diff --git "a/checkpoint-4000/trainer_state.json" "b/checkpoint-4000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-4000/trainer_state.json"
@@ -0,0 +1,24019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7815934736944946,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.9887,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.7994,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.7656,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.7299,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.817,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 1.6701,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.6423,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.806,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 1.873,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.4258,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.7725,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 1.7807,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.6943,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.7889,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.704,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.4792,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.2699,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-05,
+      "loss": 1.6088,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-05,
+      "loss": 1.4344,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.5443,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4e-05,
+      "loss": 1.2563,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.3136,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.4096,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.6e-05,
+      "loss": 1.5926,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.2075,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.4512,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.4723,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.2824,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000116,
+      "loss": 1.172,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012,
+      "loss": 1.2105,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000124,
+      "loss": 1.352,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.0482,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000132,
+      "loss": 1.1847,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.0802,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014,
+      "loss": 1.3188,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000144,
+      "loss": 1.0549,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000148,
+      "loss": 1.6955,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000152,
+      "loss": 1.4828,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.5531,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016,
+      "loss": 1.5357,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000164,
+      "loss": 1.2398,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000168,
+      "loss": 1.3306,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000172,
+      "loss": 1.6565,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.1426,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018,
+      "loss": 1.2146,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.319,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000188,
+      "loss": 1.3445,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000192,
+      "loss": 1.366,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000196,
+      "loss": 1.5138,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.3596,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 1.0246,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 1.1757,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021200000000000003,
+      "loss": 1.2443,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021600000000000002,
+      "loss": 1.3469,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022000000000000003,
+      "loss": 1.299,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 1.2945,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.9243,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000232,
+      "loss": 1.115,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000236,
+      "loss": 1.3578,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024,
+      "loss": 1.3032,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000244,
+      "loss": 1.2485,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000248,
+      "loss": 1.1845,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000252,
+      "loss": 1.4783,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 1.3757,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 1.4436,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000264,
+      "loss": 0.9979,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000268,
+      "loss": 1.285,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 1.2615,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000276,
+      "loss": 1.4968,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028,
+      "loss": 1.1454,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000284,
+      "loss": 1.4971,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000288,
+      "loss": 0.9982,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000292,
+      "loss": 1.4359,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000296,
+      "loss": 1.0167,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 0.9362,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000304,
+      "loss": 1.0421,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000308,
+      "loss": 1.5118,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00031200000000000005,
+      "loss": 1.2671,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00031600000000000004,
+      "loss": 1.492,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00032,
+      "loss": 1.376,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000324,
+      "loss": 1.2943,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000328,
+      "loss": 1.3435,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000332,
+      "loss": 1.2868,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000336,
+      "loss": 1.2065,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00034,
+      "loss": 1.3566,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000344,
+      "loss": 1.3207,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000348,
+      "loss": 1.0947,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 1.2313,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00035600000000000003,
+      "loss": 1.2427,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00036,
+      "loss": 1.2776,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000364,
+      "loss": 1.1623,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 1.2151,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00037200000000000004,
+      "loss": 1.2164,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000376,
+      "loss": 1.3346,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00038,
+      "loss": 1.2365,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000384,
+      "loss": 1.0202,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000388,
+      "loss": 1.2316,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000392,
+      "loss": 1.1914,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 1.3235,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 1.4001,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000399999960788674,
+      "loss": 1.4078,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999984315471136,
+      "loss": 1.3857,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999964709815816,
+      "loss": 1.22,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999993726190913,
+      "loss": 1.1953,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999901971761844,
+      "loss": 1.136,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999985883938779,
+      "loss": 1.2634,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999807864803886,
+      "loss": 1.2271,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999749048030115,
+      "loss": 1.0238,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999968238908955,
+      "loss": 1.1472,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999960788800832,
+      "loss": 1.074,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999525544815633,
+      "loss": 1.3325,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999943535954379,
+      "loss": 1.0271,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999337332228136,
+      "loss": 1.3176,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999923146290713,
+      "loss": 1.1819,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039999117751622274,
+      "loss": 1.2116,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039998996198418156,
+      "loss": 1.0165,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039998866803342436,
+      "loss": 1.1161,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999872956644586,
+      "loss": 1.173,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039998584487782227,
+      "loss": 1.3781,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999843156740844,
+      "loss": 1.0384,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039998270805384446,
+      "loss": 1.077,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039998102201773295,
+      "loss": 0.9497,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999792575664109,
+      "loss": 1.1894,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039997741470057025,
+      "loss": 0.7906,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999754934209335,
+      "loss": 1.1553,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999734937282541,
+      "loss": 1.1726,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999714156233162,
+      "loss": 1.2578,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999692591069345,
+      "loss": 1.5058,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999670241799547,
+      "loss": 1.0702,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999647108432531,
+      "loss": 1.1127,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999623190977369,
+      "loss": 1.2201,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999598489443438,
+      "loss": 1.2033,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999573003840424,
+      "loss": 1.2268,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999546734178321,
+      "loss": 1.3701,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999519680467429,
+      "loss": 1.1456,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039994918427183567,
+      "loss": 1.1357,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039994632209420193,
+      "loss": 1.214,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999433815149639,
+      "loss": 1.0102,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999403625352748,
+      "loss": 1.2684,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999372651563183,
+      "loss": 0.9517,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039993408937930887,
+      "loss": 1.141,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999308352054919,
+      "loss": 1.3739,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999275026361433,
+      "loss": 1.4468,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039992409167256983,
+      "loss": 1.389,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039992060231610905,
+      "loss": 1.1805,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039991703456812913,
+      "loss": 1.4139,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399913388430029,
+      "loss": 1.2785,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999096639032383,
+      "loss": 1.0883,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999058609892177,
+      "loss": 1.0075,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039990197968945817,
+      "loss": 1.0009,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998980200054817,
+      "loss": 1.145,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998939819388408,
+      "loss": 1.2862,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998898654911191,
+      "loss": 1.2401,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998856706639305,
+      "loss": 1.3843,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998813974589199,
+      "loss": 1.2574,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998770458777629,
+      "loss": 1.012,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039987261592216587,
+      "loss": 1.1339,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998681075938658,
+      "loss": 1.1074,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039986352089463036,
+      "loss": 1.2205,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039985885582625823,
+      "loss": 1.2594,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039985411239057856,
+      "loss": 1.0544,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039984929058945135,
+      "loss": 1.1792,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998443904247672,
+      "loss": 1.1686,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998394118984477,
+      "loss": 1.3172,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039983435501244485,
+      "loss": 1.1999,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039982921976874155,
+      "loss": 1.0079,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039982400616935145,
+      "loss": 1.3275,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998187142163188,
+      "loss": 1.3412,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039981334391171877,
+      "loss": 1.1768,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998078952576569,
+      "loss": 1.11,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039980236825626987,
+      "loss": 1.1575,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997967629097249,
+      "loss": 1.2903,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997910792202198,
+      "loss": 1.1977,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039978531718998326,
+      "loss": 1.178,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997794768212747,
+      "loss": 1.0352,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997735581163842,
+      "loss": 1.2894,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039976756107763247,
+      "loss": 1.0364,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039976148570737113,
+      "loss": 1.2585,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003997553320079824,
+      "loss": 1.4905,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039974909998187914,
+      "loss": 1.3823,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997427896315051,
+      "loss": 1.079,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039973640095933467,
+      "loss": 1.126,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039972993396787286,
+      "loss": 1.1305,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997233886596555,
+      "loss": 1.2446,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039971676503724915,
+      "loss": 1.2293,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997100631032508,
+      "loss": 1.3964,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039970328286028864,
+      "loss": 1.0242,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039969642431102116,
+      "loss": 1.0485,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039968948745813766,
+      "loss": 1.2055,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996824723043583,
+      "loss": 1.4947,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996753788524337,
+      "loss": 1.3421,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039966820710514526,
+      "loss": 0.9612,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039966095706530524,
+      "loss": 1.3824,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996536287357564,
+      "loss": 1.2348,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039964622211937235,
+      "loss": 1.3612,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996387372190572,
+      "loss": 1.3525,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996311740377459,
+      "loss": 1.3522,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996235325784043,
+      "loss": 1.4302,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996158128440284,
+      "loss": 1.0885,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039960801483764535,
+      "loss": 1.2563,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996001385623129,
+      "loss": 1.2558,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995921840211194,
+      "loss": 1.0408,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995841512171839,
+      "loss": 1.6237,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039957604015365624,
+      "loss": 1.0799,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995678508337167,
+      "loss": 0.9232,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995595832605767,
+      "loss": 1.0494,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995512374374778,
+      "loss": 1.0462,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995428133676927,
+      "loss": 1.0942,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995343110545245,
+      "loss": 1.0402,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039952573050130704,
+      "loss": 0.9961,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995170717114049,
+      "loss": 1.4105,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003995083346882134,
+      "loss": 1.414,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994995194351583,
+      "loss": 1.2834,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994906259556962,
+      "loss": 1.3716,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994816542533145,
+      "loss": 1.2308,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994726043315309,
+      "loss": 1.252,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994634761938942,
+      "loss": 0.9373,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994542698439835,
+      "loss": 1.1687,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994449852854088,
+      "loss": 1.3147,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039943562252181075,
+      "loss": 1.0666,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994261815568605,
+      "loss": 0.9707,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003994166623942601,
+      "loss": 1.3308,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399407065037742,
+      "loss": 1.1298,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993973894910696,
+      "loss": 1.2518,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993876357580368,
+      "loss": 0.9028,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039937780384246803,
+      "loss": 1.4411,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039936789374821864,
+      "loss": 1.2699,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993579054791744,
+      "loss": 1.1833,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039934783903925196,
+      "loss": 1.0511,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003993376944323984,
+      "loss": 1.4113,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993274716625917,
+      "loss": 1.2889,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039931717073384015,
+      "loss": 1.1903,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399306791650183,
+      "loss": 0.7803,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039929633441569,
+      "loss": 1.0534,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992857990344616,
+      "loss": 1.0636,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992751855106289,
+      "loss": 1.3234,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992644938483535,
+      "loss": 1.1037,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992537240518277,
+      "loss": 1.1833,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039924287612527466,
+      "loss": 0.963,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992319500729479,
+      "loss": 1.4008,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039922094589913163,
+      "loss": 1.0209,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992098636081408,
+      "loss": 1.2867,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003991987032043209,
+      "loss": 1.3323,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039918746469204804,
+      "loss": 1.5347,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039917614807572903,
+      "loss": 1.374,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039916475335980133,
+      "loss": 1.3625,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003991532805487328,
+      "loss": 1.387,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003991417296470222,
+      "loss": 1.0278,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039913010065919873,
+      "loss": 1.4229,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003991183935898223,
+      "loss": 1.4442,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039910660844348346,
+      "loss": 0.8023,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003990947452248032,
+      "loss": 1.1311,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003990828039384333,
+      "loss": 1.2557,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039907078458905617,
+      "loss": 1.1586,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039905868718138464,
+      "loss": 1.2231,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039904651172016235,
+      "loss": 1.036,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039903425821016347,
+      "loss": 1.13,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039902192665619264,
+      "loss": 1.1742,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039900951706308536,
+      "loss": 1.298,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039899702943570756,
+      "loss": 1.0841,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003989844637789557,
+      "loss": 1.0977,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039897182009775715,
+      "loss": 1.243,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039895909839706946,
+      "loss": 1.3182,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003989462986818811,
+      "loss": 1.411,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039893342095721095,
+      "loss": 0.9376,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003989204652281085,
+      "loss": 0.9839,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039890743149965404,
+      "loss": 1.5705,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039889431977695807,
+      "loss": 0.8626,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000398881130065162,
+      "loss": 1.4965,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039886786236943754,
+      "loss": 0.9286,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039885451669498733,
+      "loss": 1.007,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988410930470443,
+      "loss": 1.0725,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988275914308719,
+      "loss": 1.1517,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988140118517645,
