      "step": 1949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000802590692014526,
+      "loss": 3.2995,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008023827552403815,
+      "loss": 3.3751,
+      "step": 1951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008021747359792403,
+      "loss": 3.3041,
+      "step": 1952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008019666342878479,
+      "loss": 3.3719,
+      "step": 1953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008017584502229723,
+      "loss": 3.2941,
+      "step": 1954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008015501838414038,
+      "loss": 3.2941,
+      "step": 1955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008013418351999561,
+      "loss": 3.3175,
+      "step": 1956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008011334043554639,
+      "loss": 3.2465,
+      "step": 1957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008009248913647855,
+      "loss": 3.2828,
+      "step": 1958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008007162962848009,
+      "loss": 3.2684,
+      "step": 1959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008005076191724128,
+      "loss": 3.4251,
+      "step": 1960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008002988600845464,
+      "loss": 3.3131,
+      "step": 1961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008000900190781489,
+      "loss": 3.2152,
+      "step": 1962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007998810962101902,
+      "loss": 3.3632,
+      "step": 1963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007996720915376623,
+      "loss": 3.3837,
+      "step": 1964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007994630051175795,
+      "loss": 3.2259,
+      "step": 1965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007992538370069787,
+      "loss": 3.2372,
+      "step": 1966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007990445872629188,
+      "loss": 3.2951,
+      "step": 1967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007988352559424808,
+      "loss": 3.3707,
+      "step": 1968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007986258431027684,
+      "loss": 3.302,
+      "step": 1969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007984163488009076,
+      "loss": 3.2188,
+      "step": 1970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007982067730940457,
+      "loss": 3.1943,
+      "step": 1971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007979971160393534,
+      "loss": 3.2798,
+      "step": 1972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000797787377694023,
+      "loss": 3.3238,
+      "step": 1973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007975775581152687,
+      "loss": 3.4319,
+      "step": 1974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007973676573603274,
+      "loss": 3.3202,
+      "step": 1975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000797157675486458,
+      "loss": 3.3417,
+      "step": 1976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007969476125509414,
+      "loss": 3.4365,
+      "step": 1977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007967374686110807,
+      "loss": 3.3685,
+      "step": 1978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796527243724201,
+      "loss": 3.2208,
+      "step": 1979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007963169379476495,
+      "loss": 3.1208,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007961065513387956,
+      "loss": 3.2225,
+      "step": 1981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007958960839550307,
+      "loss": 3.2271,
+      "step": 1982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007956855358537682,
+      "loss": 3.2955,
+      "step": 1983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007954749070924434,
+      "loss": 3.2951,
+      "step": 1984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952641977285137,
+      "loss": 3.2528,
+      "step": 1985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007950534078194589,
+      "loss": 3.3092,
+      "step": 1986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007948425374227799,
+      "loss": 3.1577,
+      "step": 1987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007946315865960004,
+      "loss": 3.4106,
+      "step": 1988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007944205553966653,
+      "loss": 3.1926,
+      "step": 1989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007942094438823421,
+      "loss": 3.2895,
+      "step": 1990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007939982521106198,
+      "loss": 3.2897,
+      "step": 1991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007937869801391095,
+      "loss": 3.3033,
+      "step": 1992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000793575628025444,
+      "loss": 3.244,
+      "step": 1993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007933641958272782,
+      "loss": 3.2956,
+      "step": 1994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007931526836022884,
+      "loss": 3.3539,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792941091408173,
+      "loss": 3.2385,
+      "step": 1996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007927294193026529,
+      "loss": 3.2053,
+      "step": 1997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007925176673434693,
+      "loss": 3.3122,
+      "step": 1998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007923058355883864,
+      "loss": 3.3112,
+      "step": 1999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007920939240951899,
+      "loss": 3.3024,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007918819329216871,
+      "loss": 3.2744,
+      "step": 2001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007916698621257068,
+      "loss": 3.4105,
+      "step": 2002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007914577117651001,
+      "loss": 3.3182,
+      "step": 2003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007912454818977393,
+      "loss": 3.424,
+      "step": 2004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007910331725815185,
+      "loss": 3.1176,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007908207838743541,
+      "loss": 3.2679,
+      "step": 2006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007906083158341831,
+      "loss": 3.3404,
+      "step": 2007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903957685189649,
+      "loss": 3.2217,
+      "step": 2008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007901831419866799,
+      "loss": 3.354,
+      "step": 2009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000789970436295331,
+      "loss": 3.1425,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007897576515029421,
+      "loss": 3.3965,
+      "step": 2011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007895447876675583,
+      "loss": 3.2766,
+      "step": 2012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007893318448472474,
+      "loss": 3.4282,
+      "step": 2013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891188231000977,
+      "loss": 3.2903,
+      "step": 2014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007889057224842193,
+      "loss": 3.3053,
+      "step": 2015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007886925430577443,
+      "loss": 3.4105,
+      "step": 2016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007884792848788256,
+      "loss": 3.2865,
+      "step": 2017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000788265948005638,
+      "loss": 3.3477,
+      "step": 2018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007880525324963776,
+      "loss": 3.1795,
+      "step": 2019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878390384092622,
+      "loss": 3.2196,
+      "step": 2020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007876254658025307,
+      "loss": 3.2707,
+      "step": 2021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007874118147344436,
+      "loss": 3.2618,
+      "step": 2022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007871980852632829,
+      "loss": 3.3785,
+      "step": 2023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007869842774473517,
+      "loss": 3.3158,
+      "step": 2024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007867703913449747,
+      "loss": 3.3477,
+      "step": 2025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007865564270144978,
+      "loss": 3.04,
+      "step": 2026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007863423845142886,
+      "loss": 3.3698,
+      "step": 2027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007861282639027354,
+      "loss": 3.4027,
+      "step": 2028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007859140652382486,
+      "loss": 3.2539,
+      "step": 2029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007856997885792589,
+      "loss": 3.2992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007854854339842195,
+      "loss": 3.2591,
+      "step": 2031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007852710015116036,
+      "loss": 3.4617,
+      "step": 2032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007850564912199066,
+      "loss": 3.2176,
+      "step": 2033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007848419031676445,
+      "loss": 3.2108,
+      "step": 2034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007846272374133549,
+      "loss": 3.2394,
+      "step": 2035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007844124940155967,
+      "loss": 3.2863,
+      "step": 2036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007841976730329494,
+      "loss": 3.533,
+      "step": 2037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007839827745240143,
+      "loss": 3.3082,
+      "step": 2038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007837677985474132,
+      "loss": 3.4071,
+      "step": 2039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835527451617899,
+      "loss": 3.2687,
+      "step": 2040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007833376144258085,
+      "loss": 3.2481,
+      "step": 2041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007831224063981543,
+      "loss": 3.1558,
+      "step": 2042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007829071211375345,
+      "loss": 3.3342,
+      "step": 2043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007826917587026763,
+      "loss": 3.3242,
+      "step": 2044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007824763191523285,
+      "loss": 3.1648,
+      "step": 2045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000782260802545261,
+      "loss": 3.3914,
+      "step": 2046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007820452089402645,
+      "loss": 3.2113,
+      "step": 2047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818295383961507,
+      "loss": 3.2991,
+      "step": 2048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007816137909717526,
+      "loss": 3.3813,
+      "step": 2049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007813979667259238,
+      "loss": 3.4531,
+      "step": 2050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007811820657175389,
+      "loss": 3.1524,
+      "step": 2051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007809660880054936,
+      "loss": 3.291,
+      "step": 2052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007807500336487046,
+      "loss": 3.314,
+      "step": 2053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007805339027061093,
+      "loss": 3.3191,
+      "step": 2054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000780317695236666,
+      "loss": 3.2411,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007801014112993539,
+      "loss": 3.3343,
+      "step": 2056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007798850509531732,
+      "loss": 3.3003,
+      "step": 2057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779668614257145,
+      "loss": 3.3541,
+      "step": 2058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007794521012703106,
+      "loss": 3.1875,
+      "step": 2059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779235512051733,
+      "loss": 3.3534,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007790188466604955,
+      "loss": 3.3571,
+      "step": 2061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007788021051557022,
+      "loss": 3.3341,
+      "step": 2062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007785852875964778,
+      "loss": 3.1624,
+      "step": 2063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007783683940419684,
+      "loss": 3.3886,
+      "step": 2064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007781514245513401,
+      "loss": 3.2168,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007779343791837803,
+      "loss": 3.2119,
+      "step": 2066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007777172579984963,
+      "loss": 3.2798,
+      "step": 2067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007775000610547169,
+      "loss": 3.3222,
+      "step": 2068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007772827884116915,
+      "loss": 3.3426,
+      "step": 2069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007770654401286896,
+      "loss": 3.177,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007768480162650016,
+      "loss": 3.2953,
+      "step": 2071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007766305168799385,
+      "loss": 3.2255,
+      "step": 2072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007764129420328326,
+      "loss": 3.2328,
+      "step": 2073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007761952917830353,
+      "loss": 3.3376,
+      "step": 2074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007759775661899198,
+      "loss": 3.3519,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007757597653128796,
+      "loss": 3.3258,
+      "step": 2076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007755418892113282,
+      "loss": 3.2734,
+      "step": 2077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007753239379447005,
+      "loss": 3.266,
+      "step": 2078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007751059115724511,
+      "loss": 3.3698,
+      "step": 2079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007748878101540556,
+      "loss": 3.4313,
+      "step": 2080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007746696337490098,
+      "loss": 3.2772,
+      "step": 2081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007744513824168301,
+      "loss": 3.2807,
+      "step": 2082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007742330562170532,
+      "loss": 3.3006,
+      "step": 2083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007740146552092364,
+      "loss": 3.2362,
+      "step": 2084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007737961794529573,
+      "loss": 3.4251,
+      "step": 2085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007735776290078138,
+      "loss": 3.3316,
+      "step": 2086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007733590039334242,
+      "loss": 3.3105,
+      "step": 2087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007731403042894276,
+      "loss": 3.2811,
+      "step": 2088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007729215301354828,
+      "loss": 3.3068,
+      "step": 2089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007727026815312694,
+      "loss": 3.4135,
+      "step": 2090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007724837585364868,
+      "loss": 3.2548,
+      "step": 2091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007722647612108554,
+      "loss": 3.4309,
+      "step": 2092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000772045689614115,
+      "loss": 3.1607,
+      "step": 2093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007718265438060266,
+      "loss": 3.2631,
+      "step": 2094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007716073238463709,
+      "loss": 3.1701,
+      "step": 2095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007713880297949488,
+      "loss": 3.3182,
+      "step": 2096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007711686617115814,
+      "loss": 3.4139,
+      "step": 2097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007709492196561103,
+      "loss": 3.2526,
+      "step": 2098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007707297036883972,
+      "loss": 3.4242,
+      "step": 2099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007705101138683237,
+      "loss": 3.2089,
+      "step": 2100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007702904502557919,
+      "loss": 3.3798,
+      "step": 2101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007700707129107234,
+      "loss": 3.3468,
+      "step": 2102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000769850901893061,
+      "loss": 3.2908,
+      "step": 2103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007696310172627664,
+      "loss": 3.352,
+      "step": 2104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000769411059079822,
+      "loss": 3.1928,
+      "step": 2105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007691910274042306,
+      "loss": 3.3302,
+      "step": 2106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007689709222960144,
+      "loss": 3.2059,
+      "step": 2107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007687507438152157,
+      "loss": 3.2943,
+      "step": 2108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007685304920218972,
+      "loss": 3.2335,
+      "step": 2109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007683101669761412,
+      "loss": 3.2784,
+      "step": 2110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007680897687380504,
+      "loss": 3.4007,
+      "step": 2111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007678692973677472,
+      "loss": 3.3147,
+      "step": 2112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767648752925374,
+      "loss": 3.165,
+      "step": 2113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767428135471093,
+      "loss": 3.4647,
+      "step": 2114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007672074450650863,
+      "loss": 3.2766,
+      "step": 2115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007669866817675564,
+      "loss": 3.5702,
+      "step": 2116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007667658456387251,
+      "loss": 3.3298,
+      "step": 2117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007665449367388346,
+      "loss": 3.3949,
+      "step": 2118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007663239551281461,
+      "loss": 3.2642,
+      "step": 2119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007661029008669417,
+      "loss": 3.3703,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007658817740155224,
+      "loss": 3.2931,
+      "step": 2121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007656605746342096,
+      "loss": 3.2367,
+      "step": 2122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007654393027833443,
+      "loss": 3.2409,
+      "step": 2123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007652179585232871,
+      "loss": 3.3024,
+      "step": 2124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007649965419144187,
+      "loss": 3.4153,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007647750530171393,
+      "loss": 3.3285,
+      "step": 2126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007645534918918687,
+      "loss": 3.2449,
+      "step": 2127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007643318585990465,
+      "loss": 3.3336,
+      "step": 2128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007641101531991324,
+      "loss": 3.2487,
+      "step": 2129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007638883757526052,
+      "loss": 3.2595,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007636665263199637,
+      "loss": 3.1952,
+      "step": 2131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007634446049617261,
+      "loss": 3.3774,
+      "step": 2132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632226117384302,
+      "loss": 3.3321,
+      "step": 2133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007630005467106338,
+      "loss": 3.3728,
+      "step": 2134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762778409938914,
+      "loss": 3.2154,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007625562014838673,
+      "loss": 3.1877,
+      "step": 2136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007623339214061101,
+      "loss": 3.2775,
+      "step": 2137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007621115697662782,
+      "loss": 3.299,
+      "step": 2138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007618891466250267,
+      "loss": 3.2165,
+      "step": 2139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007616666520430304,
+      "loss": 3.3403,
+      "step": 2140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007614440860809841,
+      "loss": 3.3165,
+      "step": 2141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000761221448799601,
+      "loss": 3.2329,
+      "step": 2142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007609987402596146,
+      "loss": 3.3127,
+      "step": 2143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007607759605217777,
+      "loss": 3.3359,
+      "step": 2144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000760553109646862,
+      "loss": 3.368,
+      "step": 2145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007603301876956592,
+      "loss": 3.3018,
+      "step": 2146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601071947289803,
+      "loss": 3.2157,
+      "step": 2147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007598841308076553,
+      "loss": 3.2489,
+      "step": 2148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007596609959925342,
+      "loss": 3.3598,
+      "step": 2149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007594377903444856,
+      "loss": 3.2148,
+      "step": 2150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007592145139243977,
+      "loss": 3.2342,
+      "step": 2151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007589911667931785,
+      "loss": 3.4123,
+      "step": 2152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007587677490117546,
+      "loss": 3.1564,
+      "step": 2153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007585442606410721,
+      "loss": 3.2405,
+      "step": 2154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007583207017420966,
+      "loss": 3.3394,
+      "step": 2155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007580970723758123,
+      "loss": 3.1147,
+      "step": 2156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007578733726032234,
+      "loss": 3.2013,
+      "step": 2157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007576496024853532,
+      "loss": 3.3885,
+      "step": 2158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007574257620832435,
+      "loss": 3.0919,
+      "step": 2159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007572018514579557,
+      "loss": 3.2372,
+      "step": 2160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007569778706705708,
+      "loss": 3.1643,
+      "step": 2161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007567538197821882,
+      "loss": 3.3225,
+      "step": 2162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000756529698853927,
+      "loss": 3.2682,
+      "step": 2163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000756305507946925,
+      "loss": 3.3258,
+      "step": 2164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000756081247122339,
+      "loss": 3.156,
+      "step": 2165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007558569164413456,
+      "loss": 3.3805,
+      "step": 2166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007556325159651399,
+      "loss": 3.2488,
+      "step": 2167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007554080457549355,
+      "loss": 3.2599,
+      "step": 2168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007551835058719663,
+      "loss": 3.3745,
+      "step": 2169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007549588963774843,
+      "loss": 3.3421,
+      "step": 2170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007547342173327606,
+      "loss": 3.2082,
+      "step": 2171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007545094687990857,
+      "loss": 3.3218,
+      "step": 2172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542846508377687,
+      "loss": 3.3511,
+      "step": 2173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007540597635101371,
+      "loss": 3.1408,
+      "step": 2174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000753834806877539,
+      "loss": 3.1676,
+      "step": 2175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007536097810013395,
+      "loss": 3.3126,
+      "step": 2176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007533846859429234,
+      "loss": 3.0507,
+      "step": 2177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000753159521763695,
+      "loss": 3.4192,
+      "step": 2178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007529342885250763,
+      "loss": 3.2173,
+      "step": 2179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007527089862885088,
+      "loss": 3.3216,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007524836151154528,
+      "loss": 3.2036,
+      "step": 2181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007522581750673871,
+      "loss": 3.1914,
+      "step": 2182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007520326662058099,
+      "loss": 3.2593,
+      "step": 2183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007518070885922374,
+      "loss": 3.2114,
+      "step": 2184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007515814422882051,
+      "loss": 3.2457,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007513557273552669,
+      "loss": 3.0602,
+      "step": 2186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511299438549957,
+      "loss": 3.3307,
+      "step": 2187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007509040918489829,
+      "loss": 3.3034,
+      "step": 2188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007506781713988389,
+      "loss": 3.1123,
+      "step": 2189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007504521825661923,
+      "loss": 3.0999,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007502261254126905,
+      "loss": 3.0838,
+      "step": 2191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075,
+      "loss": 3.3044,
+      "step": 2192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007497738063898053,
+      "loss": 3.3323,
+      "step": 2193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007495475446438098,
+      "loss": 3.3135,
+      "step": 2194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007493212148237356,
+      "loss": 3.2394,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000749094816991323,
+      "loss": 3.3311,
+      "step": 2196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007488683512083309,
+      "loss": 3.2943,
+      "step": 2197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007486418175365376,
+      "loss": 3.2907,
+      "step": 2198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007484152160377385,
+      "loss": 3.2355,
+      "step": 2199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007481885467737486,
+      "loss": 3.337,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007479618098064009,
+      "loss": 3.2714,
+      "step": 2201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007477350051975468,
+      "loss": 3.3108,
+      "step": 2202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007475081330090567,
+      "loss": 3.1877,
+      "step": 2203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007472811933028191,
+      "loss": 3.2094,
+      "step": 2204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007470541861407404,
+      "loss": 3.1647,
+      "step": 2205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007468271115847463,
+      "loss": 3.2426,
+      "step": 2206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465999696967804,
+      "loss": 3.2321,
+      "step": 2207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007463727605388044,
+      "loss": 3.2257,
+      "step": 2208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007461454841727992,
+      "loss": 3.3371,
+      "step": 2209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007459181406607633,
+      "loss": 3.2265,
+      "step": 2210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007456907300647137,
+      "loss": 3.2353,
+      "step": 2211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007454632524466858,
+      "loss": 3.2555,
+      "step": 2212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007452357078687332,
+      "loss": 3.2839,
+      "step": 2213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007450080963929277,
+      "loss": 3.1861,
+      "step": 2214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007447804180813596,
+      "loss": 3.2644,
+      "step": 2215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007445526729961372,
+      "loss": 3.3463,
+      "step": 2216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007443248611993872,
+      "loss": 3.3303,
+      "step": 2217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000744096982753254,
+      "loss": 3.3225,
+      "step": 2218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007438690377199011,
+      "loss": 3.2817,
+      "step": 2219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007436410261615092,
+      "loss": 3.2854,
+      "step": 2220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007434129481402777,
+      "loss": 3.4136,
+      "step": 2221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007431848037184242,
+      "loss": 3.2205,
+      "step": 2222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742956592958184,
+      "loss": 3.368,
+      "step": 2223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007427283159218107,
+      "loss": 3.1307,
+      "step": 2224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007424999726715763,
+      "loss": 3.377,
+      "step": 2225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007422715632697702,
+      "loss": 3.3145,
+      "step": 2226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007420430877787003,
+      "loss": 3.3853,
+      "step": 2227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007418145462606925,
+      "loss": 3.1315,
+      "step": 2228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007415859387780906,
+      "loss": 3.1234,
+      "step": 2229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007413572653932566,
+      "loss": 3.4594,
+      "step": 2230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007411285261685701,
+      "loss": 3.3256,
+      "step": 2231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740899721166429,
+      "loss": 3.301,
+      "step": 2232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740670850449249,
+      "loss": 3.3586,
+      "step": 2233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007404419140794634,
+      "loss": 3.2743,
+      "step": 2234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007402129121195245,
+      "loss": 3.221,
+      "step": 2235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007399838446319011,
+      "loss": 3.1987,
+      "step": 2236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007397547116790807,
+      "loss": 3.1278,
+      "step": 2237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007395255133235686,
+      "loss": 3.3863,
+      "step": 2238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007392962496278879,
+      "loss": 3.2034,
+      "step": 2239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007390669206545791,
+      "loss": 3.2001,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007388375264662011,
+      "loss": 3.3175,
+      "step": 2241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007386080671253305,
+      "loss": 3.1897,
+      "step": 2242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007383785426945611,
+      "loss": 3.0812,
+      "step": 2243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007381489532365051,
+      "loss": 3.3124,
+      "step": 2244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007379192988137924,
+      "loss": 3.0756,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007376895794890699,
+      "loss": 3.2923,
+      "step": 2246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374597953250034,
+      "loss": 3.3694,
+      "step": 2247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007372299463842753,
+      "loss": 3.2766,
+      "step": 2248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000737000032729586,
+      "loss": 3.2082,
+      "step": 2249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007367700544236541,
+      "loss": 3.3897,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365400115292151,
+      "loss": 3.151,
+      "step": 2251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007363099041090223,
+      "loss": 3.2438,
+      "step": 2252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000736079732225847,
+      "loss": 3.2622,
+      "step": 2253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007358494959424775,
+      "loss": 3.3762,
+      "step": 2254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356191953217202,
+      "loss": 3.244,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007353888304263987,
+      "loss": 3.2934,
+      "step": 2256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007351584013193543,
+      "loss": 3.2493,
+      "step": 2257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007349279080634456,
+      "loss": 3.2075,
+      "step": 2258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007346973507215491,
+      "loss": 3.1229,
+      "step": 2259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007344667293565582,
+      "loss": 3.4161,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007342360440313845,
+      "loss": 3.4066,
+      "step": 2261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007340052948089564,
+      "loss": 3.1915,
+      "step": 2262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00073377448175222,
+      "loss": 3.3072,
+      "step": 2263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007335436049241391,
+      "loss": 3.3976,
+      "step": 2264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000733312664387694,
+      "loss": 3.2397,
+      "step": 2265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007330816602058835,
+      "loss": 3.2002,
+      "step": 2266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007328505924417231,
+      "loss": 3.176,
+      "step": 2267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007326194611582457,
+      "loss": 3.3226,
+      "step": 2268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007323882664185016,
+      "loss": 3.2193,
+      "step": 2269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321570082855584,
+      "loss": 3.1747,
+      "step": 2270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007319256868225011,
+      "loss": 3.211,
+      "step": 2271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000731694302092432,
+      "loss": 3.3152,
+      "step": 2272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007314628541584703,
+      "loss": 3.3018,
+      "step": 2273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007312313430837528,
+      "loss": 3.2751,
+      "step": 2274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007309997689314333,
+      "loss": 3.3193,
+      "step": 2275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007307681317646831,
+      "loss": 3.2053,
+      "step": 2276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007305364316466902,
+      "loss": 3.313,
+      "step": 2277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007303046686406606,
+      "loss": 3.4025,
+      "step": 2278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007300728428098165,
+      "loss": 3.2414,
+      "step": 2279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007298409542173979,
+      "loss": 3.2211,
+      "step": 2280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007296090029266613,
+      "loss": 3.2886,
+      "step": 2281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293769890008813,
+      "loss": 3.3432,
+      "step": 2282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007291449125033485,
+      "loss": 3.3046,
+      "step": 2283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007289127734973713,
+      "loss": 3.28,
+      "step": 2284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007286805720462749,
+      "loss": 3.197,
+      "step": 2285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007284483082134013,
+      "loss": 3.1771,
+      "step": 2286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007282159820621101,
+      "loss": 3.2587,
+      "step": 2287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007279835936557773,
+      "loss": 3.1347,
+      "step": 2288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007277511430577961,
+      "loss": 3.2187,
+      "step": 2289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007275186303315769,
+      "loss": 3.3697,
+      "step": 2290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007272860555405468,
+      "loss": 3.3054,
+      "step": 2291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007270534187481498,
+      "loss": 3.2354,
+      "step": 2292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726820720017847,
+      "loss": 3.3119,
+      "step": 2293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007265879594131161,
+      "loss": 3.2453,
+      "step": 2294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007263551369974522,
+      "loss": 3.3854,
+      "step": 2295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007261222528343665,
+      "loss": 3.157,
+      "step": 2296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007258893069873878,
+      "loss": 3.1915,
+      "step": 2297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007256562995200614,
+      "loss": 3.2659,
+      "step": 2298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007254232304959491,
+      "loss": 3.1568,
+      "step": 2299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007251900999786303,
+      "loss": 3.2489,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007249569080317002,
+      "loss": 3.3377,
+      "step": 2301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007247236547187715,
+      "loss": 3.2611,
+      "step": 2302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007244903401034733,
+      "loss": 3.2788,
+      "step": 2303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007242569642494516,
+      "loss": 3.3779,
+      "step": 2304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007240235272203689,
+      "loss": 3.2913,
+      "step": 2305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007237900290799044,
+      "loss": 3.2563,
+      "step": 2306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007235564698917542,
+      "loss": 3.1967,
+      "step": 2307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007233228497196309,
+      "loss": 3.3739,
+      "step": 2308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230891686272636,
+      "loss": 3.3081,
+      "step": 2309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007228554266783985,
+      "loss": 3.1338,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007226216239367977,
+      "loss": 3.1505,
+      "step": 2311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007223877604662403,
+      "loss": 3.2499,
+      "step": 2312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007221538363305223,
+      "loss": 3.2739,
+      "step": 2313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007219198515934556,
+      "loss": 3.3302,
+      "step": 2314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216858063188687,
+      "loss": 3.4127,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007214517005706073,
+      "loss": 3.4279,
+      "step": 2316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007212175344125327,
+      "loss": 3.2479,
+      "step": 2317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007209833079085231,
+      "loss": 3.3085,
+      "step": 2318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007207490211224736,
+      "loss": 3.2179,
+      "step": 2319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007205146741182946,
+      "loss": 3.457,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007202802669599143,
+      "loss": 3.1284,
+      "step": 2321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007200457997112761,
+      "loss": 3.2329,
+      "step": 2322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007198112724363407,
+      "loss": 3.226,
+      "step": 2323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007195766851990846,
+      "loss": 3.2632,
+      "step": 2324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000719342038063501,
+      "loss": 3.2964,
+      "step": 2325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000719107331093599,
+      "loss": 3.2298,
+      "step": 2326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007188725643534047,
+      "loss": 3.1379,
+      "step": 2327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00071863773790696,
+      "loss": 3.1072,
+      "step": 2328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000718402851818323,
+      "loss": 3.2268,
+      "step": 2329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007181679061515684,
+      "loss": 3.2713,
+      "step": 2330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007179329009707872,
+      "loss": 3.2938,
+      "step": 2331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007176978363400864,
+      "loss": 3.3029,
+      "step": 2332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007174627123235891,
+      "loss": 3.2513,
+      "step": 2333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007172275289854349,
+      "loss": 3.4377,
+      "step": 2334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007169922863897795,
+      "loss": 3.3368,
+      "step": 2335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007167569846007946,
+      "loss": 3.1661,
+      "step": 2336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007165216236826684,
+      "loss": 3.3538,
+      "step": 2337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007162862036996048,
+      "loss": 3.2639,
+      "step": 2338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007160507247158243,
+      "loss": 3.2069,
+      "step": 2339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007158151867955629,
+      "loss": 3.2416,
+      "step": 2340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000715579590003073,
+      "loss": 3.1853,
+      "step": 2341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007153439344026233,
+      "loss": 3.3641,
+      "step": 2342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007151082200584982,
+      "loss": 3.173,
+      "step": 2343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007148724470349981,
+      "loss": 3.2297,
+      "step": 2344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007146366153964399,
+      "loss": 3.1906,
+      "step": 2345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007144007252071555,
+      "loss": 3.227,
+      "step": 2346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007141647765314938,
+      "loss": 3.3371,
+      "step": 2347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007139287694338192,
+      "loss": 3.2033,
+      "step": 2348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000713692703978512,
+      "loss": 3.2108,
+      "step": 2349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007134565802299686,
+      "loss": 3.309,
+      "step": 2350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007132203982526012,
+      "loss": 3.2436,
+      "step": 2351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007129841581108376,
+      "loss": 3.4498,
+      "step": 2352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007127478598691222,
+      "loss": 3.1987,
+      "step": 2353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007125115035919147,
+      "loss": 3.2257,
+      "step": 2354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007122750893436904,
+      "loss": 3.2829,
+      "step": 2355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007120386171889409,
+      "loss": 3.4217,
+      "step": 2356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007118020871921736,
+      "loss": 3.211,
+      "step": 2357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007115654994179115,
+      "loss": 3.4212,
+      "step": 2358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007113288539306932,
+      "loss": 3.2172,
+      "step": 2359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007110921507950733,
+      "loss": 3.2645,
+      "step": 2360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007108553900756222,
+      "loss": 3.2434,
+      "step": 2361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007106185718369258,
+      "loss": 3.3187,
+      "step": 2362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007103816961435858,
+      "loss": 3.3843,
+      "step": 2363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007101447630602192,
+      "loss": 3.2455,
+      "step": 2364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007099077726514592,
+      "loss": 3.2321,
+      "step": 2365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007096707249819545,
+      "loss": 3.2395,
+      "step": 2366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007094336201163692,
+      "loss": 3.1378,
+      "step": 2367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007091964581193833,
+      "loss": 3.2701,
+      "step": 2368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007089592390556919,
+      "loss": 3.2501,
+      "step": 2369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007087219629900066,
+      "loss": 3.1448,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007084846299870532,
+      "loss": 3.2279,
+      "step": 2371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007082472401115742,
+      "loss": 3.2373,
+      "step": 2372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007080097934283274,
+      "loss": 3.1273,
+      "step": 2373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007077722900020853,
+      "loss": 3.2881,
+      "step": 2374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007075347298976369,
+      "loss": 3.2207,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007072971131797861,
+      "loss": 3.4982,
+      "step": 2376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007070594399133524,
+      "loss": 3.1521,
+      "step": 2377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007068217101631705,
+      "loss": 3.2129,
+      "step": 2378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007065839239940911,
+      "loss": 3.2518,
+      "step": 2379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007063460814709795,
+      "loss": 3.3066,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007061081826587169,
+      "loss": 3.2251,
+      "step": 2381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007058702276221998,
+      "loss": 3.2024,
+      "step": 2382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00070563221642634,
+      "loss": 3.3269,
+      "step": 2383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007053941491360642,
+      "loss": 3.1851,
+      "step": 2384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007051560258163152,
+      "loss": 3.1468,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007049178465320506,
+      "loss": 3.2766,
+      "step": 2386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007046796113482431,
+      "loss": 3.2584,
+      "step": 2387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007044413203298812,
+      "loss": 3.2294,
+      "step": 2388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007042029735419681,
+      "loss": 3.3003,
+      "step": 2389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007039645710495224,
+      "loss": 3.2018,
+      "step": 2390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007037261129175781,
+      "loss": 3.1521,
+      "step": 2391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034875992111839,
+      "loss": 3.3059,
+      "step": 2392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007032490299954041,
+      "loss": 3.2848,
+      "step": 2393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007030104053353184,
+      "loss": 3.2664,
+      "step": 2394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007027717252960205,
+      "loss": 3.318,
+      "step": 2395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007025329899426205,
+      "loss": 3.2827,
+      "step": 2396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007022941993402428,
+      "loss": 3.1678,
+      "step": 2397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000702055353554027,
+      "loss": 3.2703,
+      "step": 2398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007018164526491281,
+      "loss": 3.1449,
+      "step": 2399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007015774966907157,
+      "loss": 3.2486,
+      "step": 2400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007013384857439746,
+      "loss": 3.2396,
+      "step": 2401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007010994198741046,
+      "loss": 3.2147,
+      "step": 2402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007008602991463207,
+      "loss": 3.3983,
+      "step": 2403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007006211236258523,
+      "loss": 3.1631,
+      "step": 2404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007003818933779444,
+      "loss": 3.1418,
+      "step": 2405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007001426084678563,
+      "loss": 3.2172,
+      "step": 2406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006999032689608629,
+      "loss": 3.1305,
+      "step": 2407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006996638749222534,
+      "loss": 3.2526,
+      "step": 2408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006994244264173321,
+      "loss": 3.2125,
+      "step": 2409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006991849235114183,
+      "loss": 3.3355,
+      "step": 2410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006989453662698458,
+      "loss": 3.2209,
+      "step": 2411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006987057547579636,
+      "loss": 3.2027,
+      "step": 2412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006984660890411353,
+      "loss": 3.131,
+      "step": 2413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006982263691847393,
+      "loss": 3.2785,
+      "step": 2414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006979865952541687,
+      "loss": 3.2359,
+      "step": 2415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006977467673148315,
+      "loss": 3.2334,
+      "step": 2416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006975068854321504,
+      "loss": 3.3546,
+      "step": 2417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000697266949671563,
+      "loss": 3.2672,
+      "step": 2418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006970269600985208,
+      "loss": 3.2393,
+      "step": 2419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006967869167784909,
+      "loss": 3.2753,
+      "step": 2420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965468197769547,
+      "loss": 3.321,
+      "step": 2421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006963066691594084,
+      "loss": 3.1718,
+      "step": 2422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006960664649913628,
+      "loss": 3.1256,
+      "step": 2423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006958262073383424,
+      "loss": 3.1845,
+      "step": 2424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006955858962658881,
+      "loss": 3.2462,
+      "step": 2425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006953455318395538,
+      "loss": 3.3439,
+      "step": 2426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006951051141249086,
+      "loss": 3.2748,
+      "step": 2427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006948646431875362,
+      "loss": 3.0447,
+      "step": 2428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006946241190930345,
+      "loss": 3.4248,
+      "step": 2429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006943835419070161,
+      "loss": 3.3493,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006941429116951081,
+      "loss": 3.1887,
+      "step": 2431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000693902228522952,
+      "loss": 3.2811,
+      "step": 2432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006936614924562038,
+      "loss": 3.1866,
+      "step": 2433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006934207035605338,
+      "loss": 3.0892,
+      "step": 2434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006931798619016269,
+      "loss": 3.2421,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006929389675451823,
+      "loss": 3.3325,
+      "step": 2436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006926980205569135,
+      "loss": 3.3181,
+      "step": 2437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006924570210025485,
+      "loss": 3.1922,
+      "step": 2438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006922159689478297,
+      "loss": 3.0082,
+      "step": 2439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006919748644585133,
+      "loss": 3.1829,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006917337076003706,
+      "loss": 3.1297,
+      "step": 2441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006914924984391866,
+      "loss": 3.1293,
+      "step": 2442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006912512370407608,
+      "loss": 3.2214,
+      "step": 2443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910099234709069,
+      "loss": 3.2349,
+      "step": 2444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006907685577954528,
+      "loss": 3.2974,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006905271400802405,
+      "loss": 3.3025,
+      "step": 2446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006902856703911266,
+      "loss": 3.2207,
+      "step": 2447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006900441487939817,
+      "loss": 3.1736,
+      "step": 2448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006898025753546902,
+      "loss": 3.2478,
+      "step": 2449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006895609501391509,
+      "loss": 3.3099,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000689319273213277,
+      "loss": 3.3642,
+      "step": 2451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006890775446429955,
+      "loss": 3.2618,
+      "step": 2452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006888357644942472,
+      "loss": 3.0745,
+      "step": 2453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006885939328329878,
+      "loss": 3.2057,
+      "step": 2454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006883520497251861,
+      "loss": 3.2935,
+      "step": 2455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006881101152368259,
+      "loss": 3.366,
+      "step": 2456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006878681294339042,
+      "loss": 3.2241,
+      "step": 2457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006876260923824322,
+      "loss": 3.2848,
+      "step": 2458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006873840041484356,
+      "loss": 3.275,
+      "step": 2459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006871418647979532,
+      "loss": 3.1394,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006868996743970386,
+      "loss": 3.2407,
+      "step": 2461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006866574330117585,
+      "loss": 3.1837,
+      "step": 2462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006864151407081943,
+      "loss": 3.2849,
+      "step": 2463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006861727975524407,
+      "loss": 3.2006,
+      "step": 2464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006859304036106067,
+      "loss": 3.2295,
+      "step": 2465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856879589488147,
+      "loss": 3.3242,
+      "step": 2466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006854454636332014,
+      "loss": 3.1532,
+      "step": 2467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852029177299169,
+      "loss": 3.2337,
+      "step": 2468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006849603213051255,
+      "loss": 3.3271,
+      "step": 2469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006847176744250049,
+      "loss": 3.2234,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000684474977155747,
+      "loss": 3.2485,
+      "step": 2471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006842322295635569,
+      "loss": 3.279,
+      "step": 2472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006839894317146538,
+      "loss": 3.0611,
+      "step": 2473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006837465836752706,
+      "loss": 3.2719,
+      "step": 2474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006835036855116539,
+      "loss": 3.2059,
+      "step": 2475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006832607372900636,
+      "loss": 3.2122,
+      "step": 2476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000683017739076774,
+      "loss": 3.3313,
+      "step": 2477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006827746909380722,
+      "loss": 3.074,
+      "step": 2478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006825315929402594,
+      "loss": 3.2372,
+      "step": 2479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006822884451496505,
+      "loss": 3.153,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006820452476325733,
+      "loss": 3.2201,
+      "step": 2481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006818020004553702,
+      "loss": 3.2218,
+      "step": 2482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006815587036843964,
+      "loss": 3.2461,
+      "step": 2483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006813153573860209,
+      "loss": 3.3799,
+      "step": 2484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006810719616266259,
+      "loss": 3.4,
+      "step": 2485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006808285164726076,
+      "loss": 3.1708,
+      "step": 2486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006805850219903751,
+      "loss": 3.2029,
+      "step": 2487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006803414782463516,
+      "loss": 3.131,
+      "step": 2488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006800978853069733,
+      "loss": 3.3197,
+      "step": 2489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006798542432386897,
+      "loss": 3.3108,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006796105521079643,
+      "loss": 3.1706,
+      "step": 2491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006793668119812735,
+      "loss": 3.3288,
+      "step": 2492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006791230229251067,
+      "loss": 3.2508,
+      "step": 2493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788791850059679,
+      "loss": 3.1876,
+      "step": 2494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006786352982903732,
+      "loss": 3.2854,
+      "step": 2495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006783913628448523,
+      "loss": 3.3168,
+      "step": 2496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006781473787359488,
+      "loss": 3.2681,
+      "step": 2497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006779033460302189,
+      "loss": 3.2489,
+      "step": 2498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006776592647942322,
+      "loss": 3.0535,
+      "step": 2499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006774151350945717,
+      "loss": 3.2502,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006771709569978337,
+      "loss": 3.1901,
+      "step": 2501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006769267305706272,
+      "loss": 3.1329,
+      "step": 2502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676682455879575,
+      "loss": 3.2133,
+      "step": 2503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676438132991313,
+      "loss": 3.1973,
+      "step": 2504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006761937619724894,
+      "loss": 3.3134,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006759493428897667,
+      "loss": 3.3521,
+      "step": 2506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006757048758098199,
+      "loss": 3.2364,
+      "step": 2507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000675460360799337,
+      "loss": 3.3886,
+      "step": 2508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006752157979250195,
+      "loss": 3.308,
+      "step": 2509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006749711872535816,
+      "loss": 3.0457,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006747265288517505,
+      "loss": 3.2852,
+      "step": 2511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006744818227862668,
+      "loss": 3.2673,
+      "step": 2512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742370691238836,
+      "loss": 3.2516,
+      "step": 2513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006739922679313677,
+      "loss": 3.2373,
+      "step": 2514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006737474192754979,
+      "loss": 3.2346,
+      "step": 2515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006735025232230668,
+      "loss": 3.0965,
+      "step": 2516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006732575798408797,
+      "loss": 3.2551,
+      "step": 2517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006730125891957543,
+      "loss": 3.1609,
+      "step": 2518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006727675513545219,
+      "loss": 3.1434,
+      "step": 2519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006725224663840264,
+      "loss": 3.3367,
+      "step": 2520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006722773343511244,
+      "loss": 3.1853,
+      "step": 2521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006720321553226854,
+      "loss": 3.3621,
+      "step": 2522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006717869293655919,
+      "loss": 3.2767,
+      "step": 2523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006715416565467392,
+      "loss": 3.3447,
+      "step": 2524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006712963369330351,
+      "loss": 3.1636,
+      "step": 2525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006710509705914005,
+      "loss": 3.2909,
+      "step": 2526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006708055575887687,
+      "loss": 3.2877,
+      "step": 2527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006705600979920861,
+      "loss": 3.3157,
+      "step": 2528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006703145918683116,
+      "loss": 3.2956,
+      "step": 2529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006700690392844166,
+      "loss": 3.3955,
+      "step": 2530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006698234403073856,
+      "loss": 3.3652,
+      "step": 2531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006695777950042155,
+      "loss": 3.3426,
+      "step": 2532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000669332103441916,
+      "loss": 3.336,
+      "step": 2533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000669086365687509,
+      "loss": 3.2321,
+      "step": 2534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006688405818080298,
+      "loss": 3.3056,
+      "step": 2535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006685947518705253,
+      "loss": 3.1641,
+      "step": 2536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006683488759420555,
+      "loss": 3.2236,
+      "step": 2537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006681029540896934,
+      "loss": 3.2956,
+      "step": 2538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006678569863805234,
+      "loss": 3.2407,
+      "step": 2539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006676109728816434,
+      "loss": 3.1761,
+      "step": 2540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006673649136601634,
+      "loss": 3.2503,
+      "step": 2541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006671188087832056,
+      "loss": 3.1361,
+      "step": 2542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006668726583179054,
+      "loss": 3.2491,
+      "step": 2543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00066662646233141,
+      "loss": 3.3543,
+      "step": 2544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666380220890879,
+      "loss": 3.165,
+      "step": 2545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006661339340634848,
+      "loss": 3.2246,
+      "step": 2546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006658876019164122,
+      "loss": 3.1885,
+      "step": 2547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006656412245168578,
+      "loss": 3.1899,
+      "step": 2548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006653948019320309,
+      "loss": 3.3509,
+      "step": 2549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006651483342291535,
+      "loss": 3.2327,
+      "step": 2550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006649018214754591,
+      "loss": 3.2743,
+      "step": 2551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006646552637381942,
+      "loss": 3.316,
+      "step": 2552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006644086610846171,
+      "loss": 3.2112,
+      "step": 2553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006641620135819985,
+      "loss": 3.1752,
+      "step": 2554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006639153212976217,
+      "loss": 3.2628,
+      "step": 2555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006636685842987816,
+      "loss": 3.3566,
+      "step": 2556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006634218026527856,
+      "loss": 3.2893,
+      "step": 2557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006631749764269536,
+      "loss": 3.2105,
+      "step": 2558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006629281056886167,
+      "loss": 3.2792,
+      "step": 2559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006626811905051194,
+      "loss": 3.292,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006624342309438176,
+      "loss": 3.3569,
+      "step": 2561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006621872270720789,
+      "loss": 3.0615,
+      "step": 2562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006619401789572841,
+      "loss": 3.3385,
+      "step": 2563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006616930866668252,
+      "loss": 3.3599,
+      "step": 2564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006614459502681062,
+      "loss": 3.1784,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000661198769828544,
+      "loss": 3.2052,
+      "step": 2566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006609515454155668,
+      "loss": 3.1359,
+      "step": 2567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006607042770966147,
+      "loss": 3.1847,
+      "step": 2568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006604569649391402,
+      "loss": 3.2265,
+      "step": 2569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006602096090106077,
+      "loss": 3.2389,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000659962209378493,
+      "loss": 3.232,
+      "step": 2571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000659714766110285,
+      "loss": 3.3135,
+      "step": 2572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006594672792734831,
+      "loss": 3.3107,
+      "step": 2573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006592197489355993,
+      "loss": 3.2807,
+      "step": 2574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006589721751641578,
+      "loss": 3.3102,
+      "step": 2575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006587245580266937,
+      "loss": 3.2197,
+      "step": 2576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584768975907552,
+      "loss": 3.3875,
+      "step": 2577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006582291939239008,
+      "loss": 3.2672,
+      "step": 2578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006579814470937021,
+      "loss": 3.2274,
+      "step": 2579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000657733657167742,
+      "loss": 3.2129,
+      "step": 2580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006574858242136146,
+      "loss": 3.3573,
+      "step": 2581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006572379482989269,
+      "loss": 3.3165,
+      "step": 2582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006569900294912968,
+      "loss": 3.2058,
+      "step": 2583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006567420678583536,
+      "loss": 3.2085,
+      "step": 2584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006564940634677392,
+      "loss": 3.2628,
+      "step": 2585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006562460163871068,
+      "loss": 3.3337,
+      "step": 2586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006559979266841209,
+      "loss": 3.1905,
+      "step": 2587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006557497944264578,
+      "loss": 3.2537,
+      "step": 2588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006555016196818062,
+      "loss": 3.3512,
+      "step": 2589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006552534025178647,
+      "loss": 3.2943,
+      "step": 2590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006550051430023452,
+      "loss": 3.2082,
+      "step": 2591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006547568412029701,
+      "loss": 3.1531,
+      "step": 2592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006545084971874737,
+      "loss": 3.1906,
+      "step": 2593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006542601110236018,
+      "loss": 3.2806,
+      "step": 2594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006540116827791118,
+      "loss": 3.3476,
+      "step": 2595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006537632125217722,
+      "loss": 3.2024,
+      "step": 2596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006535147003193634,
+      "loss": 3.2417,
+      "step": 2597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006532661462396768,
+      "loss": 3.2972,
+      "step": 2598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006530175503505157,
+      "loss": 3.1001,
+      "step": 2599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006527689127196945,
+      "loss": 3.2191,
+      "step": 2600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006525202334150392,
+      "loss": 3.3669,
+      "step": 2601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006522715125043868,
+      "loss": 3.3027,
+      "step": 2602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006520227500555858,
+      "loss": 3.2313,
+      "step": 2603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006517739461364965,
+      "loss": 3.3859,
+      "step": 2604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00065152510081499,
+      "loss": 3.1639,
+      "step": 2605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006512762141589486,
+      "loss": 3.175,
+      "step": 2606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006510272862362662,
+      "loss": 3.1064,
+      "step": 2607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006507783171148479,
+      "loss": 3.3029,
+      "step": 2608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00065052930686261,
+      "loss": 3.2576,
+      "step": 2609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006502802555474801,
+      "loss": 3.2599,
+      "step": 2610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006500311632373967,
+      "loss": 3.1116,
+      "step": 2611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006497820300003098,
+      "loss": 3.2654,
+      "step": 2612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006495328559041805,
+      "loss": 3.166,
+      "step": 2613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006492836410169809,
+      "loss": 3.3195,
+      "step": 2614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006490343854066945,
+      "loss": 3.186,
+      "step": 2615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006487850891413157,
+      "loss": 3.2709,
+      "step": 2616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006485357522888496,
+      "loss": 3.2833,
+      "step": 2617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006482863749173134,
+      "loss": 3.1862,
+      "step": 2618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006480369570947347,
+      "loss": 3.2558,
+      "step": 2619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006477874988891518,
+      "loss": 3.2192,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006475380003686147,
+      "loss": 3.3221,
+      "step": 2621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000647288461601184,
+      "loss": 3.1782,
+      "step": 2622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006470388826549314,
+      "loss": 3.3355,
+      "step": 2623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006467892635979396,
+      "loss": 3.0389,
+      "step": 2624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006465396044983023,
+      "loss": 3.3462,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006462899054241237,
+      "loss": 3.2048,
+      "step": 2626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006460401664435195,
+      "loss": 3.2257,
+      "step": 2627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006457903876246156,
+      "loss": 3.2329,
+      "step": 2628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006455405690355497,
+      "loss": 3.2402,
+      "step": 2629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006452907107444696,
+      "loss": 3.0645,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006450408128195338,
+      "loss": 3.3536,
+      "step": 2631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006447908753289127,
+      "loss": 3.0372,
+      "step": 2632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006445408983407861,
+      "loss": 3.1113,
+      "step": 2633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006442908819233453,
+      "loss": 3.1191,
+      "step": 2634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006440408261447927,
+      "loss": 3.2081,
+      "step": 2635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006437907310733406,
+      "loss": 3.3645,
+      "step": 2636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006435405967772126,
+      "loss": 3.2132,
+      "step": 2637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006432904233246428,
+      "loss": 3.2448,
+      "step": 2638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006430402107838762,
+      "loss": 3.1957,
+      "step": 2639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000642789959223168,
+      "loss": 3.1748,
+      "step": 2640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425396687107845,
+      "loss": 3.1541,
+      "step": 2641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006422893393150024,
+      "loss": 3.1863,
+      "step": 2642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006420389711041091,
+      "loss": 3.1469,
+      "step": 2643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006417885641464026,
+      "loss": 3.1419,
+      "step": 2644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641538118510191,
+      "loss": 3.238,
+      "step": 2645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641287634263794,
+      "loss": 3.1513,
+      "step": 2646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006410371114755409,
+      "loss": 3.3119,
+      "step": 2647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006407865502137717,
+      "loss": 3.2069,
+      "step": 2648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006405359505468372,
+      "loss": 3.3763,
+      "step": 2649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006402853125430985,
+      "loss": 3.2821,
+      "step": 2650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000640034636270927,
+      "loss": 3.2266,
+      "step": 2651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000639783921798705,
+      "loss": 3.1684,
+      "step": 2652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006395331691948243,
+      "loss": 3.119,
+      "step": 2653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006392823785276882,
+      "loss": 3.2322,
+      "step": 2654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006390315498657098,
+      "loss": 3.2483,
+      "step": 2655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006387806832773125,
+      "loss": 3.3167,
+      "step": 2656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006385297788309303,
+      "loss": 3.1301,
+      "step": 2657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006382788365950076,
+      "loss": 3.3424,
+      "step": 2658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380278566379987,
+      "loss": 3.2576,
+      "step": 2659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006377768390283683,
+      "loss": 3.2985,
+      "step": 2660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006375257838345921,
+      "loss": 3.1934,
+      "step": 2661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006372746911251548,
+      "loss": 3.1864,
+      "step": 2662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006370235609685522,
+      "loss": 3.3021,
+      "step": 2663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006367723934332903,
+      "loss": 3.4077,
+      "step": 2664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006365211885878848,
+      "loss": 3.3459,
+      "step": 2665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006362699465008619,
+      "loss": 3.2732,
+      "step": 2666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006360186672407583,
+      "loss": 3.3125,
+      "step": 2667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00063576735087612,
+      "loss": 3.2195,
+      "step": 2668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006355159974755039,
+      "loss": 3.2436,
+      "step": 2669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006352646071074767,
+      "loss": 3.2048,
+      "step": 2670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006350131798406149,
+      "loss": 3.2943,
+      "step": 2671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006347617157435057,
+      "loss": 3.2185,
+      "step": 2672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006345102148847458,
+      "loss": 3.2188,
+      "step": 2673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006342586773329421,
+      "loss": 3.2676,
+      "step": 2674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006340071031567115,
+      "loss": 3.117,
+      "step": 2675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006337554924246812,
+      "loss": 3.3023,
+      "step": 2676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006335038452054877,
+      "loss": 3.258,
+      "step": 2677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006332521615677782,
+      "loss": 3.345,
+      "step": 2678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006330004415802095,
+      "loss": 3.0818,
+      "step": 2679
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006327486853114478,
+      "loss": 3.3204,
+      "step": 2680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006324968928301702,
+      "loss": 3.3297,
+      "step": 2681
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000632245064205063,
+      "loss": 3.0557,
+      "step": 2682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006319931995048226,
+      "loss": 3.208,
+      "step": 2683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000631741298798155,
+      "loss": 3.3319,
+      "step": 2684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006314893621537765,
+      "loss": 3.4352,
+      "step": 2685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006312373896404126,
+      "loss": 2.9215,
+      "step": 2686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006309853813267989,
+      "loss": 3.3112,
+      "step": 2687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006307333372816811,
+      "loss": 3.203,
+      "step": 2688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000630481257573814,
+      "loss": 3.2017,
+      "step": 2689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006302291422719623,
+      "loss": 3.2631,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000629976991444901,
+      "loss": 3.3584,
+      "step": 2691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006297248051614137,
+      "loss": 3.1867,
+      "step": 2692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006294725834902948,
+      "loss": 3.1731,
+      "step": 2693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006292203265003479,
+      "loss": 3.1717,
+      "step": 2694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006289680342603858,
+      "loss": 3.2928,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006287157068392315,
+      "loss": 3.2615,
+      "step": 2696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006284633443057175,
+      "loss": 3.2823,
+      "step": 2697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006282109467286856,
+      "loss": 3.3115,
+      "step": 2698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006279585141769874,
+      "loss": 3.3554,
+      "step": 2699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006277060467194841,
+      "loss": 3.3139,
+      "step": 2700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006274535444250462,
+      "loss": 3.2183,
+      "step": 2701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006272010073625538,
+      "loss": 3.2839,
+      "step": 2702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006269484356008963,
+      "loss": 3.2324,
+      "step": 2703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006266958292089732,
+      "loss": 3.1865,
+      "step": 2704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264431882556927,
+      "loss": 3.1163,
+      "step": 2705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006261905128099726,
+      "loss": 3.2799,
+      "step": 2706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006259378029407406,
+      "loss": 3.2576,
+      "step": 2707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006256850587169334,
+      "loss": 3.2953,
+      "step": 2708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006254322802074965,
+      "loss": 3.2036,
+      "step": 2709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006251794674813862,
+      "loss": 3.2261,
+      "step": 2710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006249266206075668,
+      "loss": 3.2979,
+      "step": 2711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006246737396550124,
+      "loss": 3.1635,
+      "step": 2712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006244208246927066,
+      "loss": 3.2663,
+      "step": 2713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006241678757896419,
+      "loss": 3.3069,
+      "step": 2714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006239148930148203,
+      "loss": 3.2022,
+      "step": 2715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623661876437253,
+      "loss": 3.3226,
+      "step": 2716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006234088261259604,
+      "loss": 3.1836,
+      "step": 2717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623155742149972,
+      "loss": 3.476,
+      "step": 2718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006229026245783269,
+      "loss": 3.2331,
+      "step": 2719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006226494734800725,
+      "loss": 3.2609,
+      "step": 2720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006223962889242663,
+      "loss": 3.282,
+      "step": 2721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006221430709799745,
+      "loss": 3.2676,
+      "step": 2722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006218898197162723,
+      "loss": 3.3403,
+      "step": 2723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006216365352022441,
+      "loss": 3.1192,
+      "step": 2724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006213832175069837,
+      "loss": 3.2273,
+      "step": 2725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006211298666995932,
+      "loss": 3.1297,
+      "step": 2726
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006208764828491844,
+      "loss": 3.3427,
+      "step": 2727
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006206230660248777,
+      "loss": 3.3057,
+      "step": 2728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006203696162958029,
+      "loss": 3.1609,
+      "step": 2729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006201161337310986,
+      "loss": 3.2403,
+      "step": 2730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006198626183999121,
+      "loss": 3.1572,
+      "step": 2731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006196090703713998,
+      "loss": 3.2182,
+      "step": 2732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006193554897147275,
+      "loss": 3.2134,
+      "step": 2733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006191018764990689,
+      "loss": 3.4599,
+      "step": 2734
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006188482307936076,
+      "loss": 3.1646,
+      "step": 2735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006185945526675355,
+      "loss": 3.2858,
+      "step": 2736
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006183408421900533,
+      "loss": 3.2918,
+      "step": 2737
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006180870994303707,
+      "loss": 3.2482,
+      "step": 2738
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006178333244577066,
+      "loss": 3.371,
+      "step": 2739
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006175795173412877,
+      "loss": 3.418,
+      "step": 2740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006173256781503505,
+      "loss": 3.0703,
+      "step": 2741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006170718069541395,
+      "loss": 3.3643,
+      "step": 2742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006168179038219083,
+      "loss": 3.1886,
+      "step": 2743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006165639688229193,
+      "loss": 3.1264,
+      "step": 2744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006163100020264434,
+      "loss": 3.1647,
+      "step": 2745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00061605600350176,
+      "loss": 3.3302,
+      "step": 2746
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006158019733181573,
+      "loss": 3.3022,
+      "step": 2747
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006155479115449328,
+      "loss": 3.1757,
+      "step": 2748
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006152938182513912,
+      "loss": 3.3729,
+      "step": 2749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006150396935068471,
+      "loss": 3.251,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006147855373806233,
+      "loss": 3.1539,
+      "step": 2751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006145313499420506,
+      "loss": 3.2478,
+      "step": 2752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006142771312604692,
+      "loss": 3.2511,
+      "step": 2753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006140228814052273,
+      "loss": 3.2792,
+      "step": 2754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006137686004456816,
+      "loss": 3.1717,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006135142884511975,
+      "loss": 3.2796,
+      "step": 2756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006132599454911489,
+      "loss": 3.179,
+      "step": 2757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006130055716349178,
+      "loss": 3.2311,
+      "step": 2758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006127511669518949,
+      "loss": 3.1269,
+      "step": 2759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006124967315114795,
+      "loss": 3.1684,
+      "step": 2760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006122422653830786,
+      "loss": 3.2389,
+      "step": 2761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006119877686361084,
+      "loss": 3.1949,
+      "step": 2762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000611733241339993,
+      "loss": 3.211,
+      "step": 2763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006114786835641647,
+      "loss": 3.2839,
+      "step": 2764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006112240953780644,
+      "loss": 3.4735,
+      "step": 2765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006109694768511414,
+      "loss": 3.2524,
+      "step": 2766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000610714828052853,
+      "loss": 3.2904,
+      "step": 2767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006104601490526649,
+      "loss": 3.2764,
+      "step": 2768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006102054399200507,
+      "loss": 3.1725,
+      "step": 2769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099507007244929,
+      "loss": 3.2126,
+      "step": 2770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006096959315354815,
+      "loss": 3.263,
+      "step": 2771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006094411324225152,
+      "loss": 3.1602,
+      "step": 2772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006091863034551006,
+      "loss": 3.2716,
+      "step": 2773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006089314447027525,
+      "loss": 3.1168,
+      "step": 2774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006086765562349937,
+      "loss": 3.2818,
+      "step": 2775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006084216381213553,
+      "loss": 3.1495,
+      "step": 2776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006081666904313766,
+      "loss": 3.2694,
+      "step": 2777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006079117132346043,
+      "loss": 3.2119,
+      "step": 2778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006076567066005944,
+      "loss": 3.2002,
+      "step": 2779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006074016705989094,
+      "loss": 3.2411,
+      "step": 2780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006071466052991209,
+      "loss": 3.1096,
+      "step": 2781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006068915107708084,
+      "loss": 3.2159,
+      "step": 2782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006066363870835587,
+      "loss": 3.3051,
+      "step": 2783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006063812343069672,
+      "loss": 3.236,
+      "step": 2784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061260525106372,
+      "loss": 3.0784,
+      "step": 2785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006058708417641795,
+      "loss": 3.2405,
+      "step": 2786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006056156021372131,
+      "loss": 3.3285,
+      "step": 2787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006053603336993649,
+      "loss": 3.2654,
+      "step": 2788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006051050365202695,
+      "loss": 3.3196,
+      "step": 2789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006048497106695694,
+      "loss": 3.1975,
+      "step": 2790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006045943562169151,
+      "loss": 3.2366,
+      "step": 2791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006043389732319645,
+      "loss": 3.4659,
+      "step": 2792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006040835617843836,
+      "loss": 3.2157,
+      "step": 2793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006038281219438463,
+      "loss": 3.3577,
+      "step": 2794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006035726537800339,
+      "loss": 3.3213,
+      "step": 2795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006033171573626355,
+      "loss": 3.1815,
+      "step": 2796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006030616327613482,
+      "loss": 3.2646,
+      "step": 2797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006028060800458762,
+      "loss": 3.2252,
+      "step": 2798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006025504992859321,
+      "loss": 3.1211,
+      "step": 2799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006022948905512355,
+      "loss": 3.3619,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006020392539115138,
+      "loss": 3.089,
+      "step": 2801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006017835894365027,
+      "loss": 3.2625,
+      "step": 2802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006015278971959444,
+      "loss": 3.1854,
+      "step": 2803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006012721772595893,
+      "loss": 3.2827,
+      "step": 2804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006010164296971953,
+      "loss": 3.3445,
+      "step": 2805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006007606545785278,
+      "loss": 3.2105,
+      "step": 2806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006005048519733596,
+      "loss": 3.2327,
+      "step": 2807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006002490219514712,
+      "loss": 3.3197,
+      "step": 2808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005999931645826505,
+      "loss": 3.1442,
+      "step": 2809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005997372799366927,
+      "loss": 3.2893,
+      "step": 2810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005994813680834008,
+      "loss": 3.1903,
+      "step": 2811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005992254290925847,
+      "loss": 3.3819,
+      "step": 2812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005989694630340622,
+      "loss": 3.248,
+      "step": 2813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005987134699776583,
+      "loss": 3.1616,
+      "step": 2814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005984574499932052,
+      "loss": 3.2481,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005982014031505427,
+      "loss": 3.1431,
+      "step": 2816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005979453295195179,
+      "loss": 3.256,
+      "step": 2817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005976892291699848,
+      "loss": 3.2701,
+      "step": 2818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005974331021718056,
+      "loss": 3.3866,
+      "step": 2819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005971769485948488,
+      "loss": 3.1709,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005969207685089901,
+      "loss": 3.2827,
+      "step": 2821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005966645619841139,
+      "loss": 3.3235,
+      "step": 2822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005964083290901101,
+      "loss": 3.3103,
+      "step": 2823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005961520698968766,
+      "loss": 3.4071,
+      "step": 2824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005958957844743182,
+      "loss": 3.3132,
+      "step": 2825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005956394728923474,
+      "loss": 3.2692,
+      "step": 2826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005953831352208831,
+      "loss": 3.2395,
+      "step": 2827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005951267715298516,
+      "loss": 3.3205,
+      "step": 2828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005948703818891869,
+      "loss": 3.2166,
+      "step": 2829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005946139663688288,
+      "loss": 3.2329,
+      "step": 2830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005943575250387253,
+      "loss": 3.3753,
+      "step": 2831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005941010579688312,
+      "loss": 3.2393,
+      "step": 2832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005938445652291078,
+      "loss": 3.2916,
+      "step": 2833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005935880468895239,
+      "loss": 3.2894,
+      "step": 2834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000593331503020055,
+      "loss": 3.1294,
+      "step": 2835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005930749336906841,
+      "loss": 3.1199,
+      "step": 2836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005928183389714004,
+      "loss": 3.3385,
+      "step": 2837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005925617189322004,
+      "loss": 3.154,
+      "step": 2838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005923050736430876,
+      "loss": 3.1568,
+      "step": 2839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005920484031740722,
+      "loss": 3.242,
+      "step": 2840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005917917075951714,
+      "loss": 3.1087,
+      "step": 2841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005915349869764092,
+      "loss": 3.2438,
+      "step": 2842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005912782413878164,
+      "loss": 3.3033,
+      "step": 2843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910214708994306,
+      "loss": 3.411,
+      "step": 2844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005907646755812963,
+      "loss": 3.2507,
+      "step": 2845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005905078555034645,
+      "loss": 3.2788,
+      "step": 2846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005902510107359934,
+      "loss": 3.1932,
+      "step": 2847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005899941413489477,
+      "loss": 3.2109,
+      "step": 2848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005897372474123985,
+      "loss": 3.3343,
+      "step": 2849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005894803289964242,
+      "loss": 3.2959,
+      "step": 2850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005892233861711094,
+      "loss": 3.2676,
+      "step": 2851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005889664190065456,
+      "loss": 3.3498,
+      "step": 2852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005887094275728309,
+      "loss": 3.2089,
+      "step": 2853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00058845241194007,
+      "loss": 3.2367,
+      "step": 2854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005881953721783743,
+      "loss": 3.09,
+      "step": 2855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005879383083578614,
+      "loss": 3.234,
+      "step": 2856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587681220548656,
+      "loss": 3.3249,
+      "step": 2857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587424108820889,
+      "loss": 3.1046,
+      "step": 2858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005871669732446978,
+      "loss": 3.2977,
+      "step": 2859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005869098138902265,
+      "loss": 3.1625,
+      "step": 2860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005866526308276258,
+      "loss": 3.2562,
+      "step": 2861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005863954241270524,
+      "loss": 3.3571,
+      "step": 2862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005861381938586699,
+      "loss": 3.3367,
+      "step": 2863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000585880940092648,
+      "loss": 3.1936,
+      "step": 2864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000585623662899163,
+      "loss": 3.2304,
+      "step": 2865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005853663623483977,
+      "loss": 3.2672,
+      "step": 2866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005851090385105409,
+      "loss": 3.0635,
+      "step": 2867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005848516914557881,
+      "loss": 3.2419,
+      "step": 2868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000584594321254341,
+      "loss": 3.284,
+      "step": 2869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005843369279764076,
+      "loss": 3.113,
+      "step": 2870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005840795116922022,
+      "loss": 3.2406,
+      "step": 2871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005838220724719454,
+      "loss": 3.2539,
+      "step": 2872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005835646103858641,
+      "loss": 3.2641,
+      "step": 2873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005833071255041914,
+      "loss": 3.1373,
+      "step": 2874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005830496178971665,
+      "loss": 3.2945,
+      "step": 2875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005827920876350349,
+      "loss": 3.3018,
+      "step": 2876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005825345347880484,
+      "loss": 3.1548,
+      "step": 2877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005822769594264649,
+      "loss": 3.4063,
+      "step": 2878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000582019361620548,
+      "loss": 3.2583,
+      "step": 2879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005817617414405684,
+      "loss": 3.3856,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005815040989568018,
+      "loss": 3.1103,
+      "step": 2881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005812464342395307,
+      "loss": 3.1975,
+      "step": 2882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005809887473590435,
+      "loss": 3.3603,
+      "step": 2883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005807310383856345,
+      "loss": 3.2184,
+      "step": 2884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005804733073896044,
+      "loss": 3.2268,
+      "step": 2885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005802155544412593,
+      "loss": 3.191,
+      "step": 2886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005799577796109117,
+      "loss": 3.1723,
+      "step": 2887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005796999829688802,
+      "loss": 3.0655,
+      "step": 2888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000579442164585489,
+      "loss": 3.1588,
+      "step": 2889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005791843245310683,
+      "loss": 3.3678,
+      "step": 2890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005789264628759545,
+      "loss": 3.0815,
+      "step": 2891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005786685796904897,
+      "loss": 3.2768,
+      "step": 2892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005784106750450215,
+      "loss": 3.391,
+      "step": 2893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005781527490099042,
+      "loss": 3.2984,
+      "step": 2894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005778948016554971,
+      "loss": 3.1969,
+      "step": 2895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005776368330521656,
+      "loss": 3.2379,
+      "step": 2896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005773788432702814,
+      "loss": 3.3042,
+      "step": 2897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005771208323802213,
+      "loss": 3.2272,
+      "step": 2898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005768628004523678,
+      "loss": 3.2949,
+      "step": 2899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005766047475571098,
+      "loss": 3.3118,
+      "step": 2900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005763466737648415,
+      "loss": 3.2878,
+      "step": 2901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005760885791459627,
+      "loss": 3.2965,
+      "step": 2902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005758304637708791,
+      "loss": 3.1834,
+      "step": 2903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005755723277100021,
+      "loss": 3.2982,
+      "step": 2904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005753141710337486,
+      "loss": 3.2503,
+      "step": 2905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005750559938125411,
+      "loss": 3.2491,
+      "step": 2906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005747977961168078,
+      "loss": 3.2223,
+      "step": 2907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005745395780169826,
+      "loss": 3.1516,
+      "step": 2908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005742813395835045,
+      "loss": 3.339,
+      "step": 2909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005740230808868185,
+      "loss": 3.105,
+      "step": 2910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005737648019973753,
+      "loss": 3.2541,
+      "step": 2911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005735065029856304,
+      "loss": 3.4053,
+      "step": 2912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005732481839220453,
+      "loss": 3.1824,
+      "step": 2913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005729898448770869,
+      "loss": 3.2087,
+      "step": 2914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005727314859212275,
+      "loss": 3.1342,
+      "step": 2915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005724731071249449,
+      "loss": 3.0929,
+      "step": 2916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005722147085587223,
+      "loss": 3.2516,
+      "step": 2917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005719562902930479,
+      "loss": 3.4588,
+      "step": 2918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000571697852398416,
+      "loss": 3.0319,
+      "step": 2919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005714393949453258,
+      "loss": 3.1345,
+      "step": 2920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005711809180042819,
+      "loss": 3.2252,
+      "step": 2921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005709224216457941,
+      "loss": 3.3952,
+      "step": 2922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005706639059403778,
+      "loss": 3.0421,
+      "step": 2923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005704053709585533,
+      "loss": 3.3343,
+      "step": 2924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005701468167708467,
+      "loss": 3.2704,
+      "step": 2925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005698882434477887,
+      "loss": 3.1814,
+      "step": 2926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005696296510599156,
+      "loss": 3.289,
+      "step": 2927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005693710396777687,
+      "loss": 3.3108,
+      "step": 2928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000569112409371895,
+      "loss": 3.0992,
+      "step": 2929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005688537602128458,
+      "loss": 3.159,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005685950922711782,
+      "loss": 3.1889,
+      "step": 2931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683364056174545,
+      "loss": 3.2205,
+      "step": 2932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005680777003222413,
+      "loss": 3.3713,
+      "step": 2933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005678189764561113,
+      "loss": 3.2425,
+      "step": 2934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005675602340896415,
+      "loss": 3.3798,
+      "step": 2935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005673014732934143,
+      "loss": 3.4012,
+      "step": 2936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005670426941380172,
+      "loss": 3.2409,
+      "step": 2937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005667838966940423,
+      "loss": 3.2229,
+      "step": 2938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005665250810320871,
+      "loss": 3.2976,
+      "step": 2939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000566266247222754,
+      "loss": 3.1528,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005660073953366503,
+      "loss": 3.188,
+      "step": 2941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005657485254443881,
+      "loss": 3.3084,
+      "step": 2942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005654896376165846,
+      "loss": 3.1222,
+      "step": 2943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005652307319238617,
+      "loss": 3.1588,
+      "step": 2944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005649718084368463,
+      "loss": 3.2015,
+      "step": 2945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005647128672261702,
+      "loss": 3.1567,
+      "step": 2946
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005644539083624701,
+      "loss": 3.1609,
+      "step": 2947
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005641949319163872,
+      "loss": 3.1626,
+      "step": 2948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005639359379585679,
+      "loss": 3.0694,
+      "step": 2949
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005636769265596628,
+      "loss": 3.2203,
+      "step": 2950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000563417897790328,
+      "loss": 3.3084,
+      "step": 2951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005631588517212238,
+      "loss": 3.2316,
+      "step": 2952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005628997884230154,
+      "loss": 3.1063,
+      "step": 2953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005626407079663726,
+      "loss": 3.1869,
+      "step": 2954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005623816104219701,
+      "loss": 3.227,
+      "step": 2955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005621224958604868,
+      "loss": 3.2836,
+      "step": 2956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005618633643526072,
+      "loss": 3.243,
+      "step": 2957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005616042159690193,
+      "loss": 3.2233,
+      "step": 2958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005613450507804161,
+      "loss": 3.1671,
+      "step": 2959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005610858688574958,
+      "loss": 3.1712,
+      "step": 2960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005608266702709602,
+      "loss": 3.2488,
+      "step": 2961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005605674550915161,
+      "loss": 3.2241,
+      "step": 2962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005603082233898751,
+      "loss": 3.163,
+      "step": 2963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005600489752367528,
+      "loss": 3.2125,
+      "step": 2964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005597897107028694,
+      "loss": 3.1831,
+      "step": 2965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00055953042985895,
+      "loss": 3.1,
+      "step": 2966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005592711327757236,
+      "loss": 2.9714,
+      "step": 2967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005590118195239238,
+      "loss": 3.3674,
+      "step": 2968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005587524901742889,
+      "loss": 3.3734,
+      "step": 2969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005584931447975613,
+      "loss": 3.194,
+      "step": 2970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005582337834644875,
+      "loss": 3.2746,
+      "step": 2971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005579744062458191,
+      "loss": 3.1882,
+      "step": 2972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005577150132123115,
+      "loss": 3.1259,
+      "step": 2973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005574556044347245,
+      "loss": 3.2761,
+      "step": 2974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005571961799838223,
+      "loss": 3.2351,
+      "step": 2975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000556936739930373,
+      "loss": 3.1233,
+      "step": 2976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005566772843451495,
+      "loss": 3.1026,
+      "step": 2977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005564178132989288,
+      "loss": 3.3462,
+      "step": 2978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005561583268624917,
+      "loss": 3.1032,
+      "step": 2979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000555898825106624,
+      "loss": 3.2915,
+      "step": 2980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005556393081021145,
+      "loss": 3.0523,
+      "step": 2981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005553797759197574,
+      "loss": 3.2151,
+      "step": 2982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005551202286303504,
+      "loss": 3.1518,
+      "step": 2983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005548606663046951,
+      "loss": 3.2045,
+      "step": 2984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005546010890135979,
+      "loss": 3.1733,
+      "step": 2985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005543414968278687,
+      "loss": 3.2085,
+      "step": 2986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005540818898183216,
+      "loss": 3.2113,
+      "step": 2987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005538222680557749,
+      "loss": 3.0879,
+      "step": 2988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000553562631611051,
+      "loss": 3.3749,
+      "step": 2989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005533029805549756,
+      "loss": 3.232,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005530433149583795,
+      "loss": 3.3065,
+      "step": 2991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005527836348920965,
+      "loss": 3.2166,
+      "step": 2992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552523940426965,
+      "loss": 3.0705,
+      "step": 2993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005522642316338268,
+      "loss": 3.3184,
+      "step": 2994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552004508583528,
+      "loss": 3.216,
+      "step": 2995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005517447713469185,
+      "loss": 3.2676,
+      "step": 2996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005514850199948519,
+      "loss": 3.3614,
+      "step": 2997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000551225254598186,
+      "loss": 3.2384,
+      "step": 2998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005509654752277819,
+      "loss": 3.2557,
+      "step": 2999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005507056819545049,
+      "loss": 3.3166,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005504458748492243,
+      "loss": 3.1723,
+      "step": 3001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005501860539828125,
+      "loss": 3.1236,
+      "step": 3002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005499262194261462,
+      "loss": 3.2639,
+      "step": 3003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005496663712501057,
+      "loss": 3.3128,
+      "step": 3004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005494065095255751,
+      "loss": 3.2574,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005491466343234418,
+      "loss": 3.1444,
+      "step": 3006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005488867457145974,
+      "loss": 3.2533,
+      "step": 3007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005486268437699368,
+      "loss": 3.2704,
+      "step": 3008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005483669285603585,
+      "loss": 3.2861,
+      "step": 3009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005481070001567653,
+      "loss": 3.3433,
+      "step": 3010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005478470586300625,
+      "loss": 3.2027,
+      "step": 3011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005475871040511597,
+      "loss": 3.1678,
+      "step": 3012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005473271364909702,
+      "loss": 3.3192,
+      "step": 3013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005470671560204104,
+      "loss": 3.199,
+      "step": 3014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005468071627104001,
+      "loss": 3.3145,
+      "step": 3015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005465471566318633,
+      "loss": 3.195,
+      "step": 3016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005462871378557267,
+      "loss": 3.218,
+      "step": 3017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005460271064529211,
+      "loss": 3.1642,
+      "step": 3018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005457670624943805,
+      "loss": 3.1151,
+      "step": 3019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005455070060510419,
+      "loss": 3.3324,
+      "step": 3020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005452469371938464,
+      "loss": 3.1659,
+      "step": 3021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005449868559937385,
+      "loss": 3.1428,
+      "step": 3022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000544726762521665,
+      "loss": 3.3396,
+      "step": 3023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005444666568485774,
+      "loss": 3.1478,
+      "step": 3024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005442065390454296,
+      "loss": 3.1821,
+      "step": 3025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005439464091831795,
+      "loss": 3.0375,
+      "step": 3026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005436862673327877,
+      "loss": 3.3832,