+      "loss": 1.0705,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003988003543150468,
+      "loss": 0.9155,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000398786618826074,
+      "loss": 1.3305,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000398772805390232,
+      "loss": 1.2028,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003987589140129372,
+      "loss": 1.0305,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039874494469963674,
+      "loss": 1.1503,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039873089745580805,
+      "loss": 1.1282,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039871677228695923,
+      "loss": 1.2126,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000398702569198629,
+      "loss": 1.3774,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039868828819638664,
+      "loss": 1.5071,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986739292858317,
+      "loss": 1.1821,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039865949247259476,
+      "loss": 1.0666,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986449777623365,
+      "loss": 1.0486,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986303851607485,
+      "loss": 1.2113,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986157146735525,
+      "loss": 1.1278,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986009663065011,
+      "loss": 1.0706,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985861400653774,
+      "loss": 1.482,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985712359559949,
+      "loss": 1.0938,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039855625398419766,
+      "loss": 1.1436,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985411941558603,
+      "loss": 1.2525,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985260564768881,
+      "loss": 1.2264,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985108409532167,
+      "loss": 1.2647,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003984955475908122,
+      "loss": 1.0336,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039848017639567145,
+      "loss": 0.9665,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039846472737382167,
+      "loss": 1.5644,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003984492005313206,
+      "loss": 0.8903,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003984335958742565,
+      "loss": 1.4631,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003984179134087483,
+      "loss": 1.2745,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003984021531409451,
+      "loss": 1.2507,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039838631507702684,
+      "loss": 1.2827,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983703992232038,
+      "loss": 1.3398,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039835440558571683,
+      "loss": 1.1426,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039833833417083725,
+      "loss": 1.199,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983221849848668,
+      "loss": 1.3096,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039830595803413786,
+      "loss": 0.9878,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039828965332501314,
+      "loss": 1.2342,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003982732708638861,
+      "loss": 1.1846,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039825681065718037,
+      "loss": 1.1255,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039824027271135026,
+      "loss": 1.178,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039822365703288053,
+      "loss": 1.1266,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003982069636282864,
+      "loss": 0.9682,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003981901925041136,
+      "loss": 1.0544,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039817334366693826,
+      "loss": 0.958,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003981564171233671,
+      "loss": 1.1739,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003981394128800372,
+      "loss": 1.1927,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039812233094361615,
+      "loss": 1.1473,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000398105171320802,
+      "loss": 1.3942,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980879340183233,
+      "loss": 1.3453,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980706190429389,
+      "loss": 1.1546,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980532264014384,
+      "loss": 1.3081,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980357561006416,
+      "loss": 1.1298,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003980182081473989,
+      "loss": 1.1021,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039800058254859093,
+      "loss": 1.2194,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003979828793111291,
+      "loss": 1.2085,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000397965098441955,
+      "loss": 1.1911,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039794723994804077,
+      "loss": 1.3557,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003979293038363889,
+      "loss": 1.1175,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003979112901140324,
+      "loss": 1.602,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003978931987880348,
+      "loss": 1.0221,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978750298654898,
+      "loss": 1.0024,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978567833535218,
+      "loss": 0.8084,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039783845925928535,
+      "loss": 1.1718,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039782005758996574,
+      "loss": 1.2516,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978015783527784,
+      "loss": 1.2351,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003977830215549693,
+      "loss": 0.9891,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003977643872038148,
+      "loss": 1.1998,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003977456753066218,
+      "loss": 1.0335,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003977268858707272,
+      "loss": 1.1657,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039770801890349886,
+      "loss": 1.3366,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003976890744123347,
+      "loss": 0.9495,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039767005240466305,
+      "loss": 1.2014,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003976509528879427,
+      "loss": 1.1499,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003976317758696628,
+      "loss": 1.3681,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003976125213573431,
+      "loss": 1.1694,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039759318935853326,
+      "loss": 1.2523,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039757377988081387,
+      "loss": 1.3283,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003975542929317954,
+      "loss": 1.3693,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039753472851911917,
+      "loss": 1.3649,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039751508665045654,
+      "loss": 1.3445,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039749536733350936,
+      "loss": 1.1628,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003974755705760098,
+      "loss": 0.9889,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003974556963857205,
+      "loss": 0.9554,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039743574477043425,
+      "loss": 1.1845,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039741571573797456,
+      "loss": 1.0012,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039739560929619497,
+      "loss": 1.3621,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039737542545297943,
+      "loss": 1.2549,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039735516421624234,
+      "loss": 1.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039733482559392846,
+      "loss": 1.1141,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003973144095940127,
+      "loss": 1.3007,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039729391622450057,
+      "loss": 1.2732,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003972733454934277,
+      "loss": 0.9662,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003972526974088603,
+      "loss": 1.2102,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003972319719788945,
+      "loss": 1.016,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003972111692116573,
+      "loss": 1.0781,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003971902891153055,
+      "loss": 1.1852,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003971693316980266,
+      "loss": 1.2531,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003971482969680382,
+      "loss": 1.0714,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003971271849335884,
+      "loss": 1.3018,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039710599560295547,
+      "loss": 1.4692,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039708472898444797,
+      "loss": 1.0331,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003970633850864049,
+      "loss": 1.2851,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039704196391719546,
+      "loss": 1.2833,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003970204654852191,
+      "loss": 1.2106,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039699888979890575,
+      "loss": 1.3475,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039697723686671544,
+      "loss": 1.383,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039695550669713856,
+      "loss": 1.1795,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039693369929869597,
+      "loss": 1.278,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039691181467993843,
+      "loss": 1.0538,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003968898528494473,
+      "loss": 1.4773,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003968678138158341,
+      "loss": 1.4556,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039684569758774056,
+      "loss": 0.9972,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003968235041738388,
+      "loss": 1.2519,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003968012335828312,
+      "loss": 0.9984,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039677888582345027,
+      "loss": 1.1659,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039675646090445887,
+      "loss": 1.3487,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039673395883465016,
+      "loss": 1.195,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003967113796228475,
+      "loss": 1.139,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003966887232779044,
+      "loss": 1.5205,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039666598980870476,
+      "loss": 1.2067,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039664317922416276,
+      "loss": 1.2486,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003966202915332227,
+      "loss": 1.2718,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039659732674485903,
+      "loss": 1.336,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003965742848680766,
+      "loss": 1.2165,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039655116591191065,
+      "loss": 1.2909,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039652796988542613,
+      "loss": 1.2913,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003965046967977186,
+      "loss": 1.2588,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039648134665791387,
+      "loss": 1.1336,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003964579194751676,
+      "loss": 1.3975,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003964344152586662,
+      "loss": 1.1979,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003964108340176258,
+      "loss": 1.1705,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039638717576129283,
+      "loss": 1.0904,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003963634404989442,
+      "loss": 1.1098,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039633962823988676,
+      "loss": 1.1726,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003963157389934575,
+      "loss": 1.1827,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039629177276902396,
+      "loss": 1.2665,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003962677295759834,
+      "loss": 1.121,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003962436094237635,
+      "loss": 1.2562,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003962194123218221,
+      "loss": 1.2932,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003961951382796473,
+      "loss": 1.4318,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039617078730675717,
+      "loss": 1.0694,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039614635941270007,
+      "loss": 1.4911,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039612185460705454,
+      "loss": 1.0522,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003960972728994292,
+      "loss": 1.1211,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003960726142994629,
+      "loss": 1.2501,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039604787881682455,
+      "loss": 1.1557,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003960230664612134,
+      "loss": 1.0853,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039599817724235843,
+      "loss": 1.3463,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039597321117001934,
+      "loss": 1.0142,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959481682539854,
+      "loss": 1.0591,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003959230485040764,
+      "loss": 1.1959,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003958978519301422,
+      "loss": 1.487,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003958725785420626,
+      "loss": 1.2754,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003958472283497476,
+      "loss": 1.1854,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039582180136313745,
+      "loss": 0.9861,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039579629759220236,
+      "loss": 1.1677,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039577071704694264,
+      "loss": 1.404,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003957450597373889,
+      "loss": 1.2453,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003957193256736016,
+      "loss": 1.2848,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003956935148656714,
+      "loss": 1.0572,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003956676273237191,
+      "loss": 1.1688,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039564166305789555,
+      "loss": 1.5483,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039561562207838174,
+      "loss": 1.2882,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039558950439538853,
+      "loss": 1.1499,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039556331001915714,
+      "loss": 1.1007,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039553703895995874,
+      "loss": 1.418,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039551069122809447,
+      "loss": 1.0723,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003954842668338957,
+      "loss": 0.9946,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003954577657877237,
+      "loss": 1.2226,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039543118809996997,
+      "loss": 1.099,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003954045337810559,
+      "loss": 1.0353,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003953778028414331,
+      "loss": 1.146,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000395350995291583,
+      "loss": 1.1502,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039532411114201737,
+      "loss": 1.4111,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039529715040327763,
+      "loss": 1.1497,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003952701130859356,
+      "loss": 1.257,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039524299920059295,
+      "loss": 0.838,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039521580875788133,
+      "loss": 1.1173,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003951885417684625,
+      "loss": 1.451,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003951611982430282,
+      "loss": 1.015,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003951337781923002,
+      "loss": 1.2794,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003951062816270303,
+      "loss": 1.1732,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003950787085580002,
+      "loss": 1.1328,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003950510589960217,
+      "loss": 1.2637,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003950233329519366,
+      "loss": 1.2957,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039499553043661665,
+      "loss": 1.0357,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039496765146096344,
+      "loss": 1.5246,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039493969603590884,
+      "loss": 1.0832,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003949116641724145,
+      "loss": 1.3073,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003948835558814721,
+      "loss": 1.1683,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003948553711741033,
+      "loss": 1.4835,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039482711006135954,
+      "loss": 1.0243,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039479877255432257,
+      "loss": 1.2557,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003947703586641038,
+      "loss": 1.2528,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003947418684018447,
+      "loss": 1.5079,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039471330177871664,
+      "loss": 1.0428,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039468465880592113,
+      "loss": 1.3686,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003946559394946893,
+      "loss": 1.0172,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039462714385628245,
+      "loss": 1.5552,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003945982719019917,
+      "loss": 1.0887,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039456932364313816,
+      "loss": 1.2052,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039454029909107274,
+      "loss": 1.146,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003945111982571764,
+      "loss": 1.2577,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003944820211528601,
+      "loss": 1.1647,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039445276778956435,
+      "loss": 1.276,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003944234381787599,
+      "loss": 1.2987,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003943940323319473,
+      "loss": 1.1758,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003943645502606569,
+      "loss": 1.3176,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000394334991976449,
+      "loss": 0.963,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039430535749091384,
+      "loss": 1.1901,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039427564681567153,
+      "loss": 1.0747,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000394245859962372,
+      "loss": 0.9175,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039421599694269496,
+      "loss": 1.1743,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039418605776835025,
+      "loss": 1.1763,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003941560424510773,
+      "loss": 1.3974,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039412595100264566,
+      "loss": 1.3525,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003940957834348545,
+      "loss": 1.0265,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003940655397595328,
+      "loss": 1.2594,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003940352199885397,
+      "loss": 0.9566,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003940048241337639,
+      "loss": 1.0901,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039397435220712404,
+      "loss": 0.9579,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003939438042205685,
+      "loss": 1.2461,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039391318018607566,
+      "loss": 1.391,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003938824801156535,
+      "loss": 0.9878,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039385170402134,
+      "loss": 1.2161,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039382085191520284,
+      "loss": 1.2693,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003937899238093396,
+      "loss": 1.1717,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003937589197158775,
+      "loss": 1.184,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003937278396469737,
+      "loss": 1.7093,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039369668361481517,
+      "loss": 0.9683,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003936654516316185,
+      "loss": 1.3175,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039363414370963026,
+      "loss": 1.1481,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003936027598611266,
+      "loss": 1.2939,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003935713000984136,
+      "loss": 1.323,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003935397644338271,
+      "loss": 1.2684,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039350815287973257,
+      "loss": 1.0274,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039347646544852534,
+      "loss": 1.3879,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003934447021526305,
+      "loss": 1.2625,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039341286300450287,
+      "loss": 1.1891,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039338094801662693,
+      "loss": 0.9302,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000393348957201517,
+      "loss": 1.1729,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003933168905717171,
+      "loss": 1.2388,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003932847481398011,
+      "loss": 1.4643,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003932525299183723,
+      "loss": 1.1767,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003932202359200639,
+      "loss": 1.0861,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000393187866157539,
+      "loss": 1.0404,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039315542064349,
+      "loss": 1.4746,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003931228993906393,
+      "loss": 1.2396,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039309030241173895,
+      "loss": 1.0537,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003930576297195706,
+      "loss": 0.9747,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003930248813269456,
+      "loss": 1.0674,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003929920572467052,
+      "loss": 1.0652,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039295915749172,
+      "loss": 1.5682,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039292618207489047,