+      "step": 3027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000543426113565218,
+      "loss": 3.2798,
+      "step": 3028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005431659479514383,
+      "loss": 3.219,
+      "step": 3029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005429057705624191,
+      "loss": 3.2747,
+      "step": 3030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005426455814691337,
+      "loss": 3.165,
+      "step": 3031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423853807425596,
+      "loss": 3.0539,
+      "step": 3032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005421251684536769,
+      "loss": 3.2583,
+      "step": 3033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005418649446734684,
+      "loss": 3.1761,
+      "step": 3034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005416047094729214,
+      "loss": 3.2972,
+      "step": 3035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005413444629230248,
+      "loss": 3.1143,
+      "step": 3036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005410842050947714,
+      "loss": 3.1442,
+      "step": 3037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005408239360591572,
+      "loss": 3.3191,
+      "step": 3038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005405636558871808,
+      "loss": 3.3667,
+      "step": 3039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005403033646498439,
+      "loss": 3.2234,
+      "step": 3040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005400430624181516,
+      "loss": 3.1733,
+      "step": 3041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005397827492631116,
+      "loss": 3.104,
+      "step": 3042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005395224252557347,
+      "loss": 3.1591,
+      "step": 3043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005392620904670349,
+      "loss": 3.0896,
+      "step": 3044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005390017449680287,
+      "loss": 3.3104,
+      "step": 3045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005387413888297359,
+      "loss": 3.2191,
+      "step": 3046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005384810221231789,
+      "loss": 3.2861,
+      "step": 3047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005382206449193833,
+      "loss": 3.0633,
+      "step": 3048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005379602572893774,
+      "loss": 3.2495,
+      "step": 3049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005376998593041921,
+      "loss": 3.2491,
+      "step": 3050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005374394510348617,
+      "loss": 3.1534,
+      "step": 3051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005371790325524228,
+      "loss": 3.2734,
+      "step": 3052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000536918603927915,
+      "loss": 3.2062,
+      "step": 3053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005366581652323807,
+      "loss": 3.2644,
+      "step": 3054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005363977165368649,
+      "loss": 3.188,
+      "step": 3055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005361372579124152,
+      "loss": 3.1875,
+      "step": 3056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005358767894300824,
+      "loss": 3.303,
+      "step": 3057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005356163111609198,
+      "loss": 3.1102,
+      "step": 3058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005353558231759828,
+      "loss": 3.2855,
+      "step": 3059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005350953255463304,
+      "loss": 3.2437,
+      "step": 3060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005348348183430236,
+      "loss": 3.3033,
+      "step": 3061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005345743016371261,
+      "loss": 3.2723,
+      "step": 3062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005343137754997042,
+      "loss": 3.1542,
+      "step": 3063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005340532400018269,
+      "loss": 3.2267,
+      "step": 3064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005337926952145658,
+      "loss": 3.3288,
+      "step": 3065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005335321412089947,
+      "loss": 3.2769,
+      "step": 3066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005332715780561904,
+      "loss": 3.1983,
+      "step": 3067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005330110058272316,
+      "loss": 3.356,
+      "step": 3068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005327504245932,
+      "loss": 3.3256,
+      "step": 3069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005324898344251796,
+      "loss": 3.2436,
+      "step": 3070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005322292353942568,
+      "loss": 3.1418,
+      "step": 3071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005319686275715202,
+      "loss": 3.3643,
+      "step": 3072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005317080110280613,
+      "loss": 3.0293,
+      "step": 3073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005314473858349733,
+      "loss": 3.2588,
+      "step": 3074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005311867520633526,
+      "loss": 3.2127,
+      "step": 3075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005309261097842974,
+      "loss": 3.1018,
+      "step": 3076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005306654590689079,
+      "loss": 3.3698,
+      "step": 3077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005304047999882874,
+      "loss": 3.2209,
+      "step": 3078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005301441326135412,
+      "loss": 3.3358,
+      "step": 3079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005298834570157763,
+      "loss": 3.1811,
+      "step": 3080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005296227732661028,
+      "loss": 3.3593,
+      "step": 3081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005293620814356326,
+      "loss": 3.1968,
+      "step": 3082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005291013815954795,
+      "loss": 3.2258,
+      "step": 3083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005288406738167601,
+      "loss": 3.3108,
+      "step": 3084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000528579958170593,
+      "loss": 3.1465,
+      "step": 3085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005283192347280984,
+      "loss": 3.1123,
+      "step": 3086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005280585035603993,
+      "loss": 3.3332,
+      "step": 3087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005277977647386207,
+      "loss": 3.1474,
+      "step": 3088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005275370183338893,
+      "loss": 3.1415,
+      "step": 3089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005272762644173342,
+      "loss": 3.1128,
+      "step": 3090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005270155030600866,
+      "loss": 3.105,
+      "step": 3091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005267547343332793,
+      "loss": 3.1629,
+      "step": 3092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005264939583080477,
+      "loss": 3.2143,
+      "step": 3093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005262331750555287,
+      "loss": 3.2167,
+      "step": 3094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005259723846468615,
+      "loss": 3.3578,
+      "step": 3095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005257115871531873,
+      "loss": 3.2177,
+      "step": 3096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005254507826456483,
+      "loss": 3.1446,
+      "step": 3097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005251899711953903,
+      "loss": 3.2223,
+      "step": 3098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005249291528735594,
+      "loss": 3.1324,
+      "step": 3099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005246683277513044,
+      "loss": 3.2095,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005244074958997762,
+      "loss": 3.2255,
+      "step": 3101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005241466573901267,
+      "loss": 3.2232,
+      "step": 3102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005238858122935099,
+      "loss": 3.2391,
+      "step": 3103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005236249606810821,
+      "loss": 3.1954,
+      "step": 3104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000523364102624001,
+      "loss": 3.3927,
+      "step": 3105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005231032381934256,
+      "loss": 3.2091,
+      "step": 3106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005228423674605179,
+      "loss": 3.1559,
+      "step": 3107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005225814904964402,
+      "loss": 3.2105,
+      "step": 3108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005223206073723573,
+      "loss": 3.3645,
+      "step": 3109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005220597181594356,
+      "loss": 3.1628,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005217988229288431,
+      "loss": 3.1492,
+      "step": 3111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005215379217517492,
+      "loss": 3.1273,
+      "step": 3112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005212770146993254,
+      "loss": 3.1202,
+      "step": 3113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005210161018427443,
+      "loss": 3.1059,
+      "step": 3114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005207551832531803,
+      "loss": 3.2784,
+      "step": 3115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005204942590018098,
+      "loss": 3.3378,
+      "step": 3116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005202333291598099,
+      "loss": 3.1373,
+      "step": 3117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005199723937983599,
+      "loss": 3.1663,
+      "step": 3118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005197114529886402,
+      "loss": 3.3453,
+      "step": 3119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005194505068018328,
+      "loss": 3.0295,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005191895553091216,
+      "loss": 3.4432,
+      "step": 3121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000518928598581691,
+      "loss": 3.2766,
+      "step": 3122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005186676366907278,
+      "loss": 3.105,
+      "step": 3123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005184066697074197,
+      "loss": 3.1081,
+      "step": 3124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000518145697702956,
+      "loss": 3.3497,
+      "step": 3125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005178847207485268,
+      "loss": 3.2287,
+      "step": 3126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005176237389153246,
+      "loss": 3.159,
+      "step": 3127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005173627522745422,
+      "loss": 3.2109,
+      "step": 3128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005171017608973744,
+      "loss": 3.1443,
+      "step": 3129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000516840764855017,
+      "loss": 3.2476,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005165797642186671,
+      "loss": 3.0899,
+      "step": 3131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005163187590595229,
+      "loss": 3.2179,
+      "step": 3132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005160577494487843,
+      "loss": 3.2804,
+      "step": 3133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005157967354576518,
+      "loss": 3.3433,
+      "step": 3134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005155357171573276,
+      "loss": 3.3916,
+      "step": 3135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000515274694619015,
+      "loss": 3.2238,
+      "step": 3136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000515013667913918,
+      "loss": 3.1708,
+      "step": 3137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005147526371132424,
+      "loss": 3.1294,
+      "step": 3138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005144916022881949,
+      "loss": 3.2582,
+      "step": 3139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142305635099826,
+      "loss": 3.2925,
+      "step": 3140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005139695208498147,
+      "loss": 3.245,
+      "step": 3141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005137084743789013,
+      "loss": 3.1815,
+      "step": 3142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005134474241684526,
+      "loss": 3.3845,
+      "step": 3143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005131863702896809,
+      "loss": 3.1579,
+      "step": 3144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005129253128137992,
+      "loss": 3.1075,
+      "step": 3145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000512664251812021,
+      "loss": 3.1062,
+      "step": 3146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005124031873555613,
+      "loss": 3.1656,
+      "step": 3147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005121421195156361,
+      "loss": 3.0344,
+      "step": 3148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005118810483634619,
+      "loss": 3.2101,
+      "step": 3149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005116199739702563,
+      "loss": 3.24,
+      "step": 3150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005113588964072378,
+      "loss": 3.1664,
+      "step": 3151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005110978157456257,
+      "loss": 3.3337,
+      "step": 3152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005108367320566403,
+      "loss": 3.2461,
+      "step": 3153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005105756454115027,
+      "loss": 3.0171,
+      "step": 3154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005103145558814344,
+      "loss": 3.1236,
+      "step": 3155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005100534635376583,
+      "loss": 3.1632,
+      "step": 3156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005097923684513979,
+      "loss": 3.1155,
+      "step": 3157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005095312706938768,
+      "loss": 3.1794,
+      "step": 3158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005092701703363205,
+      "loss": 3.209,
+      "step": 3159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005090090674499542,
+      "loss": 3.1605,
+      "step": 3160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005087479621060042,
+      "loss": 3.1132,
+      "step": 3161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005084868543756974,
+      "loss": 3.2622,
+      "step": 3162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005082257443302615,
+      "loss": 3.0867,
+      "step": 3163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005079646320409249,
+      "loss": 3.05,
+      "step": 3164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000507703517578916,
+      "loss": 3.3022,
+      "step": 3165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005074424010154643,
+      "loss": 3.3712,
+      "step": 3166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005071812824218001,
+      "loss": 3.2997,
+      "step": 3167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005069201618691537,
+      "loss": 3.2909,
+      "step": 3168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005066590394287562,
+      "loss": 3.3085,
+      "step": 3169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005063979151718393,
+      "loss": 3.1279,
+      "step": 3170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000506136789169635,
+      "loss": 3.277,
+      "step": 3171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005058756614933757,
+      "loss": 3.3897,
+      "step": 3172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005056145322142948,
+      "loss": 3.1298,
+      "step": 3173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005053534014036256,
+      "loss": 3.197,
+      "step": 3174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005050922691326017,
+      "loss": 3.2081,
+      "step": 3175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005048311354724578,
+      "loss": 3.2495,
+      "step": 3176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005045700004944283,
+      "loss": 3.273,
+      "step": 3177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005043088642697482,
+      "loss": 3.2238,
+      "step": 3178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005040477268696532,
+      "loss": 3.1562,
+      "step": 3179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005037865883653785,
+      "loss": 3.342,
+      "step": 3180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005035254488281604,
+      "loss": 3.3034,
+      "step": 3181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005032643083292351,
+      "loss": 3.2256,
+      "step": 3182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005030031669398389,
+      "loss": 3.2956,
+      "step": 3183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005027420247312089,
+      "loss": 3.1793,
+      "step": 3184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000502480881774582,
+      "loss": 3.1414,
+      "step": 3185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005022197381411951,
+      "loss": 3.1477,
+      "step": 3186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005019585939022859,
+      "loss": 3.2471,
+      "step": 3187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005016974491290918,
+      "loss": 3.1409,
+      "step": 3188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005014363038928506,
+      "loss": 3.3576,
+      "step": 3189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005011751582648,
+      "loss": 3.0862,
+      "step": 3190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000500914012316178,
+      "loss": 3.2099,
+      "step": 3191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005006528661182225,
+      "loss": 3.2117,
+      "step": 3192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005003917197421717,
+      "loss": 3.1058,
+      "step": 3193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005001305732592636,
+      "loss": 3.3008,
+      "step": 3194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004998694267407364,
+      "loss": 3.1614,
+      "step": 3195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004996082802578285,
+      "loss": 3.2821,
+      "step": 3196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004993471338817775,
+      "loss": 3.2097,
+      "step": 3197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499085987683822,
+      "loss": 3.3142,
+      "step": 3198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004988248417352,
+      "loss": 3.237,
+      "step": 3199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004985636961071495,
+      "loss": 2.9857,
+      "step": 3200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004983025508709082,
+      "loss": 3.1696,
+      "step": 3201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004980414060977142,
+      "loss": 3.2288,
+      "step": 3202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004977802618588049,
+      "loss": 3.1491,
+      "step": 3203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497519118225418,
+      "loss": 3.1794,
+      "step": 3204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497257975268791,
+      "loss": 3.365,
+      "step": 3205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004969968330601611,
+      "loss": 3.1326,
+      "step": 3206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000496735691670765,
+      "loss": 3.2382,
+      "step": 3207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004964745511718397,
+      "loss": 3.1295,
+      "step": 3208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004962134116346215,
+      "loss": 3.1542,
+      "step": 3209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000495952273130347,
+      "loss": 3.1851,
+      "step": 3210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004956911357302517,
+      "loss": 3.2107,
+      "step": 3211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004954299995055719,
+      "loss": 3.1284,
+      "step": 3212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004951688645275423,
+      "loss": 3.0539,
+      "step": 3213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004949077308673984,
+      "loss": 3.0944,
+      "step": 3214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004946465985963746,
+      "loss": 3.2093,
+      "step": 3215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004943854677857054,
+      "loss": 3.1784,
+      "step": 3216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004941243385066243,
+      "loss": 3.186,
+      "step": 3217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004938632108303651,
+      "loss": 3.1289,
+      "step": 3218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004936020848281609,
+      "loss": 3.252,
+      "step": 3219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000493340960571244,
+      "loss": 3.3504,
+      "step": 3220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004930798381308464,
+      "loss": 3.3246,
+      "step": 3221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004928187175782001,
+      "loss": 3.3063,
+      "step": 3222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004925575989845356,
+      "loss": 3.2332,
+      "step": 3223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000492296482421084,
+      "loss": 3.1433,
+      "step": 3224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004920353679590753,
+      "loss": 3.2586,
+      "step": 3225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004917742556697385,
+      "loss": 3.163,
+      "step": 3226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004915131456243026,
+      "loss": 3.1047,
+      "step": 3227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000491252037893996,
+      "loss": 3.2494,
+      "step": 3228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004909909325500459,
+      "loss": 3.1794,
+      "step": 3229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004907298296636795,
+      "loss": 3.0601,
+      "step": 3230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004904687293061232,
+      "loss": 3.3469,
+      "step": 3231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004902076315486023,
+      "loss": 3.2351,
+      "step": 3232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004899465364623418,
+      "loss": 3.3783,
+      "step": 3233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004896854441185658,
+      "loss": 3.1779,
+      "step": 3234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004894243545884976,
+      "loss": 3.1827,
+      "step": 3235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004891632679433597,
+      "loss": 3.1433,
+      "step": 3236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004889021842543744,
+      "loss": 3.2206,
+      "step": 3237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004886411035927622,
+      "loss": 3.2492,
+      "step": 3238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004883800260297439,
+      "loss": 3.3205,
+      "step": 3239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004881189516365383,
+      "loss": 3.1091,
+      "step": 3240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048785788048436404,
+      "loss": 3.2434,
+      "step": 3241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048759681264443864,
+      "loss": 3.2045,
+      "step": 3242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004873357481879791,
+      "loss": 3.1103,
+      "step": 3243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048707468718620095,
+      "loss": 3.2026,
+      "step": 3244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004868136297103192,
+      "loss": 3.1947,
+      "step": 3245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048655257583154755,
+      "loss": 3.2089,
+      "step": 3246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000486291525621099,
+      "loss": 3.1493,
+      "step": 3247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004860304791501852,
+      "loss": 3.1183,
+      "step": 3248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048576943649001746,
+      "loss": 3.1506,
+      "step": 3249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048550839771180533,
+      "loss": 3.209,
+      "step": 3250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048524736288675766,
+      "loss": 3.0394,
+      "step": 3251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000484986332086082,
+      "loss": 3.1961,
+      "step": 3252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004847253053809852,
+      "loss": 3.0844,
+      "step": 3253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048446428284267234,
+      "loss": 3.2307,
+      "step": 3254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004842032645423483,
+      "loss": 3.0909,
+      "step": 3255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004839422505512158,
+      "loss": 3.2183,
+      "step": 3256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004836812409404772,
+      "loss": 3.2456,
+      "step": 3257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004834202357813331,
+      "loss": 3.1626,
+      "step": 3258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048315923514498317,
+      "loss": 2.9725,
+      "step": 3259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004828982391026258,
+      "loss": 3.2855,
+      "step": 3260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004826372477254579,
+      "loss": 3.2049,
+      "step": 3261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004823762610846755,
+      "loss": 3.1008,
+      "step": 3262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048211527925147326,
+      "loss": 3.1315,
+      "step": 3263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004818543022970442,
+      "loss": 3.228,
+      "step": 3264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004815933302925804,
+      "loss": 3.3829,
+      "step": 3265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048133236330927227,
+      "loss": 3.1658,
+      "step": 3266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048107140141830893,
+      "loss": 3.0786,
+      "step": 3267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048081044469087855,
+      "loss": 3.2617,
+      "step": 3268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004805494931981672,
+      "loss": 3.2626,
+      "step": 3269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048028854701135987,
+      "loss": 3.2426,
+      "step": 3270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048002760620164027,
+      "loss": 3.3305,
+      "step": 3271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047976667084019016,
+      "loss": 3.1159,
+      "step": 3272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047950574099819013,
+      "loss": 3.3174,
+      "step": 3273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047924481674681957,
+      "loss": 3.2665,
+      "step": 3274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004789838981572558,
+      "loss": 3.153,
+      "step": 3275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004787229853006747,
+      "loss": 3.1022,
+      "step": 3276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004784620782482509,
+      "loss": 3.1786,
+      "step": 3277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047820117707115704,
+      "loss": 2.9901,
+      "step": 3278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047794028184056434,
+      "loss": 3.2805,
+      "step": 3279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047767939262764264,
+      "loss": 3.2803,
+      "step": 3280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004774185095035599,
+      "loss": 3.3162,
+      "step": 3281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004771576325394822,
+      "loss": 3.2557,
+      "step": 3282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047689676180657435,
+      "loss": 3.1711,
+      "step": 3283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047663589737599924,
+      "loss": 3.1379,
+      "step": 3284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004763750393189181,
+      "loss": 3.2339,
+      "step": 3285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047611418770649014,
+      "loss": 3.2799,
+      "step": 3286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047585334260987345,
+      "loss": 3.3689,
+      "step": 3287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004755925041002239,
+      "loss": 3.2719,
+      "step": 3288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047533167224869563,
+      "loss": 3.2448,
+      "step": 3289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047507084712644074,
+      "loss": 3.1812,
+      "step": 3290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047481002880461,
+      "loss": 3.1863,
+      "step": 3291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004745492173543517,
+      "loss": 3.3076,
+      "step": 3292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004742884128468129,
+      "loss": 3.3589,
+      "step": 3293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047402761535313853,
+      "loss": 3.1985,
+      "step": 3294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004737668249444714,
+      "loss": 3.1538,
+      "step": 3295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047350604169195236,
+      "loss": 3.3158,
+      "step": 3296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047324526566672084,
+      "loss": 3.2434,
+      "step": 3297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047298449693991345,
+      "loss": 3.1582,
+      "step": 3298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004727237355826657,
+      "loss": 3.227,
+      "step": 3299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047246298166611075,
+      "loss": 3.1504,
+      "step": 3300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047220223526137935,
+      "loss": 3.112,
+      "step": 3301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004719414964396007,
+      "loss": 3.3051,
+      "step": 3302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047168076527190177,
+      "loss": 3.1122,
+      "step": 3303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047142004182940714,
+      "loss": 3.2138,
+      "step": 3304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047115932618323984,
+      "loss": 3.1657,
+      "step": 3305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047089861840452055,
+      "loss": 3.1503,
+      "step": 3306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004706379185643675,
+      "loss": 3.3249,
+      "step": 3307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047037722673389724,
+      "loss": 3.0006,
+      "step": 3308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047011654298422385,
+      "loss": 3.2079,
+      "step": 3309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000469855867386459,
+      "loss": 3.2883,
+      "step": 3310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004695952000117125,
+      "loss": 3.0878,
+      "step": 3311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004693345409310922,
+      "loss": 3.2655,
+      "step": 3312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004690738902157028,
+      "loss": 3.3669,
+      "step": 3313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004688132479366475,
+      "loss": 3.0848,
+      "step": 3314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004685526141650267,
+      "loss": 3.0766,
+      "step": 3315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004682919889719389,
+      "loss": 3.1671,
+      "step": 3316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004680313724284798,
+      "loss": 3.1254,
+      "step": 3317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046777076460574336,
+      "loss": 3.1751,
+      "step": 3318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004675101655748204,
+      "loss": 3.2538,
+      "step": 3319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004672495754068001,
+      "loss": 3.1657,
+      "step": 3320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046698899417276846,
+      "loss": 3.3485,
+      "step": 3321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046672842194380983,
+      "loss": 3.251,
+      "step": 3322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004664678587910053,
+      "loss": 3.2865,
+      "step": 3323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004662073047854343,
+      "loss": 3.2871,
+      "step": 3324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046594675999817313,
+      "loss": 3.3369,
+      "step": 3325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046568622450029594,
+      "loss": 3.1117,
+      "step": 3326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004654256983628741,
+      "loss": 3.2471,
+      "step": 3327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004651651816569766,
+      "loss": 3.2726,
+      "step": 3328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004649046744536696,
+      "loss": 3.2909,
+      "step": 3329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046464417682401724,
+      "loss": 3.2507,
+      "step": 3330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046438368883908036,
+      "loss": 3.2136,
+      "step": 3331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046412321056991766,
+      "loss": 3.2698,
+      "step": 3332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004638627420875849,
+      "loss": 3.2456,
+      "step": 3333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046360228346313537,
+      "loss": 3.1461,
+      "step": 3334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004633418347676195,
+      "loss": 3.2197,
+      "step": 3335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004630813960720849,
+      "loss": 3.2401,
+      "step": 3336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046282096744757725,
+      "loss": 2.9968,
+      "step": 3337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004625605489651384,
+      "loss": 3.1799,
+      "step": 3338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046230014069580794,
+      "loss": 3.1964,
+      "step": 3339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004620397427106229,
+      "loss": 3.3756,
+      "step": 3340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046177935508061687,
+      "loss": 3.2419,
+      "step": 3341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004615189778768211,
+      "loss": 3.3482,
+      "step": 3342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004612586111702642,
+      "loss": 3.2245,
+      "step": 3343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046099825503197146,
+      "loss": 3.2362,
+      "step": 3344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046073790953296524,
+      "loss": 3.0774,
+      "step": 3345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046047757474426546,
+      "loss": 3.0815,
+      "step": 3346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046021725073688857,
+      "loss": 3.2171,
+      "step": 3347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045995693758184845,
+      "loss": 3.2208,
+      "step": 3348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045969663535015616,
+      "loss": 3.1965,
+      "step": 3349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045943634411281926,
+      "loss": 3.1778,
+      "step": 3350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004591760639408429,
+      "loss": 3.3007,
+      "step": 3351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045891579490522867,
+      "loss": 3.0539,
+      "step": 3352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004586555370769753,
+      "loss": 3.2757,
+      "step": 3353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045839529052707864,
+      "loss": 3.2285,
+      "step": 3354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004581350553265315,
+      "loss": 3.2487,
+      "step": 3355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004578748315463232,
+      "loss": 3.2526,
+      "step": 3356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004576146192574405,
+      "loss": 3.1446,
+      "step": 3357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004573544185308664,
+      "loss": 3.1207,
+      "step": 3358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004570942294375811,
+      "loss": 3.1406,
+      "step": 3359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045683405204856184,
+      "loss": 3.0804,
+      "step": 3360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004565738864347819,
+      "loss": 3.2368,
+      "step": 3361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045631373266721246,
+      "loss": 3.1443,
+      "step": 3362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045605359081682064,
+      "loss": 3.2279,
+      "step": 3363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004557934609545705,
+      "loss": 3.1636,
+      "step": 3364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045553334315142274,
+      "loss": 3.061,
+      "step": 3365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045527323747833514,
+      "loss": 3.1988,
+      "step": 3366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004550131440062617,
+      "loss": 3.2118,
+      "step": 3367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045475306280615347,
+      "loss": 3.173,
+      "step": 3368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045449299394895816,
+      "loss": 3.0572,
+      "step": 3369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004542329375056197,
+      "loss": 3.0886,
+      "step": 3370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000453972893547079,
+      "loss": 3.2411,
+      "step": 3371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045371286214427345,
+      "loss": 3.3076,
+      "step": 3372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004534528433681368,
+      "loss": 3.3572,
+      "step": 3373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045319283728959984,
+      "loss": 3.2711,
+      "step": 3374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004529328439795898,
+      "loss": 3.2829,
+      "step": 3375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004526728635090298,
+      "loss": 3.2396,
+      "step": 3376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004524128959488403,
+      "loss": 3.318,
+      "step": 3377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004521529413699377,
+      "loss": 3.3001,
+      "step": 3378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045189299984323486,
+      "loss": 3.1894,
+      "step": 3379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004516330714396414,
+      "loss": 3.2905,
+      "step": 3380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045137315623006337,
+      "loss": 3.1209,
+      "step": 3381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004511132542854027,
+      "loss": 3.3547,
+      "step": 3382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045085336567655834,
+      "loss": 3.2633,
+      "step": 3383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000450593490474425,
+      "loss": 3.1031,
+      "step": 3384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045033362874989435,
+      "loss": 3.1281,
+      "step": 3385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045007378057385376,
+      "loss": 3.2569,
+      "step": 3386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004498139460171876,
+      "loss": 3.1097,
+      "step": 3387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004495541251507758,
+      "loss": 3.2054,
+      "step": 3388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044929431804549514,
+      "loss": 3.1308,
+      "step": 3389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044903452477221826,
+      "loss": 2.993,
+      "step": 3390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044877474540181426,
+      "loss": 3.2576,
+      "step": 3391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044851498000514806,
+      "loss": 3.1196,
+      "step": 3392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004482552286530816,
+      "loss": 3.2188,
+      "step": 3393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044799549141647205,
+      "loss": 3.2312,
+      "step": 3394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044773576836617336,
+      "loss": 3.0987,
+      "step": 3395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004474760595730352,
+      "loss": 3.0337,
+      "step": 3396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044721636510790366,
+      "loss": 3.3296,
+      "step": 3397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004469566850416206,
+      "loss": 3.1772,
+      "step": 3398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044669701944502444,
+      "loss": 3.1926,
+      "step": 3399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004464373683889492,
+      "loss": 3.2088,
+      "step": 3400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004461777319442252,
+      "loss": 3.148,
+      "step": 3401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004459181101816785,
+      "loss": 3.1503,
+      "step": 3402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044565850317213155,
+      "loss": 3.0292,
+      "step": 3403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044539891098640215,
+      "loss": 3.2398,
+      "step": 3404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044513933369530484,
+      "loss": 3.1354,
+      "step": 3405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044487977136964976,
+      "loss": 3.2474,
+      "step": 3406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004446202240802427,
+      "loss": 3.2825,
+      "step": 3407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044436069189788554,
+      "loss": 3.2114,
+      "step": 3408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044410117489337633,
+      "loss": 3.1577,
+      "step": 3409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044384167313750835,
+      "loss": 3.1737,
+      "step": 3410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004435821867010712,
+      "loss": 3.1093,
+      "step": 3411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004433227156548505,
+      "loss": 3.207,
+      "step": 3412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004430632600696272,
+      "loss": 3.2113,
+      "step": 3413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004428038200161779,
+      "loss": 3.0631,
+      "step": 3414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044254439556527564,
+      "loss": 3.2515,
+      "step": 3415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004422849867876886,
+      "loss": 3.2185,
+      "step": 3416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004420255937541808,
+      "loss": 3.1429,
+      "step": 3417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044176621653551246,
+      "loss": 3.4016,
+      "step": 3418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044150685520243885,
+      "loss": 3.1982,
+      "step": 3419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044124750982571113,
+      "loss": 3.273,
+      "step": 3420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004409881804760763,
+      "loss": 3.1384,
+      "step": 3421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004407288672242766,
+      "loss": 3.1849,
+      "step": 3422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044046957014105007,
+      "loss": 3.0956,
+      "step": 3423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044021028929713065,
+      "loss": 3.4236,
+      "step": 3424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043995102476324734,
+      "loss": 3.1925,
+      "step": 3425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043969177661012503,
+      "loss": 3.1946,
+      "step": 3426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000439432544908484,
+      "loss": 3.2926,
+      "step": 3427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043917332972904,
+      "loss": 3.1886,
+      "step": 3428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043891413114250424,
+      "loss": 3.2028,
+      "step": 3429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004386549492195838,
+      "loss": 3.2794,
+      "step": 3430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043839578403098077,
+      "loss": 3.2413,
+      "step": 3431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004381366356473929,
+      "loss": 3.2096,
+      "step": 3432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004378775041395132,
+      "loss": 3.2268,
+      "step": 3433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004376183895780301,
+      "loss": 3.095,
+      "step": 3434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043735929203362765,
+      "loss": 3.1742,
+      "step": 3435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043710021157698477,
+      "loss": 3.2431,
+      "step": 3436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004368411482787763,
+      "loss": 3.1533,
+      "step": 3437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043658210220967207,
+      "loss": 3.477,
+      "step": 3438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004363230734403372,
+      "loss": 3.2855,
+      "step": 3439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043606406204143227,
+      "loss": 3.2613,
+      "step": 3440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043580506808361296,
+      "loss": 3.1322,
+      "step": 3441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004355460916375299,
+      "loss": 3.2311,
+      "step": 3442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004352871327738297,
+      "loss": 3.2778,
+      "step": 3443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043502819156315375,
+      "loss": 3.1577,
+      "step": 3444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004347692680761384,
+      "loss": 3.2708,
+      "step": 3445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004345103623834156,
+      "loss": 3.2704,
+      "step": 3446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043425147455561206,
+      "loss": 3.1493,
+      "step": 3447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043399260466334973,
+      "loss": 3.2697,
+      "step": 3448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004337337527772459,
+      "loss": 3.1715,
+      "step": 3449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043347491896791293,
+      "loss": 3.2397,
+      "step": 3450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004332161033059578,
+      "loss": 3.1042,
+      "step": 3451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000432957305861983,
+      "loss": 3.2512,
+      "step": 3452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043269852670658576,
+      "loss": 3.2441,
+      "step": 3453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004324397659103586,
+      "loss": 3.0104,
+      "step": 3454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043218102354388875,
+      "loss": 3.2199,
+      "step": 3455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043192229967775874,
+      "loss": 3.3412,
+      "step": 3456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004316635943825456,
+      "loss": 3.1961,
+      "step": 3457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004314049077288218,
+      "loss": 3.0986,
+      "step": 3458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043114623978715425,
+      "loss": 3.0902,
+      "step": 3459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043088759062810515,
+      "loss": 3.2653,
+      "step": 3460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004306289603222312,
+      "loss": 2.9927,
+      "step": 3461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004303703489400845,
+      "loss": 3.2053,
+      "step": 3462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043011175655221134,
+      "loss": 3.1761,
+      "step": 3463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004298531832291535,
+      "loss": 3.33,
+      "step": 3464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004295946290414467,
+      "loss": 3.1031,
+      "step": 3465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004293360940596224,
+      "loss": 3.2573,
+      "step": 3466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042907757835420593,
+      "loss": 3.1434,
+      "step": 3467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042881908199571825,
+      "loss": 3.2677,
+      "step": 3468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004285606050546742,
+      "loss": 3.1098,
+      "step": 3469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004283021476015841,
+      "loss": 3.1216,
+      "step": 3470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004280437097069522,
+      "loss": 3.1876,
+      "step": 3471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000427785291441278,
+      "loss": 3.3451,
+      "step": 3472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004275268928750551,
+      "loss": 3.2029,
+      "step": 3473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004272685140787724,
+      "loss": 3.254,
+      "step": 3474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004270101551229131,
+      "loss": 3.1195,
+      "step": 3475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004267518160779549,
+      "loss": 3.2315,
+      "step": 3476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004264934970143697,
+      "loss": 3.1434,
+      "step": 3477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004262351980026249,
+      "loss": 3.0753,
+      "step": 3478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042597691911318134,
+      "loss": 3.1686,
+      "step": 3479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004257186604164955,
+      "loss": 3.2603,
+      "step": 3480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004254604219830175,
+      "loss": 3.2399,
+      "step": 3481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042520220388319224,
+      "loss": 3.1753,
+      "step": 3482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042494400618745893,
+      "loss": 3.0867,
+      "step": 3483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004246858289662515,
+      "loss": 3.2331,
+      "step": 3484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042442767228999804,
+      "loss": 3.3531,
+      "step": 3485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042416953622912085,
+      "loss": 3.2081,
+      "step": 3486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004239114208540374,
+      "loss": 3.0914,
+      "step": 3487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042365332623515864,
+      "loss": 3.3112,
+      "step": 3488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042339525244289034,
+      "loss": 3.1849,
+      "step": 3489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004231371995476324,
+      "loss": 3.21,
+      "step": 3490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000422879167619779,
+      "loss": 3.3596,
+      "step": 3491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004226211567297186,
+      "loss": 3.0654,
+      "step": 3492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042236316694783434,
+      "loss": 3.2633,
+      "step": 3493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042210519834450297,
+      "loss": 3.1809,
+      "step": 3494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042184725099009594,
+      "loss": 3.2025,
+      "step": 3495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004215893249549786,
+      "loss": 3.285,
+      "step": 3496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042133142030951044,
+      "loss": 3.2173,
+      "step": 3497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004210735371240454,
+      "loss": 3.3562,
+      "step": 3498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042081567546893166,
+      "loss": 3.2351,
+      "step": 3499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000420557835414511,
+      "loss": 3.103,
+      "step": 3500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004203000170311199,
+      "loss": 3.204,
+      "step": 3501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042004222038908846,
+      "loss": 3.0736,
+      "step": 3502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004197844455587409,
+      "loss": 3.1707,
+      "step": 3503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004195266926103959,
+      "loss": 3.2559,
+      "step": 3504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004192689616143655,
+      "loss": 3.1687,
+      "step": 3505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004190112526409565,
+      "loss": 3.216,
+      "step": 3506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004187535657604694,
+      "loss": 3.3032,
+      "step": 3507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041849590104319834,
+      "loss": 3.155,
+      "step": 3508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004182382585594318,
+      "loss": 3.1065,
+      "step": 3509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041798063837945214,
+      "loss": 3.2242,
+      "step": 3510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041772304057353525,
+      "loss": 3.357,
+      "step": 3511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004174654652119516,
+      "loss": 3.1096,
+      "step": 3512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004172079123649652,
+      "loss": 3.1568,
+      "step": 3513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004169503821028336,
+      "loss": 3.0949,
+      "step": 3514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004166928744958088,
+      "loss": 3.0884,
+      "step": 3515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004164353896141361,
+      "loss": 3.2861,
+      "step": 3516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041617792752805463,
+      "loss": 3.0788,
+      "step": 3517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004159204883077978,
+      "loss": 3.1022,
+      "step": 3518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004156630720235925,
+      "loss": 3.1972,
+      "step": 3519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041540567874565905,
+      "loss": 3.0318,
+      "step": 3520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041514830854421205,
+      "loss": 3.191,
+      "step": 3521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041489096148945913,
+      "loss": 3.1712,
+      "step": 3522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004146336376516024,
+      "loss": 3.1524,
+      "step": 3523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000414376337100837,
+      "loss": 3.3731,
+      "step": 3524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004141190599073521,
+      "loss": 3.3135,
+      "step": 3525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004138618061413302,
+      "loss": 3.2335,
+      "step": 3526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004136045758729477,
+      "loss": 3.1564,
+      "step": 3527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041334736917237426,
+      "loss": 3.1694,
+      "step": 3528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041309018610977355,
+      "loss": 3.1674,
+      "step": 3529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004128330267553022,
+      "loss": 3.1642,
+      "step": 3530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004125758911791112,
+      "loss": 3.164,
+      "step": 3531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041231877945134413,
+      "loss": 2.9476,
+      "step": 3532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004120616916421387,
+      "loss": 3.1589,
+      "step": 3533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041180462782162584,
+      "loss": 3.2706,
+      "step": 3534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004115475880599301,
+      "loss": 3.2866,
+      "step": 3535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041129057242716906,
+      "loss": 3.1801,
+      "step": 3536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041103358099345446,
+      "loss": 3.2429,
+      "step": 3537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041077661382889067,
+      "loss": 3.162,
+      "step": 3538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000410519671003576,
+      "loss": 3.1663,
+      "step": 3539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004102627525876016,
+      "loss": 3.2447,
+      "step": 3540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041000585865105257,
+      "loss": 3.0776,
+      "step": 3541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004097489892640066,
+      "loss": 3.3406,
+      "step": 3542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004094921444965354,
+      "loss": 3.1378,
+      "step": 3543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004092353244187038,
+      "loss": 3.0408,
+      "step": 3544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004089785291005695,
+      "loss": 3.0257,
+      "step": 3545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004087217586121836,
+      "loss": 3.1976,
+      "step": 3546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004084650130235909,
+      "loss": 3.2189,
+      "step": 3547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040820829240482856,
+      "loss": 3.2109,
+      "step": 3548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040795159682592775,
+      "loss": 3.1772,
+      "step": 3549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040769492635691246,
+      "loss": 3.2018,
+      "step": 3550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040743828106779977,
+      "loss": 3.3782,
+      "step": 3551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040718166102859975,
+      "loss": 3.0642,
+      "step": 3552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040692506630931615,
+      "loss": 3.1237,
+      "step": 3553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040666849697994505,
+      "loss": 3.0363,
+      "step": 3554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004064119531104762,
+      "loss": 3.182,
+      "step": 3555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040615543477089236,
+      "loss": 3.2085,
+      "step": 3556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004058989420311689,
+      "loss": 3.1791,
+      "step": 3557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004056424749612747,
+      "loss": 3.2958,
+      "step": 3558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004053860336311714,
+      "loss": 3.2603,
+      "step": 3559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004051296181108134,
+      "loss": 3.1953,
+      "step": 3560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004048732284701483,
+      "loss": 3.2972,
+      "step": 3561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004046168647791171,
+      "loss": 3.1484,
+      "step": 3562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004043605271076527,
+      "loss": 3.24,
+      "step": 3563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004041042155256819,
+      "loss": 3.1752,
+      "step": 3564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040384793010312363,
+      "loss": 3.2012,
+      "step": 3565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040359167090989005,
+      "loss": 3.3491,
+      "step": 3566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040333543801588615,
+      "loss": 3.0724,
+      "step": 3567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040307923149100977,
+      "loss": 3.06,
+      "step": 3568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004028230514051514,
+      "loss": 3.3532,
+      "step": 3569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040256689782819453,
+      "loss": 3.2151,
+      "step": 3570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004023107708300153,
+      "loss": 3.1697,
+      "step": 3571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004020546704804823,
+      "loss": 3.1717,
+      "step": 3572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004017985968494573,
+      "loss": 3.042,
+      "step": 3573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040154255000679484,
+      "loss": 3.146,
+      "step": 3574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040128653002234176,
+      "loss": 3.4314,
+      "step": 3575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004010305369659379,
+      "loss": 3.2727,
+      "step": 3576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040077457090741537,
+      "loss": 3.1477,
+      "step": 3577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040051863191659944,
+      "loss": 3.1486,
+      "step": 3578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040026272006330745,
+      "loss": 3.2335,
+      "step": 3579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040000683541734956,
+      "loss": 3.2524,
+      "step": 3580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039975097804852876,
+      "loss": 3.2035,
+      "step": 3581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003994951480266405,
+      "loss": 3.2949,
+      "step": 3582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003992393454214723,
+      "loss": 3.1868,
+      "step": 3583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039898357030280487,
+      "loss": 3.0569,
+      "step": 3584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000398727822740411,
+      "loss": 3.1112,
+      "step": 3585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039847210280405577,
+      "loss": 3.2703,
+      "step": 3586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003982164105634974,
+      "loss": 3.1283,
+      "step": 3587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003979607460884862,
+      "loss": 3.1639,
+      "step": 3588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003977051094487647,
+      "loss": 3.3186,
+      "step": 3589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039744950071406816,
+      "loss": 3.1536,
+      "step": 3590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003971939199541239,
+      "loss": 3.0353,
+      "step": 3591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039693836723865176,
+      "loss": 3.0424,
+      "step": 3592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039668284263736443,
+      "loss": 3.1944,
+      "step": 3593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039642734621996614,
+      "loss": 3.3091,
+      "step": 3594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003961718780561537,
+      "loss": 3.2013,
+      "step": 3595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003959164382156164,
+      "loss": 3.2532,
+      "step": 3596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003956610267680356,
+      "loss": 3.0643,
+      "step": 3597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039540564378308493,
+      "loss": 3.0848,
+      "step": 3598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039515028933043064,
+      "loss": 3.2112,
+      "step": 3599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039489496347973064,
+      "loss": 3.1576,
+      "step": 3600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003946396663006352,
+      "loss": 3.1201,
+      "step": 3601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000394384397862787,
+      "loss": 3.2159,
+      "step": 3602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039412915823582056,
+      "loss": 3.1391,
+      "step": 3603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003938739474893629,
+      "loss": 3.1118,
+      "step": 3604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039361876569303267,
+      "loss": 3.0487,
+      "step": 3605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003933636129164413,
+      "loss": 3.1398,
+      "step": 3606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003931084892291917,
+      "loss": 3.1341,
+      "step": 3607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003928533947008791,
+      "loss": 3.162,
+      "step": 3608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003925983294010907,
+      "loss": 3.2357,
+      "step": 3609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039234329339940587,
+      "loss": 3.1344,
+      "step": 3610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039208828676539557,
+      "loss": 3.2706,
+      "step": 3611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003918333095686235,
+      "loss": 3.0659,
+      "step": 3612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039157836187864474,
+      "loss": 3.3347,
+      "step": 3613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039132344376500647,
+      "loss": 3.2705,
+      "step": 3614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039106855529724764,
+      "loss": 3.1605,
+      "step": 3615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039081369654489956,
+      "loss": 3.2678,
+      "step": 3616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003905588675774848,
+      "loss": 3.1127,
+      "step": 3617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003903040684645184,
+      "loss": 3.264,
+      "step": 3618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039004929927550716,
+      "loss": 2.9704,
+      "step": 3619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003897945600799493,
+      "loss": 3.1316,
+      "step": 3620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003895398509473352,
+      "loss": 3.3078,
+      "step": 3621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038928517194714707,
+      "loss": 3.4036,
+      "step": 3622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038903052314885855,
+      "loss": 3.279,
+      "step": 3623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003887759046219355,
+      "loss": 3.5028,
+      "step": 3624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003885213164358354,
+      "loss": 3.0752,
+      "step": 3625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003882667586600071,
+      "loss": 3.2094,
+      "step": 3626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038801223136389167,
+      "loss": 3.2529,
+      "step": 3627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038775773461692157,
+      "loss": 3.2987,
+      "step": 3628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038750326848852073,
+      "loss": 3.1633,
+      "step": 3629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038724883304810503,
+      "loss": 3.2604,
+      "step": 3630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038699442836508227,
+      "loss": 3.1565,
+      "step": 3631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038674005450885115,
+      "loss": 3.2225,
+      "step": 3632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003864857115488025,
+      "loss": 3.3834,
+      "step": 3633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038623139955431857,
+      "loss": 3.174,
+      "step": 3634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038597711859477286,
+      "loss": 3.2362,
+      "step": 3635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003857228687395308,
+      "loss": 3.2945,
+      "step": 3636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003854686500579494,
+      "loss": 3.1396,
+      "step": 3637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003852144626193768,
+      "loss": 3.1494,
+      "step": 3638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038496030649315296,
+      "loss": 3.2615,
+      "step": 3639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000384706181748609,
+      "loss": 3.0965,
+      "step": 3640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003844520884550675,
+      "loss": 3.1378,
+      "step": 3641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003841980266818426,
+      "loss": 3.1798,
+      "step": 3642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003839439964982402,
+      "loss": 3.1281,
+      "step": 3643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038368999797355676,
+      "loss": 3.1302,
+      "step": 3644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003834360311770808,
+      "loss": 3.0452,
+      "step": 3645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003831820961780917,
+      "loss": 3.1079,
+      "step": 3646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003829281930458607,
+      "loss": 3.3207,
+      "step": 3647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003826743218496496,
+      "loss": 3.2192,
+      "step": 3648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038242048265871233,
+      "loss": 3.1293,
+      "step": 3649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003821666755422935,
+      "loss": 3.024,
+      "step": 3650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038191290056962927,
+      "loss": 3.1154,
+      "step": 3651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003816591578099468,
+      "loss": 3.1187,
+      "step": 3652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003814054473324647,
+      "loss": 3.1588,
+      "step": 3653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038115176920639263,
+      "loss": 3.1915,
+      "step": 3654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003808981235009311,
+      "loss": 3.1378,
+      "step": 3655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038064451028527267,
+      "loss": 3.0514,
+      "step": 3656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003803909296286002,
+      "loss": 3.156,
+      "step": 3657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000380137381600088,
+      "loss": 3.3135,
+      "step": 3658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003798838662689016,
+      "loss": 3.1641,
+      "step": 3659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037963038370419715,
+      "loss": 3.106,
+      "step": 3660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037937693397512224,
+      "loss": 3.0765,
+      "step": 3661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003791235171508157,
+      "loss": 3.2622,
+      "step": 3662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003788701333004069,
+      "loss": 3.1377,
+      "step": 3663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003786167824930164,
+      "loss": 3.2103,
+      "step": 3664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003783634647977559,
+      "loss": 3.2589,
+      "step": 3665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037811018028372775,
+      "loss": 3.0918,
+      "step": 3666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003778569290200254,
+      "loss": 3.219,
+      "step": 3667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003776037110757336,
+      "loss": 3.2095,
+      "step": 3668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003773505265199275,
+      "loss": 3.4172,
+      "step": 3669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003770973754216732,
+      "loss": 3.2771,
+      "step": 3670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000376844257850028,
+      "loss": 3.1861,
+      "step": 3671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003765911738740397,
+      "loss": 3.2049,
+      "step": 3672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037633812356274694,
+      "loss": 3.225,
+      "step": 3673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037608510698517974,
+      "loss": 3.2639,
+      "step": 3674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037583212421035824,
+      "loss": 3.1461,
+      "step": 3675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003755791753072935,
+      "loss": 3.1747,
+      "step": 3676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037532626034498775,
+      "loss": 3.2179,
+      "step": 3677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037507337939243334,
+      "loss": 3.2245,
+      "step": 3678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000374820532518614,
+      "loss": 3.1879,
+      "step": 3679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037456771979250334,
+      "loss": 3.1053,
+      "step": 3680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037431494128306673,
+      "loss": 3.2661,
+      "step": 3681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003740621970592594,
+      "loss": 3.15,
+      "step": 3682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037380948719002743,
+      "loss": 3.1301,
+      "step": 3683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037355681174430745,
+      "loss": 3.2726,
+      "step": 3684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037330417079102697,
+      "loss": 3.2972,
+      "step": 3685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037305156439910363,
+      "loss": 3.2773,
+      "step": 3686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003727989926374463,
+      "loss": 3.0526,
+      "step": 3687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003725464555749539,
+      "loss": 3.1856,
+      "step": 3688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037229395328051604,
+      "loss": 3.108,
+      "step": 3689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720414858230126,
+      "loss": 3.247,
+      "step": 3690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003717890532713145,
+      "loss": 3.2252,
+      "step": 3691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003715366556942825,
+      "loss": 3.23,
+      "step": 3692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037128429316076844,
+      "loss": 3.2128,
+      "step": 3693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003710319657396143,
+      "loss": 3.1424,
+      "step": 3694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003707796734996522,
+      "loss": 3.1431,
+      "step": 3695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037052741650970525,
+      "loss": 3.1507,
+      "step": 3696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003702751948385864,
+      "loss": 3.0993,
+      "step": 3697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003700230085550992,
+      "loss": 3.0506,
+      "step": 3698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003697708577280377,
+      "loss": 3.1036,
+      "step": 3699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003695187424261862,
+      "loss": 3.2103,
+      "step": 3700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000369266662718319,
+      "loss": 3.1875,
+      "step": 3701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003690146186732012,
+      "loss": 3.0278,
+      "step": 3702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036876261035958765,
+      "loss": 3.2181,
+      "step": 3703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036851063784622376,
+      "loss": 3.2134,
+      "step": 3704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036825870120184496,
+      "loss": 3.0586,
+      "step": 3705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036800680049517756,
+      "loss": 3.3617,
+      "step": 3706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000367754935794937,
+      "loss": 3.1478,
+      "step": 3707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036750310716982996,
+      "loss": 3.1093,
+      "step": 3708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003672513146885523,
+      "loss": 3.1908,
+      "step": 3709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003669995584197908,
+      "loss": 3.1753,
+      "step": 3710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036674783843222177,
+      "loss": 3.3199,
+      "step": 3711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003664961547945123,
+      "loss": 3.1907,
+      "step": 3712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003662445075753189,
+      "loss": 3.0726,
+      "step": 3713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003659928968432886,
+      "loss": 3.132,
+      "step": 3714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036574132266705805,
+      "loss": 3.3089,
+      "step": 3715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003654897851152544,
+      "loss": 3.2814,
+      "step": 3716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003652382842564943,
+      "loss": 3.0118,
+      "step": 3717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003649868201593851,
+      "loss": 3.133,
+      "step": 3718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003647353928925234,
+      "loss": 3.2615,
+      "step": 3719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036448400252449617,
+      "loss": 3.1365,
+      "step": 3720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036423264912388,
+      "loss": 3.0687,
+      "step": 3721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003639813327592419,
+      "loss": 3.2587,
+      "step": 3722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000363730053499138,
+      "loss": 3.2239,
+      "step": 3723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003634788114121154,
+      "loss": 3.183,
+      "step": 3724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003632276065667098,
+      "loss": 3.0569,
+      "step": 3725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036297643903144796,
+      "loss": 3.3382,
+      "step": 3726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036272530887484535,
+      "loss": 3.4445,
+      "step": 3727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036247421616540816,
+      "loss": 3.1898,
+      "step": 3728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003622231609716317,
+      "loss": 3.3619,
+      "step": 3729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036197214336200137,
+      "loss": 3.177,
+      "step": 3730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003617211634049925,
+      "loss": 3.2446,
+      "step": 3731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003614702211690698,
+      "loss": 3.1348,
+      "step": 3732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003612193167226876,
+      "loss": 3.0741,
+      "step": 3733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003609684501342905,
+      "loss": 3.167,
+      "step": 3734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036071762147231196,
+      "loss": 3.0626,
+      "step": 3735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036046683080517573,
+      "loss": 3.1588,
+      "step": 3736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003602160782012952,
+      "loss": 3.1651,
+      "step": 3737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003599653637290731,
+      "loss": 3.1758,
+      "step": 3738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003597146874569015,
+      "loss": 3.2,
+      "step": 3739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003594640494531628,
+      "loss": 3.1079,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003592134497862283,
+      "loss": 3.1873,
+      "step": 3741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035896288852445903,
+      "loss": 3.1328,
+      "step": 3742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000358712365736206,
+      "loss": 3.2796,
+      "step": 3743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035846188148980896,
+      "loss": 3.2468,
+      "step": 3744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003582114358535976,
+      "loss": 3.425,
+      "step": 3745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003579610288958911,
+      "loss": 3.2191,
+      "step": 3746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035771066068499767,
+      "loss": 3.1012,
+      "step": 3747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003574603312892155,
+      "loss": 3.0041,
+      "step": 3748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035721004077683206,
+      "loss": 3.1309,
+      "step": 3749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035695978921612383,
+      "loss": 3.1496,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003567095766753572,
+      "loss": 3.2345,
+      "step": 3751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003564594032227875,
+      "loss": 3.1334,
+      "step": 3752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003562092689266595,
+      "loss": 3.1289,
+      "step": 3753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035595917385520756,
+      "loss": 3.1095,
+      "step": 3754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003557091180766547,
+      "loss": 3.0637,
+      "step": 3755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000355459101659214,
+      "loss": 3.27,
+      "step": 3756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035520912467108747,
+      "loss": 3.078,
+      "step": 3757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035495918718046624,
+      "loss": 3.2856,
+      "step": 3758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003547092892555306,
+      "loss": 3.1458,
+      "step": 3759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035445943096445044,
+      "loss": 3.2951,
+      "step": 3760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035420961237538437,
+      "loss": 3.348,
+      "step": 3761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035395983355648067,
+      "loss": 3.2594,
+      "step": 3762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035371009457587645,
+      "loss": 3.293,
+      "step": 3763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035346039550169785,
+      "loss": 3.2117,
+      "step": 3764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003532107364020605,
+      "loss": 3.1945,
+      "step": 3765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035296111734506877,
+      "loss": 3.1235,
+      "step": 3766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035271153839881605,
+      "loss": 3.2138,
+      "step": 3767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035246199963138536,
+      "loss": 3.137,
+      "step": 3768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035221250111084837,
+      "loss": 3.1597,
+      "step": 3769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035196304290526545,
+      "loss": 3.1508,
+      "step": 3770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035171362508268665,
+      "loss": 3.1114,
+      "step": 3771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003514642477111505,
+      "loss": 3.0672,
+      "step": 3772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003512149108586845,
+      "loss": 3.3381,
+      "step": 3773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035096561459330557,
+      "loss": 3.1665,
+      "step": 3774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035071635898301914,
+      "loss": 3.0854,
+      "step": 3775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003504671440958195,
+      "loss": 3.2345,
+      "step": 3776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003502179699996903,
+      "loss": 3.1385,
+      "step": 3777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034996883676260347,
+      "loss": 3.0541,
+      "step": 3778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003497197444525201,
+      "loss": 3.2339,
+      "step": 3779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034947069313738993,
+      "loss": 3.3471,
+      "step": 3780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003492216828851521,
+      "loss": 3.0803,
+      "step": 3781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034897271376373385,
+      "loss": 3.2488,
+      "step": 3782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003487237858410516,
+      "loss": 3.2869,
+      "step": 3783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003484748991850102,
+      "loss": 3.1695,
+      "step": 3784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003482260538635036,
+      "loss": 3.2834,
+      "step": 3785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034797724994441414,
+      "loss": 2.9703,
+      "step": 3786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003477284874956134,
+      "loss": 3.1104,
+      "step": 3787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034747976658496097,
+      "loss": 3.1409,
+      "step": 3788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034723108728030563,
+      "loss": 3.2163,
+      "step": 3789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034698244964948433,
+      "loss": 3.2635,
+      "step": 3790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034673385376032335,
+      "loss": 3.0444,
+      "step": 3791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003464852996806367,
+      "loss": 3.0745,
+      "step": 3792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034623678747822785,
+      "loss": 3.1054,
+      "step": 3793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034598831722088826,
+      "loss": 3.29,
+      "step": 3794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003457398889763982,
+      "loss": 3.0743,
+      "step": 3795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034549150281252633,
+      "loss": 3.1037,
+      "step": 3796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034524315879703006,
+      "loss": 3.1515,
+      "step": 3797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034499485699765486,
+      "loss": 3.3224,
+      "step": 3798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003447465974821352,
+      "loss": 3.1269,
+      "step": 3799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034449838031819403,
+      "loss": 3.0513,
+      "step": 3800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003442502055735421,
+      "loss": 3.1829,
+      "step": 3801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034400207331587924,
+      "loss": 3.1715,
+      "step": 3802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003437539836128935,
+      "loss": 3.3458,
+      "step": 3803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034350593653226095,
+      "loss": 3.2647,
+      "step": 3804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003432579321416464,
+      "loss": 3.1644,
+      "step": 3805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003430099705087034,
+      "loss": 3.1215,
+      "step": 3806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003427620517010732,
+      "loss": 3.2238,
+      "step": 3807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003425141757863854,
+      "loss": 3.2699,
+      "step": 3808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003422663428322582,
+      "loss": 3.1745,
+      "step": 3809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000342018552906298,
+      "loss": 3.1356,
+      "step": 3810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003417708060760992,
+      "loss": 3.1268,
+      "step": 3811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034152310240924503,
+      "loss": 3.1308,
+      "step": 3812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034127544197330636,
+      "loss": 3.0954,
+      "step": 3813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034102782483584235,
+      "loss": 3.2691,
+      "step": 3814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003407802510644008,
+      "loss": 3.141,
+      "step": 3815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003405327207265171,
+      "loss": 3.2359,
+      "step": 3816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003402852338897151,
+      "loss": 3.1622,
+      "step": 3817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003400377906215069,
+      "loss": 3.111,
+      "step": 3818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003397903909893924,
+      "loss": 3.1455,
+      "step": 3819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033954303506085985,
+      "loss": 3.1247,
+      "step": 3820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003392957229033855,
+      "loss": 3.222,
+      "step": 3821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003390484545844334,
+      "loss": 3.2638,
+      "step": 3822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003388012301714559,
+      "loss": 3.1266,
+      "step": 3823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033855404973189376,
+      "loss": 3.0327,
+      "step": 3824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033830691333317496,
+      "loss": 3.3177,
+      "step": 3825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033805982104271606,
+      "loss": 3.0392,
+      "step": 3826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003378127729279212,
+      "loss": 3.1377,
+      "step": 3827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003375657690561826,
+      "loss": 3.1744,
+      "step": 3828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003373188094948807,
+      "loss": 3.2019,
+      "step": 3829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033707189431138323,
+      "loss": 3.0339,
+      "step": 3830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003368250235730466,
+      "loss": 3.1042,
+      "step": 3831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003365781973472144,
+      "loss": 3.0907,
+      "step": 3832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003363314157012185,
+      "loss": 3.2083,
+      "step": 3833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003360846787023785,
+      "loss": 3.1847,
+      "step": 3834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033583798641800166,
+      "loss": 3.2091,
+      "step": 3835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000335591338915383,
+      "loss": 3.3147,
+      "step": 3836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033534473626180587,
+      "loss": 3.2378,
+      "step": 3837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033509817852454094,
+      "loss": 3.0241,
+      "step": 3838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003348516657708466,
+      "loss": 3.2588,
+      "step": 3839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033460519806796906,
+      "loss": 3.0929,
+      "step": 3840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003343587754831424,
+      "loss": 3.2418,
+      "step": 3841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033411239808358787,
+      "loss": 3.302,
+      "step": 3842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033386606593651503,
+      "loss": 3.0182,
+      "step": 3843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033361977910912103,
+      "loss": 3.1076,
+      "step": 3844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003333735376685901,
+      "loss": 3.1627,
+      "step": 3845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003331273416820947,
+      "loss": 3.0565,
+      "step": 3846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003328811912167945,
+      "loss": 3.2785,
+      "step": 3847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033263508633983677,
+      "loss": 3.1276,
+      "step": 3848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003323890271183566,
+      "loss": 3.1734,
+      "step": 3849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003321430136194766,
+      "loss": 3.3011,
+      "step": 3850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033189704591030675,
+      "loss": 3.2495,
+      "step": 3851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003316511240579445,
+      "loss": 3.139,
+      "step": 3852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003314052481294748,
+      "loss": 3.3046,
+      "step": 3853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033115941819197045,
+      "loss": 3.2697,
+      "step": 3854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033091363431249094,
+      "loss": 3.0877,
+      "step": 3855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033066789655808416,
+      "loss": 3.1205,
+      "step": 3856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033042220499578454,
+      "loss": 3.1523,
+      "step": 3857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003301765596926145,
+      "loss": 3.0972,
+      "step": 3858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003299309607155835,
+      "loss": 2.9794,
+      "step": 3859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003296854081316887,
+      "loss": 3.2037,
+      "step": 3860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032943990200791395,
+      "loss": 3.1062,
+      "step": 3861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032919444241123134,
+      "loss": 3.3388,
+      "step": 3862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003289490294085996,
+      "loss": 3.3456,
+      "step": 3863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032870366306696495,
+      "loss": 3.1253,
+      "step": 3864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032845834345326085,
+      "loss": 3.0594,
+      "step": 3865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003282130706344082,
+      "loss": 3.205,
+      "step": 3866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032796784467731466,
+      "loss": 3.2603,
+      "step": 3867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032772266564887566,
+      "loss": 3.3066,
+      "step": 3868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032747753361597363,
+      "loss": 3.086,
+      "step": 3869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003272324486454782,
+      "loss": 3.249,
+      "step": 3870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032698741080424576,
+      "loss": 3.0652,
+      "step": 3871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003267424201591205,
+      "loss": 3.218,
+      "step": 3872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032649747677693307,
+      "loss": 3.12,
+      "step": 3873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032625258072450203,
+      "loss": 3.1962,
+      "step": 3874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032600773206863245,
+      "loss": 3.1178,
+      "step": 3875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003257629308761164,
+      "loss": 3.239,
+      "step": 3876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032551817721373333,
+      "loss": 3.1471,
+      "step": 3877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003252734711482497,
+      "loss": 3.1145,
+      "step": 3878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003250288127464186,
+      "loss": 3.2499,
+      "step": 3879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003247842020749805,
+      "loss": 3.063,
+      "step": 3880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000324539639200663,
+      "loss": 3.1301,
+      "step": 3881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032429512419018027,
+      "loss": 3.2871,
+      "step": 3882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003240506571102334,
+      "loss": 3.0852,
+      "step": 3883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032380623802751073,
+      "loss": 3.2337,
+      "step": 3884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032356186700868727,
+      "loss": 3.1967,
+      "step": 3885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003233175441204249,
+      "loss": 3.2311,
+      "step": 3886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003230732694293728,
+      "loss": 3.2081,
+      "step": 3887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003228290430021664,
+      "loss": 3.1216,
+      "step": 3888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032258486490542836,
+      "loss": 3.1377,
+      "step": 3889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000322340735205768,
+      "loss": 3.1451,
+      "step": 3890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003220966539697813,
+      "loss": 3.0131,
+      "step": 3891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032185262126405113,
+      "loss": 3.2489,
+      "step": 3892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032160863715514763,
+      "loss": 3.2848,
+      "step": 3893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032136470170962686,
+      "loss": 3.3124,
+      "step": 3894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003211208149940321,
+      "loss": 3.3503,
+      "step": 3895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032087697707489327,
+      "loss": 3.198,
+      "step": 3896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003206331880187267,
+      "loss": 3.1232,
+      "step": 3897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003203894478920356,
+      "loss": 3.1928,
+      "step": 3898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003201457567613102,
+      "loss": 3.2272,
+      "step": 3899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003199021146930268,
+      "loss": 3.2764,
+      "step": 3900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003196585217536485,
+      "loss": 3.1136,
+      "step": 3901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031941497800962496,
+      "loss": 3.1489,
+      "step": 3902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003191714835273927,
+      "loss": 3.0662,
+      "step": 3903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031892803837337436,
+      "loss": 3.1778,
+      "step": 3904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031868464261397924,
+      "loss": 3.0942,
+      "step": 3905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003184412963156036,
+      "loss": 3.1304,
+      "step": 3906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003181979995446298,
+      "loss": 3.1426,
+      "step": 3907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031795475236742667,
+      "loss": 3.26,
+      "step": 3908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031771155485034973,
+      "loss": 3.2953,
+      "step": 3909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003174684070597408,
+      "loss": 3.2263,
+      "step": 3910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003172253090619279,
+      "loss": 3.1705,
+      "step": 3911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000316982260923226,
+      "loss": 3.072,
+      "step": 3912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031673926270993634,
+      "loss": 3.2148,
+      "step": 3913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031649631448834616,
+      "loss": 3.1464,
+      "step": 3914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003162534163247295,
+      "loss": 3.0728,
+      "step": 3915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031601056828534633,
+      "loss": 3.1059,
+      "step": 3916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031576777043644316,
+      "loss": 3.319,
+      "step": 3917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031552502284425306,
+      "loss": 3.195,
+      "step": 3918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031528232557499514,
+      "loss": 3.0563,
+      "step": 3919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031503967869487453,
+      "loss": 3.2843,
+      "step": 3920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003147970822700832,
+      "loss": 3.1263,
+      "step": 3921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031455453636679867,
+      "loss": 3.2415,
+      "step": 3922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031431204105118515,
+      "loss": 3.0898,
+      "step": 3923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003140695963893933,
+      "loss": 3.1328,
+      "step": 3924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003138272024475593,
+      "loss": 3.2403,
+      "step": 3925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003135848592918057,
+      "loss": 3.2497,
+      "step": 3926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003133425669882416,
+      "loss": 3.2576,
+      "step": 3927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031310032560296155,
+      "loss": 3.1809,
+      "step": 3928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003128581352020469,
+      "loss": 3.1822,
+      "step": 3929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031261599585156443,
+      "loss": 3.1691,
+      "step": 3930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003123739076175678,
+      "loss": 3.2921,
+      "step": 3931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003121318705660959,
+      "loss": 3.2874,
+      "step": 3932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003118898847631742,
+      "loss": 3.2254,
+      "step": 3933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031164795027481383,
+      "loss": 3.1937,
+      "step": 3934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003114060671670124,
+      "loss": 3.0856,
+      "step": 3935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003111642355057528,
+      "loss": 3.2967,
+      "step": 3936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031092245535700464,
+      "loss": 3.1986,
+      "step": 3937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003106807267867231,
+      "loss": 3.2297,
+      "step": 3938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031043904986084926,
+      "loss": 3.1417,
+      "step": 3939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031019742464531,
+      "loss": 3.0704,
+      "step": 3940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030995585120601854,
+      "loss": 3.1322,
+      "step": 3941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030971432960887334,
+      "loss": 3.1744,
+      "step": 3942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003094728599197595,
+      "loss": 3.1892,
+      "step": 3943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003092314422045474,
+      "loss": 2.9808,
+      "step": 3944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030899007652909326,
+      "loss": 3.1168,
+      "step": 3945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003087487629592393,
+      "loss": 3.0731,
+      "step": 3946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003085075015608135,
+      "loss": 3.1754,
+      "step": 3947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030826629239962943,
+      "loss": 3.2921,
+      "step": 3948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030802513554148664,
+      "loss": 3.2581,
+      "step": 3949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030778403105217046,
+      "loss": 3.1403,
+      "step": 3950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003075429789974515,
+      "loss": 3.1663,
+      "step": 3951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003073019794430866,
+      "loss": 3.1007,
+      "step": 3952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003070610324548179,
+      "loss": 2.9963,
+      "step": 3953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030682013809837325,
+      "loss": 3.2805,
+      "step": 3954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003065792964394662,
+      "loss": 3.2961,
+      "step": 3955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030633850754379635,
+      "loss": 3.2686,
+      "step": 3956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030609777147704806,
+      "loss": 3.2565,
+      "step": 3957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000305857088304892,
+      "loss": 3.2475,
+      "step": 3958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003056164580929841,
+      "loss": 3.17,
+      "step": 3959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003053758809069657,
+      "loss": 3.2915,
+      "step": 3960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003051353568124638,
+      "loss": 3.0746,
+      "step": 3961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003048948858750914,
+      "loss": 3.0767,
+      "step": 3962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003046544681604462,
+      "loss": 3.2315,
+      "step": 3963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030441410373411193,
+      "loss": 3.3243,
+      "step": 3964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003041737926616576,
+      "loss": 3.1168,
+      "step": 3965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030393353500863754,
+      "loss": 3.1595,
+      "step": 3966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003036933308405915,
+      "loss": 3.2305,
+      "step": 3967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003034531802230452,
+      "loss": 3.0983,
+      "step": 3968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003032130832215091,
+      "loss": 3.1213,
+      "step": 3969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003029730399014794,
+      "loss": 3.1739,
+      "step": 3970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030273305032843724,
+      "loss": 3.0616,
+      "step": 3971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030249311456784965,
+      "loss": 3.1848,
+      "step": 3972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003022532326851685,
+      "loss": 3.2177,
+      "step": 3973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030201340474583137,
+      "loss": 3.2138,
+      "step": 3974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003017736308152608,
+      "loss": 3.2187,
+      "step": 3975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003015339109588648,
+      "loss": 3.1413,
+      "step": 3976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003012942452420364,
+      "loss": 3.0542,
+      "step": 3977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030105463373015427,
+      "loss": 3.1738,
+      "step": 3978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000300815076488582,
+      "loss": 3.3473,
+      "step": 3979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030057557358266794,
+      "loss": 3.1101,
+      "step": 3980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030033612507774667,
+      "loss": 3.1499,
+      "step": 3981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003000967310391373,
+      "loss": 3.0242,
+      "step": 3982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029985739153214373,
+      "loss": 3.1692,
+      "step": 3983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002996181066220558,
+      "loss": 3.1013,
+      "step": 3984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002993788763741479,
+      "loss": 3.1263,
+      "step": 3985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002991397008536794,
+      "loss": 3.1486,
+      "step": 3986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002989005801258954,
+      "loss": 3.3533,
+      "step": 3987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002986615142560255,
+      "loss": 3.0696,
+      "step": 3988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002984225033092844,
+      "loss": 3.2083,
+      "step": 3989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002981835473508721,
+      "loss": 3.2213,
+      "step": 3990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029794464644597305,
+      "loss": 3.1267,
+      "step": 3991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002977058006597572,
+      "loss": 3.2145,
+      "step": 3992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002974670100573795,
+      "loss": 3.1627,
+      "step": 3993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029722827470397953,
+      "loss": 3.0899,
+      "step": 3994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969895946646818,
+      "loss": 3.1466,
+      "step": 3995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029675097000459594,
+      "loss": 3.2256,
+      "step": 3996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002965124007888163,
+      "loss": 3.1676,
+      "step": 3997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029627388708242195,
+      "loss": 3.2112,
+      "step": 3998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002960354289504776,
+      "loss": 3.1592,
+      "step": 3999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002957970264580321,
+      "loss": 3.2051,
+      "step": 4000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029555867967011887,
+      "loss": 2.9887,
+      "step": 4001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029532038865175695,
+      "loss": 3.1266,
+      "step": 4002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002950821534679495,
+      "loss": 3.1663,
+      "step": 4003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029484397418368493,
+      "loss": 3.1082,
+      "step": 4004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029460585086393576,
+      "loss": 3.0419,
+      "step": 4005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029436778357366014,
+      "loss": 3.1903,
+      "step": 4006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029412977237780024,
+      "loss": 3.1059,
+      "step": 4007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002938918173412832,
+      "loss": 3.2009,
+      "step": 4008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002936539185290206,
+      "loss": 3.132,
+      "step": 4009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002934160760059091,
+      "loss": 3.2861,
+      "step": 4010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002931782898368294,
+      "loss": 3.2226,
+      "step": 4011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002929405600866476,
+      "loss": 3.2351,
+      "step": 4012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002927028868202139,
+      "loss": 3.0392,
+      "step": 4013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002924652701023631,
+      "loss": 3.0767,
+      "step": 4014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029222770999791473,
+      "loss": 3.0831,
+      "step": 4015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002919902065716728,
+      "loss": 3.2114,
+      "step": 4016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002917527598884256,
+      "loss": 3.3294,
+      "step": 4017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002915153700129468,
+      "loss": 3.2263,
+      "step": 4018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029127803700999355,
+      "loss": 3.2039,
+      "step": 4019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000291040760944308,
+      "loss": 3.2107,
+      "step": 4020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000290803541880617,
+      "loss": 3.2709,
+      "step": 4021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000290566379883631,
+      "loss": 3.13,
+      "step": 4022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029032927501804553,
+      "loss": 3.2467,
+      "step": 4023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002900922273485409,
+      "loss": 3.2125,
+      "step": 4024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289855236939781,
+      "loss": 3.0606,
+      "step": 4025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002896183038564144,
+      "loss": 3.146,
+      "step": 4026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893814281630744,
+      "loss": 3.335,
+      "step": 4027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028914460992437784,
+      "loss": 3.2915,
+      "step": 4028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028890784920492673,
+      "loss": 3.148,
+      "step": 4029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002886711460693069,
+      "loss": 3.1647,
+      "step": 4030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028843450058208865,
+      "loss": 3.1223,
+      "step": 4031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002881979128078264,
+      "loss": 3.1942,
+      "step": 4032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002879613828110591,
+      "loss": 3.2868,
+      "step": 4033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002877249106563099,
+      "loss": 3.1564,
+      "step": 4034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002874884964080856,
+      "loss": 3.3212,
+      "step": 4035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028725214013087787,
+      "loss": 3.1215,
+      "step": 4036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028701584188916234,
+      "loss": 3.1798,
+      "step": 4037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002867796017473989,
+      "loss": 3.2662,
+      "step": 4038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002865434197700314,
+      "loss": 3.0221,
+      "step": 4039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028630729602148816,
+      "loss": 3.2844,
+      "step": 4040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028607123056618094,
+      "loss": 3.268,
+      "step": 4041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002858352234685063,
+      "loss": 3.0573,
+      "step": 4042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002855992747928446,
+      "loss": 3.2745,
+      "step": 4043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002853633846035603,
+      "loss": 3.1147,
+      "step": 4044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002851275529650018,
+      "loss": 3.1835,
+      "step": 4045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028489177994150196,
+      "loss": 3.1462,
+      "step": 4046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028465606559737675,
+      "loss": 3.3475,
+      "step": 4047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028442040999692705,
+      "loss": 3.252,
+      "step": 4048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002841848132044372,
+      "loss": 3.1954,
+      "step": 4049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002839492752841758,
+      "loss": 3.2729,
+      "step": 4050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002837137963003952,
+      "loss": 3.0788,
+      "step": 4051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002834783763173318,
+      "loss": 3.2465,
+      "step": 4052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002832430153992055,
+      "loss": 3.1525,
+      "step": 4053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002830077136102207,
+      "loss": 3.2683,
+      "step": 4054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002827724710145653,
+      "loss": 3.122,
+      "step": 4055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028253728767641104,
+      "loss": 3.1916,
+      "step": 4056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002823021636599137,
+      "loss": 3.025,
+      "step": 4057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028206709902921294,
+      "loss": 3.2628,
+      "step": 4058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028183209384843167,
+      "loss": 3.2141,
+      "step": 4059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028159714818167713,
+      "loss": 3.1428,
+      "step": 4060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028136226209304015,
+      "loss": 3.258,
+      "step": 4061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028112743564659534,
+      "loss": 2.9183,
+      "step": 4062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002808926689064009,
+      "loss": 3.0787,
+      "step": 4063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028065796193649917,
+      "loss": 3.0015,
+      "step": 4064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002804233148009155,
+      "loss": 3.1877,
+      "step": 4065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002801887275636594,
+      "loss": 3.2151,
+      "step": 4066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002799542002887239,
+      "loss": 3.2113,
+      "step": 4067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002797197330400858,
+      "loss": 3.0123,
+      "step": 4068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002794853258817053,
+      "loss": 3.1978,
+      "step": 4069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027925097887752666,
+      "loss": 3.2574,
+      "step": 4070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002790166920914769,
+      "loss": 3.0455,
+      "step": 4071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002787824655874674,
+      "loss": 3.13,
+      "step": 4072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027854829942939273,
+      "loss": 3.174,
+      "step": 4073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002783141936811312,
+      "loss": 3.2969,
+      "step": 4074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027808014840654437,
+      "loss": 3.2942,
+      "step": 4075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002778461636694778,
+      "loss": 3.0967,
+      "step": 4076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002776122395337597,
+      "loss": 3.1659,
+      "step": 4077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027737837606320244,
+      "loss": 3.2761,
+      "step": 4078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002771445733216017,
+      "loss": 3.2108,
+      "step": 4079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027691083137273645,
+      "loss": 3.1669,
+      "step": 4080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002766771502803692,
+      "loss": 3.2361,
+      "step": 4081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000276443530108246,
+      "loss": 3.1721,
+      "step": 4082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002762099709200958,
+      "loss": 3.2816,
+      "step": 4083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002759764727796313,
+      "loss": 3.1641,
+      "step": 4084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027574303575054847,
+      "loss": 3.1325,
+      "step": 4085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027550965989652664,
+      "loss": 3.0987,
+      "step": 4086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002752763452812285,
+      "loss": 3.2683,
+      "step": 4087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027504309196829966,
+      "loss": 3.2042,
+      "step": 4088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027480990002136987,
+      "loss": 3.1936,
+      "step": 4089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002745767695040509,
+      "loss": 3.2831,
+      "step": 4090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002743437004799387,
+      "loss": 3.1679,
+      "step": 4091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027411069301261213,
+      "loss": 3.1541,
+      "step": 4092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027387774716563346,
+      "loss": 3.1048,
+      "step": 4093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027364486300254787,
+      "loss": 3.2002,