+      "loss": 1.1858,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039289313100914674,
+      "loss": 1.1616,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003928600043074485,
+      "loss": 1.219,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003928268019827853,
+      "loss": 1.2048,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039279352404817606,
+      "loss": 1.0229,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003927601705166696,
+      "loss": 1.0653,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039272674140134427,
+      "loss": 1.162,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039269323671530806,
+      "loss": 1.342,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003926596564716985,
+      "loss": 1.201,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039262600068368307,
+      "loss": 1.3626,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039259226936445846,
+      "loss": 1.2054,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039255846252725125,
+      "loss": 1.3337,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003925245801853175,
+      "loss": 1.0409,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000392490622351943,
+      "loss": 1.1616,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039245658904044295,
+      "loss": 1.2069,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039242248026416244,
+      "loss": 1.0586,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039238829603647574,
+      "loss": 1.161,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039235403637078716,
+      "loss": 1.1759,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039231970128053025,
+      "loss": 1.0165,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039228529077916824,
+      "loss": 1.1026,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039225080488019404,
+      "loss": 1.2867,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039221624359712996,
+      "loss": 1.1867,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039218160694352795,
+      "loss": 1.3941,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003921468949329695,
+      "loss": 1.257,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003921121075790657,
+      "loss": 1.31,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000392077244895457,
+      "loss": 1.1858,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039204230689581366,
+      "loss": 1.0339,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039200729359383527,
+      "loss": 1.1613,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000391972205003251,
+      "loss": 1.3972,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003919370411378196,
+      "loss": 1.0933,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003919018020113292,
+      "loss": 1.138,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003918664876375976,
+      "loss": 1.0548,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000391831098030472,
+      "loss": 0.7899,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003917956332038292,
+      "loss": 0.9507,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003917600931715754,
+      "loss": 1.3815,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003917244779476463,
+      "loss": 0.8621,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039168878754600704,
+      "loss": 1.5691,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039165302198065243,
+      "loss": 1.1682,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039161718126560645,
+      "loss": 1.2582,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039158126541492295,
+      "loss": 1.2293,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039154527444268487,
+      "loss": 1.3579,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039150920836300465,
+      "loss": 1.318,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003914730671900245,
+      "loss": 1.1305,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003914368509379157,
+      "loss": 1.001,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003914005596208792,
+      "loss": 0.9722,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039136419325314526,
+      "loss": 1.1744,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039132775184897356,
+      "loss": 1.4034,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003912912354226534,
+      "loss": 1.0346,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039125464398850326,
+      "loss": 1.2753,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003912179775608712,
+      "loss": 1.3861,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003911812361541345,
+      "loss": 1.3507,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039114441978270004,
+      "loss": 1.1945,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039110752846100393,
+      "loss": 1.0553,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039107056220351186,
+      "loss": 1.0685,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003910335210247187,
+      "loss": 1.0416,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003909964049391488,
+      "loss": 1.2612,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003909592139613559,
+      "loss": 1.5501,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000390921948105923,
+      "loss": 0.9467,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039088460738746273,
+      "loss": 1.0128,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003908471918206166,
+      "loss": 1.0382,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000390809701420056,
+      "loss": 1.2181,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039077213620048124,
+      "loss": 1.1086,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039073449617662225,
+      "loss": 0.9224,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039069678136323816,
+      "loss": 0.9627,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039065899177511743,
+      "loss": 1.0851,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039062112742707775,
+      "loss": 1.1702,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003905831883339665,
+      "loss": 1.3536,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039054517451065984,
+      "loss": 1.2522,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003905070859720636,
+      "loss": 0.9294,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003904689227331129,
+      "loss": 1.2989,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039043068480877185,
+      "loss": 1.2796,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039039237221403416,
+      "loss": 1.19,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039035398496392273,
+      "loss": 0.9731,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003903155230734896,
+      "loss": 1.3007,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039027698655781634,
+      "loss": 0.9268,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003902383754320135,
+      "loss": 0.8783,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003901996897112211,
+      "loss": 1.0703,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003901609294106083,
+      "loss": 1.104,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039012209454537345,
+      "loss": 1.1009,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039008318513074434,
+      "loss": 0.9415,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003900442011819778,
+      "loss": 0.9715,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039000514271435993,
+      "loss": 0.9704,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003899660097432062,
+      "loss": 1.1094,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038992680228386095,
+      "loss": 1.1316,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038988752035169815,
+      "loss": 1.0946,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003898481639621206,
+      "loss": 1.2194,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038980873313056066,
+      "loss": 0.9559,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003897692278724795,
+      "loss": 1.2283,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003897296482033677,
+      "loss": 1.0747,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003896899941387451,
+      "loss": 1.1144,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038965026569416037,
+      "loss": 1.2952,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038961046288519174,
+      "loss": 0.9456,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003895705857274463,
+      "loss": 1.1458,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038953063423656053,
+      "loss": 1.1702,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038949060842819984,
+      "loss": 1.3011,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003894505083180589,
+      "loss": 1.0087,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038941033392186143,
+      "loss": 1.1936,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038937008525536055,
+      "loss": 1.1719,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003893297623343381,
+      "loss": 1.0481,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038928936517460536,
+      "loss": 1.3264,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038924889379200246,
+      "loss": 1.0813,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003892083482023988,
+      "loss": 1.0196,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038916772842169297,
+      "loss": 1.4213,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038912703446581244,
+      "loss": 1.4053,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003890862663507138,
+      "loss": 1.116,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003890454240923828,
+      "loss": 1.4398,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003890045077068343,
+      "loss": 1.2635,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000388963517210112,
+      "loss": 0.8584,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003889224526182891,
+      "loss": 0.8592,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038888131394746725,
+      "loss": 1.3589,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038884010121377764,
+      "loss": 1.1295,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038879881443338026,
+      "loss": 0.9472,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038875745362246424,
+      "loss": 0.9955,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038871601879724777,
+      "loss": 1.0132,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038867450997397785,
+      "loss": 1.1086,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003886329271689308,
+      "loss": 1.1512,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003885912703984117,
+      "loss": 1.2654,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038854953967875467,
+      "loss": 0.979,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038850773502632297,
+      "loss": 1.1965,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003884658564575087,
+      "loss": 1.3423,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038842390398873304,
+      "loss": 1.398,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038838187763644614,
+      "loss": 0.999,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000388339777417127,
+      "loss": 1.1471,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003882976033472837,
+      "loss": 1.1227,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003882553554434533,
+      "loss": 0.9914,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003882130337222018,
+      "loss": 1.1912,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038817063820012395,
+      "loss": 1.3185,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038812816889384374,
+      "loss": 1.0942,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038808562582001396,
+      "loss": 1.1122,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038804300899531613,
+      "loss": 1.0294,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000388000318436461,
+      "loss": 1.1165,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003879575541601882,
+      "loss": 1.0432,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000387914716183266,
+      "loss": 1.2856,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038787180452249184,
+      "loss": 0.8364,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038782881919469195,
+      "loss": 1.2317,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038778576021672136,
+      "loss": 1.1464,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038774262760546413,
+      "loss": 1.2551,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003876994213778331,
+      "loss": 1.0051,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003876561415507701,
+      "loss": 1.2758,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038761278814124553,
+      "loss": 0.9655,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003875693611662591,
+      "loss": 1.2343,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038752586064283884,
+      "loss": 1.1657,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038748228658804206,
+      "loss": 1.0113,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003874386390189547,
+      "loss": 1.0402,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003873949179526915,
+      "loss": 0.9541,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000387351123406396,
+      "loss": 1.1698,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038730725539724083,
+      "loss": 1.2394,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000387263313942427,
+      "loss": 1.2014,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003872192990591847,
+      "loss": 0.9173,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003871752107647726,
+      "loss": 0.9348,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003871310490764785,
+      "loss": 0.9029,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003870868140116186,
+      "loss": 1.1448,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003870425055875382,
+      "loss": 1.0387,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000386998123821611,
+      "loss": 1.2264,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038695366873123997,
+      "loss": 1.1751,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038690914033385636,
+      "loss": 1.2026,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038686453864692036,
+      "loss": 1.365,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038681986368792093,
+      "loss": 1.1339,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003867751154743757,
+      "loss": 1.0776,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000386730294023831,
+      "loss": 1.1448,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038668539935386194,
+      "loss": 1.3364,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003866404314820723,
+      "loss": 1.2377,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038659539042609466,
+      "loss": 0.9484,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038655027620359005,
+      "loss": 0.9454,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038650508883224856,
+      "loss": 1.2155,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003864598283297886,
+      "loss": 1.3749,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003864144947139575,
+      "loss": 1.1404,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003863690880025311,
+      "loss": 1.1285,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038632360821331407,
+      "loss": 1.1885,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003862780553641395,
+      "loss": 0.7371,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003862324294728694,
+      "loss": 1.2077,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038618673055739426,
+      "loss": 1.1394,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038614095863563315,
+      "loss": 1.073,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038609511372553396,
+      "loss": 1.1773,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000386049195845073,
+      "loss": 1.0149,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038600320501225527,
+      "loss": 1.0413,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003859571412451145,
+      "loss": 1.1869,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038591100456171273,
+      "loss": 1.2536,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038586479498014094,
+      "loss": 1.2079,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038581851251851847,
+      "loss": 1.239,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003857721571949932,
+      "loss": 1.1729,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003857257290277418,
+      "loss": 1.138,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003856792280349692,
+      "loss": 1.1402,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003856326542349092,
+      "loss": 0.9048,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038558600764582396,
+      "loss": 1.4733,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038553928828600424,
+      "loss": 0.8463,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038549249617376925,
+      "loss": 1.2829,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003854456313274669,
+      "loss": 1.2487,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038539869376547347,
+      "loss": 1.2804,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003853516835061938,
+      "loss": 1.0734,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038530460056806126,
+      "loss": 1.2288,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003852574449695376,
+      "loss": 1.093,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038521021672911326,
+      "loss": 0.928,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000385162915865307,
+      "loss": 1.0319,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003851155423966662,
+      "loss": 0.7286,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038506809634176656,
+      "loss": 1.0882,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038502057771921234,
+      "loss": 1.0756,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038497298654763615,
+      "loss": 0.8429,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003849253228456991,
+      "loss": 1.4406,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038487758663209093,
+      "loss": 1.1619,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038482977792552955,
+      "loss": 1.1641,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003847818967447613,
+      "loss": 1.4605,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038473394310856114,
+      "loss": 1.1747,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038468591703573224,
+      "loss": 1.1988,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038463781854510644,
+      "loss": 1.2558,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003845896476555435,
+      "loss": 1.4377,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003845414043859321,
+      "loss": 1.0059,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038449308875518904,
+      "loss": 1.162,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003844447007822594,
+      "loss": 1.2795,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003843962404861168,
+      "loss": 1.412,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038434770788576324,
+      "loss": 1.1093,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038429910300022895,
+      "loss": 1.2322,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038425042584857254,
+      "loss": 0.9994,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038420167644988095,
+      "loss": 1.192,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003841528548232695,
+      "loss": 1.1044,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003841039609878817,
+      "loss": 1.2112,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003840549949628897,
+      "loss": 1.2885,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003840059567674935,
+      "loss": 1.0466,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003839568464209217,
+      "loss": 1.3142,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003839076639424311,
+      "loss": 0.9909,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038385840935130686,
+      "loss": 1.1802,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038380908266686227,
+      "loss": 1.3882,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003837596839084391,
+      "loss": 1.0363,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038371021309540715,
+      "loss": 1.2876,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038366067024716465,
+      "loss": 1.2632,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000383611055383138,
+      "loss": 1.189,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038356136852278177,
+      "loss": 1.0299,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003835116096855789,
+      "loss": 1.3577,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003834617788910405,
+      "loss": 1.2476,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038341187615870587,
+      "loss": 1.1567,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038336190150814253,
+      "loss": 0.8092,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003833118549589462,
+      "loss": 1.1087,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003832617365307408,
+      "loss": 1.1113,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038321154624317835,
+      "loss": 1.0268,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003831612841159393,
+      "loss": 0.8065,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000383110950168732,
+      "loss": 1.3197,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038306054442129305,
+      "loss": 1.5737,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003830100668933872,
+      "loss": 1.0348,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003829595176048074,
+      "loss": 1.1218,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003829088965753747,
+      "loss": 1.2894,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038285820382493824,
+      "loss": 1.1233,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038280743937337534,
+      "loss": 1.0981,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038275660324059137,
+      "loss": 0.9755,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003827056954465199,
+      "loss": 1.2779,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038265471601112253,
+      "loss": 1.2683,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000382603664954389,
+      "loss": 1.1203,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003825525422963371,
+      "loss": 1.0245,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003825013480570127,
+      "loss": 1.2775,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003824500822564897,
+      "loss": 1.3334,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003823987449148702,
+      "loss": 1.2607,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003823473360522841,
+      "loss": 1.2484,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038229585568888964,