+      "step": 4094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000273412040586884,
+      "loss": 3.1037,
+      "step": 4095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002731792799821532,
+      "loss": 3.1418,
+      "step": 4096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002729465812518503,
+      "loss": 3.3329,
+      "step": 4097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027271394445945346,
+      "loss": 3.0457,
+      "step": 4098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002724813696684231,
+      "loss": 3.2609,
+      "step": 4099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002722488569422039,
+      "loss": 3.0507,
+      "step": 4100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002720164063442229,
+      "loss": 3.0593,
+      "step": 4101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002717840179378901,
+      "loss": 3.1612,
+      "step": 4102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027155169178659874,
+      "loss": 3.3641,
+      "step": 4103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027131942795372536,
+      "loss": 3.2591,
+      "step": 4104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002710872265026286,
+      "loss": 3.1082,
+      "step": 4105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027085508749665144,
+      "loss": 3.2559,
+      "step": 4106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002706230109991188,
+      "loss": 3.0895,
+      "step": 4107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002703909970733387,
+      "loss": 3.3199,
+      "step": 4108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002701590457826023,
+      "loss": 3.0681,
+      "step": 4109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002699271571901837,
+      "loss": 3.1658,
+      "step": 4110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026969533135933946,
+      "loss": 3.0492,
+      "step": 4111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002694635683533096,
+      "loss": 3.1969,
+      "step": 4112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026923186823531707,
+      "loss": 3.1702,
+      "step": 4113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002690002310685669,
+      "loss": 3.1702,
+      "step": 4114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002687686569162474,
+      "loss": 2.9403,
+      "step": 4115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002685371458415298,
+      "loss": 3.1884,
+      "step": 4116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026830569790756804,
+      "loss": 3.1893,
+      "step": 4117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002680743131774987,
+      "loss": 3.1849,
+      "step": 4118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002678429917144417,
+      "loss": 3.1613,
+      "step": 4119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002676117335814985,
+      "loss": 3.1853,
+      "step": 4120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026738053884175437,
+      "loss": 3.2112,
+      "step": 4121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026714940755827695,
+      "loss": 3.1344,
+      "step": 4122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002669183397941166,
+      "loss": 3.2412,
+      "step": 4123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002666873356123059,
+      "loss": 3.2354,
+      "step": 4124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002664563950758611,
+      "loss": 3.1895,
+      "step": 4125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026622551824778,
+      "loss": 3.2522,
+      "step": 4126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002659947051910436,
+      "loss": 3.2679,
+      "step": 4127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026576395596861554,
+      "loss": 3.1023,
+      "step": 4128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002655332706434419,
+      "loss": 3.1887,
+      "step": 4129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002653026492784509,
+      "loss": 2.9679,
+      "step": 4130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026507209193655444,
+      "loss": 3.2127,
+      "step": 4131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026484159868064584,
+      "loss": 3.0054,
+      "step": 4132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002646111695736013,
+      "loss": 3.3069,
+      "step": 4133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643808046782797,
+      "loss": 3.2137,
+      "step": 4134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002641505040575226,
+      "loss": 3.0031,
+      "step": 4135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002639202677741529,
+      "loss": 3.1172,
+      "step": 4136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002636900958909776,
+      "loss": 3.1372,
+      "step": 4137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000263459988470785,
+      "loss": 3.0422,
+      "step": 4138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002632299455763459,
+      "loss": 3.3007,
+      "step": 4139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002629999672704139,
+      "loss": 2.9857,
+      "step": 4140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026277005361572493,
+      "loss": 3.0287,
+      "step": 4141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026254020467499664,
+      "loss": 3.0796,
+      "step": 4142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002623104205109299,
+      "loss": 3.1754,
+      "step": 4143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026208070118620777,
+      "loss": 2.9953,
+      "step": 4144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026185104676349494,
+      "loss": 3.1351,
+      "step": 4145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000261621457305439,
+      "loss": 3.2094,
+      "step": 4146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002613919328746698,
+      "loss": 3.2105,
+      "step": 4147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000261162473533799,
+      "loss": 3.2661,
+      "step": 4148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002609330793454208,
+      "loss": 3.1949,
+      "step": 4149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026070375037211225,
+      "loss": 3.2443,
+      "step": 4150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002604744866764314,
+      "loss": 3.2915,
+      "step": 4151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026024528832091926,
+      "loss": 3.0836,
+      "step": 4152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026001615536809915,
+      "loss": 3.1686,
+      "step": 4153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002597870878804758,
+      "loss": 3.1607,
+      "step": 4154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025955808592053643,
+      "loss": 3.179,
+      "step": 4155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025932914955075127,
+      "loss": 3.1991,
+      "step": 4156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002591002788335711,
+      "loss": 3.27,
+      "step": 4157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025887147383142997,
+      "loss": 3.1548,
+      "step": 4158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002586427346067436,
+      "loss": 3.1206,
+      "step": 4159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002584140612219095,
+      "loss": 3.2813,
+      "step": 4160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025818545373930743,
+      "loss": 3.2768,
+      "step": 4161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002579569122212998,
+      "loss": 3.1401,
+      "step": 4162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002577284367302299,
+      "loss": 3.4463,
+      "step": 4163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025750002732842384,
+      "loss": 3.3511,
+      "step": 4164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002572716840781894,
+      "loss": 3.0053,
+      "step": 4165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025704340704181614,
+      "loss": 3.0359,
+      "step": 4166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002568151962815757,
+      "loss": 3.0519,
+      "step": 4167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002565870518597223,
+      "loss": 3.1631,
+      "step": 4168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025635897383849095,
+      "loss": 3.0362,
+      "step": 4169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000256130962280099,
+      "loss": 3.242,
+      "step": 4170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025590301724674595,
+      "loss": 3.0901,
+      "step": 4171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002556751388006131,
+      "loss": 3.1802,
+      "step": 4172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002554473270038629,
+      "loss": 3.2809,
+      "step": 4173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002552195819186405,
+      "loss": 3.1039,
+      "step": 4174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025499190360707234,
+      "loss": 3.05,
+      "step": 4175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002547642921312669,
+      "loss": 3.1561,
+      "step": 4176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025453674755331426,
+      "loss": 3.1667,
+      "step": 4177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025430926993528645,
+      "loss": 3.1744,
+      "step": 4178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002540818593392369,
+      "loss": 3.2322,
+      "step": 4179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025385451582720086,
+      "loss": 3.1559,
+      "step": 4180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002536272394611956,
+      "loss": 3.1086,
+      "step": 4181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025340003030321977,
+      "loss": 3.0673,
+      "step": 4182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025317288841525377,
+      "loss": 3.1716,
+      "step": 4183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025294581385925974,
+      "loss": 3.2887,
+      "step": 4184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002527188066971811,
+      "loss": 3.1829,
+      "step": 4185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025249186699094326,
+      "loss": 3.0651,
+      "step": 4186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002522649948024531,
+      "loss": 3.1439,
+      "step": 4187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002520381901935992,
+      "loss": 3.1099,
+      "step": 4188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025181145322625146,
+      "loss": 3.1224,
+      "step": 4189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002515847839622617,
+      "loss": 3.0274,
+      "step": 4190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002513581824634626,
+      "loss": 3.1903,
+      "step": 4191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025113164879166886,
+      "loss": 3.2002,
+      "step": 4192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025090518300867717,
+      "loss": 3.1281,
+      "step": 4193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025067878517626445,
+      "loss": 3.1395,
+      "step": 4194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002504524553561901,
+      "loss": 3.2152,
+      "step": 4195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002502261936101948,
+      "loss": 3.1826,
+      "step": 4196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 3.2082,
+      "step": 4197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024977387458730954,
+      "loss": 3.1214,
+      "step": 4198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024954781743380785,
+      "loss": 3.1125,
+      "step": 4199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024932182860116115,
+      "loss": 3.0731,
+      "step": 4200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000249095908151017,
+      "loss": 2.8992,
+      "step": 4201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024887005614500444,
+      "loss": 3.2321,
+      "step": 4202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002486442726447332,
+      "loss": 3.1698,
+      "step": 4203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002484185577117951,
+      "loss": 3.2945,
+      "step": 4204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002481929114077626,
+      "loss": 3.2349,
+      "step": 4205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024796733379419013,
+      "loss": 3.2539,
+      "step": 4206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002477418249326128,
+      "loss": 3.1775,
+      "step": 4207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002475163848845474,
+      "loss": 3.1465,
+      "step": 4208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002472910137114914,
+      "loss": 3.2704,
+      "step": 4209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002470657114749238,
+      "loss": 3.1569,
+      "step": 4210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002468404782363051,
+      "loss": 3.0388,
+      "step": 4211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024661531405707656,
+      "loss": 3.1313,
+      "step": 4212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002463902189986606,
+      "loss": 3.2158,
+      "step": 4213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024616519312246123,
+      "loss": 3.2347,
+      "step": 4214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024594023648986284,
+      "loss": 3.1266,
+      "step": 4215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024571534916223155,
+      "loss": 3.2098,
+      "step": 4216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002454905312009144,
+      "loss": 3.2493,
+      "step": 4217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002452657826672394,
+      "loss": 3.0256,
+      "step": 4218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002450411036225158,
+      "loss": 3.2188,
+      "step": 4219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002448164941280337,
+      "loss": 3.1295,
+      "step": 4220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024459195424506464,
+      "loss": 3.1127,
+      "step": 4221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024436748403486037,
+      "loss": 3.1494,
+      "step": 4222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002441430835586544,
+      "loss": 3.1721,
+      "step": 4223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002439187528776609,
+      "loss": 3.0297,
+      "step": 4224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024369449205307504,
+      "loss": 3.2309,
+      "step": 4225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024347030114607295,
+      "loss": 3.0079,
+      "step": 4226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024324618021781186,
+      "loss": 3.052,
+      "step": 4227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002430221293294293,
+      "loss": 3.2588,
+      "step": 4228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024279814854204435,
+      "loss": 3.2389,
+      "step": 4229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002425742379167567,
+      "loss": 3.1093,
+      "step": 4230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024235039751464694,
+      "loss": 3.3073,
+      "step": 4231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024212662739677655,
+      "loss": 3.0591,
+      "step": 4232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024190292762418786,
+      "loss": 3.1953,
+      "step": 4233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002416792982579037,
+      "loss": 3.3459,
+      "step": 4234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024145573935892802,
+      "loss": 3.2357,
+      "step": 4235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024123225098824548,
+      "loss": 3.0579,
+      "step": 4236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024100883320682148,
+      "loss": 3.1938,
+      "step": 4237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024078548607560214,
+      "loss": 3.2051,
+      "step": 4238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024056220965551457,
+      "loss": 3.1117,
+      "step": 4239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024033900400746589,
+      "loss": 3.0689,
+      "step": 4240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024011586919234462,
+      "loss": 3.1101,
+      "step": 4241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002398928052710197,
+      "loss": 3.0901,
+      "step": 4242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002396698123043407,
+      "loss": 3.1744,
+      "step": 4243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000239446890353138,
+      "loss": 3.227,
+      "step": 4244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023922403947822252,
+      "loss": 3.3615,
+      "step": 4245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023900125974038545,
+      "loss": 3.2235,
+      "step": 4246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023877855120039905,
+      "loss": 3.2234,
+      "step": 4247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002385559139190162,
+      "loss": 3.4154,
+      "step": 4248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023833334795696955,
+      "loss": 3.2047,
+      "step": 4249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002381108533749734,
+      "loss": 3.2145,
+      "step": 4250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023788843023372209,
+      "loss": 3.1631,
+      "step": 4251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023766607859389006,
+      "loss": 3.1779,
+      "step": 4252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002374437985161328,
+      "loss": 3.171,
+      "step": 4253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023722159006108606,
+      "loss": 3.3265,
+      "step": 4254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002369994532893661,
+      "loss": 3.2081,
+      "step": 4255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002367773882615697,
+      "loss": 3.1658,
+      "step": 4256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023655539503827407,
+      "loss": 3.0066,
+      "step": 4257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023633347368003639,
+      "loss": 3.1828,
+      "step": 4258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002361116242473948,
+      "loss": 3.2217,
+      "step": 4259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023588984680086755,
+      "loss": 3.2927,
+      "step": 4260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023566814140095344,
+      "loss": 3.1804,
+      "step": 4261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023544650810813138,
+      "loss": 3.2444,
+      "step": 4262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023522494698286097,
+      "loss": 2.9836,
+      "step": 4263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023500345808558144,
+      "loss": 3.2735,
+      "step": 4264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023478204147671294,
+      "loss": 3.2904,
+      "step": 4265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002345606972166558,
+      "loss": 3.1513,
+      "step": 4266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023433942536579038,
+      "loss": 3.2664,
+      "step": 4267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023411822598447756,
+      "loss": 3.2827,
+      "step": 4268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002338970991330585,
+      "loss": 3.0973,
+      "step": 4269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023367604487185394,
+      "loss": 3.0368,
+      "step": 4270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002334550632611655,
+      "loss": 3.0457,
+      "step": 4271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023323415436127482,
+      "loss": 3.082,
+      "step": 4272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002330133182324437,
+      "loss": 3.153,
+      "step": 4273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002327925549349136,
+      "loss": 3.0703,
+      "step": 4274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023257186452890706,
+      "loss": 3.1036,
+      "step": 4275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002323512470746262,
+      "loss": 3.3104,
+      "step": 4276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023213070263225282,
+      "loss": 3.0323,
+      "step": 4277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023191023126194955,
+      "loss": 3.063,
+      "step": 4278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023168983302385894,
+      "loss": 3.0741,
+      "step": 4279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023146950797810285,
+      "loss": 3.0996,
+      "step": 4280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023124925618478432,
+      "loss": 3.1348,
+      "step": 4281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002310290777039858,
+      "loss": 3.1846,
+      "step": 4282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023080897259576943,
+      "loss": 3.1313,
+      "step": 4283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002305889409201779,
+      "loss": 3.1428,
+      "step": 4284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023036898273723382,
+      "loss": 3.228,
+      "step": 4285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023014909810693907,
+      "loss": 3.0927,
+      "step": 4286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022992928708927645,
+      "loss": 3.1049,
+      "step": 4287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022970954974420828,
+      "loss": 3.051,
+      "step": 4288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022948988613167632,
+      "loss": 3.287,
+      "step": 4289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022927029631160278,
+      "loss": 2.9917,
+      "step": 4290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022905078034388983,
+      "loss": 3.2523,
+      "step": 4291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022883133828841858,
+      "loss": 3.1499,
+      "step": 4292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022861197020505126,
+      "loss": 3.0912,
+      "step": 4293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022839267615362924,
+      "loss": 3.1126,
+      "step": 4294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022817345619397338,
+      "loss": 3.1517,
+      "step": 4295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022795431038588498,
+      "loss": 3.1545,
+      "step": 4296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022773523878914494,
+      "loss": 3.2291,
+      "step": 4297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022751624146351336,
+      "loss": 3.2285,
+      "step": 4298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022729731846873064,
+      "loss": 3.1711,
+      "step": 4299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002270784698645173,
+      "loss": 3.1627,
+      "step": 4300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022685969571057248,
+      "loss": 3.1063,
+      "step": 4301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022664099606657578,
+      "loss": 3.3439,
+      "step": 4302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022642237099218648,
+      "loss": 3.0449,
+      "step": 4303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022620382054704298,
+      "loss": 3.0033,
+      "step": 4304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002259853447907636,
+      "loss": 3.1899,
+      "step": 4305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002257669437829469,
+      "loss": 3.1683,
+      "step": 4306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022554861758316996,
+      "loss": 3.2902,
+      "step": 4307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002253303662509902,
+      "loss": 3.1518,
+      "step": 4308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022511218984594438,
+      "loss": 3.1665,
+      "step": 4309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022489408842754898,
+      "loss": 3.057,
+      "step": 4310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022467606205529945,
+      "loss": 3.3021,
+      "step": 4311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022445811078867185,
+      "loss": 3.2694,
+      "step": 4312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022424023468712058,
+      "loss": 3.2542,
+      "step": 4313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022402243381008024,
+      "loss": 3.1886,
+      "step": 4314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022380470821696476,
+      "loss": 3.1301,
+      "step": 4315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002235870579671677,
+      "loss": 3.1813,
+      "step": 4316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022336948312006127,
+      "loss": 3.2524,
+      "step": 4317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002231519837349985,
+      "loss": 3.2157,
+      "step": 4318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022293455987131052,
+      "loss": 3.0789,
+      "step": 4319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022271721158830855,
+      "loss": 3.061,
+      "step": 4320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000222499938945283,
+      "loss": 3.0671,
+      "step": 4321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022228274200150383,
+      "loss": 3.1777,
+      "step": 4322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022206562081621996,
+      "loss": 3.1716,
+      "step": 4323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022184857544865994,
+      "loss": 3.1554,
+      "step": 4324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002216316059580316,
+      "loss": 3.1727,
+      "step": 4325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022141471240352212,
+      "loss": 3.0826,
+      "step": 4326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022119789484429785,
+      "loss": 3.1011,
+      "step": 4327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002209811533395047,
+      "loss": 3.1726,
+      "step": 4328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022076448794826708,
+      "loss": 3.2459,
+      "step": 4329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022054789872968928,
+      "loss": 3.1833,
+      "step": 4330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022033138574285515,
+      "loss": 3.1993,
+      "step": 4331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022011494904682682,
+      "loss": 3.1781,
+      "step": 4332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002198985887006461,
+      "loss": 3.1976,
+      "step": 4333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021968230476333424,
+      "loss": 3.1598,
+      "step": 4334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021946609729389088,
+      "loss": 3.0296,
+      "step": 4335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002192499663512953,
+      "loss": 3.1073,
+      "step": 4336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002190339119945064,
+      "loss": 3.2129,
+      "step": 4337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021881793428246116,
+      "loss": 3.0633,
+      "step": 4338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021860203327407624,
+      "loss": 3.1796,
+      "step": 4339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021838620902824758,
+      "loss": 3.1979,
+      "step": 4340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021817046160384934,
+      "loss": 3.0947,
+      "step": 4341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021795479105973542,
+      "loss": 3.133,
+      "step": 4342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021773919745473908,
+      "loss": 3.2023,
+      "step": 4343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021752368084767155,
+      "loss": 3.2168,
+      "step": 4344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002173082412973238,
+      "loss": 3.1128,
+      "step": 4345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021709287886246577,
+      "loss": 3.357,
+      "step": 4346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021687759360184577,
+      "loss": 3.208,
+      "step": 4347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021666238557419178,
+      "loss": 3.1786,
+      "step": 4348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021644725483821025,
+      "loss": 3.1689,
+      "step": 4349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021623220145258683,
+      "loss": 3.0795,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002160172254759858,
+      "loss": 3.1103,
+      "step": 4351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021580232696705076,
+      "loss": 3.2286,
+      "step": 4352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021558750598440346,
+      "loss": 3.2055,
+      "step": 4353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021537276258664511,
+      "loss": 3.1061,
+      "step": 4354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002151580968323556,
+      "loss": 3.1011,
+      "step": 4355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002149435087800935,
+      "loss": 3.1195,
+      "step": 4356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021472899848839645,
+      "loss": 3.3282,
+      "step": 4357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021451456601578056,
+      "loss": 3.0536,
+      "step": 4358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021430021142074113,
+      "loss": 3.302,
+      "step": 4359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002140859347617516,
+      "loss": 2.993,
+      "step": 4360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021387173609726463,
+      "loss": 3.0391,
+      "step": 4361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002136576154857115,
+      "loss": 3.2748,
+      "step": 4362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002134435729855022,
+      "loss": 3.2154,
+      "step": 4363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021322960865502533,
+      "loss": 3.1719,
+      "step": 4364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002130157225526485,
+      "loss": 3.1404,
+      "step": 4365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002128019147367173,
+      "loss": 3.2187,
+      "step": 4366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021258818526555647,
+      "loss": 3.3178,
+      "step": 4367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021237453419746934,
+      "loss": 3.1361,
+      "step": 4368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021216096159073784,
+      "loss": 3.0917,
+      "step": 4369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021194746750362236,
+      "loss": 3.2578,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021173405199436217,
+      "loss": 3.1958,
+      "step": 4371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021152071512117455,
+      "loss": 3.0908,
+      "step": 4372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021130745694225578,
+      "loss": 3.1061,
+      "step": 4373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021109427751578064,
+      "loss": 3.1406,
+      "step": 4374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021088117689990234,
+      "loss": 2.9409,
+      "step": 4375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021066815515275255,
+      "loss": 3.176,
+      "step": 4376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021045521233244169,
+      "loss": 3.2612,
+      "step": 4377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021024234849705809,
+      "loss": 3.2818,
+      "step": 4378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021002956370466902,
+      "loss": 3.1456,
+      "step": 4379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020981685801332013,
+      "loss": 3.0804,
+      "step": 4380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020960423148103525,
+      "loss": 3.2385,
+      "step": 4381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020939168416581695,
+      "loss": 3.1754,
+      "step": 4382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020917921612564606,
+      "loss": 3.1608,
+      "step": 4383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020896682741848143,
+      "loss": 3.1386,
+      "step": 4384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020875451810226081,
+      "loss": 3.1207,
+      "step": 4385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020854228823490001,
+      "loss": 3.2492,
+      "step": 4386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020833013787429323,
+      "loss": 3.2126,
+      "step": 4387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020811806707831299,
+      "loss": 3.202,
+      "step": 4388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020790607590481019,
+      "loss": 3.202,
+      "step": 4389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020769416441161366,
+      "loss": 3.2607,
+      "step": 4390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020748233265653084,
+      "loss": 3.1455,
+      "step": 4391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002072705806973473,
+      "loss": 3.1702,
+      "step": 4392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020705890859182692,
+      "loss": 3.1232,
+      "step": 4393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002068473163977117,
+      "loss": 3.2072,
+      "step": 4394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002066358041727221,
+      "loss": 3.213,
+      "step": 4395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020642437197455617,
+      "loss": 3.1671,
+      "step": 4396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020621301986089064,
+      "loss": 3.1399,
+      "step": 4397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020600174788938026,
+      "loss": 3.1052,
+      "step": 4398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020579055611765797,
+      "loss": 3.1146,
+      "step": 4399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020557944460333478,
+      "loss": 3.1345,
+      "step": 4400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020536841340399988,
+      "loss": 3.0773,
+      "step": 4401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020515746257722024,
+      "loss": 3.1409,
+      "step": 4402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020494659218054124,
+      "loss": 3.009,
+      "step": 4403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020473580227148625,
+      "loss": 3.301,
+      "step": 4404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020452509290755667,
+      "loss": 3.0811,
+      "step": 4405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020431446414623185,
+      "loss": 3.1263,
+      "step": 4406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020410391604496947,
+      "loss": 3.2726,
+      "step": 4407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002038934486612045,
+      "loss": 3.1985,
+      "step": 4408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002036830620523506,
+      "loss": 3.217,
+      "step": 4409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020347275627579915,
+      "loss": 3.0197,
+      "step": 4410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020326253138891936,
+      "loss": 3.1509,
+      "step": 4411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020305238744905857,
+      "loss": 3.2235,
+      "step": 4412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020284232451354189,
+      "loss": 3.056,
+      "step": 4413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020263234263967266,
+      "loss": 3.1263,
+      "step": 4414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020242244188473142,
+      "loss": 3.051,
+      "step": 4415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020221262230597716,
+      "loss": 3.2285,
+      "step": 4416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020200288396064658,
+      "loss": 3.3223,
+      "step": 4417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020179322690595426,
+      "loss": 3.3599,
+      "step": 4418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002015836511990925,
+      "loss": 3.2719,
+      "step": 4419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002013741568972316,
+      "loss": 3.2008,
+      "step": 4420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020116474405751922,
+      "loss": 3.0324,
+      "step": 4421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020095541273708128,
+      "loss": 3.2074,
+      "step": 4422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020074616299302144,
+      "loss": 3.0828,
+      "step": 4423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002005369948824204,
+      "loss": 3.3061,
+      "step": 4424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020032790846233762,
+      "loss": 3.1345,
+      "step": 4425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020011890378980983,
+      "loss": 3.1573,
+      "step": 4426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019990998092185108,
+      "loss": 3.1563,
+      "step": 4427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019970113991545364,
+      "loss": 3.1301,
+      "step": 4428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019949238082758735,
+      "loss": 3.1447,
+      "step": 4429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001992837037151991,
+      "loss": 3.1626,
+      "step": 4430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019907510863521449,
+      "loss": 3.1769,
+      "step": 4431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019886659564453618,
+      "loss": 3.037,
+      "step": 4432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000198658164800044,
+      "loss": 3.1764,
+      "step": 4433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001984498161585961,
+      "loss": 3.0076,
+      "step": 4434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019824154977702795,
+      "loss": 3.2423,
+      "step": 4435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019803336571215212,
+      "loss": 3.1217,
+      "step": 4436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019782526402075963,
+      "loss": 3.1258,
+      "step": 4437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001976172447596185,
+      "loss": 3.3219,
+      "step": 4438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019740930798547407,
+      "loss": 3.1885,
+      "step": 4439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001972014537550495,
+      "loss": 3.1222,
+      "step": 4440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019699368212504554,
+      "loss": 3.0396,
+      "step": 4441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019678599315213968,
+      "loss": 3.2,
+      "step": 4442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019657838689298796,
+      "loss": 3.1946,
+      "step": 4443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019637086340422327,
+      "loss": 3.1946,
+      "step": 4444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019616342274245564,
+      "loss": 2.9829,
+      "step": 4445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019595606496427293,
+      "loss": 3.2215,
+      "step": 4446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001957487901262403,
+      "loss": 3.1195,
+      "step": 4447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001955415982849004,
+      "loss": 3.1998,
+      "step": 4448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019533448949677262,
+      "loss": 3.2898,
+      "step": 4449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019512746381835483,
+      "loss": 3.1303,
+      "step": 4450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019492052130612103,
+      "loss": 3.1352,
+      "step": 4451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019471366201652328,
+      "loss": 3.2404,
+      "step": 4452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019450688600599077,
+      "loss": 3.3935,
+      "step": 4453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019430019333093002,
+      "loss": 3.1514,
+      "step": 4454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019409358404772425,
+      "loss": 3.2582,
+      "step": 4455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019388705821273512,
+      "loss": 3.2168,
+      "step": 4456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001936806158823003,
+      "loss": 3.1718,
+      "step": 4457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019347425711273525,
+      "loss": 3.0814,
+      "step": 4458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019326798196033274,
+      "loss": 3.1877,
+      "step": 4459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001930617904813627,
+      "loss": 3.2138,
+      "step": 4460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019285568273207156,
+      "loss": 3.2122,
+      "step": 4461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019264965876868395,
+      "loss": 3.1165,
+      "step": 4462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019244371864740118,
+      "loss": 3.1591,
+      "step": 4463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001922378624244014,
+      "loss": 3.1526,
+      "step": 4464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019203209015584012,
+      "loss": 3.3269,
+      "step": 4465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001918264018978503,
+      "loss": 3.1421,
+      "step": 4466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019162079770654105,
+      "loss": 3.1378,
+      "step": 4467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019141527763799965,
+      "loss": 3.0,
+      "step": 4468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019120984174828999,
+      "loss": 3.3038,
+      "step": 4469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001910044900934525,
+      "loss": 3.0214,
+      "step": 4470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001907992227295053,
+      "loss": 3.158,
+      "step": 4471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001905940397124434,
+      "loss": 2.9944,
+      "step": 4472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019038894109823834,
+      "loss": 3.2414,
+      "step": 4473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019018392694283905,
+      "loss": 3.1629,
+      "step": 4474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018997899730217177,
+      "loss": 3.3381,
+      "step": 4475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001897741522321388,
+      "loss": 3.0663,
+      "step": 4476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018956939178861994,
+      "loss": 3.0867,
+      "step": 4477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000189364716027472,
+      "loss": 3.3121,
+      "step": 4478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018916012500452817,
+      "loss": 3.1079,
+      "step": 4479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001889556187755988,
+      "loss": 3.3009,
+      "step": 4480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018875119739647172,
+      "loss": 3.1446,
+      "step": 4481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018854686092291046,
+      "loss": 3.2041,
+      "step": 4482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018834260941065611,
+      "loss": 3.1649,
+      "step": 4483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018813844291542677,
+      "loss": 3.1978,
+      "step": 4484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018793436149291647,
+      "loss": 3.1597,
+      "step": 4485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018773036519879676,
+      "loss": 3.1231,
+      "step": 4486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018752645408871626,
+      "loss": 3.292,
+      "step": 4487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018732262821829931,
+      "loss": 3.1241,
+      "step": 4488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018711888764314777,
+      "loss": 3.1956,
+      "step": 4489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018691523241884029,
+      "loss": 3.1702,
+      "step": 4490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018671166260093147,
+      "loss": 2.9963,
+      "step": 4491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001865081782449533,
+      "loss": 3.2701,
+      "step": 4492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018630477940641472,
+      "loss": 3.2058,
+      "step": 4493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018610146614080037,
+      "loss": 3.0433,
+      "step": 4494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018589823850357224,
+      "loss": 3.0629,
+      "step": 4495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018569509655016886,
+      "loss": 3.2751,
+      "step": 4496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018549204033600547,
+      "loss": 3.2882,
+      "step": 4497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018528906991647343,
+      "loss": 3.2119,
+      "step": 4498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018508618534694121,
+      "loss": 3.1328,
+      "step": 4499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018488338668275369,
+      "loss": 3.1851,
+      "step": 4500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018468067397923231,
+      "loss": 3.1355,
+      "step": 4501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018447804729167518,
+      "loss": 3.1353,
+      "step": 4502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018427550667535686,
+      "loss": 3.0431,
+      "step": 4503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018407305218552815,
+      "loss": 3.0502,
+      "step": 4504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018387068387741679,
+      "loss": 3.0584,
+      "step": 4505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018366840180622675,
+      "loss": 3.2285,
+      "step": 4506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018346620602713865,
+      "loss": 3.0916,
+      "step": 4507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018326409659530945,
+      "loss": 3.1314,
+      "step": 4508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018306207356587274,
+      "loss": 3.3056,
+      "step": 4509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018286013699393805,
+      "loss": 3.2546,
+      "step": 4510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001826582869345919,
+      "loss": 3.2522,
+      "step": 4511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018245652344289686,
+      "loss": 3.1361,
+      "step": 4512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018225484657389206,
+      "loss": 3.2023,
+      "step": 4513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001820532563825929,
+      "loss": 3.1521,
+      "step": 4514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018185175292399143,
+      "loss": 3.0769,
+      "step": 4515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018165033625305526,
+      "loss": 3.104,
+      "step": 4516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018144900642472905,
+      "loss": 3.1297,
+      "step": 4517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000181247763493934,
+      "loss": 3.0331,
+      "step": 4518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001810466075155666,
+      "loss": 3.1865,
+      "step": 4519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018084553854450048,
+      "loss": 2.997,
+      "step": 4520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018064455663558532,
+      "loss": 3.3253,
+      "step": 4521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001804436618436467,
+      "loss": 3.2662,
+      "step": 4522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018024285422348679,
+      "loss": 3.2048,
+      "step": 4523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018004213382988405,
+      "loss": 3.1431,
+      "step": 4524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017984150071759288,
+      "loss": 3.0896,
+      "step": 4525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017964095494134403,
+      "loss": 3.113,
+      "step": 4526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017944049655584454,
+      "loss": 3.1699,
+      "step": 4527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017924012561577714,
+      "loss": 3.3035,
+      "step": 4528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017903984217580116,
+      "loss": 3.225,
+      "step": 4529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000178839646290552,
+      "loss": 3.1091,
+      "step": 4530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000178639538014641,
+      "loss": 3.1416,
+      "step": 4531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017843951740265578,
+      "loss": 3.1746,
+      "step": 4532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001782395845091601,
+      "loss": 3.1843,
+      "step": 4533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001780397393886933,
+      "loss": 3.2838,
+      "step": 4534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017783998209577135,
+      "loss": 3.1648,
+      "step": 4535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017764031268488595,
+      "loss": 3.1085,
+      "step": 4536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017744073121050508,
+      "loss": 3.1781,
+      "step": 4537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001772412377270724,
+      "loss": 3.2897,
+      "step": 4538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017704183228900805,
+      "loss": 3.1062,
+      "step": 4539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001768425149507074,
+      "loss": 3.1701,
+      "step": 4540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001766432857665425,
+      "loss": 3.1701,
+      "step": 4541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017644414479086102,
+      "loss": 3.1457,
+      "step": 4542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017624509207798662,
+      "loss": 3.0751,
+      "step": 4543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001760461276822189,
+      "loss": 3.1069,
+      "step": 4544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017584725165783354,
+      "loss": 3.1885,
+      "step": 4545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017564846405908164,
+      "loss": 3.0473,
+      "step": 4546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017544976494019056,
+      "loss": 3.057,
+      "step": 4547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017525115435536347,
+      "loss": 3.2509,
+      "step": 4548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017505263235877944,
+      "loss": 3.1918,
+      "step": 4549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001748541990045931,
+      "loss": 3.0851,
+      "step": 4550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017465585434693525,
+      "loss": 3.047,
+      "step": 4551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017445759843991244,
+      "loss": 3.2151,
+      "step": 4552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017425943133760657,
+      "loss": 3.0513,
+      "step": 4553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017406135309407583,
+      "loss": 3.148,
+      "step": 4554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017386336376335398,
+      "loss": 3.1805,
+      "step": 4555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001736654633994505,
+      "loss": 3.1815,
+      "step": 4556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017346765205635068,
+      "loss": 3.1679,
+      "step": 4557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017326992978801563,
+      "loss": 3.1974,
+      "step": 4558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001730722966483817,
+      "loss": 3.1465,
+      "step": 4559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017287475269136133,
+      "loss": 3.1986,
+      "step": 4560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017267729797084265,
+      "loss": 3.129,
+      "step": 4561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017247993254068921,
+      "loss": 3.1412,
+      "step": 4562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001722826564547404,
+      "loss": 3.1509,
+      "step": 4563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017208546976681138,
+      "loss": 3.1893,
+      "step": 4564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017188837253069228,
+      "loss": 3.1064,
+      "step": 4565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017169136480014947,
+      "loss": 3.0647,
+      "step": 4566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017149444662892472,
+      "loss": 3.2034,
+      "step": 4567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001712976180707353,
+      "loss": 3.1724,
+      "step": 4568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017110087917927407,
+      "loss": 2.9871,
+      "step": 4569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017090423000820965,
+      "loss": 3.1078,
+      "step": 4570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001707076706111856,
+      "loss": 3.1918,
+      "step": 4571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001705112010418215,
+      "loss": 3.1108,
+      "step": 4572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017031482135371252,
+      "loss": 3.251,
+      "step": 4573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017011853160042855,
+      "loss": 3.3603,
+      "step": 4574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016992233183551596,
+      "loss": 3.0229,
+      "step": 4575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001697262221124961,
+      "loss": 3.0078,
+      "step": 4576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016953020248486545,
+      "loss": 3.1344,
+      "step": 4577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016933427300609626,
+      "loss": 3.0572,
+      "step": 4578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016913843372963645,
+      "loss": 3.1619,
+      "step": 4579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016894268470890844,
+      "loss": 3.0956,
+      "step": 4580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016874702599731106,
+      "loss": 3.0275,
+      "step": 4581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016855145764821817,
+      "loss": 3.1206,
+      "step": 4582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016835597971497847,
+      "loss": 3.1665,
+      "step": 4583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001681605922509165,
+      "loss": 3.2499,
+      "step": 4584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001679652953093321,
+      "loss": 3.2056,
+      "step": 4585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016777008894350027,
+      "loss": 3.1612,
+      "step": 4586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016757497320667137,
+      "loss": 3.1377,
+      "step": 4587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016737994815207126,
+      "loss": 3.0569,
+      "step": 4588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016718501383290035,
+      "loss": 3.0067,
+      "step": 4589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016699017030233509,
+      "loss": 3.0986,
+      "step": 4590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016679541761352673,
+      "loss": 3.1474,
+      "step": 4591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016660075581960198,
+      "loss": 3.0609,
+      "step": 4592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016640618497366255,
+      "loss": 3.1537,
+      "step": 4593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016621170512878559,
+      "loss": 3.1354,
+      "step": 4594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001660173163380229,
+      "loss": 3.0581,
+      "step": 4595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016582301865440213,
+      "loss": 3.3011,
+      "step": 4596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016562881213092557,
+      "loss": 3.0772,
+      "step": 4597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016543469682057105,
+      "loss": 3.1538,
+      "step": 4598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001652406727762908,
+      "loss": 3.0836,
+      "step": 4599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001650467400510131,
+      "loss": 3.0759,
+      "step": 4600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001648528986976409,
+      "loss": 3.1682,
+      "step": 4601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016465914876905192,
+      "loss": 3.1532,
+      "step": 4602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001644654903180992,
+      "loss": 3.1793,
+      "step": 4603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016427192339761115,
+      "loss": 3.1389,
+      "step": 4604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016407844806039034,
+      "loss": 3.2438,
+      "step": 4605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016388506435921542,
+      "loss": 3.1604,
+      "step": 4606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016369177234683952,
+      "loss": 3.2712,
+      "step": 4607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016349857207599046,
+      "loss": 3.1298,
+      "step": 4608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016330546359937144,
+      "loss": 3.0876,
+      "step": 4609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016311244696966075,
+      "loss": 3.1551,
+      "step": 4610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016291952223951094,
+      "loss": 3.056,
+      "step": 4611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016272668946155038,
+      "loss": 3.0544,
+      "step": 4612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016253394868838195,
+      "loss": 3.0717,
+      "step": 4613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016234129997258308,
+      "loss": 3.0896,
+      "step": 4614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001621487433667066,
+      "loss": 3.0597,
+      "step": 4615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016195627892328014,
+      "loss": 3.1656,
+      "step": 4616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016176390669480568,
+      "loss": 3.0709,
+      "step": 4617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001615716267337608,
+      "loss": 3.0979,
+      "step": 4618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016137943909259766,
+      "loss": 3.1539,
+      "step": 4619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016118734382374278,
+      "loss": 3.1911,
+      "step": 4620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000160995340979598,
+      "loss": 3.1721,
+      "step": 4621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016080343061253993,
+      "loss": 3.2664,
+      "step": 4622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001606116127749194,
+      "loss": 3.1529,
+      "step": 4623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001604198875190626,
+      "loss": 3.0861,
+      "step": 4624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016022825489727055,
+      "loss": 2.9629,
+      "step": 4625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016003671496181833,
+      "loss": 3.1157,
+      "step": 4626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015984526776495627,
+      "loss": 3.2161,
+      "step": 4627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001596539133589094,
+      "loss": 3.1533,
+      "step": 4628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015946265179587698,
+      "loss": 3.0633,
+      "step": 4629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015927148312803324,
+      "loss": 3.0741,
+      "step": 4630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001590804074075276,
+      "loss": 3.2475,
+      "step": 4631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015888942468648309,
+      "loss": 3.0978,
+      "step": 4632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001586985350169981,
+      "loss": 3.2043,
+      "step": 4633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015850773845114534,
+      "loss": 3.178,
+      "step": 4634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001583170350409725,
+      "loss": 3.0867,
+      "step": 4635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015812642483850094,
+      "loss": 3.1799,
+      "step": 4636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015793590789572792,
+      "loss": 2.9196,
+      "step": 4637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015774548426462416,
+      "loss": 3.3066,
+      "step": 4638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015755515399713533,
+      "loss": 3.2982,
+      "step": 4639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015736491714518175,
+      "loss": 3.1342,
+      "step": 4640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015717477376065824,
+      "loss": 3.2029,
+      "step": 4641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001569847238954336,
+      "loss": 3.1513,
+      "step": 4642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015679476760135209,
+      "loss": 3.3632,
+      "step": 4643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015660490493023144,
+      "loss": 3.0427,
+      "step": 4644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015641513593386448,
+      "loss": 3.0676,
+      "step": 4645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015622546066401826,
+      "loss": 3.1386,
+      "step": 4646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001560358791724345,
+      "loss": 3.2193,
+      "step": 4647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015584639151082873,
+      "loss": 3.126,
+      "step": 4648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001556569977308916,
+      "loss": 3.0454,
+      "step": 4649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015546769788428772,
+      "loss": 3.0527,
+      "step": 4650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015527849202265625,
+      "loss": 3.1509,
+      "step": 4651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015508938019761066,
+      "loss": 3.0953,
+      "step": 4652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001549003624607389,
+      "loss": 3.111,
+      "step": 4653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015471143886360277,
+      "loss": 3.1105,
+      "step": 4654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001545226094577389,
+      "loss": 2.9829,
+      "step": 4655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015433387429465833,
+      "loss": 3.1464,
+      "step": 4656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001541452334258458,
+      "loss": 3.3385,
+      "step": 4657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015395668690276076,
+      "loss": 3.0799,
+      "step": 4658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015376823477683698,
+      "loss": 3.0852,
+      "step": 4659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000153579877099482,
+      "loss": 3.1752,
+      "step": 4660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015339161392207795,
+      "loss": 3.116,
+      "step": 4661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015320344529598162,
+      "loss": 3.1615,
+      "step": 4662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015301537127252295,
+      "loss": 3.2399,
+      "step": 4663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015282739190300697,
+      "loss": 3.2792,
+      "step": 4664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001526395072387126,
+      "loss": 3.1532,
+      "step": 4665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001524517173308927,
+      "loss": 3.0397,
+      "step": 4666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001522640222307744,
+      "loss": 3.0568,
+      "step": 4667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015207642198955957,
+      "loss": 3.1239,
+      "step": 4668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015188891665842314,
+      "loss": 3.0319,
+      "step": 4669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015170150628851493,
+      "loss": 3.0856,
+      "step": 4670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015151419093095876,
+      "loss": 3.1884,
+      "step": 4671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015132697063685203,
+      "loss": 3.0688,
+      "step": 4672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015113984545726673,
+      "loss": 3.2576,
+      "step": 4673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001509528154432488,
+      "loss": 3.242,
+      "step": 4674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001507658806458181,
+      "loss": 3.2179,
+      "step": 4675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001505790411159686,
+      "loss": 3.2161,
+      "step": 4676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015039229690466845,
+      "loss": 3.2918,
+      "step": 4677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015020564806285925,
+      "loss": 3.0003,
+      "step": 4678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015001909464145707,
+      "loss": 3.082,
+      "step": 4679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014983263669135187,
+      "loss": 3.3381,
+      "step": 4680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014964627426340755,
+      "loss": 3.1087,
+      "step": 4681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014946000740846183,
+      "loss": 3.2561,
+      "step": 4682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014927383617732666,
+      "loss": 3.3689,
+      "step": 4683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014908776062078733,
+      "loss": 3.2829,
+      "step": 4684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014890178078960366,
+      "loss": 3.2179,
+      "step": 4685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014871589673450897,
+      "loss": 3.1368,
+      "step": 4686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014853010850621062,
+      "loss": 3.1453,
+      "step": 4687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014834441615538984,
+      "loss": 3.0794,
+      "step": 4688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014815881973270157,
+      "loss": 3.2073,
+      "step": 4689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014797331928877484,
+      "loss": 3.072,
+      "step": 4690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000147787914874212,
+      "loss": 3.0164,
+      "step": 4691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014760260653958973,
+      "loss": 3.0295,
+      "step": 4692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014741739433545825,
+      "loss": 3.1035,
+      "step": 4693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014723227831234159,
+      "loss": 3.2157,
+      "step": 4694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001470472585207376,
+      "loss": 3.1863,
+      "step": 4695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001468623350111179,
+      "loss": 3.125,
+      "step": 4696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014667750783392753,
+      "loss": 3.0151,
+      "step": 4697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001464927770395857,
+      "loss": 3.2317,
+      "step": 4698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014630814267848502,
+      "loss": 3.179,
+      "step": 4699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000146123604800992,
+      "loss": 3.1452,
+      "step": 4700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014593916345744667,
+      "loss": 3.1667,
+      "step": 4701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014575481869816292,
+      "loss": 3.1468,
+      "step": 4702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014557057057342792,
+      "loss": 3.0361,
+      "step": 4703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014538641913350287,
+      "loss": 3.0457,
+      "step": 4704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014520236442862238,
+      "loss": 3.218,
+      "step": 4705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014501840650899477,
+      "loss": 3.2558,
+      "step": 4706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001448345454248019,
+      "loss": 3.2224,
+      "step": 4707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014465078122619945,
+      "loss": 3.0929,
+      "step": 4708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014446711396331614,
+      "loss": 3.2538,
+      "step": 4709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014428354368625467,
+      "loss": 3.0254,
+      "step": 4710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014410007044509123,
+      "loss": 3.0677,
+      "step": 4711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001439166942898754,
+      "loss": 3.1195,
+      "step": 4712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001437334152706305,
+      "loss": 3.243,
+      "step": 4713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001435502334373533,
+      "loss": 3.1226,
+      "step": 4714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014336714884001362,
+      "loss": 3.1822,
+      "step": 4715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014318416152855535,
+      "loss": 3.2075,
+      "step": 4716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014300127155289572,
+      "loss": 3.0089,
+      "step": 4717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014281847896292484,
+      "loss": 3.3334,
+      "step": 4718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014263578380850706,
+      "loss": 3.0239,
+      "step": 4719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014245318613947998,
+      "loss": 3.0748,
+      "step": 4720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014227068600565395,
+      "loss": 3.1174,
+      "step": 4721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014208828345681345,
+      "loss": 3.019,
+      "step": 4722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014190597854271602,
+      "loss": 3.2834,
+      "step": 4723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014172377131309267,
+      "loss": 3.1406,
+      "step": 4724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001415416618176476,
+      "loss": 3.1112,
+      "step": 4725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014135965010605878,
+      "loss": 3.1568,
+      "step": 4726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001411777362279767,
+      "loss": 3.0292,
+      "step": 4727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014099592023302594,
+      "loss": 3.1026,
+      "step": 4728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014081420217080403,
+      "loss": 3.1332,
+      "step": 4729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014063258209088186,
+      "loss": 3.119,
+      "step": 4730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014045106004280362,
+      "loss": 3.0669,
+      "step": 4731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001402696360760868,
+      "loss": 3.2539,
+      "step": 4732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014008831024022166,
+      "loss": 3.3096,
+      "step": 4733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013990708258467243,
+      "loss": 3.2309,
+      "step": 4734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001397259531588761,
+      "loss": 3.0476,
+      "step": 4735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013954492201224289,
+      "loss": 3.0703,
+      "step": 4736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013936398919415645,
+      "loss": 3.0459,
+      "step": 4737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013918315475397336,
+      "loss": 3.1504,
+      "step": 4738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001390024187410236,
+      "loss": 3.1693,
+      "step": 4739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001388217812046098,
+      "loss": 3.2382,
+      "step": 4740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013864124219400837,
+      "loss": 3.1005,
+      "step": 4741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001384608017584686,
+      "loss": 3.0691,
+      "step": 4742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013828045994721238,
+      "loss": 3.193,
+      "step": 4743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013810021680943563,
+      "loss": 3.2702,
+      "step": 4744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013792007239430687,
+      "loss": 3.0813,
+      "step": 4745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013774002675096743,
+      "loss": 3.175,
+      "step": 4746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013756007992853209,
+      "loss": 3.1572,
+      "step": 4747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013738023197608866,
+      "loss": 3.1296,
+      "step": 4748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001372004829426975,
+      "loss": 3.2493,
+      "step": 4749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001370208328773927,
+      "loss": 3.227,
+      "step": 4750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001368412818291811,
+      "loss": 3.2034,
+      "step": 4751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013666182984704207,
+      "loss": 3.1294,
+      "step": 4752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013648247697992848,
+      "loss": 3.1629,
+      "step": 4753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001363032232767662,
+      "loss": 3.1215,
+      "step": 4754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013612406878645334,
+      "loss": 3.1865,
+      "step": 4755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013594501355786197,
+      "loss": 3.1684,
+      "step": 4756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001357660576398365,
+      "loss": 3.3268,
+      "step": 4757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013558720108119404,
+      "loss": 3.1857,
+      "step": 4758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013540844393072505,
+      "loss": 3.1692,
+      "step": 4759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013522978623719279,
+      "loss": 3.1543,
+      "step": 4760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001350512280493329,
+      "loss": 3.2336,
+      "step": 4761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001348727694158547,
+      "loss": 3.1581,
+      "step": 4762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013469441038543994,
+      "loss": 3.1,
+      "step": 4763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013451615100674286,
+      "loss": 3.195,
+      "step": 4764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013433799132839098,
+      "loss": 3.0023,
+      "step": 4765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001341599313989847,
+      "loss": 3.1391,
+      "step": 4766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001339819712670966,
+      "loss": 3.1696,
+      "step": 4767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013380411098127244,
+      "loss": 2.9665,
+      "step": 4768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013362635059003126,
+      "loss": 3.1726,
+      "step": 4769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013344869014186378,
+      "loss": 3.1854,
+      "step": 4770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001332711296852342,
+      "loss": 3.0604,
+      "step": 4771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013309366926857923,
+      "loss": 3.1276,
+      "step": 4772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001329163089403085,
+      "loss": 3.1862,
+      "step": 4773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013273904874880354,
+      "loss": 3.0583,
+      "step": 4774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013256188874241986,
+      "loss": 3.2679,
+      "step": 4775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001323848289694845,
+      "loss": 3.1481,
+      "step": 4776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013220786947829778,
+      "loss": 3.1598,
+      "step": 4777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013203101031713239,
+      "loss": 3.2316,
+      "step": 4778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013185425153423391,
+      "loss": 3.264,
+      "step": 4779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013167759317782003,
+      "loss": 3.1103,
+      "step": 4780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013150103529608183,
+      "loss": 3.0893,
+      "step": 4781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013132457793718218,
+      "loss": 3.1284,
+      "step": 4782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013114822114925701,
+      "loss": 3.2,
+      "step": 4783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013097196498041474,
+      "loss": 3.1236,
+      "step": 4784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013079580947873638,
+      "loss": 3.1939,
+      "step": 4785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013061975469227493,
+      "loss": 3.1774,
+      "step": 4786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001304438006690571,
+      "loss": 3.1344,
+      "step": 4787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013026794745708077,
+      "loss": 3.1989,
+      "step": 4788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001300921951043172,
+      "loss": 3.1928,
+      "step": 4789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012991654365870987,
+      "loss": 3.1758,
+      "step": 4790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012974099316817483,
+      "loss": 3.2121,
+      "step": 4791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001295655436806003,
+      "loss": 3.1831,
+      "step": 4792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012939019524384703,
+      "loss": 3.1475,
+      "step": 4793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001292149479057489,
+      "loss": 3.1872,
+      "step": 4794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012903980171411107,
+      "loss": 3.3097,
+      "step": 4795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012886475671671188,
+      "loss": 3.1364,
+      "step": 4796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000128689812961302,
+      "loss": 3.1821,
+      "step": 4797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012851497049560406,
+      "loss": 3.2509,
+      "step": 4798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012834022936731332,
+      "loss": 3.0814,
+      "step": 4799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012816558962409785,
+      "loss": 3.1439,
+      "step": 4800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012799105131359719,
+      "loss": 3.2606,
+      "step": 4801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001278166144834238,
+      "loss": 3.1709,
+      "step": 4802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012764227918116245,
+      "loss": 3.215,
+      "step": 4803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001274680454543698,
+      "loss": 3.1228,
+      "step": 4804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000127293913350575,
+      "loss": 3.2092,
+      "step": 4805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001271198829172801,
+      "loss": 3.1427,
+      "step": 4806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012694595420195836,
+      "loss": 3.0412,
+      "step": 4807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012677212725205594,
+      "loss": 3.0979,
+      "step": 4808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012659840211499124,
+      "loss": 3.1395,
+      "step": 4809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012642477883815446,
+      "loss": 3.2791,
+      "step": 4810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001262512574689083,
+      "loss": 3.0985,
+      "step": 4811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012607783805458812,
+      "loss": 3.1743,
+      "step": 4812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012590452064250053,
+      "loss": 3.1178,
+      "step": 4813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012573130527992492,
+      "loss": 3.2321,
+      "step": 4814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001255581920141129,
+      "loss": 2.9761,
+      "step": 4815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012538518089228774,
+      "loss": 3.1456,
+      "step": 4816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001252122719616453,
+      "loss": 3.2656,
+      "step": 4817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001250394652693534,
+      "loss": 3.2394,
+      "step": 4818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012486676086255195,
+      "loss": 3.1443,
+      "step": 4819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000124694158788353,
+      "loss": 3.0552,
+      "step": 4820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001245216590938409,
+      "loss": 3.1,
+      "step": 4821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012434926182607143,
+      "loss": 3.0593,
+      "step": 4822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012417696703207303,
+      "loss": 3.1442,
+      "step": 4823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012400477475884604,
+      "loss": 3.1415,
+      "step": 4824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012383268505336277,
+      "loss": 3.1526,
+      "step": 4825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012366069796256756,
+      "loss": 3.1983,
+      "step": 4826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001234888135333768,
+      "loss": 3.1195,
+      "step": 4827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000123317031812679,
+      "loss": 3.0281,
+      "step": 4828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012314535284733413,
+      "loss": 3.1642,
+      "step": 4829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012297377668417476,
+      "loss": 3.2705,
+      "step": 4830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012280230337000504,
+      "loss": 3.2031,
+      "step": 4831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001226309329516012,
+      "loss": 3.2556,
+      "step": 4832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012245966547571146,
+      "loss": 3.203,
+      "step": 4833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012228850098905598,
+      "loss": 3.052,
+      "step": 4834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012211743953832637,
+      "loss": 3.1725,
+      "step": 4835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012194648117018676,
+      "loss": 3.173,
+      "step": 4836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012177562593127272,
+      "loss": 3.1276,
+      "step": 4837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012160487386819202,
+      "loss": 3.0708,
+      "step": 4838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012143422502752399,
+      "loss": 3.1303,
+      "step": 4839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012126367945582017,
+      "loss": 3.1881,
+      "step": 4840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012109323719960336,
+      "loss": 3.0731,
+      "step": 4841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012092289830536857,
+      "loss": 3.144,
+      "step": 4842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012075266281958269,
+      "loss": 3.2424,
+      "step": 4843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001205825307886842,
+      "loss": 3.3722,
+      "step": 4844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001204125022590834,
+      "loss": 3.1274,
+      "step": 4845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012024257727716253,
+      "loss": 3.1859,
+      "step": 4846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012007275588927519,
+      "loss": 3.228,
+      "step": 4847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011990303814174702,
+      "loss": 3.2129,
+      "step": 4848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001197334240808754,
+      "loss": 3.084,
+      "step": 4849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001195639137529293,
+      "loss": 3.1734,
+      "step": 4850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011939450720414946,
+      "loss": 3.2095,
+      "step": 4851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011922520448074841,
+      "loss": 3.0224,
+      "step": 4852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011905600562891006,
+      "loss": 3.1346,
+      "step": 4853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011888691069479018,
+      "loss": 3.0641,
+      "step": 4854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011871791972451628,
+      "loss": 3.0481,
+      "step": 4855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011854903276418738,
+      "loss": 3.1751,
+      "step": 4856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011838024985987411,
+      "loss": 3.0715,
+      "step": 4857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011821157105761898,
+      "loss": 3.0544,
+      "step": 4858
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011804299640343551,
+      "loss": 3.1181,
+      "step": 4859
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011787452594330944,
+      "loss": 3.4344,
+      "step": 4860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011770615972319771,
+      "loss": 3.0966,
+      "step": 4861
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011753789778902902,
+      "loss": 3.2174,
+      "step": 4862
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001173697401867035,
+      "loss": 3.1807,
+      "step": 4863
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011720168696209305,
+      "loss": 3.0867,
+      "step": 4864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011703373816104062,
+      "loss": 3.3051,
+      "step": 4865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011686589382936108,
+      "loss": 3.0291,
+      "step": 4866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011669815401284073,
+      "loss": 3.2071,
+      "step": 4867
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011653051875723725,
+      "loss": 3.0775,
+      "step": 4868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011636298810827994,
+      "loss": 3.2901,
+      "step": 4869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001161955621116696,
+      "loss": 3.1684,
+      "step": 4870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011602824081307805,
+      "loss": 3.0837,
+      "step": 4871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001158610242581491,
+      "loss": 2.9655,
+      "step": 4872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011569391249249766,
+      "loss": 3.1339,
+      "step": 4873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011552690556171019,
+      "loss": 3.1537,
+      "step": 4874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011536000351134457,
+      "loss": 3.3432,
+      "step": 4875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011519320638692993,
+      "loss": 3.1176,
+      "step": 4876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011502651423396698,
+      "loss": 3.2177,
+      "step": 4877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011485992709792753,
+      "loss": 3.1806,
+      "step": 4878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011469344502425488,
+      "loss": 3.1027,
+      "step": 4879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011452706805836376,
+      "loss": 3.0263,
+      "step": 4880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011436079624564011,
+      "loss": 3.0376,
+      "step": 4881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011419462963144129,
+      "loss": 3.2135,
+      "step": 4882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011402856826109592,
+      "loss": 3.2435,
+      "step": 4883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011386261217990368,
+      "loss": 3.1854,
+      "step": 4884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001136967614331359,
+      "loss": 3.3745,
+      "step": 4885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011353101606603489,
+      "loss": 3.1521,
+      "step": 4886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011336537612381448,
+      "loss": 2.9812,
+      "step": 4887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011319984165165947,
+      "loss": 3.1545,
+      "step": 4888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001130344126947262,
+      "loss": 3.121,
+      "step": 4889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011286908929814176,
+      "loss": 3.2406,
+      "step": 4890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011270387150700489,
+      "loss": 3.1395,
+      "step": 4891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011253875936638542,
+      "loss": 3.1566,
+      "step": 4892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011237375292132395,
+      "loss": 3.1297,
+      "step": 4893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011220885221683297,
+      "loss": 3.0869,
+      "step": 4894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011204405729789574,
+      "loss": 3.2534,
+      "step": 4895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011187936820946642,
+      "loss": 3.2705,
+      "step": 4896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011171478499647064,
+      "loss": 3.0887,
+      "step": 4897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011155030770380525,
+      "loss": 3.0535,
+      "step": 4898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001113859363763376,
+      "loss": 3.0782,
+      "step": 4899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001112216710589069,
+      "loss": 3.0684,
+      "step": 4900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011105751179632318,
+      "loss": 3.0915,
+      "step": 4901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011089345863336703,
+      "loss": 3.1393,
+      "step": 4902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011072951161479083,
+      "loss": 3.0918,
+      "step": 4903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011056567078531771,
+      "loss": 3.1304,
+      "step": 4904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001104019361896414,
+      "loss": 3.1763,
+      "step": 4905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001102383078724275,
+      "loss": 3.2028,
+      "step": 4906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011007478587831222,
+      "loss": 3.2097,
+      "step": 4907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010991137025190239,
+      "loss": 3.0088,
+      "step": 4908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010974806103777635,
+      "loss": 3.0237,
+      "step": 4909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001095848582804832,
+      "loss": 3.1475,
+      "step": 4910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010942176202454302,
+      "loss": 3.2141,
+      "step": 4911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010925877231444687,
+      "loss": 3.0692,
+      "step": 4912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001090958891946568,
+      "loss": 3.2114,
+      "step": 4913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010893311270960542,
+      "loss": 3.2398,
+      "step": 4914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001087704429036967,
+      "loss": 3.1053,
+      "step": 4915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010860787982130538,
+      "loss": 3.217,
+      "step": 4916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001084454235067771,
+      "loss": 3.0064,
+      "step": 4917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000108283074004428,
+      "loss": 3.1373,
+      "step": 4918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010812083135854589,
+      "loss": 3.1779,
+      "step": 4919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001079586956133885,
+      "loss": 3.0326,
+      "step": 4920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010779666681318507,
+      "loss": 3.078,
+      "step": 4921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010763474500213539,
+      "loss": 3.0738,
+      "step": 4922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010747293022441024,
+      "loss": 3.1753,
+      "step": 4923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010731122252415065,
+      "loss": 3.1369,
+      "step": 4924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001071496219454695,
+      "loss": 3.1852,
+      "step": 4925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010698812853244932,
+      "loss": 3.0958,
+      "step": 4926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010682674232914407,
+      "loss": 3.0674,
+      "step": 4927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010666546337957828,
+      "loss": 3.1281,
+      "step": 4928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001065042917277474,
+      "loss": 3.2644,
+      "step": 4929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010634322741761699,
+      "loss": 3.0315,
+      "step": 4930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010618227049312423,
+      "loss": 3.2259,
+      "step": 4931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010602142099817658,
+      "loss": 3.0497,
+      "step": 4932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010586067897665186,
+      "loss": 3.1686,
+      "step": 4933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010570004447239906,
+      "loss": 3.0573,
+      "step": 4934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010553951752923779,
+      "loss": 3.1807,
+      "step": 4935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010537909819095781,
+      "loss": 3.3023,
+      "step": 4936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010521878650132027,
+      "loss": 3.1906,
+      "step": 4937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010505858250405664,
+      "loss": 3.2826,
+      "step": 4938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010489848624286868,
+      "loss": 3.0047,
+      "step": 4939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010473849776142924,
+      "loss": 3.2077,
+      "step": 4940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010457861710338167,
+      "loss": 3.1718,
+      "step": 4941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001044188443123395,
+      "loss": 3.1703,
+      "step": 4942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010425917943188728,
+      "loss": 3.1818,
+      "step": 4943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010409962250558025,
+      "loss": 3.0449,
+      "step": 4944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010394017357694368,
+      "loss": 3.0185,
+      "step": 4945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010378083268947369,
+      "loss": 3.1774,
+      "step": 4946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010362159988663699,
+      "loss": 3.0863,
+      "step": 4947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010346247521187058,
+      "loss": 3.0478,
+      "step": 4948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010330345870858194,
+      "loss": 2.9725,
+      "step": 4949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010314455042014964,
+      "loss": 3.0298,
+      "step": 4950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010298575038992186,
+      "loss": 3.1567,
+      "step": 4951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010282705866121778,
+      "loss": 3.0787,
+      "step": 4952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010266847527732714,
+      "loss": 3.2465,
+      "step": 4953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010251000028150954,
+      "loss": 3.0574,
+      "step": 4954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010235163371699541,
+      "loss": 3.1112,
+      "step": 4955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010219337562698594,
+      "loss": 3.1573,
+      "step": 4956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010203522605465204,
+      "loss": 3.094,
+      "step": 4957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010187718504313537,
+      "loss": 3.1952,
+      "step": 4958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010171925263554815,
+      "loss": 3.1114,
+      "step": 4959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010156142887497244,
+      "loss": 3.099,
+      "step": 4960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000101403713804461,
+      "loss": 3.211,
+      "step": 4961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010124610746703738,
+      "loss": 3.1278,
+      "step": 4962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010108860990569452,
+      "loss": 3.1937,
+      "step": 4963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010093122116339642,
+      "loss": 3.2154,
+      "step": 4964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010077394128307704,
+      "loss": 3.1238,
+      "step": 4965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010061677030764105,
+      "loss": 3.1657,
+      "step": 4966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001004597082799627,
+      "loss": 3.0789,
+      "step": 4967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010030275524288718,
+      "loss": 3.1792,
+      "step": 4968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010014591123922962,
+      "loss": 3.237,
+      "step": 4969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.998917631177557e-05,
+      "loss": 3.006,
+      "step": 4970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.983255050328077e-05,
+      "loss": 3.1297,
+      "step": 4971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.967603385647128e-05,
+      "loss": 3.1067,
+      "step": 4972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.9519626414043e-05,
+      "loss": 3.2081,
+      "step": 4973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.936332821866239e-05,
+      "loss": 3.2674,
+      "step": 4974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.92071393129662e-05,
+      "loss": 3.0694,
+      "step": 4975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.905105973956107e-05,
+      "loss": 3.2677,
+      "step": 4976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.8895089541024e-05,
+      "loss": 3.1751,
+      "step": 4977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.87392287599022e-05,
+      "loss": 3.0518,
+      "step": 4978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.858347743871277e-05,
+      "loss": 3.1192,
+      "step": 4979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.842783561994295e-05,
+      "loss": 3.2888,
+      "step": 4980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.827230334605086e-05,
+      "loss": 3.1125,
+      "step": 4981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.811688065946361e-05,
+      "loss": 3.3754,
+      "step": 4982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.796156760257912e-05,
+      "loss": 3.1284,
+      "step": 4983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78063642177654e-05,
+      "loss": 3.0273,
+      "step": 4984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.765127054736011e-05,
+      "loss": 3.1209,
+      "step": 4985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.749628663367127e-05,
+      "loss": 2.9797,
+      "step": 4986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.734141251897733e-05,
+      "loss": 3.3267,
+      "step": 4987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.718664824552604e-05,
+      "loss": 3.1869,
+      "step": 4988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.703199385553558e-05,
+      "loss": 3.1266,
+      "step": 4989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.68774493911943e-05,
+      "loss": 3.0127,
+      "step": 4990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.672301489466023e-05,
+      "loss": 3.1857,
+      "step": 4991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.656869040806155e-05,
+      "loss": 3.2729,
+      "step": 4992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.641447597349651e-05,
+      "loss": 3.2776,
+      "step": 4993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.626037163303319e-05,
+      "loss": 3.0817,
+      "step": 4994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.610637742870987e-05,
+      "loss": 3.0875,
+      "step": 4995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.595249340253459e-05,
+      "loss": 3.168,
+      "step": 4996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.579871959648523e-05,
+      "loss": 2.8711,
+      "step": 4997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.564505605250984e-05,
+      "loss": 2.9325,
+      "step": 4998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-05,
+      "loss": 3.0769,
+      "step": 4999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.533805991842242e-05,
+      "loss": 3.1756,
+      "step": 5000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.518472741205586e-05,
+      "loss": 3.1727,
+      "step": 5001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.503150533525435e-05,
+      "loss": 3.1319,
+      "step": 5002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.487839372981505e-05,
+      "loss": 3.1683,
+      "step": 5003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.472539263750546e-05,
+      "loss": 3.1614,
+      "step": 5004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.457250210006274e-05,
+      "loss": 3.0727,
+      "step": 5005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.441972215919387e-05,
+      "loss": 3.1181,
+      "step": 5006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.426705285657567e-05,
+      "loss": 3.1921,
+      "step": 5007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.411449423385499e-05,
+      "loss": 3.108,
+      "step": 5008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.396204633264798e-05,
+      "loss": 3.0903,
+      "step": 5009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.380970919454113e-05,
+      "loss": 3.0797,
+      "step": 5010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.365748286109044e-05,
+      "loss": 3.0541,
+      "step": 5011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.35053673738217e-05,
+      "loss": 3.1747,
+      "step": 5012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.335336277423051e-05,
+      "loss": 3.0877,
+      "step": 5013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.320146910378224e-05,
+      "loss": 3.2224,
+      "step": 5014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.30496864039121e-05,
+      "loss": 3.2575,
+      "step": 5015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.289801471602455e-05,
+      "loss": 3.0143,
+      "step": 5016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.274645408149435e-05,
+      "loss": 3.1051,
+      "step": 5017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.25950045416657e-05,
+      "loss": 3.1506,
+      "step": 5018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.244366613785249e-05,
+      "loss": 3.1688,
+      "step": 5019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.229243891133832e-05,
+      "loss": 3.0318,
+      "step": 5020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.214132290337663e-05,
+      "loss": 3.1099,
+      "step": 5021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.19903181551901e-05,
+      "loss": 3.1499,
+      "step": 5022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.183942470797141e-05,
+      "loss": 3.1946,
+      "step": 5023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.168864260288285e-05,
+      "loss": 3.2483,
+      "step": 5024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.153797188105623e-05,
+      "loss": 3.1258,
+      "step": 5025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.138741258359295e-05,
+      "loss": 2.9609,
+      "step": 5026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.123696475156434e-05,
+      "loss": 2.9939,
+      "step": 5027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.108662842601079e-05,
+      "loss": 3.0845,
+      "step": 5028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.093640364794258e-05,
+      "loss": 3.0058,
+      "step": 5029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.078629045833964e-05,
+      "loss": 3.0417,
+      "step": 5030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.063628889815128e-05,
+      "loss": 3.0204,
+      "step": 5031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.048639900829642e-05,
+      "loss": 3.0949,
+      "step": 5032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.03366208296637e-05,
+      "loss": 3.0728,
+      "step": 5033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.018695440311087e-05,
+      "loss": 3.1579,
+      "step": 5034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.003739976946552e-05,
+      "loss": 3.1137,
+      "step": 5035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.988795696952462e-05,
+      "loss": 3.0077,
+      "step": 5036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.97386260440548e-05,
+      "loss": 3.1329,
+      "step": 5037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.958940703379192e-05,
+      "loss": 3.3283,
+      "step": 5038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.944029997944165e-05,
+      "loss": 3.1693,
+      "step": 5039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.929130492167864e-05,
+      "loss": 3.1059,
+      "step": 5040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.91424219011473e-05,
+      "loss": 3.0857,
+      "step": 5041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.899365095846167e-05,
+      "loss": 3.2368,
+      "step": 5042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.88449921342045e-05,
+      "loss": 3.0072,
+      "step": 5043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.869644546892891e-05,
+      "loss": 3.053,
+      "step": 5044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.854801100315685e-05,
+      "loss": 3.1801,
+      "step": 5045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.839968877737958e-05,
+      "loss": 3.201,
+      "step": 5046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.825147883205804e-05,
+      "loss": 3.121,
+      "step": 5047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.810338120762235e-05,
+      "loss": 3.1162,
+      "step": 5048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.795539594447216e-05,
+      "loss": 3.1618,
+      "step": 5049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.780752308297629e-05,
+      "loss": 3.1336,
+      "step": 5050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.76597626634732e-05,
+      "loss": 3.0342,
+      "step": 5051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.751211472627014e-05,
+      "loss": 3.1021,
+      "step": 5052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.736457931164416e-05,
+      "loss": 3.2496,
+      "step": 5053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.721715645984135e-05,
+      "loss": 2.9835,
+      "step": 5054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.706984621107733e-05,
+      "loss": 3.2831,
+      "step": 5055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.692264860553673e-05,
+      "loss": 3.249,
+      "step": 5056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.677556368337386e-05,
+      "loss": 2.9502,
+      "step": 5057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.662859148471164e-05,
+      "loss": 3.0645,
+      "step": 5058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.648173204964277e-05,
+      "loss": 3.3553,
+      "step": 5059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.633498541822909e-05,
+      "loss": 3.0614,
+      "step": 5060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.61883516305016e-05,
+      "loss": 2.975,
+      "step": 5061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.604183072646055e-05,
+      "loss": 3.1529,
+      "step": 5062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.589542274607543e-05,
+      "loss": 3.2732,
+      "step": 5063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.574912772928461e-05,
+      "loss": 3.1847,
+      "step": 5064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.56029457159962e-05,
+      "loss": 3.0164,
+      "step": 5065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.545687674608704e-05,
+      "loss": 3.0532,
+      "step": 5066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.531092085940345e-05,
+      "loss": 3.1011,
+      "step": 5067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.516507809576041e-05,
+      "loss": 3.01,
+      "step": 5068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.50193484949427e-05,
+      "loss": 3.2182,
+      "step": 5069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.487373209670391e-05,
+      "loss": 3.0786,
+      "step": 5070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.472822894076643e-05,
+      "loss": 3.1203,
+      "step": 5071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.458283906682229e-05,
+      "loss": 3.1015,
+      "step": 5072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.443756251453249e-05,
+      "loss": 3.2932,
+      "step": 5073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.429239932352666e-05,
+      "loss": 3.1192,
+      "step": 5074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.414734953340419e-05,
+      "loss": 3.2206,
+      "step": 5075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.400241318373331e-05,
+      "loss": 3.1854,
+      "step": 5076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.385759031405082e-05,
+      "loss": 3.3214,
+      "step": 5077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.371288096386321e-05,
+      "loss": 3.2587,
+      "step": 5078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.35682851726458e-05,
+      "loss": 3.1495,
+      "step": 5079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.342380297984253e-05,
+      "loss": 3.0329,
+      "step": 5080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.327943442486708e-05,
+      "loss": 3.114,
+      "step": 5081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.31351795471017e-05,
+      "loss": 3.0755,
+      "step": 5082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.29910383858975e-05,
+      "loss": 3.2667,
+      "step": 5083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.284701098057485e-05,
+      "loss": 3.1781,
+      "step": 5084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.270309737042308e-05,
+      "loss": 3.3113,
+      "step": 5085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.255929759470004e-05,
+      "loss": 3.1095,
+      "step": 5086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.241561169263329e-05,
+      "loss": 3.2347,
+      "step": 5087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.227203970341879e-05,
+      "loss": 3.0496,
+      "step": 5088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.21285816662214e-05,
+      "loss": 3.0478,
+      "step": 5089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.198523762017512e-05,
+      "loss": 3.002,
+      "step": 5090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.184200760438298e-05,
+      "loss": 3.2932,
+      "step": 5091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.169889165791633e-05,
+      "loss": 3.1013,
+      "step": 5092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.155588981981583e-05,
+      "loss": 3.1271,
+      "step": 5093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.141300212909131e-05,
+      "loss": 3.2487,
+      "step": 5094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.127022862472077e-05,
+      "loss": 3.1907,
+      "step": 5095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.112756934565146e-05,
+      "loss": 2.9575,
+      "step": 5096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.098502433079963e-05,
+      "loss": 3.2369,
+      "step": 5097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.084259361904977e-05,
+      "loss": 3.0907,
+      "step": 5098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.070027724925565e-05,
+      "loss": 3.0067,
+      "step": 5099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.055807526024005e-05,
+      "loss": 3.1305,
+      "step": 5100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.041598769079395e-05,
+      "loss": 3.1364,
+      "step": 5101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.027401457967748e-05,
+      "loss": 3.2659,
+      "step": 5102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.013215596561957e-05,
+      "loss": 3.1394,
+      "step": 5103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.999041188731787e-05,
+      "loss": 3.1791,
+      "step": 5104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.984878238343846e-05,
+      "loss": 3.0557,
+      "step": 5105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.970726749261687e-05,
+      "loss": 2.9985,
+      "step": 5106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.95658672534566e-05,
+      "loss": 3.1085,
+      "step": 5107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.942458170453043e-05,
+      "loss": 3.0313,
+      "step": 5108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.928341088437952e-05,
+      "loss": 3.0882,
+      "step": 5109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.914235483151405e-05,
+      "loss": 3.1335,
+      "step": 5110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.900141358441232e-05,
+      "loss": 3.1116,
+      "step": 5111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.886058718152222e-05,
+      "loss": 3.0612,
+      "step": 5112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.871987566125938e-05,
+      "loss": 3.1548,
+      "step": 5113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.857927906200863e-05,
+      "loss": 3.0652,
+      "step": 5114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.843879742212334e-05,
+      "loss": 3.3375,
+      "step": 5115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.82984307799256e-05,
+      "loss": 3.1351,
+      "step": 5116
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.815817917370577e-05,
+      "loss": 3.4055,
+      "step": 5117
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.801804264172313e-05,
+      "loss": 3.0711,
+      "step": 5118
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.787802122220583e-05,
+      "loss": 2.9377,
+      "step": 5119
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.773811495334999e-05,
+      "loss": 3.3156,
+      "step": 5120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.759832387332078e-05,
+      "loss": 3.1588,
+      "step": 5121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.745864802025194e-05,
+      "loss": 3.1414,
+      "step": 5122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.73190874322453e-05,
+      "loss": 3.1297,
+      "step": 5123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.717964214737178e-05,
+      "loss": 3.1544,
+      "step": 5124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.704031220367086e-05,
+      "loss": 3.1058,
+      "step": 5125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.690109763915004e-05,
+      "loss": 3.2174,
+      "step": 5126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.676199849178583e-05,
+      "loss": 3.2697,
+      "step": 5127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.662301479952317e-05,
+      "loss": 3.0165,
+      "step": 5128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.64841466002752e-05,
+      "loss": 3.0727,
+      "step": 5129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.63453939319238e-05,
+      "loss": 3.1139,
+      "step": 5130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.620675683231959e-05,
+      "loss": 3.0211,
+      "step": 5131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.60682353392811e-05,
+      "loss": 3.2531,
+      "step": 5132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.592982949059567e-05,
+      "loss": 3.1231,
+      "step": 5133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.579153932401928e-05,
+      "loss": 3.2348,
+      "step": 5134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.56533648772757e-05,
+      "loss": 3.1262,
+      "step": 5135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.551530618805768e-05,
+      "loss": 3.1831,
+      "step": 5136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.53773632940265e-05,
+      "loss": 3.192,
+      "step": 5137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.523953623281132e-05,
+      "loss": 3.1427,
+      "step": 5138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.510182504201013e-05,
+      "loss": 3.112,
+      "step": 5139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.496422975918915e-05,
+      "loss": 3.0814,
+      "step": 5140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.482675042188292e-05,
+      "loss": 3.0157,
+      "step": 5141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.468938706759448e-05,
+      "loss": 3.1684,
+      "step": 5142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.455213973379516e-05,
+      "loss": 2.9667,
+      "step": 5143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.441500845792471e-05,
+      "loss": 3.2422,
+      "step": 5144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.42779932773911e-05,
+      "loss": 2.9819,
+      "step": 5145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.414109422957088e-05,
+      "loss": 3.1231,
+      "step": 5146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.400431135180852e-05,
+      "loss": 3.0518,
+      "step": 5147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.38676446814171e-05,
+      "loss": 3.061,
+      "step": 5148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.373109425567797e-05,
+      "loss": 3.0037,
+      "step": 5149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.359466011184068e-05,
+      "loss": 3.1321,
+      "step": 5150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.345834228712311e-05,
+      "loss": 3.102,
+      "step": 5151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.332214081871141e-05,
+      "loss": 3.0638,
+      "step": 5152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.318605574376014e-05,
+      "loss": 3.0269,
+      "step": 5153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.30500870993916e-05,
+      "loss": 3.1972,
+      "step": 5154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.291423492269694e-05,
+      "loss": 3.1706,
+      "step": 5155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.277849925073515e-05,
+      "loss": 3.2558,
+      "step": 5156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.264288012053366e-05,
+      "loss": 3.2528,
+      "step": 5157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.250737756908794e-05,