+      "loss": 0.9712,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003822443038448729,
+      "loss": 1.1914,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038219268054044794,
+      "loss": 1.1138,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038214098579585716,
+      "loss": 1.2609,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038208921963137055,
+      "loss": 1.3366,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038203738206728645,
+      "loss": 0.9697,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381985473123931,
+      "loss": 1.4803,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003819334928216583,
+      "loss": 1.1559,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038188144118085073,
+      "loss": 1.3246,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003818293182219182,
+      "loss": 1.2751,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381777123965299,
+      "loss": 0.8997,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381724858431459,
+      "loss": 0.894,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003816725216408924,
+      "loss": 1.4176,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381620113614121,
+      "loss": 1.0055,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038156763437169474,
+      "loss": 1.0604,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038151508393419143,
+      "loss": 0.9006,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038151508393419143,
+      "loss": 1.2544,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003814624623222168,
+      "loss": 1.1533,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038140976955640453,
+      "loss": 1.2244,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000381357005657416,
+      "loss": 1.212,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003813041706459408,
+      "loss": 0.9663,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038125126454269617,
+      "loss": 0.9962,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038119828736842725,
+      "loss": 1.2438,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038114523914390715,
+      "loss": 1.0999,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003810921198899368,
+      "loss": 1.2779,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038103892962734493,
+      "loss": 1.1167,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038098566837698817,
+      "loss": 1.4277,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038093233615975086,
+      "loss": 1.1559,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038087893299654544,
+      "loss": 1.1273,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038082545890831185,
+      "loss": 1.1827,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003807719139160181,
+      "loss": 1.0012,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003807182980406598,
+      "loss": 1.1193,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003806646113032605,
+      "loss": 1.1586,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038061085372487153,
+      "loss": 1.1183,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038055702532657185,
+      "loss": 0.9418,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003805031261294683,
+      "loss": 1.1668,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038044915615469545,
+      "loss": 1.1753,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003803951154234158,
+      "loss": 1.2128,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003803410039568192,
+      "loss": 1.4828,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003802868217761237,
+      "loss": 1.2271,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003802325689025747,
+      "loss": 1.1336,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003801782453574456,
+      "loss": 0.9645,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038012385116203723,
+      "loss": 1.0583,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003800693863376784,
+      "loss": 1.0348,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038001485090572546,
+      "loss": 1.0957,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037996024488756236,
+      "loss": 1.3504,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000379905568304601,
+      "loss": 1.2574,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037985082117828075,
+      "loss": 1.0698,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037979600353006866,
+      "loss": 1.291,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037974111538145944,
+      "loss": 1.2868,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003796861567539755,
+      "loss": 0.9827,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003796311276691668,
+      "loss": 0.9238,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037957602814861106,
+      "loss": 1.157,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003795208582139134,
+      "loss": 1.0137,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037946561788670686,
+      "loss": 1.2525,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037941030718865174,
+      "loss": 1.1235,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003793549261414362,
+      "loss": 1.0313,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037929947476677585,
+      "loss": 1.4533,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037924395308641386,
+      "loss": 1.1855,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003791883611221211,
+      "loss": 1.339,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003791326988956959,
+      "loss": 1.149,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003790769664289641,
+      "loss": 1.2451,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037902116374377917,
+      "loss": 1.4959,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003789652908620221,
+      "loss": 1.1932,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003789093478056014,
+      "loss": 1.0408,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00037885333459645304,
+      "loss": 0.9749,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003787972512565406,
+      "loss": 1.2659,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000378741097807855,
+      "loss": 1.1978,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037868487427241483,
+      "loss": 1.0881,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003786285806722661,
+      "loss": 1.1109,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037857221702948226,
+      "loss": 1.2945,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003785157833661642,
+      "loss": 1.2665,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037845927970444037,
+      "loss": 1.2989,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037840270606646654,
+      "loss": 1.0401,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000378346062474426,
+      "loss": 1.1805,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003782893489505295,
+      "loss": 1.0664,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037823256551701515,
+      "loss": 1.2912,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037817571219614843,
+      "loss": 1.1273,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003781187890102223,
+      "loss": 0.9925,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037806179598155726,
+      "loss": 1.3739,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003780047331325007,
+      "loss": 1.2597,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037794760048542804,
+      "loss": 0.9924,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037789039806274154,
+      "loss": 1.1225,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003778331258868712,
+      "loss": 1.4085,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037777578398027404,
+      "loss": 1.2164,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037771837236543463,
+      "loss": 1.2097,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037766089106486484,
+      "loss": 1.1332,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000377603340101104,
+      "loss": 1.2233,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003775457194967183,
+      "loss": 1.2537,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003774880292743017,
+      "loss": 1.4306,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003774302694564754,
+      "loss": 1.421,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037737244006588765,
+      "loss": 1.153,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037731454112521415,
+      "loss": 1.0943,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003772565726571579,
+      "loss": 1.2551,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037719853468444904,
+      "loss": 0.8635,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000377140427229845,
+      "loss": 1.2615,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003770822503161306,
+      "loss": 1.1095,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003770240039661177,
+      "loss": 1.3115,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003769656882026455,
+      "loss": 1.2981,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037690730304858037,
+      "loss": 1.2343,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037684884852681586,
+      "loss": 1.3345,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037679032466027284,
+      "loss": 0.979,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037673173147189916,
+      "loss": 1.1173,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037667306898467014,
+      "loss": 1.3023,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003766143372215881,
+      "loss": 1.0767,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003765555362056825,
+      "loss": 1.1044,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037649666596000994,
+      "loss": 1.3592,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037643772650765436,
+      "loss": 1.1186,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003763787178717266,
+      "loss": 1.1019,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003763196400753647,
+      "loss": 1.2361,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037626049314173394,
+      "loss": 1.1686,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003762012770940266,
+      "loss": 1.2372,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003761419919554621,
+      "loss": 0.9912,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003760826377492868,
+      "loss": 0.8527,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003760232144987744,
+      "loss": 1.2391,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037596372222722544,
+      "loss": 1.2002,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037590416095796775,
+      "loss": 1.0221,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037584453071435605,
+      "loss": 1.1407,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003757848315197721,
+      "loss": 1.1188,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037572506339762475,
+      "loss": 1.1895,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037566522637135,
+      "loss": 1.4682,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037560532046441053,
+      "loss": 1.2183,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037554534570029643,
+      "loss": 1.3253,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003754853021025245,
+      "loss": 1.116,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003754251896946386,
+      "loss": 1.1413,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037536500850020977,
+      "loss": 0.9866,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003753047585428356,
+      "loss": 1.3307,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037524443984614117,
+      "loss": 0.9217,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037518405243377805,
+      "loss": 1.2211,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000375123596329425,
+      "loss": 1.4421,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003750630715567877,
+      "loss": 0.8873,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037500247813959865,
+      "loss": 1.3018,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003749418161016174,
+      "loss": 1.2742,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003748810854666303,
+      "loss": 1.5223,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003748202862584506,
+      "loss": 1.0743,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037475941850091855,
+      "loss": 1.2772,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003746984822179011,
+      "loss": 1.2477,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037463747743329227,
+      "loss": 1.3745,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037457640417101285,
+      "loss": 0.8157,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003745152624550105,
+      "loss": 1.3773,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003744540523092596,
+      "loss": 1.0182,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037439277375776145,
+      "loss": 1.6209,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003743314268245443,
+      "loss": 1.1952,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003742700115336629,
+      "loss": 0.946,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003742085279091993,
+      "loss": 0.9825,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003741469759752619,
+      "loss": 0.9462,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037408535575598583,
+      "loss": 0.9567,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003740236672755335,
+      "loss": 0.9778,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003739619105580936,
+      "loss": 1.0339,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003739000856278819,
+      "loss": 0.971,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037383819250914055,
+      "loss": 1.2212,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037377623122613886,
+      "loss": 0.9543,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003737142018031726,
+      "loss": 1.0813,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037365210426456436,
+      "loss": 1.0844,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037358993863466334,
+      "loss": 0.9866,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037352770493784564,
+      "loss": 1.4293,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003734654031985137,
+      "loss": 1.1738,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003734030334410971,
+      "loss": 1.0252,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037334059569005165,
+      "loss": 1.0502,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003732780899698602,
+      "loss": 1.144,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037321551630503197,
+      "loss": 1.3813,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037315287472010285,
+      "loss": 1.1943,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003730901652396356,
+      "loss": 0.9506,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003730273878882193,
+      "loss": 0.9542,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003729645426904699,
+      "loss": 1.2661,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037290162967102975,
+      "loss": 1.1637,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003728386488545679,
+      "loss": 1.0211,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037277560026578,
+      "loss": 1.3428,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037271248392938813,
+      "loss": 1.0019,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003726492998701412,
+      "loss": 1.0647,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037258604811281435,
+      "loss": 1.1667,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037252272868220965,
+      "loss": 1.0877,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003724593416031553,
+      "loss": 1.2383,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003723958869005062,
+      "loss": 1.1612,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003723323645991439,
+      "loss": 0.7366,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003722687747239763,
+      "loss": 1.1841,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003722051172999378,
+      "loss": 1.2867,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037214139235198937,
+      "loss": 1.3673,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003720775999051184,
+      "loss": 1.0275,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037201373998433865,
+      "loss": 0.9159,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003719498126146906,
+      "loss": 1.1435,
+      "step": 957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003718858178212409,
+      "loss": 1.2578,
+      "step": 958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003718217556290828,
+      "loss": 1.3414,
+      "step": 959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037175762606333595,
+      "loss": 1.4157,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037169342914914643,
+      "loss": 1.2321,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037162916491168663,
+      "loss": 0.968,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003715648333761554,
+      "loss": 1.1025,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037150043456777806,
+      "loss": 0.9843,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003714359685118062,
+      "loss": 1.3214,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037137143523351785,
+      "loss": 1.1491,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003713068347582173,
+      "loss": 1.0515,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037124216711123537,
+      "loss": 0.969,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003711774323179289,
+      "loss": 1.1379,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037111263040368145,
+      "loss": 1.3747,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003710477613939026,
+      "loss": 1.0135,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037098282531402844,
+      "loss": 1.1064,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003709178221895212,
+      "loss": 0.9658,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037085275204586945,
+      "loss": 1.2839,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037078761490858816,
+      "loss": 1.1293,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037072241080321835,
+      "loss": 1.2691,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003706571397553274,
+      "loss": 1.1507,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037059180179050907,
+      "loss": 0.8424,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037052639693438324,
+      "loss": 1.2257,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003704609252125959,
+      "loss": 1.2359,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037039538665081947,
+      "loss": 1.215,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037032978127475244,
+      "loss": 1.1194,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037026410911011957,
+      "loss": 1.2506,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037019837018267186,
+      "loss": 1.2468,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003701325645181863,
+      "loss": 1.08,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037006669214246623,
+      "loss": 1.1519,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037000075308134104,
+      "loss": 1.3219,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036993474736066634,
+      "loss": 1.3724,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003698686750063238,
+      "loss": 1.0225,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036980253604422135,
+      "loss": 1.4939,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003697363305002929,
+      "loss": 0.9938,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003696700584004986,
+      "loss": 0.9052,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00036960371977082446,
+      "loss": 1.1438,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003695373146372829,
+      "loss": 0.9388,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003694708430259121,
+      "loss": 1.0525,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003694043049627766,
+      "loss": 1.2485,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003693377004739667,
+      "loss": 1.093,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003692710295855991,
+      "loss": 0.9621,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036920429232381623,
+      "loss": 0.993,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003691374887147867,
+      "loss": 1.0119,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000369070618784705,
+      "loss": 1.2026,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003690036825597918,
+      "loss": 1.3449,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003689366800662936,
+      "loss": 1.0224,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036886961133048306,
+      "loss": 1.2056,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003688024763786587,
+      "loss": 1.4487,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000368735275237145,
+      "loss": 1.1407,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003686680079322924,
+      "loss": 1.1494,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003686006744904774,
+      "loss": 1.3017,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036853327493810217,
+      "loss": 1.0581,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036846580930159514,
+      "loss": 0.923,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036839827760741035,
+      "loss": 1.186,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003683306798820279,
+      "loss": 1.0743,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003682630161519538,
+      "loss": 1.1003,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003681952864437199,
+      "loss": 1.2893,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003681274907838839,
+      "loss": 1.0891,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003680596291990293,
+      "loss": 1.3426,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036799170171576565,
+      "loss": 1.1255,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003679237083607282,
+      "loss": 1.0142,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000367855649160578,
+      "loss": 1.1091,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003677875241420019,
+      "loss": 1.287,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036771933333171273,
+      "loss": 1.2994,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000367651076756449,
+      "loss": 1.0266,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003675827544429749,
+      "loss": 1.0001,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036751436641808076,
+      "loss": 1.4208,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003674459127085822,
+      "loss": 1.089,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000367377393341321,
+      "loss": 1.4864,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003673088083431643,
+      "loss": 1.3168,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036724015774100533,
+      "loss": 1.1053,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003671714415617629,
+      "loss": 1.1824,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036710265983238156,
+      "loss": 1.4769,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036703381257983145,
+      "loss": 1.0945,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003669648998311085,
+      "loss": 1.1224,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003668959216132344,
+      "loss": 1.1406,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003668268779532564,
+      "loss": 0.9152,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036675776887824736,
+      "loss": 1.0805,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036668859441530593,
+      "loss": 1.143,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003666193545915563,
+      "loss": 1.4526,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036655004943414834,
+      "loss": 1.0523,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003664806789702575,
+      "loss": 1.395,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003664112432270849,
+      "loss": 1.1989,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003663417422318572,
+      "loss": 1.2365,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036627217601182665,
+      "loss": 1.2004,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036620254459427114,
+      "loss": 1.2267,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036613284800649393,
+      "loss": 1.2281,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003660630862758241,
+      "loss": 1.2472,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036599325942961614,
+      "loss": 1.2256,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003659233674952501,
+      "loss": 0.9445,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003658534105001314,
+      "loss": 1.3284,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036578338847169123,
+      "loss": 1.0688,
+      "step": 1049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003657133014373862,
+      "loss": 1.3117,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003656431494246983,
+      "loss": 1.1613,
+      "step": 1051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036557293246113493,
+      "loss": 1.0607,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036550265057422933,
+      "loss": 1.1814,
+      "step": 1053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036543230379153984,
+      "loss": 1.1707,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003653618921406503,
+      "loss": 0.9765,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036529141564917026,
+      "loss": 0.8707,
+      "step": 1056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036522087434473423,
+      "loss": 0.9407,
+      "step": 1057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003651502682550026,
+      "loss": 1.1004,
+      "step": 1058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036507959740766086,
+      "loss": 1.0669,
+      "step": 1059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003650088618304199,
+      "loss": 1.0919,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036493806155101627,
+      "loss": 1.0565,
+      "step": 1061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003648671965972116,
+      "loss": 1.1689,
+      "step": 1062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036479626699679297,
+      "loss": 1.1131,
+      "step": 1063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003647252727775728,
+      "loss": 1.1281,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036465421396738895,
+      "loss": 0.9793,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003645830905941044,
+      "loss": 1.1124,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036451190268560767,
+      "loss": 1.4163,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036444065026981244,
+      "loss": 1.3063,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003643693333746577,
+      "loss": 1.0752,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036429795202810785,
+      "loss": 1.3464,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003642265062581523,
+      "loss": 1.2559,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003641549960928061,
+      "loss": 1.1888,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036408342156010916,
+      "loss": 1.3183,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036401178268812684,
+      "loss": 1.2875,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003639400795049497,
+      "loss": 1.1994,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036386831203869354,
+      "loss": 1.2946,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003637964803174993,
+      "loss": 1.1676,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003637245843695332,
+      "loss": 1.198,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036365262422298646,
+      "loss": 1.1899,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036358059990607584,
+      "loss": 1.0322,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036350851144704283,
+      "loss": 1.2394,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036343635887415427,
+      "loss": 0.9391,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003633641422157023,
+      "loss": 1.0928,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003632918615000039,
+      "loss": 1.0968,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003632195167554013,
+      "loss": 1.3039,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036314710801026195,
+      "loss": 1.2615,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036307463529297807,
+      "loss": 1.2398,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036300209863196736,
+      "loss": 1.4281,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003629294980556724,
+      "loss": 1.436,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003628568335925607,
+      "loss": 1.1081,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003627841052711251,
+      "loss": 1.1067,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036271131311988326,
+      "loss": 1.0367,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036263845716737794,
+      "loss": 1.1605,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036256553744217705,
+      "loss": 0.9692,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003624925539728732,
+      "loss": 1.2426,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036241950678808426,
+      "loss": 1.177,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036234639591645304,
+      "loss": 1.1194,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003622732213866472,
+      "loss": 0.961,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036219998322735956,
+      "loss": 1.2715,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003621266814673076,
+      "loss": 0.9698,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036205331613523413,
+      "loss": 1.2391,
+      "step": 1101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036197988725990645,
+      "loss": 1.2414,
+      "step": 1102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003619063948701172,
+      "loss": 1.0104,
+      "step": 1103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036183283899468354,
+      "loss": 1.0454,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003617592196624478,
+      "loss": 1.054,
+      "step": 1105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003616855369022771,
+      "loss": 0.9333,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003616117907430634,
+      "loss": 1.1955,
+      "step": 1107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003615379812137236,
+      "loss": 0.9666,
+      "step": 1108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036146410834319925,
+      "loss": 1.1335,
+      "step": 1109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000361390172160457,
+      "loss": 1.083,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003613161726944882,
+      "loss": 1.1315,
+      "step": 1111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003612421099743091,
+      "loss": 1.0174,
+      "step": 1112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003611679840289605,
+      "loss": 1.0766,
+      "step": 1113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036109379488750824,
+      "loss": 1.2678,
+      "step": 1114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003610195425790429,
+      "loss": 1.0181,
+      "step": 1115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003609452271326798,
+      "loss": 0.9212,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000360870848577559,
+      "loss": 1.011,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003607964069428452,
+      "loss": 1.2065,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003607219022577281,
+      "loss": 0.9411,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036064733455142196,
+      "loss": 1.316,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003605727038531657,
+      "loss": 1.1334,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036049801019222305,
+      "loss": 1.2196,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003604232535978824,
+      "loss": 0.9452,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036034843409945677,
+      "loss": 1.278,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003602735517262839,
+      "loss": 0.8892,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036019860650772615,
+      "loss": 1.1297,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003601235984731705,
+      "loss": 1.1495,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003600485276520286,
+      "loss": 1.1328,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003599733940737368,
+      "loss": 0.9314,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003598981977677559,
+      "loss": 1.1294,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035982293876357137,
+      "loss": 1.3597,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035974761709069327,
+      "loss": 1.0998,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003596722327786562,
+      "loss": 1.1038,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035959678585701944,
+      "loss": 1.0637,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035952127635536667,
+      "loss": 1.0607,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003594457043033061,
+      "loss": 1.2598,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035937006973047066,
+      "loss": 1.2221,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003592943726665176,
+      "loss": 0.9576,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035921861314112876,
+      "loss": 1.1179,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035914279118401046,
+      "loss": 1.0558,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035906690682489345,
+      "loss": 1.257,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000358990960093533,
+      "loss": 1.0578,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003589149510197089,
+      "loss": 1.0976,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003588388796332252,
+      "loss": 1.1248,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035876274596391064,
+      "loss": 1.1485,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035868655004161814,
+      "loss": 1.0761,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035861029189622516,
+      "loss": 1.0524,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035853397155763357,
+      "loss": 1.0784,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035845758905576956,
+      "loss": 1.1381,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003583811444205837,
+      "loss": 1.2853,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00035830463768205087,
+      "loss": 1.2901,
+      "step": 1151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003582280688701705,
+      "loss": 1.0708,
+      "step": 1152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003581514380149662,
+      "loss": 1.191,
+      "step": 1153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000358074745146486,
+      "loss": 1.1383,
+      "step": 1154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003579979902948021,
+      "loss": 1.2729,
+      "step": 1155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003579211734900111,
+      "loss": 1.3415,
+      "step": 1156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035784429476223395,
+      "loss": 1.1419,
+      "step": 1157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035776735414161576,
+      "loss": 1.2621,
+      "step": 1158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000357690351658326,
+      "loss": 0.9822,
+      "step": 1159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003576132873425584,
+      "loss": 0.9845,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003575361612245308,
+      "loss": 1.1973,
+      "step": 1161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003574589733344855,
+      "loss": 1.0473,
+      "step": 1162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035738172370268874,
+      "loss": 1.3139,
+      "step": 1163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003573044123594313,
+      "loss": 1.3369,
+      "step": 1164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035722703933502786,
+      "loss": 1.2486,
+      "step": 1165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003571496046598174,
+      "loss": 1.0747,
+      "step": 1166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035707210836416313,
+      "loss": 1.1838,
+      "step": 1167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035699455047845244,
+      "loss": 0.9702,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003569169310330967,
+      "loss": 1.2516,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003568392500585315,
+      "loss": 1.2388,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035676150758521673,
+      "loss": 1.0161,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035668370364363614,
+      "loss": 1.1382,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003566058382642977,
+      "loss": 1.3585,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035652791147773344,
+      "loss": 1.1258,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035644992331449946,
+      "loss": 1.008,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003563718738051761,
+      "loss": 1.2755,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035629376298036746,
+      "loss": 1.0001,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035621559087070187,
+      "loss": 1.5262,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003561373575068316,
+      "loss": 1.1628,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035605906291943316,
+      "loss": 0.9307,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035598070713920677,
+      "loss": 1.048,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035590229019687674,
+      "loss": 1.1641,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035582381212319146,
+      "loss": 1.2365,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035574527294892315,
+      "loss": 1.0277,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003556666727048682,
+      "loss": 1.2817,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035558801142184665,
+      "loss": 1.0254,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003555092891307027,
+      "loss": 0.9908,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035543050586230447,
+      "loss": 1.4276,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035535166164754384,
+      "loss": 0.8255,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003552727565173366,
+      "loss": 1.2273,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003551937905026227,
+      "loss": 1.2478,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003551147636343656,
+      "loss": 1.1231,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003550356759435529,
+      "loss": 1.2562,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003549565274611958,
+      "loss": 1.5545,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003548773182183295,
+      "loss": 0.8958,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003547980482460131,
+      "loss": 0.9869,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003547187175753293,
+      "loss": 1.1297,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003546393262373848,
+      "loss": 1.2241,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035455987426330986,
+      "loss": 1.0484,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003544803616842588,
+      "loss": 1.3494,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003544007885314095,
+      "loss": 1.2658,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003543211548359636,
+      "loss": 1.2069,
+      "step": 1202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035424146062914663,
+      "loss": 1.2909,
+      "step": 1203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003541617059422076,
+      "loss": 1.1044,
+      "step": 1204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003540818908064195,
+      "loss": 1.0702,
+      "step": 1205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003540020152530788,
+      "loss": 1.2859,
+      "step": 1206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035392207931350596,
+      "loss": 1.2021,
+      "step": 1207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003538420830190447,
+      "loss": 1.1767,
+      "step": 1208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035376202640106274,
+      "loss": 1.5335,
+      "step": 1209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035368190949095136,
+      "loss": 1.2915,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003536017323201254,
+      "loss": 1.134,
+      "step": 1211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003535214949200234,
+      "loss": 1.2298,
+      "step": 1212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003534411973221075,
+      "loss": 1.0439,
+      "step": 1213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003533608395578635,
+      "loss": 0.9439,
+      "step": 1214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003532804216588008,
+      "loss": 1.3265,
+      "step": 1215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035319994365645214,
+      "loss": 1.0452,
+      "step": 1216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035311940558237415,
+      "loss": 1.1494,
+      "step": 1217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035303880746814683,
+      "loss": 0.9744,
+      "step": 1218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035295814934537385,
+      "loss": 1.1207,
+      "step": 1219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003528774312456822,
+      "loss": 0.9438,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003527966532007226,
+      "loss": 0.8724,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003527158152421692,
+      "loss": 1.4577,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003526349174017196,
+      "loss": 1.2534,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003525539597110949,
+      "loss": 1.2105,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035247294220203975,
+      "loss": 1.2792,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035239186490632215,
+      "loss": 1.1939,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003523107278557335,
+      "loss": 1.1155,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003522295310820889,
+      "loss": 1.0049,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003521482746172265,
+      "loss": 0.9368,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003520669584930081,
+      "loss": 1.1501,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035198558274131893,
+      "loss": 1.219,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003519041473940674,
+      "loss": 1.2595,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003518226524831854,
+      "loss": 1.1699,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035174109804062806,
+      "loss": 0.8531,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035165948409837417,
+      "loss": 1.1388,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035157781068842557,
+      "loss": 1.1385,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003514960778428074,
+      "loss": 1.1177,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035141428559356825,
+      "loss": 1.1985,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035133243397277996,
+      "loss": 1.3033,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035125052301253763,
+      "loss": 1.0922,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003511685527449596,
+      "loss": 1.4708,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035108652320218754,
+      "loss": 1.0354,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035100443441638625,
+      "loss": 1.3805,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003509222864197439,
+      "loss": 1.2726,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003508400792444719,
+      "loss": 1.0497,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003507578129228046,
+      "loss": 1.2441,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003506754874869997,
+      "loss": 1.0833,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035059310296933825,
+      "loss": 1.2875,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035051065940212427,
+      "loss": 1.0126,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003504281568176849,
+      "loss": 0.8941,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003503455952483705,
+      "loss": 1.0186,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035026297472655473,
+      "loss": 1.187,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035018029528463394,
+      "loss": 0.9576,
+      "step": 1253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035009755695502806,
+      "loss": 1.2393,
+      "step": 1254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035001475977017966,
+      "loss": 1.0614,
+      "step": 1255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003499319037625548,
+      "loss": 1.1292,
+      "step": 1256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034984898896464237,
+      "loss": 1.2439,
+      "step": 1257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034976601540895433,
+      "loss": 1.164,
+      "step": 1258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034968298312802574,
+      "loss": 1.1718,
+      "step": 1259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003495998921544146,
+      "loss": 1.0979,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003495167425207021,
+      "loss": 1.2347,
+      "step": 1261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003494335342594922,
+      "loss": 1.0142,
+      "step": 1262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003493502674034121,
+      "loss": 0.9921,
+      "step": 1263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003492669419851116,
+      "loss": 1.1515,