+      "loss": 3.181,
+      "step": 5158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.237199163336189e-05,
+      "loss": 3.0513,
+      "step": 5159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.223672235028727e-05,
+      "loss": 3.2261,
+      "step": 5160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.210156975676418e-05,
+      "loss": 3.0933,
+      "step": 5161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.196653388966095e-05,
+      "loss": 3.195,
+      "step": 5162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.183161478581406e-05,
+      "loss": 3.2378,
+      "step": 5163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.169681248202808e-05,
+      "loss": 3.1512,
+      "step": 5164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.15621270150758e-05,
+      "loss": 3.1431,
+      "step": 5165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.142755842169785e-05,
+      "loss": 3.2168,
+      "step": 5166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.129310673860334e-05,
+      "loss": 2.9561,
+      "step": 5167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.11587720024694e-05,
+      "loss": 3.1838,
+      "step": 5168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.102455424994108e-05,
+      "loss": 3.1856,
+      "step": 5169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.089045351763174e-05,
+      "loss": 3.112,
+      "step": 5170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.075646984212275e-05,
+      "loss": 3.0103,
+      "step": 5171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.062260325996339e-05,
+      "loss": 3.0474,
+      "step": 5172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.048885380767123e-05,
+      "loss": 3.2045,
+      "step": 5173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.035522152173168e-05,
+      "loss": 3.1105,
+      "step": 5174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.022170643859838e-05,
+      "loss": 3.2056,
+      "step": 5175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.008830859469296e-05,
+      "loss": 3.1943,
+      "step": 5176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.995502802640497e-05,
+      "loss": 3.2604,
+      "step": 5177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.982186477009194e-05,
+      "loss": 3.2211,
+      "step": 5178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.968881886207956e-05,
+      "loss": 3.1681,
+      "step": 5179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.955589033866139e-05,
+      "loss": 3.0491,
+      "step": 5180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.942307923609904e-05,
+      "loss": 3.2136,
+      "step": 5181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.929038559062201e-05,
+      "loss": 3.2473,
+      "step": 5182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.915780943842792e-05,
+      "loss": 3.1467,
+      "step": 5183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9025350815682e-05,
+      "loss": 3.1693,
+      "step": 5184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.889300975851781e-05,
+      "loss": 3.158,
+      "step": 5185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.876078630303661e-05,
+      "loss": 3.3385,
+      "step": 5186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.862868048530768e-05,
+      "loss": 3.1976,
+      "step": 5187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.849669234136813e-05,
+      "loss": 3.1504,
+      "step": 5188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.836482190722309e-05,
+      "loss": 2.9868,
+      "step": 5189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.823306921884537e-05,
+      "loss": 3.0692,
+      "step": 5190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.810143431217585e-05,
+      "loss": 3.1256,
+      "step": 5191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.796991722312318e-05,
+      "loss": 3.2791,
+      "step": 5192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.783851798756396e-05,
+      "loss": 3.0654,
+      "step": 5193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.770723664134254e-05,
+      "loss": 3.1216,
+      "step": 5194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.757607322027132e-05,
+      "loss": 3.2263,
+      "step": 5195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.744502776013018e-05,
+      "loss": 3.2164,
+      "step": 5196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.731410029666701e-05,
+      "loss": 3.2069,
+      "step": 5197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.71832908655976e-05,
+      "loss": 3.1535,
+      "step": 5198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.70525995026055e-05,
+      "loss": 3.1078,
+      "step": 5199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.692202624334187e-05,
+      "loss": 3.1529,
+      "step": 5200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.679157112342604e-05,
+      "loss": 3.1202,
+      "step": 5201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.666123417844456e-05,
+      "loss": 3.0805,
+      "step": 5202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.653101544395218e-05,
+      "loss": 3.2423,
+      "step": 5203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.640091495547129e-05,
+      "loss": 3.0891,
+      "step": 5204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.627093274849194e-05,
+      "loss": 3.1721,
+      "step": 5205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.61410688584721e-05,
+      "loss": 3.1488,
+      "step": 5206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.60113233208372e-05,
+      "loss": 2.9562,
+      "step": 5207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.588169617098071e-05,
+      "loss": 3.169,
+      "step": 5208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.575218744426348e-05,
+      "loss": 3.1151,
+      "step": 5209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.562279717601415e-05,
+      "loss": 3.0651,
+      "step": 5210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.54935254015292e-05,
+      "loss": 3.168,
+      "step": 5211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.536437215607261e-05,
+      "loss": 3.1644,
+      "step": 5212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.52353374748762e-05,
+      "loss": 3.2229,
+      "step": 5213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.510642139313933e-05,
+      "loss": 3.3354,
+      "step": 5214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.49776239460289e-05,
+      "loss": 2.9036,
+      "step": 5215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.484894516867962e-05,
+      "loss": 3.2851,
+      "step": 5216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.472038509619388e-05,
+      "loss": 3.3044,
+      "step": 5217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.45919437636413e-05,
+      "loss": 3.1882,
+      "step": 5218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.446362120605969e-05,
+      "loss": 3.0779,
+      "step": 5219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.433541745845417e-05,
+      "loss": 3.2151,
+      "step": 5220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.42073325557972e-05,
+      "loss": 3.1287,
+      "step": 5221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.407936653302926e-05,
+      "loss": 3.0623,
+      "step": 5222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.395151942505822e-05,
+      "loss": 3.2403,
+      "step": 5223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.382379126675919e-05,
+      "loss": 3.1908,
+      "step": 5224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.369618209297546e-05,
+      "loss": 3.1351,
+      "step": 5225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.356869193851755e-05,
+      "loss": 3.2411,
+      "step": 5226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.344132083816328e-05,
+      "loss": 3.1305,
+      "step": 5227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.331406882665836e-05,
+      "loss": 3.0891,
+      "step": 5228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.318693593871593e-05,
+      "loss": 3.2478,
+      "step": 5229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.305992220901624e-05,
+      "loss": 3.1533,
+      "step": 5230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.293302767220771e-05,
+      "loss": 3.1216,
+      "step": 5231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.280625236290593e-05,
+      "loss": 3.1764,
+      "step": 5232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.26795963156937e-05,
+      "loss": 3.2167,
+      "step": 5233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.255305956512159e-05,
+      "loss": 3.0672,
+      "step": 5234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.242664214570776e-05,
+      "loss": 3.258,
+      "step": 5235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.230034409193724e-05,
+      "loss": 3.2261,
+      "step": 5236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.21741654382632e-05,
+      "loss": 3.2774,
+      "step": 5237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.20481062191059e-05,
+      "loss": 3.0319,
+      "step": 5238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.19221664688529e-05,
+      "loss": 3.3377,
+      "step": 5239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.179634622185932e-05,
+      "loss": 3.2308,
+      "step": 5240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.167064551244772e-05,
+      "loss": 3.0471,
+      "step": 5241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.15450643749081e-05,
+      "loss": 3.0981,
+      "step": 5242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.14196028434974e-05,
+      "loss": 3.2595,
+      "step": 5243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.129426095244073e-05,
+      "loss": 3.0728,
+      "step": 5244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.116903873592977e-05,
+      "loss": 3.2367,
+      "step": 5245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.104393622812399e-05,
+      "loss": 3.112,
+      "step": 5246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.091895346315013e-05,
+      "loss": 2.9946,
+      "step": 5247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.079409047510231e-05,
+      "loss": 3.1833,
+      "step": 5248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.066934729804158e-05,
+      "loss": 3.0505,
+      "step": 5249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.054472396599714e-05,
+      "loss": 3.0997,
+      "step": 5250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0420220512964585e-05,
+      "loss": 3.132,
+      "step": 5251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.029583697290736e-05,
+      "loss": 3.0132,
+      "step": 5252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017157337975609e-05,
+      "loss": 3.1636,
+      "step": 5253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.004742976740868e-05,
+      "loss": 3.1478,
+      "step": 5254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9923406169729966e-05,
+      "loss": 3.0455,
+      "step": 5255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.979950262055267e-05,
+      "loss": 3.077,
+      "step": 5256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.967571915367642e-05,
+      "loss": 3.056,
+      "step": 5257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.955205580286799e-05,
+      "loss": 3.1825,
+      "step": 5258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.94285126018615e-05,
+      "loss": 3.2461,
+      "step": 5259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.930508958435848e-05,
+      "loss": 3.0501,
+      "step": 5260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.918178678402714e-05,
+      "loss": 3.1652,
+      "step": 5261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.90586042345036e-05,
+      "loss": 3.0504,
+      "step": 5262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8935541969390825e-05,
+      "loss": 3.1661,
+      "step": 5263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8812600022258745e-05,
+      "loss": 3.2866,
+      "step": 5264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8689778426644805e-05,
+      "loss": 3.1115,
+      "step": 5265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.856707721605359e-05,
+      "loss": 3.0553,
+      "step": 5266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.844449642395666e-05,
+      "loss": 3.1275,
+      "step": 5267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.832203608379272e-05,
+      "loss": 3.0551,
+      "step": 5268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8199696228968036e-05,
+      "loss": 3.1618,
+      "step": 5269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.807747689285547e-05,
+      "loss": 3.1199,
+      "step": 5270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.795537810879531e-05,
+      "loss": 3.149,
+      "step": 5271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7833399910094955e-05,
+      "loss": 3.3431,
+      "step": 5272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.771154233002862e-05,
+      "loss": 3.1143,
+      "step": 5273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7589805401837894e-05,
+      "loss": 3.1932,
+      "step": 5274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.74681891587317e-05,
+      "loss": 3.2348,
+      "step": 5275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7346693633885446e-05,
+      "loss": 3.1473,
+      "step": 5276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.722531886044191e-05,
+      "loss": 3.0744,
+      "step": 5277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.71040648715111e-05,
+      "loss": 3.1193,
+      "step": 5278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.698293170016966e-05,
+      "loss": 3.1755,
+      "step": 5279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.68619193794615e-05,
+      "loss": 3.1354,
+      "step": 5280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6741027942397885e-05,
+      "loss": 2.9966,
+      "step": 5281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.662025742195654e-05,
+      "loss": 3.2396,
+      "step": 5282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.649960785108244e-05,
+      "loss": 3.199,
+      "step": 5283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6379079262687813e-05,
+      "loss": 3.0118,
+      "step": 5284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.625867168965138e-05,
+      "loss": 3.1244,
+      "step": 5285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.613838516481912e-05,
+      "loss": 3.0405,
+      "step": 5286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.601821972100435e-05,
+      "loss": 3.1956,
+      "step": 5287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5898175390986686e-05,
+      "loss": 3.171,
+      "step": 5288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5778252207513094e-05,
+      "loss": 3.063,
+      "step": 5289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.565845020329741e-05,
+      "loss": 3.0836,
+      "step": 5290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5538769411020596e-05,
+      "loss": 3.0852,
+      "step": 5291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.54192098633301e-05,
+      "loss": 3.1848,
+      "step": 5292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.529977159284072e-05,
+      "loss": 2.9598,
+      "step": 5293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5180454632134006e-05,
+      "loss": 2.9875,
+      "step": 5294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.506125901375847e-05,
+      "loss": 3.0159,
+      "step": 5295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.494218477022939e-05,
+      "loss": 3.1757,
+      "step": 5296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.482323193402921e-05,
+      "loss": 3.2564,
+      "step": 5297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.470440053760689e-05,
+      "loss": 3.0884,
+      "step": 5298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.458569061337854e-05,
+      "loss": 3.1643,
+      "step": 5299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.446710219372697e-05,
+      "loss": 3.0618,
+      "step": 5300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.434863531100198e-05,
+      "loss": 3.161,
+      "step": 5301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4230289997520166e-05,
+      "loss": 3.1078,
+      "step": 5302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4112066285564975e-05,
+      "loss": 3.2769,
+      "step": 5303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.399396420738656e-05,
+      "loss": 2.9889,
+      "step": 5304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.387598379520203e-05,
+      "loss": 3.0165,
+      "step": 5305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.375812508119521e-05,
+      "loss": 3.0288,
+      "step": 5306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3640388097516866e-05,
+      "loss": 3.3053,
+      "step": 5307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.352277287628449e-05,
+      "loss": 3.0723,
+      "step": 5308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3405279449582345e-05,
+      "loss": 3.1045,
+      "step": 5309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3287907849461304e-05,
+      "loss": 3.1271,
+      "step": 5310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.317065810793931e-05,
+      "loss": 3.0603,
+      "step": 5311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3053530257000826e-05,
+      "loss": 3.0677,
+      "step": 5312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.29365243285973e-05,
+      "loss": 3.2249,
+      "step": 5313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2819640354646645e-05,
+      "loss": 3.0821,
+      "step": 5314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.27028783670338e-05,
+      "loss": 3.0314,
+      "step": 5315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.258623839761012e-05,
+      "loss": 3.1329,
+      "step": 5316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.246972047819387e-05,
+      "loss": 3.1326,
+      "step": 5317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2353324640569965e-05,
+      "loss": 3.2606,
+      "step": 5318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2237050916490006e-05,
+      "loss": 3.2651,
+      "step": 5319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.212089933767239e-05,
+      "loss": 3.0767,
+      "step": 5320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.200486993580211e-05,
+      "loss": 3.1629,
+      "step": 5321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1888962742530745e-05,
+      "loss": 3.282,
+      "step": 5322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.17731777894766e-05,
+      "loss": 3.1028,
+      "step": 5323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1657515108224694e-05,
+      "loss": 3.2119,
+      "step": 5324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.154197473032668e-05,
+      "loss": 3.1275,
+      "step": 5325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.142655668730084e-05,
+      "loss": 3.0484,
+      "step": 5326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1311261010632104e-05,
+      "loss": 3.2182,
+      "step": 5327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1196087731771815e-05,
+      "loss": 3.191,
+      "step": 5328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1081036882138166e-05,
+      "loss": 3.0079,
+      "step": 5329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.096610849311589e-05,
+      "loss": 3.0101,
+      "step": 5330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.08513025960563e-05,
+      "loss": 3.0847,
+      "step": 5331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0736619222277346e-05,
+      "loss": 3.03,
+      "step": 5332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.062205840306355e-05,
+      "loss": 3.146,
+      "step": 5333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0507620169665814e-05,
+      "loss": 2.9684,
+      "step": 5334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0393304553301825e-05,
+      "loss": 3.0812,
+      "step": 5335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.027911158515569e-05,
+      "loss": 3.2237,
+      "step": 5336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.016504129637817e-05,
+      "loss": 3.1383,
+      "step": 5337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.005109371808647e-05,
+      "loss": 3.1091,
+      "step": 5338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.993726888136446e-05,
+      "loss": 3.024,
+      "step": 5339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.982356681726219e-05,
+      "loss": 3.2634,
+      "step": 5340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.970998755679662e-05,
+      "loss": 3.178,
+      "step": 5341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9596531130951026e-05,
+      "loss": 3.2132,
+      "step": 5342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.94831975706751e-05,
+      "loss": 3.197,
+      "step": 5343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.936998690688521e-05,
+      "loss": 3.0834,
+      "step": 5344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9256899170464056e-05,
+      "loss": 3.1816,
+      "step": 5345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9143934392260946e-05,
+      "loss": 3.169,
+      "step": 5346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.90310926030913e-05,
+      "loss": 3.209,
+      "step": 5347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.891837383373737e-05,
+      "loss": 3.0332,
+      "step": 5348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8805778114947744e-05,
+      "loss": 3.1495,
+      "step": 5349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8693305477437335e-05,
+      "loss": 3.179,
+      "step": 5350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.858095595188766e-05,
+      "loss": 3.1582,
+      "step": 5351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.846872956894649e-05,
+      "loss": 3.2267,
+      "step": 5352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.835662635922805e-05,
+      "loss": 3.2132,
+      "step": 5353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.824464635331294e-05,
+      "loss": 3.1774,
+      "step": 5354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8132789581748216e-05,
+      "loss": 3.0654,
+      "step": 5355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.802105607504731e-05,
+      "loss": 2.9699,
+      "step": 5356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.790944586369e-05,
+      "loss": 3.309,
+      "step": 5357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7797958978122555e-05,
+      "loss": 3.1813,
+      "step": 5358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7686595448757254e-05,
+      "loss": 3.1829,
+      "step": 5359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.757535530597307e-05,
+      "loss": 3.2379,
+      "step": 5360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.746423858011534e-05,
+      "loss": 3.1527,
+      "step": 5361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.735324530149521e-05,
+      "loss": 3.1598,
+      "step": 5362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.72423755003909e-05,
+      "loss": 3.1183,
+      "step": 5363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.713162920704656e-05,
+      "loss": 3.1192,
+      "step": 5364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.702100645167251e-05,
+      "loss": 2.943,
+      "step": 5365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.691050726444562e-05,
+      "loss": 3.2065,
+      "step": 5366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6800131675509e-05,
+      "loss": 3.0614,
+      "step": 5367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.668987971497185e-05,
+      "loss": 3.0328,
+      "step": 5368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.657975141290993e-05,
+      "loss": 3.2564,
+      "step": 5369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.646974679936527e-05,
+      "loss": 3.022,
+      "step": 5370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6359865904345765e-05,
+      "loss": 3.086,
+      "step": 5371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.625010875782598e-05,
+      "loss": 3.2474,
+      "step": 5372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.614047538974664e-05,
+      "loss": 3.3224,
+      "step": 5373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.603096583001432e-05,
+      "loss": 3.2119,
+      "step": 5374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.592158010850245e-05,
+      "loss": 3.1726,
+      "step": 5375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.581231825505033e-05,
+      "loss": 3.1038,
+      "step": 5376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570318029946341e-05,
+      "loss": 3.1587,
+      "step": 5377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.559416627151336e-05,
+      "loss": 3.1657,
+      "step": 5378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.548527620093828e-05,
+      "loss": 3.1365,
+      "step": 5379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5376510117442205e-05,
+      "loss": 3.1016,
+      "step": 5380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.526786805069549e-05,
+      "loss": 3.1437,
+      "step": 5381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5159350030334665e-05,
+      "loss": 3.38,
+      "step": 5382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.505095608596216e-05,
+      "loss": 3.0378,
+      "step": 5383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.494268624714687e-05,
+      "loss": 3.1409,
+      "step": 5384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.483454054342373e-05,
+      "loss": 3.0254,
+      "step": 5385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.472651900429392e-05,
+      "loss": 3.1104,
+      "step": 5386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.461862165922437e-05,
+      "loss": 3.0587,
+      "step": 5387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4510848537648694e-05,
+      "loss": 3.1152,
+      "step": 5388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.440319966896611e-05,
+      "loss": 3.1375,
+      "step": 5389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.429567508254223e-05,
+      "loss": 3.0635,
+      "step": 5390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4188274807708705e-05,
+      "loss": 3.1113,
+      "step": 5391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.408099887376332e-05,
+      "loss": 3.1863,
+      "step": 5392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.397384730996962e-05,
+      "loss": 3.1055,
+      "step": 5393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.386682014555776e-05,
+      "loss": 3.2366,
+      "step": 5394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.375991740972368e-05,
+      "loss": 3.0813,
+      "step": 5395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.365313913162916e-05,
+      "loss": 3.1199,
+      "step": 5396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3546485340402395e-05,
+      "loss": 3.0969,
+      "step": 5397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343995606513751e-05,
+      "loss": 3.0795,
+      "step": 5398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.333355133489442e-05,
+      "loss": 3.1548,
+      "step": 5399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.322727117869951e-05,
+      "loss": 3.094,
+      "step": 5400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3121115625544935e-05,
+      "loss": 3.0508,
+      "step": 5401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.301508470438869e-05,
+      "loss": 3.0741,
+      "step": 5402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2909178444155094e-05,
+      "loss": 3.1428,
+      "step": 5403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.280339687373436e-05,
+      "loss": 3.2297,
+      "step": 5404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.269774002198235e-05,
+      "loss": 2.9579,
+      "step": 5405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.259220791772156e-05,
+      "loss": 3.1745,
+      "step": 5406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.248680058973997e-05,
+      "loss": 3.3474,
+      "step": 5407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.238151806679158e-05,
+      "loss": 3.1116,
+      "step": 5408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.227636037759641e-05,
+      "loss": 3.1529,
+      "step": 5409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.217132755084058e-05,
+      "loss": 3.1089,
+      "step": 5410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2066419615175824e-05,
+      "loss": 3.1472,
+      "step": 5411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.196163659921992e-05,
+      "loss": 3.2151,
+      "step": 5412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.18569785315569e-05,
+      "loss": 3.1856,
+      "step": 5413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1752445440736254e-05,
+      "loss": 3.0456,
+      "step": 5414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.164803735527356e-05,
+      "loss": 3.096,
+      "step": 5415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1543754303650484e-05,
+      "loss": 3.3045,
+      "step": 5416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.143959631431415e-05,
+      "loss": 2.9486,
+      "step": 5417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1335563415677844e-05,
+      "loss": 3.1236,
+      "step": 5418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.123165563612097e-05,
+      "loss": 3.3154,
+      "step": 5419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1127873003988233e-05,
+      "loss": 3.054,
+      "step": 5420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.102421554759061e-05,
+      "loss": 3.0125,
+      "step": 5421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.092068329520493e-05,
+      "loss": 3.0724,
+      "step": 5422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.081727627507359e-05,
+      "loss": 3.0123,
+      "step": 5423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.071399451540497e-05,
+      "loss": 3.1354,
+      "step": 5424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.061083804437355e-05,
+      "loss": 3.0101,
+      "step": 5425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.050780689011912e-05,
+      "loss": 3.0255,
+      "step": 5426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.040490108074768e-05,
+      "loss": 3.2069,
+      "step": 5427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0302120644330864e-05,
+      "loss": 3.0411,
+      "step": 5428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.019946560890625e-05,
+      "loss": 3.0705,
+      "step": 5429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009693600247688e-05,
+      "loss": 3.2786,
+      "step": 5430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.999453185301211e-05,
+      "loss": 2.9021,
+      "step": 5431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9892253188446524e-05,
+      "loss": 3.276,
+      "step": 5432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.979010003668082e-05,
+      "loss": 3.0801,
+      "step": 5433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.968807242558131e-05,
+      "loss": 3.3216,
+      "step": 5434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.958617038298029e-05,
+      "loss": 3.2306,
+      "step": 5435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.948439393667536e-05,
+      "loss": 3.1317,
+      "step": 5436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.938274311443018e-05,
+      "loss": 3.206,
+      "step": 5437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9281217943974214e-05,
+      "loss": 3.1811,
+      "step": 5438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.917981845300239e-05,
+      "loss": 3.1741,
+      "step": 5439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.907854466917554e-05,
+      "loss": 3.0861,
+      "step": 5440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.897739662012017e-05,
+      "loss": 3.1293,
+      "step": 5441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8876374333428335e-05,
+      "loss": 3.0845,
+      "step": 5442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8775477836658026e-05,
+      "loss": 3.1251,
+      "step": 5443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.867470715733268e-05,
+      "loss": 3.1721,
+      "step": 5444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.857406232294164e-05,
+      "loss": 3.0012,
+      "step": 5445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8473543360939775e-05,
+      "loss": 3.2365,
+      "step": 5446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.837315029874777e-05,
+      "loss": 3.2745,
+      "step": 5447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8272883163751605e-05,
+      "loss": 3.1103,
+      "step": 5448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.817274198330323e-05,
+      "loss": 3.3087,
+      "step": 5449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.80727267847204e-05,
+      "loss": 3.1715,
+      "step": 5450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.797283759528597e-05,
+      "loss": 3.1065,
+      "step": 5451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.78730744422488e-05,
+      "loss": 3.1389,
+      "step": 5452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.777343735282346e-05,
+      "loss": 3.1071,
+      "step": 5453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.767392635418965e-05,
+      "loss": 3.0999,
+      "step": 5454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.757454147349304e-05,
+      "loss": 3.1647,
+      "step": 5455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7475282737845116e-05,
+      "loss": 2.9632,
+      "step": 5456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.737615017432239e-05,
+      "loss": 3.104,
+      "step": 5457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7277143809967274e-05,
+      "loss": 3.2383,
+      "step": 5458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.717826367178789e-05,
+      "loss": 2.9692,
+      "step": 5459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.70795097867575e-05,
+      "loss": 3.0111,
+      "step": 5460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.698088218181533e-05,
+      "loss": 3.0743,
+      "step": 5461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.688238088386592e-05,
+      "loss": 3.0761,
+      "step": 5462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.678400591977954e-05,
+      "loss": 3.1577,
+      "step": 5463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.668575731639184e-05,
+      "loss": 2.9539,
+      "step": 5464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.658763510050417e-05,
+      "loss": 3.1983,
+      "step": 5465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.64896392988831e-05,
+      "loss": 3.0652,
+      "step": 5466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6391769938261047e-05,
+      "loss": 3.1895,
+      "step": 5467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6294027045335686e-05,
+      "loss": 3.083,
+      "step": 5468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.619641064677037e-05,
+      "loss": 3.1826,
+      "step": 5469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.609892076919391e-05,
+      "loss": 3.0434,
+      "step": 5470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.600155743920058e-05,
+      "loss": 3.2173,
+      "step": 5471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.590432068334998e-05,
+      "loss": 3.163,
+      "step": 5472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5807210528167434e-05,
+      "loss": 3.1514,
+      "step": 5473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5710227000143646e-05,
+      "loss": 3.205,
+      "step": 5474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5613370125734734e-05,
+      "loss": 3.2162,
+      "step": 5475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5516639931362224e-05,
+      "loss": 3.1297,
+      "step": 5476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542003644341329e-05,
+      "loss": 3.2072,
+      "step": 5477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5323559688240246e-05,
+      "loss": 3.1577,
+      "step": 5478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5227209692161035e-05,
+      "loss": 3.262,
+      "step": 5479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5130986481458956e-05,
+      "loss": 3.1636,
+      "step": 5480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.503489008238281e-05,
+      "loss": 3.2934,
+      "step": 5481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.493892052114666e-05,
+      "loss": 3.1579,
+      "step": 5482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4843077823930146e-05,
+      "loss": 3.2004,
+      "step": 5483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.474736201687817e-05,
+      "loss": 3.1963,
+      "step": 5484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4651773126100904e-05,
+      "loss": 3.0756,
+      "step": 5485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.455631117767422e-05,
+      "loss": 3.0818,
+      "step": 5486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4460976197639047e-05,
+      "loss": 3.0706,
+      "step": 5487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.436576821200193e-05,
+      "loss": 3.1925,
+      "step": 5488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.42706872467346e-05,
+      "loss": 3.2538,
+      "step": 5489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4175733327774296e-05,
+      "loss": 3.091,
+      "step": 5490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.40809064810233e-05,
+      "loss": 2.994,
+      "step": 5491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.398620673234953e-05,
+      "loss": 3.0751,
+      "step": 5492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.389163410758622e-05,
+      "loss": 3.1426,
+      "step": 5493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3797188632531675e-05,
+      "loss": 2.9936,
+      "step": 5494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3702870332949776e-05,
+      "loss": 3.279,
+      "step": 5495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.360867923456973e-05,
+      "loss": 3.1055,
+      "step": 5496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.351461536308564e-05,
+      "loss": 3.1975,
+      "step": 5497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3420678744157384e-05,
+      "loss": 3.173,
+      "step": 5498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.33268694034099e-05,
+      "loss": 2.9665,
+      "step": 5499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3233187366433436e-05,
+      "loss": 3.1158,
+      "step": 5500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.313963265878356e-05,
+      "loss": 2.9923,
+      "step": 5501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3046205305981066e-05,
+      "loss": 3.1909,
+      "step": 5502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2952905333511865e-05,
+      "loss": 3.0754,
+      "step": 5503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.285973276682736e-05,
+      "loss": 3.1634,
+      "step": 5504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.276668763134405e-05,
+      "loss": 3.1021,
+      "step": 5505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.267376995244381e-05,
+      "loss": 3.1038,
+      "step": 5506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2580979755473586e-05,
+      "loss": 3.0182,
+      "step": 5507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.248831706574568e-05,
+      "loss": 3.2036,
+      "step": 5508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.239578190853748e-05,
+      "loss": 3.0976,
+      "step": 5509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2303374309091635e-05,
+      "loss": 3.1029,
+      "step": 5510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.221109429261615e-05,
+      "loss": 3.0883,
+      "step": 5511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2118941884283824e-05,
+      "loss": 3.1463,
+      "step": 5512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.202691710923317e-05,
+      "loss": 2.9521,
+      "step": 5513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.193501999256765e-05,
+      "loss": 3.132,
+      "step": 5514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1843250559355666e-05,
+      "loss": 3.1626,
+      "step": 5515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.175160883463113e-05,
+      "loss": 3.1504,
+      "step": 5516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.166009484339294e-05,
+      "loss": 3.0446,
+      "step": 5517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.156870861060529e-05,
+      "loss": 3.214,
+      "step": 5518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1477450161197297e-05,
+      "loss": 3.0483,
+      "step": 5519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.13863195200636e-05,
+      "loss": 3.159,
+      "step": 5520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1295316712063426e-05,
+      "loss": 2.9099,
+      "step": 5521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.120444176202153e-05,
+      "loss": 2.9941,
+      "step": 5522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.11136946947278e-05,
+      "loss": 3.1232,
+      "step": 5523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.102307553493699e-05,
+      "loss": 3.1427,
+      "step": 5524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.093258430736923e-05,
+      "loss": 3.1802,
+      "step": 5525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.084222103670964e-05,
+      "loss": 3.2014,
+      "step": 5526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.075198574760823e-05,
+      "loss": 3.175,
+      "step": 5527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.066187846468038e-05,
+      "loss": 3.1835,
+      "step": 5528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.057189921250653e-05,
+      "loss": 3.2351,
+      "step": 5529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0482048015632036e-05,
+      "loss": 3.2159,
+      "step": 5530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0392324898567414e-05,
+      "loss": 3.1304,
+      "step": 5531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.030272988578825e-05,
+      "loss": 3.0428,
+      "step": 5532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0213263001735326e-05,
+      "loss": 3.0456,
+      "step": 5533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0123924270814008e-05,
+      "loss": 2.9054,
+      "step": 5534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.003471371739519e-05,
+      "loss": 3.1082,
+      "step": 5535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.994563136581474e-05,
+      "loss": 3.1387,
+      "step": 5536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9856677240373055e-05,
+      "loss": 3.2325,
+      "step": 5537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9767851365336273e-05,
+      "loss": 3.1436,
+      "step": 5538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9679153764935184e-05,
+      "loss": 3.1744,
+      "step": 5539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9590584463365434e-05,
+      "loss": 3.107,
+      "step": 5540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9502143484787968e-05,
+      "loss": 3.083,
+      "step": 5541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9413830853328606e-05,
+      "loss": 3.1019,
+      "step": 5542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9325646593078014e-05,
+      "loss": 3.1257,
+      "step": 5543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9237590728092122e-05,
+      "loss": 3.1544,
+      "step": 5544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9149663282391713e-05,
+      "loss": 3.0667,
+      "step": 5545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.906186427996238e-05,
+      "loss": 3.2552,
+      "step": 5546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8974193744754907e-05,
+      "loss": 3.1815,
+      "step": 5547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8886651700684995e-05,
+      "loss": 3.252,
+      "step": 5548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8799238171633047e-05,
+      "loss": 3.1829,
+      "step": 5549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8711953181444707e-05,
+      "loss": 3.1914,
+      "step": 5550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8624796753930592e-05,
+      "loss": 3.0172,
+      "step": 5551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8537768912865912e-05,
+      "loss": 3.048,
+      "step": 5552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.845086968199101e-05,
+      "loss": 3.0889,
+      "step": 5553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.836409908501131e-05,
+      "loss": 3.1821,
+      "step": 5554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8277457145596653e-05,
+      "loss": 3.1015,
+      "step": 5555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8190943887382303e-05,
+      "loss": 2.9738,
+      "step": 5556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8104559333968326e-05,
+      "loss": 3.0788,
+      "step": 5557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8018303508919262e-05,
+      "loss": 3.1068,
+      "step": 5558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7932176435765066e-05,
+      "loss": 3.239,
+      "step": 5559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7846178138000333e-05,
+      "loss": 3.0694,
+      "step": 5560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7760308639084408e-05,
+      "loss": 3.0002,
+      "step": 5561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.767456796244161e-05,
+      "loss": 3.1365,
+      "step": 5562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7588956131461396e-05,
+      "loss": 3.1457,
+      "step": 5563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.750347316949764e-05,
+      "loss": 3.1996,
+      "step": 5564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.741811909986919e-05,
+      "loss": 3.0402,
+      "step": 5565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.733289394585986e-05,
+      "loss": 3.0723,
+      "step": 5566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7247797730718338e-05,
+      "loss": 3.1055,
+      "step": 5567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7162830477657717e-05,
+      "loss": 3.2497,
+      "step": 5568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7077992209856518e-05,
+      "loss": 3.1498,
+      "step": 5569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6993282950457664e-05,
+      "loss": 3.2538,
+      "step": 5570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.69087027225689e-05,
+      "loss": 3.1881,
+      "step": 5571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.682425154926299e-05,
+      "loss": 3.1852,
+      "step": 5572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6739929453577396e-05,
+      "loss": 2.9833,
+      "step": 5573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.665573645851416e-05,
+      "loss": 2.9415,
+      "step": 5574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.657167258704052e-05,
+      "loss": 3.0489,
+      "step": 5575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6487737862088136e-05,
+      "loss": 3.0258,
+      "step": 5576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6403932306553523e-05,
+      "loss": 3.1672,
+      "step": 5577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6320255943298065e-05,
+      "loss": 3.0145,
+      "step": 5578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6236708795147945e-05,
+      "loss": 3.1491,
+      "step": 5579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.615329088489371e-05,
+      "loss": 3.1024,
+      "step": 5580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6070002235291212e-05,
+      "loss": 3.2109,
+      "step": 5581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5986842869060613e-05,
+      "loss": 3.0365,
+      "step": 5582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5903812808886985e-05,
+      "loss": 3.0836,
+      "step": 5583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.58209120774201e-05,
+      "loss": 3.0877,
+      "step": 5584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5738140697274537e-05,
+      "loss": 3.2719,
+      "step": 5585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5655498691029343e-05,
+      "loss": 3.1799,
+      "step": 5586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.557298608122849e-05,
+      "loss": 3.0924,
+      "step": 5587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5490602890380687e-05,
+      "loss": 3.1799,
+      "step": 5588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.540834914095913e-05,
+      "loss": 3.1245,
+      "step": 5589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5326224855401925e-05,
+      "loss": 3.1834,
+      "step": 5590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5244230056111704e-05,
+      "loss": 3.2579,
+      "step": 5591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5162364765455804e-05,
+      "loss": 3.1161,
+      "step": 5592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5080629005766188e-05,
+      "loss": 3.1203,
+      "step": 5593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4999022799339798e-05,
+      "loss": 3.1735,
+      "step": 5594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4917546168437722e-05,
+      "loss": 3.2784,
+      "step": 5595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.483619913528612e-05,
+      "loss": 3.085,
+      "step": 5596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4754981722075744e-05,
+      "loss": 3.017,
+      "step": 5597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4673893950961645e-05,
+      "loss": 3.2078,
+      "step": 5598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4592935844063858e-05,
+      "loss": 3.1349,
+      "step": 5599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4512107423467045e-05,
+      "loss": 3.0338,
+      "step": 5600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4431408711220294e-05,
+      "loss": 3.0361,
+      "step": 5601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.435083972933744e-05,
+      "loss": 3.268,
+      "step": 5602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.427040049979695e-05,
+      "loss": 3.0204,
+      "step": 5603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.419009104454173e-05,
+      "loss": 3.1422,
+      "step": 5604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4109911385479356e-05,
+      "loss": 3.0914,
+      "step": 5605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4029861544482277e-05,
+      "loss": 3.0345,
+      "step": 5606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3949941543387087e-05,
+      "loss": 3.2027,
+      "step": 5607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3870151403995234e-05,
+      "loss": 3.1112,
+      "step": 5608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.379049114807269e-05,
+      "loss": 3.1418,
+      "step": 5609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3710960797349912e-05,
+      "loss": 3.0829,
+      "step": 5610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3631560373522044e-05,
+      "loss": 3.1345,
+      "step": 5611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.35522898982487e-05,
+      "loss": 3.2449,
+      "step": 5612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3473149393154037e-05,
+      "loss": 3.1377,
+      "step": 5613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3394138879826886e-05,
+      "loss": 3.2888,
+      "step": 5614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3315258379820516e-05,
+      "loss": 3.2535,
+      "step": 5615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.323650791465265e-05,
+      "loss": 3.1639,
+      "step": 5616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.315788750580572e-05,
+      "loss": 3.1812,
+      "step": 5617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3079397174726512e-05,
+      "loss": 3.033,
+      "step": 5618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.300103694282646e-05,
+      "loss": 3.105,
+      "step": 5619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2922806831481402e-05,
+      "loss": 3.1516,
+      "step": 5620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.284470686203177e-05,
+      "loss": 3.1781,
+      "step": 5621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2766737055782583e-05,
+      "loss": 3.1672,
+      "step": 5622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2688897434002986e-05,
+      "loss": 3.4466,
+      "step": 5623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2611188017926943e-05,
+      "loss": 3.1497,
+      "step": 5624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2533608828752827e-05,
+      "loss": 3.1041,
+      "step": 5625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.245615988764349e-05,
+      "loss": 3.1792,
+      "step": 5626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.23788412157262e-05,
+      "loss": 3.1931,
+      "step": 5627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.23016528340928e-05,
+      "loss": 2.9503,
+      "step": 5628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2224594763799344e-05,
+      "loss": 3.2422,
+      "step": 5629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2147667025866568e-05,
+      "loss": 3.1883,
+      "step": 5630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2070869641279633e-05,
+      "loss": 3.2113,
+      "step": 5631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1994202630988113e-05,
+      "loss": 3.175,
+      "step": 5632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1917666015905946e-05,
+      "loss": 3.1057,
+      "step": 5633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.184125981691165e-05,
+      "loss": 3.0862,
+      "step": 5634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1764984054847947e-05,
+      "loss": 3.2033,
+      "step": 5635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1688838750522134e-05,
+      "loss": 3.148,
+      "step": 5636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1612823924705927e-05,
+      "loss": 3.1849,
+      "step": 5637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1536939598135406e-05,
+      "loss": 3.1379,
+      "step": 5638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1461185791511072e-05,
+      "loss": 3.1795,
+      "step": 5639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1385562525497838e-05,
+      "loss": 3.0553,
+      "step": 5640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1310069820724866e-05,
+      "loss": 3.1972,
+      "step": 5641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1234707697785848e-05,
+      "loss": 3.1765,
+      "step": 5642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1159476177238846e-05,
+      "loss": 3.062,
+      "step": 5643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1084375279606273e-05,
+      "loss": 3.172,
+      "step": 5644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1009405025374904e-05,
+      "loss": 3.1337,
+      "step": 5645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0934565434995944e-05,
+      "loss": 3.0838,
+      "step": 5646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0859856528884726e-05,
+      "loss": 3.0862,
+      "step": 5647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0785278327421218e-05,
+      "loss": 3.1934,
+      "step": 5648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0710830850949547e-05,
+      "loss": 3.1889,
+      "step": 5649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0636514119778238e-05,
+      "loss": 3.1315,
+      "step": 5650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0562328154180188e-05,
+      "loss": 3.3087,
+      "step": 5651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0488272974392654e-05,
+      "loss": 3.1141,
+      "step": 5652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.041434860061697e-05,
+      "loss": 3.1611,
+      "step": 5653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0340555053019126e-05,
+      "loss": 3.149,
+      "step": 5654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0266892351729183e-05,
+      "loss": 3.0845,
+      "step": 5655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0193360516841618e-05,
+      "loss": 3.1973,
+      "step": 5656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.011995956841517e-05,
+      "loss": 2.975,
+      "step": 5657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.004668952647298e-05,
+      "loss": 3.0875,
+      "step": 5658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.997355041100224e-05,
+      "loss": 2.9051,
+      "step": 5659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9900542241954645e-05,
+      "loss": 3.085,
+      "step": 5660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9827665039246046e-05,
+      "loss": 3.2426,
+      "step": 5661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.975491882275665e-05,
+      "loss": 3.024,
+      "step": 5662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.968230361233092e-05,
+      "loss": 2.9406,
+      "step": 5663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.960981942777762e-05,
+      "loss": 3.163,
+      "step": 5664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9537466288869542e-05,
+      "loss": 3.0246,
+      "step": 5665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.946524421534396e-05,
+      "loss": 3.1116,
+      "step": 5666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9393153226902384e-05,
+      "loss": 3.1128,
+      "step": 5667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.932119334321053e-05,
+      "loss": 3.1036,
+      "step": 5668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9249364583898253e-05,
+      "loss": 3.1381,
+      "step": 5669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9177666968559825e-05,
+      "loss": 3.0413,
+      "step": 5670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9106100516753598e-05,
+      "loss": 3.0466,
+      "step": 5671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9034665248002126e-05,
+      "loss": 3.0609,
+      "step": 5672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.896336118179226e-05,
+      "loss": 3.125,
+      "step": 5673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.889218833757511e-05,
+      "loss": 3.2234,
+      "step": 5674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.882114673476587e-05,
+      "loss": 3.1876,
+      "step": 5675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8750236392743923e-05,
+      "loss": 3.209,
+      "step": 5676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8679457330853077e-05,
+      "loss": 3.1199,
+      "step": 5677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8608809568400943e-05,
+      "loss": 3.1794,
+      "step": 5678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.85382931246596e-05,
+      "loss": 3.0941,
+      "step": 5679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8467908018865275e-05,
+      "loss": 3.0452,
+      "step": 5680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8397654270218278e-05,
+      "loss": 3.1724,
+      "step": 5681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8327531897883166e-05,