+      "step": 1264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003491835580372639,
+      "loss": 1.1167,
+      "step": 1265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003491001155925649,
+      "loss": 1.1611,
+      "step": 1266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003490166146837335,
+      "loss": 1.1523,
+      "step": 1267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034893305534351146,
+      "loss": 0.876,
+      "step": 1268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003488494376046636,
+      "loss": 1.0548,
+      "step": 1269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003487657614999775,
+      "loss": 1.0373,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003486820270622636,
+      "loss": 1.2997,
+      "step": 1271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034859823432435537,
+      "loss": 0.9526,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034851438331910906,
+      "loss": 0.8999,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034843047407940366,
+      "loss": 1.2898,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034834650663814125,
+      "loss": 1.3697,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003482624810282464,
+      "loss": 0.963,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003481783972826668,
+      "loss": 1.0092,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034809425543437284,
+      "loss": 0.9918,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003480100555163574,
+      "loss": 0.9875,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003479257975616367,
+      "loss": 1.1896,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034784148160324924,
+      "loss": 1.2519,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034775710767425636,
+      "loss": 1.4018,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034767267580774233,
+      "loss": 1.0784,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034758818603681396,
+      "loss": 1.1043,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003475036383946008,
+      "loss": 1.186,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003474190329142551,
+      "loss": 1.1169,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034733436962895183,
+      "loss": 1.1678,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003472496485718885,
+      "loss": 0.9539,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034716486977628546,
+      "loss": 1.3514,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034708003327538553,
+      "loss": 0.7709,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003469951391024543,
+      "loss": 1.2613,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003469101872907798,
+      "loss": 0.8848,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003468251778736728,
+      "loss": 1.3151,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034674011088446665,
+      "loss": 1.0354,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003466549863565172,
+      "loss": 1.1201,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034656980432320295,
+      "loss": 1.1604,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034648456481792495,
+      "loss": 1.0131,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003463992678741066,
+      "loss": 0.9116,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000346313913525194,
+      "loss": 0.9276,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034622850180465584,
+      "loss": 0.8115,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003461430327459832,
+      "loss": 0.8456,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003460575063826894,
+      "loss": 1.2007,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034597192274831063,
+      "loss": 1.4241,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003458862818764054,
+      "loss": 0.9574,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003458005838005545,
+      "loss": 0.8777,
+      "step": 1305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003457148285543614,
+      "loss": 1.0881,
+      "step": 1306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003456290161714518,
+      "loss": 1.3243,
+      "step": 1307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003455431466854739,
+      "loss": 1.2858,
+      "step": 1308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003454572201300983,
+      "loss": 1.4401,
+      "step": 1309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034537123653901783,
+      "loss": 1.0988,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034528519594594797,
+      "loss": 1.2145,
+      "step": 1311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003451990983846262,
+      "loss": 1.1603,
+      "step": 1312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003451129438888126,
+      "loss": 0.9807,
+      "step": 1313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003450267324922895,
+      "loss": 0.9002,
+      "step": 1314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003449404642288615,
+      "loss": 1.1412,
+      "step": 1315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034485413913235555,
+      "loss": 1.1472,
+      "step": 1316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034476775723662085,
+      "loss": 1.1943,
+      "step": 1317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003446813185755289,
+      "loss": 1.149,
+      "step": 1318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003445948231829735,
+      "loss": 1.0622,
+      "step": 1319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003445082710928705,
+      "loss": 1.1098,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003444216623391582,
+      "loss": 1.0329,
+      "step": 1321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003443349969557971,
+      "loss": 1.1558,
+      "step": 1322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003442482749767698,
+      "loss": 1.3314,
+      "step": 1323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003441614964360811,
+      "loss": 1.0442,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000344074661367758,
+      "loss": 1.0603,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034398776980584984,
+      "loss": 1.0624,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034390082178442774,
+      "loss": 0.9251,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003438138173375853,
+      "loss": 1.1871,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034372675649943805,
+      "loss": 1.239,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003436396393041238,
+      "loss": 0.9182,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003435524657858023,
+      "loss": 0.9054,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034346523597865545,
+      "loss": 1.0939,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034337794991688714,
+      "loss": 1.4583,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034329060763472346,
+      "loss": 1.0657,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003432032091664125,
+      "loss": 1.0677,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034311575454622434,
+      "loss": 1.0149,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003430282438084511,
+      "loss": 1.0703,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034294067698740685,
+      "loss": 1.2494,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003428530541174277,
+      "loss": 1.1245,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003427653752328718,
+      "loss": 1.2655,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003426776403681193,
+      "loss": 0.9158,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034258984955757194,
+      "loss": 1.2124,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034250200283565387,
+      "loss": 1.1815,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003424141002368109,
+      "loss": 1.4238,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034232614179551083,
+      "loss": 1.201,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034223812754624323,
+      "loss": 0.9614,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034215005752351974,
+      "loss": 1.1991,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003420619317618737,
+      "loss": 1.2701,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003419737502958605,
+      "loss": 1.2855,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034188551316005725,
+      "loss": 0.8667,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034179722038906287,
+      "loss": 1.2838,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034170887201749806,
+      "loss": 1.1274,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003416204680800055,
+      "loss": 1.1717,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034153200861124954,
+      "loss": 1.0169,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034144349364591617,
+      "loss": 1.166,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003413549232187135,
+      "loss": 1.03,
+      "step": 1356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000341266297364371,
+      "loss": 1.304,
+      "step": 1357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034117761611764,
+      "loss": 0.8862,
+      "step": 1358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034108887951329376,
+      "loss": 0.9805,
+      "step": 1359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000341000087586127,
+      "loss": 1.2814,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003409112403709562,
+      "loss": 1.1895,
+      "step": 1361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003408223379026196,
+      "loss": 1.0062,
+      "step": 1362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034073338021597694,
+      "loss": 1.0828,
+      "step": 1363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003406443673459098,
+      "loss": 1.5664,
+      "step": 1364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034055529932732116,
+      "loss": 1.2374,
+      "step": 1365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034046617619513594,
+      "loss": 1.0161,
+      "step": 1366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034037699798430043,
+      "loss": 1.0078,
+      "step": 1367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003402877647297826,
+      "loss": 0.8715,
+      "step": 1368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003401984764665719,
+      "loss": 1.5004,
+      "step": 1369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003401091332296797,
+      "loss": 1.2592,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003400197350541383,
+      "loss": 0.8945,
+      "step": 1371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033993028197500216,
+      "loss": 1.2421,
+      "step": 1372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000339840774027347,
+      "loss": 1.3204,
+      "step": 1373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033975121124627005,
+      "loss": 1.2374,
+      "step": 1374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033966159366689,
+      "loss": 1.3103,
+      "step": 1375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033957192132434716,
+      "loss": 1.0489,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033948219425380326,
+      "loss": 0.9012,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033939241249044143,
+      "loss": 1.1527,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033930257606946626,
+      "loss": 1.0301,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033921268502610385,
+      "loss": 1.1934,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003391227393956016,
+      "loss": 0.9487,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033903273921322857,
+      "loss": 1.0908,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003389426845142748,
+      "loss": 1.0733,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003388525753340521,
+      "loss": 1.3386,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003387624117078934,
+      "loss": 0.8971,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033867219367115305,
+      "loss": 1.1041,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003385819212592067,
+      "loss": 0.9989,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003384915945074514,
+      "loss": 1.108,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033840121345130557,
+      "loss": 1.0741,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003383107781262087,
+      "loss": 1.0918,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033822028856762165,
+      "loss": 1.1048,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033812974481102666,
+      "loss": 1.0156,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003380391468919271,
+      "loss": 0.915,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033794849484584765,
+      "loss": 1.117,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033785778870833416,
+      "loss": 0.9707,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033776702851495363,
+      "loss": 1.2786,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003376762143012945,
+      "loss": 0.6423,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003375853461029661,
+      "loss": 1.0183,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033749442395559907,
+      "loss": 1.1975,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033740344789484524,
+      "loss": 0.9429,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003373124179563775,
+      "loss": 1.2463,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003372213341758899,
+      "loss": 1.2254,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003371301965890976,
+      "loss": 1.385,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003370390052317368,
+      "loss": 1.4896,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033694776013956494,
+      "loss": 1.0367,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033685646134836035,
+      "loss": 1.4269,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003367651088939225,
+      "loss": 1.0823,
+      "step": 1407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000336673702812072,
+      "loss": 1.1703,
+      "step": 1408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003365822431386503,
+      "loss": 1.3888,
+      "step": 1409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033649072990952,
+      "loss": 1.1914,
+      "step": 1410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033639916316056457,
+      "loss": 0.876,
+      "step": 1411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003363075429276886,
+      "loss": 1.158,
+      "step": 1412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033621586924681754,
+      "loss": 1.0331,
+      "step": 1413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033612414215389796,
+      "loss": 1.1349,
+      "step": 1414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003360323616848972,
+      "loss": 1.0579,
+      "step": 1415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033594052787580363,
+      "loss": 0.9236,
+      "step": 1416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033584864076262647,
+      "loss": 0.9726,
+      "step": 1417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003357567003813959,
+      "loss": 1.1728,
+      "step": 1418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033566470676816293,
+      "loss": 1.3181,
+      "step": 1419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003355726599589995,
+      "loss": 1.2462,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003354805599899984,
+      "loss": 1.1929,
+      "step": 1421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003353884068972733,
+      "loss": 1.2315,
+      "step": 1422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033529620071695847,
+      "loss": 1.2542,
+      "step": 1423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003352039414852094,
+      "loss": 1.3961,
+      "step": 1424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033511162923820204,
+      "loss": 1.1366,
+      "step": 1425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003350192640121332,
+      "loss": 0.8387,
+      "step": 1426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033492684584322054,
+      "loss": 0.9132,
+      "step": 1427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033483437476770255,
+      "loss": 0.9537,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003347418508218383,
+      "loss": 1.3339,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003346492740419076,
+      "loss": 1.2689,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033455664446421107,
+      "loss": 1.3972,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033446396212507006,
+      "loss": 1.1568,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003343712270608265,
+      "loss": 1.1402,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033427843930784295,
+      "loss": 1.3668,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003341855989025028,
+      "loss": 1.1537,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033409270588121006,
+      "loss": 1.2892,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003339997602803892,
+      "loss": 1.2557,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003339067621364855,
+      "loss": 1.1369,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003338137114859647,
+      "loss": 1.1729,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033372060836531334,
+      "loss": 0.9698,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003336274528110382,
+      "loss": 1.021,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003335342448596669,
+      "loss": 1.1325,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003334409845477476,
+      "loss": 1.0683,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003333476719118487,
+      "loss": 1.356,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033325430698855947,
+      "loss": 0.7988,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003331608898144895,
+      "loss": 1.0351,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003330674204262689,
+      "loss": 1.0061,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033297389886054825,
+      "loss": 1.2687,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003328803251539986,
+      "loss": 0.8196,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033278669934331146,
+      "loss": 1.3744,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033269302146519877,
+      "loss": 1.2417,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003325992915563928,
+      "loss": 1.0583,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033250550965364635,
+      "loss": 0.8831,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033241167579373246,
+      "loss": 1.2185,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033231779001344474,
+      "loss": 1.2491,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033222385234959697,
+      "loss": 1.1273,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003321298628390234,
+      "loss": 1.2482,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033203582151857853,
+      "loss": 1.2135,
+      "step": 1458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003319417284251373,
+      "loss": 1.2899,
+      "step": 1459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003318475835955948,
+      "loss": 1.1437,
+      "step": 1460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033175338706686636,
+      "loss": 1.148,
+      "step": 1461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033165913887588783,
+      "loss": 1.1972,
+      "step": 1462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033156483905961504,
+      "loss": 0.7383,
+      "step": 1463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003314704876550244,
+      "loss": 0.9826,
+      "step": 1464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033137608469911223,
+      "loss": 1.2455,
+      "step": 1465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003312816302288951,
+      "loss": 1.3128,
+      "step": 1466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033118712428141007,
+      "loss": 1.0371,
+      "step": 1467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033109256689371394,
+      "loss": 1.2007,
+      "step": 1468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033099795810288406,
+      "loss": 1.1243,
+      "step": 1469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033090329794601775,
+      "loss": 1.1391,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003308085864602325,
+      "loss": 0.7898,
+      "step": 1471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033071382368266597,
+      "loss": 1.0815,
+      "step": 1472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003306190096504759,
+      "loss": 1.0356,
+      "step": 1473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033052414440084014,
+      "loss": 1.4177,
+      "step": 1474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033042922797095654,
+      "loss": 1.1754,
+      "step": 1475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003303342603980432,
+      "loss": 1.0275,
+      "step": 1476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003302392417193381,
+      "loss": 0.8938,
+      "step": 1477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033014417197209926,
+      "loss": 1.0851,
+      "step": 1478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003300490511936049,
+      "loss": 1.296,
+      "step": 1479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032995387942115316,
+      "loss": 1.2133,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032985865669206204,
+      "loss": 0.9101,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032976338304366965,
+      "loss": 0.9238,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003296680585133341,
+      "loss": 1.1639,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003295726831384334,
+      "loss": 1.2796,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032947725695636553,
+      "loss": 1.101,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003293817800045483,
+      "loss": 1.2891,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003292862523204195,
+      "loss": 1.2151,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003291906739414368,
+      "loss": 1.0571,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003290950449050778,
+      "loss": 1.0169,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032899936524883987,
+      "loss": 1.3045,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003289036350102402,
+      "loss": 1.3203,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000328807854226816,
+      "loss": 1.2032,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003287120229361242,
+      "loss": 1.1573,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003286161411757414,
+      "loss": 1.155,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003285202089832642,
+      "loss": 1.1069,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003284242263963089,
+      "loss": 1.2333,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003283281934525114,
+      "loss": 1.2632,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032823211018952767,
+      "loss": 1.1675,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003281359766450332,
+      "loss": 1.0388,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003280397928567231,
+      "loss": 1.1391,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032794355886231247,
+      "loss": 0.8402,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003278472746995358,
+      "loss": 1.1279,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032775094040614746,
+      "loss": 1.0031,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003276545560199214,
+      "loss": 1.0491,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003275581215786512,
+      "loss": 1.1282,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003274616371201501,
+      "loss": 1.2839,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003273651026822509,
+      "loss": 0.9591,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032726851830280614,
+      "loss": 1.0022,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003271718840196877,
+      "loss": 1.3215,
+      "step": 1509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003270751998707872,
+      "loss": 1.1369,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003269784658940158,
+      "loss": 1.1737,
+      "step": 1511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003268816821273042,
+      "loss": 1.2151,
+      "step": 1512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032678484860860254,
+      "loss": 1.0848,
+      "step": 1513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032668796537588056,
+      "loss": 1.015,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032659103246712745,
+      "loss": 1.1183,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003264940499203519,
+      "loss": 1.2624,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032639701777358206,
+      "loss": 1.2344,
+      "step": 1517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003262999360648655,
+      "loss": 0.9448,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032620280483226924,
+      "loss": 1.3936,
+      "step": 1519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032610562411387975,
+      "loss": 1.3979,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032600839394780285,
+      "loss": 1.1073,
+      "step": 1521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032591111437216383,
+      "loss": 1.316,
+      "step": 1522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003258137854251072,
+      "loss": 0.8757,
+      "step": 1523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003257164071447971,
+      "loss": 1.1681,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003256189795694167,
+      "loss": 1.1686,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032552150273716863,
+      "loss": 0.9705,
+      "step": 1526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003254239766862749,
+      "loss": 1.0595,
+      "step": 1527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003253264014549768,
+      "loss": 1.1058,
+      "step": 1528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032522877708153486,
+      "loss": 0.9494,
+      "step": 1529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032513110360422886,
+      "loss": 0.9996,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003250333810613579,
+      "loss": 1.0903,
+      "step": 1531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003249356094912403,
+      "loss": 1.2025,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032483778893221347,
+      "loss": 1.1343,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032473991942263427,
+      "loss": 1.0501,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032464200100087857,
+      "loss": 1.0116,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032454403370534154,
+      "loss": 1.2105,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003244460175744374,
+      "loss": 1.0928,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032434795264659963,
+      "loss": 0.9489,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032424983896028073,
+      "loss": 1.2498,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003241516765539524,
+      "loss": 1.093,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003240534654661055,
+      "loss": 1.0413,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003239552057352497,
+      "loss": 1.0822,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003238568973999141,
+      "loss": 1.1609,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003237585404986467,
+      "loss": 1.1187,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003236601350700145,
+      "loss": 1.0982,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003235616811526036,
+      "loss": 1.3761,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003234631787850191,
+      "loss": 1.1093,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000323364628005885,
+      "loss": 1.1345,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032326602885384445,
+      "loss": 1.0082,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032316738136755946,
+      "loss": 0.9918,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000323068685585711,
+      "loss": 1.1323,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032296994154699905,
+      "loss": 1.331,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003228711492901424,
+      "loss": 1.3974,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003227723088538788,
+      "loss": 1.1592,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032267342027696497,
+      "loss": 1.3051,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003225744835981764,
+      "loss": 1.1939,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032247549885630736,
+      "loss": 1.1869,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032237646609017126,
+      "loss": 1.1633,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032227738533860005,
+      "loss": 1.1064,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032217825664044463,
+      "loss": 1.1018,
+      "step": 1560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032207908003457463,
+      "loss": 1.2293,
+      "step": 1561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032197985555987865,
+      "loss": 1.2317,
+      "step": 1562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032188058325526383,
+      "loss": 0.8852,
+      "step": 1563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003217812631596561,
+      "loss": 1.1316,
+      "step": 1564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032168189531200025,
+      "loss": 1.4239,
+      "step": 1565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032158247975125977,
+      "loss": 1.1941,
+      "step": 1566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032148301651641677,
+      "loss": 1.0941,
+      "step": 1567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032138350564647213,
+      "loss": 1.2116,
+      "step": 1568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003212839471804453,
+      "loss": 0.9964,
+      "step": 1569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003211843411573746,
+      "loss": 0.9137,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032108468761631677,
+      "loss": 1.0181,
+      "step": 1571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032098498659634733,
+      "loss": 0.656,
+      "step": 1572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032088523813656036,
+      "loss": 1.1602,
+      "step": 1573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032078544227606866,
+      "loss": 0.9315,
+      "step": 1574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032068559905400326,
+      "loss": 1.1339,
+      "step": 1575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032058570850951426,
+      "loss": 1.1436,
+      "step": 1576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032048577068176994,
+      "loss": 0.9467,
+      "step": 1577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032038578560995726,
+      "loss": 1.3432,
+      "step": 1578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003202857533332817,
+      "loss": 1.1531,
+      "step": 1579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032018567389096724,
+      "loss": 1.2634,
+      "step": 1580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003200855473222565,
+      "loss": 1.2071,
+      "step": 1581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031998537366641016,
+      "loss": 1.1977,
+      "step": 1582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003198851529627079,
+      "loss": 1.1247,
+      "step": 1583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003197848852504473,
+      "loss": 0.9402,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000319684570568945,
+      "loss": 1.302,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031958420895753545,
+      "loss": 1.0847,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031948380045557185,
+      "loss": 1.0318,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031938334510242576,
+      "loss": 1.0711,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031928284293748706,
+      "loss": 1.0582,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031918229400016384,
+      "loss": 0.9736,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031908169832988283,
+      "loss": 1.3543,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003189810559660889,
+      "loss": 1.0448,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031888036694824513,
+      "loss": 1.3689,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031877963131583315,
+      "loss": 0.9945,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031867884910835267,
+      "loss": 1.3879,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031857802036532176,
+      "loss": 0.931,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003184771451262768,
+      "loss": 0.919,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003183762234307721,
+      "loss": 1.0889,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031827525531838056,
+      "loss": 1.1982,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000318174240828693,
+      "loss": 1.3847,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003180731800013187,
+      "loss": 1.231,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031797207287588477,
+      "loss": 1.2161,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031787091949203675,
+      "loss": 1.3048,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031776971988943824,
+      "loss": 0.9392,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031766847410777095,
+      "loss": 1.1631,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003175671821867346,
+      "loss": 1.1193,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003174658441660472,
+      "loss": 0.9092,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031736446008544467,
+      "loss": 0.9157,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031726302998468105,
+      "loss": 0.8666,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031716155390352845,
+      "loss": 1.2977,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000317060031881777,
+      "loss": 0.9997,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031695846395923485,
+      "loss": 1.2235,
+      "step": 1612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000316856850175728,
+      "loss": 1.2765,
+      "step": 1613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031675519057110073,
+      "loss": 1.1136,
+      "step": 1614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031665348518521505,
+      "loss": 1.3248,
+      "step": 1615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031655173405795097,
+      "loss": 1.1099,
+      "step": 1616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003164499372292065,
+      "loss": 0.9658,
+      "step": 1617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031634809473889743,
+      "loss": 0.8747,
+      "step": 1618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031624620662695774,
+      "loss": 1.1359,
+      "step": 1619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003161442729333389,
+      "loss": 0.742,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003160422936980105,
+      "loss": 1.2613,
+      "step": 1621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031594026896096007,
+      "loss": 1.3199,
+      "step": 1622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031583819876219283,
+      "loss": 1.0635,
+      "step": 1623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003157360831417318,
+      "loss": 1.189,
+      "step": 1624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003156339221396178,
+      "loss": 1.0492,
+      "step": 1625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031553171579590977,
+      "loss": 1.3216,
+      "step": 1626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003154294641506839,
+      "loss": 0.9381,
+      "step": 1627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031532716724403455,
+      "loss": 1.2927,
+      "step": 1628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003152248251160737,
+      "loss": 1.1158,
+      "step": 1629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000315122437806931,
+      "loss": 1.4279,
+      "step": 1630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003150200053567539,
+      "loss": 1.0745,
+      "step": 1631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003149175278057075,
+      "loss": 1.2324,
+      "step": 1632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003148150051939746,
+      "loss": 1.223,
+      "step": 1633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031471243756175574,
+      "loss": 1.1551,
+      "step": 1634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000314609824949269,
+      "loss": 1.1431,
+      "step": 1635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003145071673967501,
+      "loss": 1.009,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003144044649444526,
+      "loss": 1.2728,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031430171763264733,
+      "loss": 1.0867,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003141989255016229,
+      "loss": 1.3099,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031409608859168546,
+      "loss": 1.0176,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031399320694315883,
+      "loss": 1.1742,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003138902805963841,
+      "loss": 1.1339,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003137873095917202,
+      "loss": 1.2274,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003136842939695433,
+      "loss": 1.0095,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031358123377024736,
+      "loss": 0.8373,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031347812903424353,
+      "loss": 0.9802,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031337497980196054,
+      "loss": 1.0864,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031327178611384465,
+      "loss": 1.0318,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003131685480103594,
+      "loss": 0.9073,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031306526553198584,
+      "loss": 1.0519,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003129619387192224,
+      "loss": 0.9221,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031285856761258487,
+      "loss": 1.1607,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031275515225260653,
+      "loss": 1.1798,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031265169267983775,
+      "loss": 0.9944,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003125481889348466,
+      "loss": 1.2528,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031244464105821806,
+      "loss": 1.1755,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003123410490905548,
+      "loss": 1.2017,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003122374130724765,
+      "loss": 1.1388,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031213373304462026,
+      "loss": 0.9823,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003120300090476404,
+      "loss": 1.1579,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031192624112220843,
+      "loss": 0.985,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031182242930901315,
+      "loss": 0.8356,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031171857364876055,
+      "loss": 0.9404,
+      "step": 1663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003116146741821739,
+      "loss": 1.2583,
+      "step": 1664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003115107309499933,
+      "loss": 0.979,
+      "step": 1665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003114067439929766,
+      "loss": 1.2377,
+      "step": 1666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003113027133518983,
+      "loss": 1.2107,
+      "step": 1667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003111986390675502,
+      "loss": 1.191,
+      "step": 1668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003110945211807412,
+      "loss": 0.8721,
+      "step": 1669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031099035973229726,
+      "loss": 1.0718,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003108861547630616,
+      "loss": 0.9079,
+      "step": 1671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031078190631389426,
+      "loss": 1.3147,
+      "step": 1672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031067761442567243,
+      "loss": 1.1536,
+      "step": 1673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003105732791392904,
+      "loss": 1.2583,
+      "step": 1674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003104689004956595,
+      "loss": 1.2236,
+      "step": 1675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031036447853570774,
+      "loss": 1.0161,
+      "step": 1676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031026001330038063,
+      "loss": 0.826,
+      "step": 1677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031015550483064013,
+      "loss": 1.2048,
+      "step": 1678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003100509531674655,
+      "loss": 1.4105,
+      "step": 1679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030994635835185293,
+      "loss": 1.3602,
+      "step": 1680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030984172042481533,
+      "loss": 1.0426,
+      "step": 1681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003097370394273826,
+      "loss": 0.912,
+      "step": 1682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003096323154006016,
+      "loss": 1.3722,
+      "step": 1683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030952754838553596,
+      "loss": 0.7858,
+      "step": 1684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003094227384232663,
+      "loss": 1.1181,
+      "step": 1685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003093178855548899,
+      "loss": 1.2127,
+      "step": 1686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000309212989821521,
+      "loss": 1.2257,
+      "step": 1687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003091080512642906,
+      "loss": 1.158,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030900306992434656,
+      "loss": 1.0265,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003088980458428534,
+      "loss": 1.1433,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030879297906099244,
+      "loss": 1.1096,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030868786961996174,
+      "loss": 1.1366,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030858271756097625,
+      "loss": 0.9545,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030847752292526725,
+      "loss": 0.7818,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003083722857540832,
+      "loss": 1.2642,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003082670060886888,
+      "loss": 1.0645,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030816168397036574,
+      "loss": 0.942,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030805631944041213,
+      "loss": 0.8641,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030795091254014277,
+      "loss": 1.1694,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003078454633108892,
+      "loss": 1.0831,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030773997179399945,
+      "loss": 1.6421,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003076344380308381,
+      "loss": 0.7813,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003075288620627863,
+      "loss": 1.0487,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003074232439312419,
+      "loss": 1.3504,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003073175836776191,
+      "loss": 0.9397,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030721188134334867,
+      "loss": 1.2887,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030710613696987785,
+      "loss": 0.9855,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003070003505986706,
+      "loss": 1.1876,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030689452227120696,
+      "loss": 1.2558,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003067886520289837,
+      "loss": 1.2878,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000306682739913514,
+      "loss": 0.9395,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003065767859663273,
+      "loss": 1.0234,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030647079022896965,
+      "loss": 0.9168,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003063647527430033,
+      "loss": 1.1673,
+      "step": 1714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000306258673550007,
+      "loss": 1.0,
+      "step": 1715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003061525526915758,
+      "loss": 1.1273,
+      "step": 1716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030604639020932116,
+      "loss": 1.0908,
+      "step": 1717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030594018614487056,
+      "loss": 1.2419,
+      "step": 1718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030583394053986836,
+      "loss": 1.2821,
+      "step": 1719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003057276534359746,
+      "loss": 1.0879,
+      "step": 1720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030562132487486595,
+      "loss": 1.0085,
+      "step": 1721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030551495489823536,
+      "loss": 0.9113,
+      "step": 1722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030540854354779175,
+      "loss": 1.0794,
+      "step": 1723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003053020908652605,
+      "loss": 1.1175,
+      "step": 1724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003051955968923831,
+      "loss": 1.0947,
+      "step": 1725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030508906167091716,
+      "loss": 1.1313,
+      "step": 1726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030498248524263674,
+      "loss": 1.1764,
+      "step": 1727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030487586764933175,