+      "loss": 3.0051,
+      "step": 5682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8257540920988636e-05,
+      "loss": 3.154,
+      "step": 5683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8187681358627418e-05,
+      "loss": 3.1737,
+      "step": 5684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.81179532298566e-05,
+      "loss": 3.207,
+      "step": 5685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.804835655369741e-05,
+      "loss": 3.2448,
+      "step": 5686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7978891349134828e-05,
+      "loss": 3.1649,
+      "step": 5687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7909557635118536e-05,
+      "loss": 3.2022,
+      "step": 5688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.784035543056206e-05,
+      "loss": 3.1362,
+      "step": 5689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777128475434292e-05,
+      "loss": 3.2726,
+      "step": 5690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770234562530304e-05,
+      "loss": 3.0861,
+      "step": 5691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7633538062248323e-05,
+      "loss": 3.2239,
+      "step": 5692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.756486208394864e-05,
+      "loss": 3.0944,
+      "step": 5693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7496317709138234e-05,
+      "loss": 3.1261,
+      "step": 5694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7427904956515416e-05,
+      "loss": 3.0781,
+      "step": 5695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.735962384474232e-05,
+      "loss": 3.125,
+      "step": 5696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.729147439244538e-05,
+      "loss": 3.269,
+      "step": 5697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7223456618215228e-05,
+      "loss": 3.1959,
+      "step": 5698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7155570540606236e-05,
+      "loss": 3.2378,
+      "step": 5699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7087816178137205e-05,
+      "loss": 3.1817,
+      "step": 5700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7020193549290852e-05,
+      "loss": 2.9893,
+      "step": 5701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.695270267251381e-05,
+      "loss": 2.9859,
+      "step": 5702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6885343566217016e-05,
+      "loss": 3.1793,
+      "step": 5703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.681811624877533e-05,
+      "loss": 3.0596,
+      "step": 5704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6751020738527746e-05,
+      "loss": 3.2552,
+      "step": 5705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6684057053777235e-05,
+      "loss": 3.0322,
+      "step": 5706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.66172252127908e-05,
+      "loss": 3.1178,
+      "step": 5707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6550525233799462e-05,
+      "loss": 3.3634,
+      "step": 5708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6483957134998396e-05,
+      "loss": 3.1225,
+      "step": 5709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6417520934546627e-05,
+      "loss": 3.2244,
+      "step": 5710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.635121665056738e-05,
+      "loss": 3.0655,
+      "step": 5711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285044301147634e-05,
+      "loss": 3.1764,
+      "step": 5712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621900390433878e-05,
+      "loss": 3.1152,
+      "step": 5713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61530954781558e-05,
+      "loss": 3.053,
+      "step": 5714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.608731904057792e-05,
+      "loss": 3.0433,
+      "step": 5715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6021674609548285e-05,
+      "loss": 3.1546,
+      "step": 5716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5956162202974133e-05,
+      "loss": 3.1169,
+      "step": 5717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5890781838726388e-05,
+      "loss": 3.1892,
+      "step": 5718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.582553353464039e-05,
+      "loss": 3.2242,
+      "step": 5719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.576041730851502e-05,
+      "loss": 3.0256,
+      "step": 5720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.569543317811345e-05,
+      "loss": 3.1503,
+      "step": 5721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5630581161162672e-05,
+      "loss": 3.1163,
+      "step": 5722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5565861275353755e-05,
+      "loss": 3.1321,
+      "step": 5723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5501273538341466e-05,
+      "loss": 3.0619,
+      "step": 5724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.543681796774482e-05,
+      "loss": 3.1392,
+      "step": 5725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5372494581146702e-05,
+      "loss": 3.1779,
+      "step": 5726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5308303396093682e-05,
+      "loss": 3.1552,
+      "step": 5727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5244244430096699e-05,
+      "loss": 3.2945,
+      "step": 5728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5180317700630274e-05,
+      "loss": 2.857,
+      "step": 5729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5116523225132961e-05,
+      "loss": 3.1144,
+      "step": 5730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5052861021007336e-05,
+      "loss": 3.0732,
+      "step": 5731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.498933110561984e-05,
+      "loss": 3.2471,
+      "step": 5732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.492593349630067e-05,
+      "loss": 3.153,
+      "step": 5733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4862668210344154e-05,
+      "loss": 3.1576,
+      "step": 5734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4799535265008435e-05,
+      "loss": 3.2363,
+      "step": 5735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4736534677515512e-05,
+      "loss": 3.0556,
+      "step": 5736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.467366646505125e-05,
+      "loss": 3.1518,
+      "step": 5737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4610930644765652e-05,
+      "loss": 3.0887,
+      "step": 5738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4548327233772308e-05,
+      "loss": 3.0697,
+      "step": 5739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4485856249148776e-05,
+      "loss": 3.0453,
+      "step": 5740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4423517707936595e-05,
+      "loss": 3.1049,
+      "step": 5741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4361311627140995e-05,
+      "loss": 3.2221,
+      "step": 5742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4299238023731231e-05,
+      "loss": 3.0702,
+      "step": 5743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.423729691464043e-05,
+      "loss": 3.1126,
+      "step": 5744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4175488316765407e-05,
+      "loss": 3.1977,
+      "step": 5745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4113812246967006e-05,
+      "loss": 3.1111,
+      "step": 5746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4052268722069828e-05,
+      "loss": 3.1497,
+      "step": 5747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3990857758862275e-05,
+      "loss": 3.122,
+      "step": 5748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3929579374096613e-05,
+      "loss": 3.1329,
+      "step": 5749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3868433584489137e-05,
+      "loss": 3.1188,
+      "step": 5750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3807420406719674e-05,
+      "loss": 3.1085,
+      "step": 5751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3746539857432016e-05,
+      "loss": 3.1243,
+      "step": 5752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3685791953233883e-05,
+      "loss": 3.2276,
+      "step": 5753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3625176710696519e-05,
+      "loss": 3.1399,
+      "step": 5754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.356469414635525e-05,
+      "loss": 3.1616,
+      "step": 5755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3504344276709214e-05,
+      "loss": 3.0259,
+      "step": 5756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3444127118221128e-05,
+      "loss": 3.0629,
+      "step": 5757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3384042687317632e-05,
+      "loss": 3.1944,
+      "step": 5758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3324091000389171e-05,
+      "loss": 3.1641,
+      "step": 5759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.326427207379005e-05,
+      "loss": 3.0947,
+      "step": 5760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3204585923838164e-05,
+      "loss": 3.0785,
+      "step": 5761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3145032566815318e-05,
+      "loss": 3.0591,
+      "step": 5762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.308561201896713e-05,
+      "loss": 3.0596,
+      "step": 5763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3026324296502857e-05,
+      "loss": 3.1481,
+      "step": 5764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.296716941559567e-05,
+      "loss": 3.1903,
+      "step": 5765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2908147392382496e-05,
+      "loss": 3.1231,
+      "step": 5766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2849258242963734e-05,
+      "loss": 3.0665,
+      "step": 5767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2790501983403924e-05,
+      "loss": 3.2479,
+      "step": 5768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2731878629731131e-05,
+      "loss": 3.1062,
+      "step": 5769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2673388197937231e-05,
+      "loss": 3.1694,
+      "step": 5770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2615030703977848e-05,
+      "loss": 3.0937,
+      "step": 5771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2556806163772361e-05,
+      "loss": 3.2145,
+      "step": 5772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2498714593203674e-05,
+      "loss": 3.1417,
+      "step": 5773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2440756008118725e-05,
+      "loss": 2.958,
+      "step": 5774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2382930424328087e-05,
+      "loss": 3.0864,
+      "step": 5775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.232523785760592e-05,
+      "loss": 3.1011,
+      "step": 5776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2267678323690135e-05,
+      "loss": 3.1926,
+      "step": 5777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.221025183828256e-05,
+      "loss": 3.0757,
+      "step": 5778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2152958417048388e-05,
+      "loss": 3.1547,
+      "step": 5779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.209579807561667e-05,
+      "loss": 3.1484,
+      "step": 5780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2038770829580437e-05,
+      "loss": 3.0722,
+      "step": 5781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1981876694495907e-05,
+      "loss": 3.2802,
+      "step": 5782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1925115685883336e-05,
+      "loss": 3.0681,
+      "step": 5783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.186848781922656e-05,
+      "loss": 3.1523,
+      "step": 5784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1811993109973051e-05,
+      "loss": 3.0257,
+      "step": 5785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1755631573533987e-05,
+      "loss": 3.2309,
+      "step": 5786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1699403225284289e-05,
+      "loss": 3.1925,
+      "step": 5787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1643308080562465e-05,
+      "loss": 3.1396,
+      "step": 5788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.158734615467072e-05,
+      "loss": 3.01,
+      "step": 5789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1531517462874952e-05,
+      "loss": 3.2535,
+      "step": 5790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1475822020404591e-05,
+      "loss": 3.0723,
+      "step": 5791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1420259842452817e-05,
+      "loss": 2.9622,
+      "step": 5792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1364830944176453e-05,
+      "loss": 2.8851,
+      "step": 5793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1309535340695897e-05,
+      "loss": 3.0864,
+      "step": 5794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1254373047095367e-05,
+      "loss": 3.0258,
+      "step": 5795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1199344078422491e-05,
+      "loss": 3.1879,
+      "step": 5796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1144448449688593e-05,
+      "loss": 2.9528,
+      "step": 5797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1089686175868697e-05,
+      "loss": 3.0667,
+      "step": 5798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.103505727190146e-05,
+      "loss": 2.96,
+      "step": 5799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0980561752688967e-05,
+      "loss": 3.1021,
+      "step": 5800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0926199633097156e-05,
+      "loss": 3.0839,
+      "step": 5801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0871970927955498e-05,
+      "loss": 2.8838,
+      "step": 5802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.081787565205694e-05,
+      "loss": 3.2359,
+      "step": 5803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0763913820158233e-05,
+      "loss": 3.0948,
+      "step": 5804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.071008544697949e-05,
+      "loss": 3.1211,
+      "step": 5805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0656390547204686e-05,
+      "loss": 3.1752,
+      "step": 5806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0602829135481162e-05,
+      "loss": 3.0305,
+      "step": 5807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0549401226420064e-05,
+      "loss": 3.0043,
+      "step": 5808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0496106834595897e-05,
+      "loss": 3.1328,
+      "step": 5809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0442945974546813e-05,
+      "loss": 3.3263,
+      "step": 5810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.03899186607746e-05,
+      "loss": 3.2463,
+      "step": 5811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0337024907744574e-05,
+      "loss": 2.9378,
+      "step": 5812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.028426472988564e-05,
+      "loss": 3.0917,
+      "step": 5813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0231638141590227e-05,
+      "loss": 3.2037,
+      "step": 5814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.017914515721441e-05,
+      "loss": 3.1373,
+      "step": 5815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.012678579107762e-05,
+      "loss": 3.1917,
+      "step": 5816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0074560057463044e-05,
+      "loss": 2.9772,
+      "step": 5817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.00224679706174e-05,
+      "loss": 3.194,
+      "step": 5818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.970509544750761e-06,
+      "loss": 3.0528,
+      "step": 5819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.918684794037015e-06,
+      "loss": 3.0101,
+      "step": 5820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.86699373261335e-06,
+      "loss": 3.1465,
+      "step": 5821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.815436374580544e-06,
+      "loss": 3.0981,
+      "step": 5822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.764012734002958e-06,
+      "loss": 3.1508,
+      "step": 5823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.712722824908426e-06,
+      "loss": 3.1847,
+      "step": 5824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.661566661288423e-06,
+      "loss": 3.1078,
+      "step": 5825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.610544257097731e-06,
+      "loss": 2.9871,
+      "step": 5826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.559655626254826e-06,
+      "loss": 3.2144,
+      "step": 5827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.50890078264155e-06,
+      "loss": 3.2244,
+      "step": 5828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.458279740103327e-06,
+      "loss": 3.0702,
+      "step": 5829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.407792512449109e-06,
+      "loss": 3.2553,
+      "step": 5830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.357439113451327e-06,
+      "loss": 3.1375,
+      "step": 5831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.307219556845826e-06,
+      "loss": 3.1523,
+      "step": 5832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.257133856332034e-06,
+      "loss": 3.1633,
+      "step": 5833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.207182025572802e-06,
+      "loss": 3.1104,
+      "step": 5834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.157364078194509e-06,
+      "loss": 3.063,
+      "step": 5835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.107680027787058e-06,
+      "loss": 3.2492,
+      "step": 5836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.05812988790361e-06,
+      "loss": 3.0196,
+      "step": 5837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.008713672061075e-06,
+      "loss": 3.2821,
+      "step": 5838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.959431393739726e-06,
+      "loss": 3.0367,
+      "step": 5839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.91028306638314e-06,
+      "loss": 3.0892,
+      "step": 5840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.861268703398596e-06,
+      "loss": 3.2131,
+      "step": 5841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.812388318156672e-06,
+      "loss": 3.2442,
+      "step": 5842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.763641923991483e-06,
+      "loss": 3.1646,
+      "step": 5843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.715029534200503e-06,
+      "loss": 3.2161,
+      "step": 5844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.666551162044845e-06,
+      "loss": 3.1733,
+      "step": 5845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.618206820748764e-06,
+      "loss": 3.0776,
+      "step": 5846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.569996523500212e-06,
+      "loss": 3.2024,
+      "step": 5847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.52192028345039e-06,
+      "loss": 3.0798,
+      "step": 5848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.473978113714143e-06,
+      "loss": 3.1359,
+      "step": 5849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.426170027369451e-06,
+      "loss": 3.1325,
+      "step": 5850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.378496037458106e-06,
+      "loss": 3.1476,
+      "step": 5851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.33095615698487e-06,
+      "loss": 3.0812,
+      "step": 5852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.28355039891826e-06,
+      "loss": 3.0894,
+      "step": 5853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.236278776190097e-06,
+      "loss": 3.0367,
+      "step": 5854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.189141301695568e-06,
+      "loss": 3.2506,
+      "step": 5855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.14213798829333e-06,
+      "loss": 3.1903,
+      "step": 5856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.095268848805404e-06,
+      "loss": 3.2277,
+      "step": 5857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.04853389601723e-06,
+      "loss": 3.1115,
+      "step": 5858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.001933142677664e-06,
+      "loss": 3.2206,
+      "step": 5859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.955466601498873e-06,
+      "loss": 3.2783,
+      "step": 5860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.90913428515655e-06,
+      "loss": 3.2288,
+      "step": 5861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.862936206289529e-06,
+      "loss": 3.1356,
+      "step": 5862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.816872377500395e-06,
+      "loss": 3.0367,
+      "step": 5863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.770942811354764e-06,
+      "loss": 3.1318,
+      "step": 5864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.725147520381781e-06,
+      "loss": 3.0605,
+      "step": 5865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.679486517073953e-06,
+      "loss": 3.27,
+      "step": 5866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.633959813887148e-06,
+      "loss": 3.1798,
+      "step": 5867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.588567423240544e-06,
+      "loss": 3.1512,
+      "step": 5868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.543309357516848e-06,
+      "loss": 3.0595,
+      "step": 5869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4981856290619595e-06,
+      "loss": 3.087,
+      "step": 5870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.453196250185146e-06,
+      "loss": 3.1736,
+      "step": 5871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.408341233159088e-06,
+      "loss": 2.906,
+      "step": 5872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.3636205902197776e-06,
+      "loss": 3.0359,
+      "step": 5873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.31903433356651e-06,
+      "loss": 3.2488,
+      "step": 5874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.274582475362057e-06,
+      "loss": 3.2926,
+      "step": 5875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.230265027732441e-06,
+      "loss": 3.0448,
+      "step": 5876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.186082002766991e-06,
+      "loss": 3.1184,
+      "step": 5877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.142033412518345e-06,
+      "loss": 3.1051,
+      "step": 5878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.098119269002612e-06,
+      "loss": 3.1055,
+      "step": 5879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.0543395841989875e-06,
+      "loss": 3.1803,
+      "step": 5880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.010694370050308e-06,
+      "loss": 3.1889,
+      "step": 5881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.967183638462493e-06,
+      "loss": 3.1123,
+      "step": 5882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.9238074013047715e-06,
+      "loss": 3.0366,
+      "step": 5883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.880565670409789e-06,
+      "loss": 3.0802,
+      "step": 5884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.837458457573498e-06,
+      "loss": 3.0471,
+      "step": 5885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.794485774555048e-06,
+      "loss": 2.945,
+      "step": 5886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.7516476330769514e-06,
+      "loss": 3.3361,
+      "step": 5887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.708944044825138e-06,
+      "loss": 3.0841,
+      "step": 5888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.666375021448623e-06,
+      "loss": 3.254,
+      "step": 5889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.623940574559839e-06,
+      "loss": 3.2136,
+      "step": 5890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.581640715734472e-06,
+      "loss": 3.1139,
+      "step": 5891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.539475456511512e-06,
+      "loss": 3.1468,
+      "step": 5892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.497444808393149e-06,
+      "loss": 3.0984,
+      "step": 5893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4555487828450445e-06,
+      "loss": 3.0068,
+      "step": 5894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.413787391295944e-06,
+      "loss": 2.9969,
+      "step": 5895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.372160645137903e-06,
+      "loss": 3.12,
+      "step": 5896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.330668555726393e-06,
+      "loss": 2.9608,
+      "step": 5897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.289311134380027e-06,
+      "loss": 2.9508,
+      "step": 5898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.248088392380502e-06,
+      "loss": 3.1603,
+      "step": 5899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.207000340973268e-06,
+      "loss": 3.1351,
+      "step": 5900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.166046991366525e-06,
+      "loss": 3.1398,
+      "step": 5901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1252283547320046e-06,
+      "loss": 3.0199,
+      "step": 5902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.084544442204576e-06,
+      "loss": 3.1157,
+      "step": 5903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.043995264882529e-06,
+      "loss": 3.1596,
+      "step": 5904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.003580833827127e-06,
+      "loss": 3.1201,
+      "step": 5905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.963301160063106e-06,
+      "loss": 3.2139,
+      "step": 5906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.923156254578343e-06,
+      "loss": 3.0825,
+      "step": 5907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.883146128323913e-06,
+      "loss": 3.23,
+      "step": 5908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.843270792214306e-06,
+      "loss": 3.1292,
+      "step": 5909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8035302571270435e-06,
+      "loss": 3.1517,
+      "step": 5910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.763924533902898e-06,
+      "loss": 3.1377,
+      "step": 5911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.724453633345949e-06,
+      "loss": 3.0958,
+      "step": 5912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.68511756622353e-06,
+      "loss": 3.1083,
+      "step": 5913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.645916343266056e-06,
+      "loss": 3.1886,
+      "step": 5914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.606849975167194e-06,
+      "loss": 3.0474,
+      "step": 5915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.567918472584032e-06,
+      "loss": 3.2674,
+      "step": 5916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.529121846136465e-06,
+      "loss": 3.2918,
+      "step": 5917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4904601064079126e-06,
+      "loss": 3.0303,
+      "step": 5918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.451933263945053e-06,
+      "loss": 3.1391,
+      "step": 5919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.413541329257421e-06,
+      "loss": 3.2609,
+      "step": 5920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3752843128180855e-06,
+      "loss": 3.1735,
+      "step": 5921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.337162225063141e-06,
+      "loss": 3.1581,
+      "step": 5922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.299175076391827e-06,
+      "loss": 3.1932,
+      "step": 5923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.261322877166741e-06,
+      "loss": 3.148,
+      "step": 5924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.22360563771368e-06,
+      "loss": 3.0843,
+      "step": 5925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1860233683213576e-06,
+      "loss": 3.1876,
+      "step": 5926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.148576079241907e-06,
+      "loss": 3.1561,
+      "step": 5927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1112637806906e-06,
+      "loss": 3.2201,
+      "step": 5928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.074086482845852e-06,
+      "loss": 3.1928,
+      "step": 5929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.037044195849216e-06,
+      "loss": 3.2039,
+      "step": 5930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.000136929805443e-06,
+      "loss": 3.1518,
+      "step": 5931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9633646947825905e-06,
+      "loss": 3.1451,
+      "step": 5932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.926727500811634e-06,
+      "loss": 2.879,
+      "step": 5933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.890225357886913e-06,
+      "loss": 3.1746,
+      "step": 5934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.853858275965794e-06,
+      "loss": 3.0657,
+      "step": 5935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8176262649689e-06,
+      "loss": 3.1246,
+      "step": 5936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.781529334779877e-06,
+      "loss": 3.0573,
+      "step": 5937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.74556749524574e-06,
+      "loss": 3.2034,
+      "step": 5938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.709740756176362e-06,
+      "loss": 3.2719,
+      "step": 5939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.674049127345093e-06,
+      "loss": 3.0877,
+      "step": 5940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.638492618488088e-06,
+      "loss": 2.9772,
+      "step": 5941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.603071239304924e-06,
+      "loss": 3.247,
+      "step": 5942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.567784999458147e-06,
+      "loss": 3.017,
+      "step": 5943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5326339085735025e-06,
+      "loss": 3.204,
+      "step": 5944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.49761797623982e-06,
+      "loss": 3.1675,
+      "step": 5945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.462737212009182e-06,
+      "loss": 3.1458,
+      "step": 5946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.427991625396644e-06,
+      "loss": 3.0829,
+      "step": 5947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.39338122588051e-06,
+      "loss": 3.2146,
+      "step": 5948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.358906022902065e-06,
+      "loss": 3.1524,
+      "step": 5949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.324566025865895e-06,
+      "loss": 3.144,
+      "step": 5950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.290361244139506e-06,
+      "loss": 3.1683,
+      "step": 5951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.256291687053715e-06,
+      "loss": 3.2288,
+      "step": 5952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.222357363902363e-06,
+      "loss": 3.2062,
+      "step": 5953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.188558283942323e-06,
+      "loss": 3.0767,
+      "step": 5954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.154894456393665e-06,
+      "loss": 3.1157,
+      "step": 5955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.121365890439544e-06,
+      "loss": 3.1962,
+      "step": 5956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.087972595226252e-06,
+      "loss": 3.3033,
+      "step": 5957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.054714579863117e-06,
+      "loss": 3.1311,
+      "step": 5958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.021591853422601e-06,
+      "loss": 3.0704,
+      "step": 5959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.988604424940257e-06,
+      "loss": 3.2803,
+      "step": 5960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.955752303414717e-06,
+      "loss": 3.1347,
+      "step": 5961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9230354978077584e-06,
+      "loss": 3.1817,
+      "step": 5962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.890454017044076e-06,
+      "loss": 3.012,
+      "step": 5963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8580078700117264e-06,
+      "loss": 2.9873,
+      "step": 5964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.825697065561629e-06,
+      "loss": 3.2002,
+      "step": 5965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7935216125078443e-06,
+      "loss": 3.1818,
+      "step": 5966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7614815196274633e-06,
+      "loss": 3.127,
+      "step": 5967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7295767956607717e-06,
+      "loss": 3.1516,
+      "step": 5968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6978074493110303e-06,
+      "loss": 3.1372,
+      "step": 5969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6661734892446396e-06,
+      "loss": 3.1127,
+      "step": 5970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6346749240910304e-06,
+      "loss": 2.9377,
+      "step": 5971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.603311762442607e-06,
+      "loss": 3.2213,
+      "step": 5972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5720840128550256e-06,
+      "loss": 3.1827,
+      "step": 5973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.540991683846806e-06,
+      "loss": 3.0245,
+      "step": 5974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5100347838997183e-06,
+      "loss": 3.1961,
+      "step": 5975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4792133214585074e-06,
+      "loss": 3.1742,
+      "step": 5976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.448527304930893e-06,
+      "loss": 3.0325,
+      "step": 5977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4179767426876784e-06,
+      "loss": 3.1653,
+      "step": 5978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.387561643062864e-06,
+      "loss": 3.1377,
+      "step": 5979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3572820143533133e-06,
+      "loss": 3.1327,
+      "step": 5980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3271378648190296e-06,
+      "loss": 3.0636,
+      "step": 5981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2971292026829915e-06,
+      "loss": 3.1783,
+      "step": 5982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.267256036131372e-06,
+      "loss": 3.3819,
+      "step": 5983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.237518373313153e-06,
+      "loss": 3.1377,
+      "step": 5984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2079162223405123e-06,
+      "loss": 3.2663,
+      "step": 5985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.178449591288657e-06,
+      "loss": 3.12,
+      "step": 5986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.149118488195768e-06,
+      "loss": 3.1093,
+      "step": 5987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1199229210630565e-06,
+      "loss": 3.1641,
+      "step": 5988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.090862897854818e-06,
+      "loss": 3.2216,
+      "step": 5989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.061938426498323e-06,
+      "loss": 2.9737,
+      "step": 5990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.033149514883815e-06,
+      "loss": 3.285,
+      "step": 5991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0044961708647343e-06,
+      "loss": 2.9358,
+      "step": 5992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.975978402257329e-06,
+      "loss": 3.2881,
+      "step": 5993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.947596216841042e-06,
+      "loss": 3.2222,
+      "step": 5994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9193496223581805e-06,
+      "loss": 3.1006,
+      "step": 5995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8912386265141923e-06,
+      "loss": 2.9476,
+      "step": 5996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8632632369774424e-06,
+      "loss": 3.0084,
+      "step": 5997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8354234613793275e-06,
+      "loss": 3.2239,
+      "step": 5998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8077193073142713e-06,
+      "loss": 3.1365,
+      "step": 5999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7801507823397297e-06,
+      "loss": 3.1141,
+      "step": 6000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7527178939760754e-06,
+      "loss": 3.1217,
+      "step": 6001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7254206497067114e-06,
+      "loss": 3.0611,
+      "step": 6002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.698259056978125e-06,
+      "loss": 3.2002,
+      "step": 6003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.671233123199668e-06,
+      "loss": 3.0797,
+      "step": 6004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6443428557437753e-06,
+      "loss": 3.0632,
+      "step": 6005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.617588261945747e-06,
+      "loss": 3.1875,
+      "step": 6006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.590969349104133e-06,
+      "loss": 3.2222,
+      "step": 6007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5644861244802366e-06,
+      "loss": 3.2405,
+      "step": 6008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5381385952983337e-06,
+      "loss": 3.0538,
+      "step": 6009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.511926768745898e-06,
+      "loss": 3.045,
+      "step": 6010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4858506519732095e-06,
+      "loss": 3.2631,
+      "step": 6011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.459910252093467e-06,
+      "loss": 3.1434,
+      "step": 6012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.434105576183121e-06,
+      "loss": 3.1258,
+      "step": 6013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.408436631281319e-06,
+      "loss": 3.2163,
+      "step": 6014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.382903424390348e-06,
+      "loss": 3.1349,
+      "step": 6015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3575059624754147e-06,
+      "loss": 3.0948,
+      "step": 6016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3322442524646436e-06,
+      "loss": 3.1551,
+      "step": 6017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3071183012491337e-06,
+      "loss": 3.1125,
+      "step": 6018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.282128115683124e-06,
+      "loss": 3.1164,
+      "step": 6019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.257273702583662e-06,
+      "loss": 3.0705,
+      "step": 6020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.232555068730713e-06,
+      "loss": 3.3095,
+      "step": 6021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2079722208672714e-06,
+      "loss": 3.1459,
+      "step": 6022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.183525165699418e-06,
+      "loss": 3.0908,
+      "step": 6023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.159213909895874e-06,
+      "loss": 3.0842,
+      "step": 6024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1350384600886675e-06,
+      "loss": 3.1292,
+      "step": 6025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1109988228725786e-06,
+      "loss": 3.1083,
+      "step": 6026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.087095004805306e-06,
+      "loss": 3.0559,
+      "step": 6027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.063327012407634e-06,
+      "loss": 3.1821,
+      "step": 6028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0396948521632653e-06,
+      "loss": 3.1748,
+      "step": 6029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0161985305187647e-06,
+      "loss": 3.1428,
+      "step": 6030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9928380538837278e-06,
+      "loss": 3.141,
+      "step": 6031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.969613428630668e-06,
+      "loss": 3.0225,
+      "step": 6032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9465246610949616e-06,
+      "loss": 3.2515,
+      "step": 6033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9235717575750155e-06,
+      "loss": 3.2018,
+      "step": 6034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9007547243322099e-06,
+      "loss": 3.0552,
+      "step": 6035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.878073567590788e-06,
+      "loss": 3.0947,
+      "step": 6036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8555282935378559e-06,
+      "loss": 3.2657,
+      "step": 6037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8331189083237166e-06,
+      "loss": 3.0422,
+      "step": 6038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8108454180612577e-06,
+      "loss": 3.1003,
+      "step": 6039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.788707828826508e-06,
+      "loss": 2.9106,
+      "step": 6040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7667061466584144e-06,
+      "loss": 3.107,
+      "step": 6041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7448403775587873e-06,
+      "loss": 3.0328,
+      "step": 6042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7231105274924108e-06,
+      "loss": 3.1148,
+      "step": 6043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7015166023869878e-06,
+      "loss": 2.977,
+      "step": 6044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6800586081330282e-06,
+      "loss": 3.2733,
+      "step": 6045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6587365505841833e-06,
+      "loss": 3.2722,
+      "step": 6046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.637550435556856e-06,
+      "loss": 2.9988,
+      "step": 6047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.616500268830423e-06,
+      "loss": 3.1488,
+      "step": 6048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5955860561470691e-06,
+      "loss": 3.2635,
+      "step": 6049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5748078032120638e-06,
+      "loss": 3.0553,
+      "step": 6050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5541655156934841e-06,
+      "loss": 3.2884,
+      "step": 6051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5336591992223814e-06,
+      "loss": 3.1941,
+      "step": 6052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5132888593925586e-06,
+      "loss": 3.0921,
+      "step": 6053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4930545017610152e-06,
+      "loss": 3.0898,
+      "step": 6054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.472956131847336e-06,
+      "loss": 3.0606,
+      "step": 6055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.452993755134191e-06,
+      "loss": 3.0345,
+      "step": 6056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4331673770671683e-06,
+      "loss": 3.134,
+      "step": 6057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.413477003054664e-06,
+      "loss": 3.2434,
+      "step": 6058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.393922638468048e-06,
+      "loss": 2.9231,
+      "step": 6059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3745042886414983e-06,
+      "loss": 3.0694,
+      "step": 6060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3552219588721659e-06,
+      "loss": 3.147,
+      "step": 6061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3360756544201213e-06,
+      "loss": 3.1625,
+      "step": 6062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3170653805082423e-06,
+      "loss": 3.0558,
+      "step": 6063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2981911423223248e-06,
+      "loss": 3.1825,
+      "step": 6064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2794529450111392e-06,
+      "loss": 3.0919,
+      "step": 6065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2608507936862635e-06,
+      "loss": 3.0549,
+      "step": 6066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2423846934220829e-06,
+      "loss": 3.1738,
+      "step": 6067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2240546492560678e-06,
+      "loss": 3.1592,
+      "step": 6068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2058606661884964e-06,
+      "loss": 3.1206,
+      "step": 6069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1878027491823984e-06,
+      "loss": 3.2168,
+      "step": 6070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1698809031638336e-06,
+      "loss": 3.1052,
+      "step": 6071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.15209513302178e-06,
+      "loss": 3.0692,
+      "step": 6072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.134445443607912e-06,
+      "loss": 3.0483,
+      "step": 6073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1169318397369344e-06,
+      "loss": 3.0828,
+      "step": 6074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0995543261863584e-06,
+      "loss": 3.2017,
+      "step": 6075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0823129076966142e-06,
+      "loss": 3.2117,
+      "step": 6076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.065207588971051e-06,
+      "loss": 3.0492,
+      "step": 6077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.048238374675714e-06,
+      "loss": 3.2024,
+      "step": 6078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0314052694397335e-06,
+      "loss": 3.0089,
+      "step": 6079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0147082778549367e-06,
+      "loss": 3.1902,
+      "step": 6080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.98147404476124e-07,
+      "loss": 3.2471,
+      "step": 6081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.8172265382096e-07,
+      "loss": 3.2112,
+      "step": 6082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.654340303699161e-07,
+      "loss": 3.1345,
+      "step": 6083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.492815385664377e-07,
+      "loss": 3.1793,
+      "step": 6084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.332651828166672e-07,
+      "loss": 2.969,
+      "step": 6085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.173849674897206e-07,
+      "loss": 3.2883,
+      "step": 6086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.016408969176326e-07,
+      "loss": 3.2068,
+      "step": 6087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.860329753951901e-07,
+      "loss": 3.2056,
+      "step": 6088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.705612071800429e-07,
+      "loss": 2.9873,
+      "step": 6089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.552255964928146e-07,
+      "loss": 3.0509,
+      "step": 6090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.400261475168258e-07,
+      "loss": 3.2052,
+      "step": 6091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.249628643983708e-07,
+      "loss": 3.1778,
+      "step": 6092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.100357512466072e-07,
+      "loss": 2.9292,
+      "step": 6093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.952448121333888e-07,
+      "loss": 3.1732,
+      "step": 6094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.805900510936547e-07,
+      "loss": 2.97,
+      "step": 6095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.66071472124985e-07,
+      "loss": 2.967,
+      "step": 6096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.516890791879339e-07,
+      "loss": 3.1864,
+      "step": 6097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.374428762059182e-07,
+      "loss": 3.1148,
+      "step": 6098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.233328670651073e-07,
+      "loss": 3.08,
+      "step": 6099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.093590556145891e-07,
+      "loss": 3.1764,
+      "step": 6100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.955214456663139e-07,
+      "loss": 2.9644,
+      "step": 6101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.818200409949849e-07,
+      "loss": 3.0514,
+      "step": 6102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.682548453382231e-07,
+      "loss": 3.2485,
+      "step": 6103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.548258623965131e-07,
+      "loss": 3.1937,
+      "step": 6104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.415330958330912e-07,
+      "loss": 3.1076,
+      "step": 6105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.283765492741678e-07,
+      "loss": 2.9434,
+      "step": 6106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.153562263086498e-07,
+      "loss": 3.0277,
+      "step": 6107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.02472130488363e-07,
+      "loss": 3.2009,
+      "step": 6108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.897242653279955e-07,
+      "loss": 3.1393,
+      "step": 6109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.771126343049881e-07,
+      "loss": 3.0085,
+      "step": 6110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.646372408597001e-07,
+      "loss": 3.1107,
+      "step": 6111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.522980883952422e-07,
+      "loss": 3.2047,
+      "step": 6112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.400951802777553e-07,
+      "loss": 3.2768,
+      "step": 6113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.280285198359103e-07,
+      "loss": 3.1898,
+      "step": 6114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.160981103614626e-07,
+      "loss": 3.0722,
+      "step": 6115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.043039551088646e-07,
+      "loss": 3.2404,
+      "step": 6116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.926460572954317e-07,
+      "loss": 3.098,
+      "step": 6117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.811244201013976e-07,
+      "loss": 3.1392,
+      "step": 6118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.697390466696372e-07,
+      "loss": 3.0957,
+      "step": 6119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5848994010611043e-07,
+      "loss": 3.2409,
+      "step": 6120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.473771034793628e-07,
+      "loss": 3.1426,
+      "step": 6121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3640053982085816e-07,
+      "loss": 3.0202,
+      "step": 6122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2556025212492356e-07,
+      "loss": 3.0489,
+      "step": 6123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1485624334869353e-07,
+      "loss": 3.0799,
+      "step": 6124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0428851641211016e-07,
+      "loss": 2.7815,
+      "step": 6125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9385707419792305e-07,
+      "loss": 3.1315,
+      "step": 6126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.835619195517448e-07,
+      "loss": 3.1249,
+      "step": 6127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7340305528194006e-07,
+      "loss": 3.0916,
+      "step": 6128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.633804841598476e-07,
+      "loss": 2.9497,
+      "step": 6129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5349420891939157e-07,
+      "loss": 2.9506,
+      "step": 6130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.437442322575812e-07,
+      "loss": 3.2029,
+      "step": 6131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.341305568340114e-07,
+      "loss": 3.1657,
+      "step": 6132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.246531852713064e-07,
+      "loss": 2.9914,
+      "step": 6133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1531212015467605e-07,
+      "loss": 3.1542,
+      "step": 6134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0610736403235973e-07,
+      "loss": 3.2101,
+      "step": 6135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9703891941523785e-07,
+      "loss": 3.1066,
+      "step": 6136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.881067887771649e-07,
+      "loss": 3.2224,
+      "step": 6137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.793109745547473e-07,
+      "loss": 3.1484,
+      "step": 6138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.706514791473436e-07,
+      "loss": 3.1242,
+      "step": 6139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.621283049171752e-07,
+      "loss": 3.1083,
+      "step": 6140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.537414541893268e-07,
+      "loss": 3.2423,
+      "step": 6141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4549092925157947e-07,
+      "loss": 3.1174,
+      "step": 6142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3737673235468827e-07,
+      "loss": 3.1476,
+      "step": 6143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2939886571204937e-07,
+      "loss": 3.0187,
+      "step": 6144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2155733149992196e-07,
+      "loss": 3.0417,
+      "step": 6145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1385213185748375e-07,
+      "loss": 3.1507,
+      "step": 6146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.06283268886609e-07,
+      "loss": 3.163,
+      "step": 6147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9885074465197937e-07,
+      "loss": 3.0262,
+      "step": 6148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.915545611811398e-07,
+      "loss": 3.0834,
+      "step": 6149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.843947204644425e-07,
+      "loss": 3.0183,
+      "step": 6150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7737122445493637e-07,
+      "loss": 3.1082,
+      "step": 6151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7048407506858877e-07,
+      "loss": 3.1886,
+      "step": 6152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6373327418423013e-07,
+      "loss": 3.2377,
+      "step": 6153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5711882364327635e-07,
+      "loss": 3.0895,
+      "step": 6154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5064072525017292e-07,
+      "loss": 3.2126,
+      "step": 6155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4429898077211735e-07,
+      "loss": 3.144,
+      "step": 6156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.380935919389481e-07,
+      "loss": 3.1136,
+      "step": 6157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3202456044353328e-07,
+      "loss": 3.0606,
+      "step": 6158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2609188794143744e-07,
+      "loss": 3.1164,
+      "step": 6159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2029557605097717e-07,
+      "loss": 3.0261,
+      "step": 6160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1463562635333213e-07,
+      "loss": 3.1631,
+      "step": 6161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0911204039254497e-07,
+      "loss": 3.1024,
+      "step": 6162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0372481967541036e-07,
+      "loss": 3.0692,
+      "step": 6163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.847396567141953e-08,
+      "loss": 3.0852,
+      "step": 6164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.335947981298221e-08,
+      "loss": 3.1154,
+      "step": 6165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.838136349526016e-08,
+      "loss": 3.1759,
+      "step": 6166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.353961807627819e-08,
+      "loss": 3.1643,
+      "step": 6167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.883424487681312e-08,
+      "loss": 3.1862,
+      "step": 6168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.42652451804493e-08,
+      "loss": 3.2032,
+      "step": 6169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.98326202335231e-08,
+      "loss": 3.0515,
+      "step": 6170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.553637124523393e-08,
+      "loss": 3.0979,
+      "step": 6171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.137649938758871e-08,
+      "loss": 3.2015,
+      "step": 6172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.73530057952909e-08,
+      "loss": 3.0306,
+      "step": 6173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3465891565962486e-08,
+      "loss": 3.1208,
+      "step": 6174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.971515775992197e-08,
+      "loss": 3.0952,
+      "step": 6175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.610080540035089e-08,
+      "loss": 3.1886,
+      "step": 6176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.262283547323831e-08,
+      "loss": 3.1066,
+      "step": 6177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.928124892732532e-08,
+      "loss": 3.1035,
+      "step": 6178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.607604667416054e-08,
+      "loss": 2.9709,
+      "step": 6179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.300722958810009e-08,
+      "loss": 3.1927,
+      "step": 6180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.007479850625217e-08,
+      "loss": 3.0113,
+      "step": 6181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7278754228587944e-08,
+      "loss": 3.1605,
+      "step": 6182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4619097517830646e-08,
+      "loss": 3.1529,
+      "step": 6183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.209582909945551e-08,
+      "loss": 3.0735,
+      "step": 6184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9708949661911834e-08,
+      "loss": 3.174,
+      "step": 6185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.745845985617889e-08,
+      "loss": 3.1061,
+      "step": 6186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5344360296265513e-08,
+      "loss": 3.1935,
+      "step": 6187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3366651558877063e-08,
+      "loss": 3.1264,
+      "step": 6188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1525334183415392e-08,
+      "loss": 3.1197,
+      "step": 6189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.82040867225642e-09,
+      "loss": 3.0783,
+      "step": 6190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.251875490472572e-09,
+      "loss": 3.1003,
+      "step": 6191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.8197350659437995e-09,
+      "loss": 3.2202,
+      "step": 6192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.523987789302076e-09,
+      "loss": 2.8782,
+      "step": 6193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.364634014097923e-09,
+      "loss": 3.1175,
+      "step": 6194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3416740565228586e-09,
+      "loss": 3.1302,
+      "step": 6195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4551081956314392e-09,
+      "loss": 3.117,
+      "step": 6196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7049366732857508e-09,
+      "loss": 3.1732,
+      "step": 6197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.091159694155408e-09,
+      "loss": 3.056,
+      "step": 6198
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.137774256065321e-10,
+      "loss": 3.0067,
+      "step": 6199
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.727899979793058e-10,
+      "loss": 3.1415,
+      "step": 6200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.819750414388537e-11,
+      "loss": 2.839,
+      "step": 6201
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.2192,
+      "step": 6202
+    },
+    {
+      "epoch": 1.0,
+      "step": 6202,
+      "total_flos": 1.977746612331479e+17,
+      "train_loss": 3.291977916983242,
+      "train_runtime": 15293.3269,
+      "train_samples_per_second": 38.93,
+      "train_steps_per_second": 0.406
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6202,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 1.977746612331479e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}