+      "loss": 0.9874,
+      "step": 1728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003047692089328084,
+      "loss": 1.1541,
+      "step": 1729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003046625091348889,
+      "loss": 1.1224,
+      "step": 1730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030455576829741183,
+      "loss": 1.2339,
+      "step": 1731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030444898646223153,
+      "loss": 1.5093,
+      "step": 1732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030434216367121865,
+      "loss": 1.1181,
+      "step": 1733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003042352999662598,
+      "loss": 1.0404,
+      "step": 1734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030412839538925766,
+      "loss": 1.2428,
+      "step": 1735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003040214499821309,
+      "loss": 1.1174,
+      "step": 1736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030391446378681426,
+      "loss": 1.3101,
+      "step": 1737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003038074368452585,
+      "loss": 1.4131,
+      "step": 1738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030370036919943026,
+      "loss": 0.7093,
+      "step": 1739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003035932608913121,
+      "loss": 0.9578,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030348611196290276,
+      "loss": 0.8644,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003033789224562166,
+      "loss": 1.1902,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030327169241328424,
+      "loss": 1.2839,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030316442187615177,
+      "loss": 1.0251,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030305711088688157,
+      "loss": 1.0275,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003029497594875516,
+      "loss": 1.2135,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030284236772025583,
+      "loss": 1.1406,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030273493562710393,
+      "loss": 1.1442,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003026274632502216,
+      "loss": 0.9862,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030251995063175,
+      "loss": 0.8475,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003024123978138463,
+      "loss": 1.2657,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003023048048386834,
+      "loss": 1.1376,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030219717174845,
+      "loss": 1.2807,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030208949858535043,
+      "loss": 1.2444,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003019817853916047,
+      "loss": 1.3785,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030187403220944864,
+      "loss": 0.9563,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030176623908113366,
+      "loss": 1.0895,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030165840604892696,
+      "loss": 1.0351,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003015505331551112,
+      "loss": 1.0399,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003014426204419848,
+      "loss": 0.8304,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003013346679518618,
+      "loss": 0.9187,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030122667572707177,
+      "loss": 1.1991,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030111864380996,
+      "loss": 1.0149,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000301010572242887,
+      "loss": 1.2401,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030090246106822936,
+      "loss": 1.3311,
+      "step": 1765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003007943103283786,
+      "loss": 0.9106,
+      "step": 1766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030068612006574233,
+      "loss": 1.0145,
+      "step": 1767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003005778903227432,
+      "loss": 1.3225,
+      "step": 1768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003004696211418196,
+      "loss": 1.2157,
+      "step": 1769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030036131256542533,
+      "loss": 1.1726,
+      "step": 1770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003002529646360296,
+      "loss": 1.2285,
+      "step": 1771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003001445773961171,
+      "loss": 1.2323,
+      "step": 1772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003000361508881878,
+      "loss": 0.8774,
+      "step": 1773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029992768515475733,
+      "loss": 1.2084,
+      "step": 1774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029981918023835635,
+      "loss": 1.0489,
+      "step": 1775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029971063618153124,
+      "loss": 1.2651,
+      "step": 1776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002996020530268435,
+      "loss": 1.1037,
+      "step": 1777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029949343081687004,
+      "loss": 0.7891,
+      "step": 1778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000299384769594203,
+      "loss": 0.9389,
+      "step": 1779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029927606940145004,
+      "loss": 1.2723,
+      "step": 1780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002991673302812338,
+      "loss": 0.9315,
+      "step": 1781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002990585522761924,
+      "loss": 1.3004,
+      "step": 1782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002989497354289791,
+      "loss": 1.3098,
+      "step": 1783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002988408797822625,
+      "loss": 1.0576,
+      "step": 1784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029873198537872626,
+      "loss": 1.1795,
+      "step": 1785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029862305226106936,
+      "loss": 1.1723,
+      "step": 1786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002985140804720059,
+      "loss": 0.969,
+      "step": 1787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029840507005426514,
+      "loss": 1.105,
+      "step": 1788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029829602105059164,
+      "loss": 1.3559,
+      "step": 1789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002981869335037447,
+      "loss": 0.9184,
+      "step": 1790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029807780745649934,
+      "loss": 0.9248,
+      "step": 1791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029796864295164507,
+      "loss": 1.067,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029785944003198685,
+      "loss": 1.0395,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029775019874034455,
+      "loss": 0.9651,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002976409191195531,
+      "loss": 1.089,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002975316012124626,
+      "loss": 1.1385,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002974222450619379,
+      "loss": 1.1516,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029731285071085917,
+      "loss": 1.182,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972034182021212,
+      "loss": 1.062,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002970939475786341,
+      "loss": 1.2661,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002969844388833226,
+      "loss": 1.0961,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002968748921591266,
+      "loss": 0.8134,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029676530744900093,
+      "loss": 1.2017,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029665568479591496,
+      "loss": 1.1155,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029654602424285335,
+      "loss": 1.2035,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029643632583281545,
+      "loss": 1.2447,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002963265896088154,
+      "loss": 1.0481,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002962168156138823,
+      "loss": 0.9947,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029610700389105994,
+      "loss": 1.0338,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002959971544834069,
+      "loss": 1.0412,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029588726743399665,
+      "loss": 1.2528,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002957773427859174,
+      "loss": 1.2676,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000295667380582272,
+      "loss": 1.1846,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029555738086617817,
+      "loss": 1.1813,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029544734368076815,
+      "loss": 1.3814,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029533726906918903,
+      "loss": 0.9474,
+      "step": 1816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029522715707460247,
+      "loss": 1.1424,
+      "step": 1817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029511700774018497,
+      "loss": 1.1055,
+      "step": 1818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002950068211091274,
+      "loss": 0.9551,
+      "step": 1819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002948965972246356,
+      "loss": 0.7597,
+      "step": 1820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029478633612992956,
+      "loss": 0.9747,
+      "step": 1821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029467603786824425,
+      "loss": 1.116,
+      "step": 1822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002945657024828292,
+      "loss": 1.2477,
+      "step": 1823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002944553300169481,
+      "loss": 0.9222,
+      "step": 1824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002943449205138797,
+      "loss": 0.7647,
+      "step": 1825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002942344740169169,
+      "loss": 0.9612,
+      "step": 1826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002941239905693673,
+      "loss": 1.1071,
+      "step": 1827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002940134702145529,
+      "loss": 1.3703,
+      "step": 1828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029390291299581016,
+      "loss": 1.0886,
+      "step": 1829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002937923189564901,
+      "loss": 1.0391,
+      "step": 1830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029368168813995806,
+      "loss": 1.0367,
+      "step": 1831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002935710205895939,
+      "loss": 1.2917,
+      "step": 1832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002934603163487917,
+      "loss": 0.909,
+      "step": 1833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029334957546096034,
+      "loss": 0.9997,
+      "step": 1834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029323879796952246,
+      "loss": 1.1053,
+      "step": 1835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002931279839179156,
+      "loss": 1.2384,
+      "step": 1836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002930171333495913,
+      "loss": 1.0539,
+      "step": 1837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029290624630801565,
+      "loss": 1.1419,
+      "step": 1838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029279532283666885,
+      "loss": 1.0269,
+      "step": 1839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002926843629790455,
+      "loss": 1.0609,
+      "step": 1840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002925733667786544,
+      "loss": 1.0083,
+      "step": 1841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029246233427901864,
+      "loss": 1.1389,
+      "step": 1842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029235126552367563,
+      "loss": 1.1742,
+      "step": 1843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002922401605561767,
+      "loss": 1.2319,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002921290194200878,
+      "loss": 0.8922,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029201784215898864,
+      "loss": 0.9912,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002919066288164735,
+      "loss": 1.1651,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002917953794361504,
+      "loss": 1.0881,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002916840940616419,
+      "loss": 0.9634,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002915727727365844,
+      "loss": 1.243,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002914614155046283,
+      "loss": 0.9986,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002913500224094385,
+      "loss": 1.1221,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002912385934946936,
+      "loss": 1.0394,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029112712880408623,
+      "loss": 0.8996,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002910156283813234,
+      "loss": 1.1209,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029090409227012576,
+      "loss": 1.188,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002907925205142282,
+      "loss": 0.9799,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002906809131573794,
+      "loss": 0.9039,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002905692702433421,
+      "loss": 1.3134,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029045759181589294,
+      "loss": 1.2748,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002903458779188226,
+      "loss": 1.0981,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002902341285959355,
+      "loss": 1.1143,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002901223438910501,
+      "loss": 1.3649,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029001052384799856,
+      "loss": 0.9978,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028989866851062714,
+      "loss": 1.2001,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028978677792279567,
+      "loss": 1.2196,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00028967485212837797,
+      "loss": 0.9469,
+      "step": 1867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002895628911712617,
+      "loss": 1.0567,
+      "step": 1868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028945089509534817,
+      "loss": 0.8099,
+      "step": 1869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028945089509534817,
+      "loss": 1.123,
+      "step": 1870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002893388639445526,
+      "loss": 1.0569,
+      "step": 1871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028922679776280375,
+      "loss": 1.003,
+      "step": 1872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002891146965940443,
+      "loss": 0.9546,
+      "step": 1873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028900256048223073,
+      "loss": 1.1414,
+      "step": 1874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002888903894713331,
+      "loss": 1.0351,
+      "step": 1875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002887781836053349,
+      "loss": 1.0653,
+      "step": 1876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002886659429282337,
+      "loss": 1.0038,
+      "step": 1877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028855366748404063,
+      "loss": 1.2374,
+      "step": 1878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028844135731678023,
+      "loss": 1.2323,
+      "step": 1879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028832901247049093,
+      "loss": 1.132,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028821663298922456,
+      "loss": 1.1161,
+      "step": 1881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028810421891704667,
+      "loss": 0.9467,
+      "step": 1882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002879917702980362,
+      "loss": 0.9718,
+      "step": 1883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002878792871762859,
+      "loss": 1.1584,
+      "step": 1884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002877667695959017,
+      "loss": 1.1354,
+      "step": 1885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028765421760100344,
+      "loss": 1.1454,
+      "step": 1886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002875416312357241,
+      "loss": 0.994,
+      "step": 1887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002874290105442103,
+      "loss": 1.1613,
+      "step": 1888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028731635557062223,
+      "loss": 1.0067,
+      "step": 1889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028720366635913327,
+      "loss": 1.346,
+      "step": 1890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002870909429539304,
+      "loss": 0.9365,
+      "step": 1891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000286978185399214,
+      "loss": 1.3717,
+      "step": 1892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028686539373919774,
+      "loss": 1.1675,
+      "step": 1893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002867525680181087,
+      "loss": 1.2418,
+      "step": 1894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002866397082801874,
+      "loss": 1.1002,
+      "step": 1895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028652681456968774,
+      "loss": 1.084,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002864138869308766,
+      "loss": 1.0009,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028630092540803456,
+      "loss": 0.9431,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002861879300454553,
+      "loss": 1.1149,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028607490088744585,
+      "loss": 0.8813,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002859618379783263,
+      "loss": 1.1238,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002858487413624304,
+      "loss": 1.3803,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028573561108410445,
+      "loss": 1.2947,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002856224471877086,
+      "loss": 1.2191,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028550924971761587,
+      "loss": 0.9784,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002853960187182124,
+      "loss": 1.301,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002852827542338977,
+      "loss": 1.0453,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028516945630908415,
+      "loss": 1.1204,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002850561249881974,
+      "loss": 0.8491,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028494276031567625,
+      "loss": 1.4188,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002848293623359724,
+      "loss": 1.0131,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028471593109355073,
+      "loss": 1.212,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028460246663288913,
+      "loss": 1.0126,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002844889689984785,
+      "loss": 1.1364,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028437543823482285,
+      "loss": 1.0892,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000284261874386439,
+      "loss": 0.9669,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028414827749785685,
+      "loss": 1.1882,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028403464761361937,
+      "loss": 1.1465,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028392098477828215,
+      "loss": 1.2023,
+      "step": 1919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000283807289036414,
+      "loss": 1.0616,
+      "step": 1920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002836935604325966,
+      "loss": 0.8599,
+      "step": 1921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028357979901142425,
+      "loss": 0.816,
+      "step": 1922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002834660048175045,
+      "loss": 1.0604,
+      "step": 1923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002833521778954575,
+      "loss": 1.047,
+      "step": 1924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002832383182899163,
+      "loss": 1.2057,
+      "step": 1925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002831244260455266,
+      "loss": 1.0111,
+      "step": 1926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002830105012069474,
+      "loss": 1.0089,
+      "step": 1927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002828965438188498,
+      "loss": 1.0594,
+      "step": 1928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028278255392591824,
+      "loss": 1.0711,
+      "step": 1929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002826685315728495,
+      "loss": 0.8731,
+      "step": 1930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028255447680435345,
+      "loss": 1.1281,
+      "step": 1931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028244038966515226,
+      "loss": 1.286,
+      "step": 1932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002823262701999812,
+      "loss": 1.1219,
+      "step": 1933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002822121184535878,
+      "loss": 1.1672,
+      "step": 1934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002820979344707327,
+      "loss": 1.3899,
+      "step": 1935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002819837182961889,
+      "loss": 1.0291,
+      "step": 1936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000281869469974742,
+      "loss": 1.3121,
+      "step": 1937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002817551895511903,
+      "loss": 1.3754,
+      "step": 1938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002816408770703447,
+      "loss": 1.0731,
+      "step": 1939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028152653257702864,
+      "loss": 1.117,
+      "step": 1940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002814121561160781,
+      "loss": 0.8461,
+      "step": 1941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028129774773234166,
+      "loss": 0.7918,
+      "step": 1942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002811833074706802,
+      "loss": 1.1428,
+      "step": 1943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028106883537596744,
+      "loss": 1.2813,
+      "step": 1944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028095433149308936,
+      "loss": 1.0516,
+      "step": 1945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002808397958669444,
+      "loss": 0.8452,
+      "step": 1946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002807252285424436,
+      "loss": 0.9885,
+      "step": 1947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002806106295645102,
+      "loss": 1.1161,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028049599897808007,
+      "loss": 1.1727,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002803813368281013,
+      "loss": 1.1371,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028026664315953457,
+      "loss": 1.297,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028015191801735264,
+      "loss": 1.1597,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028003716144654083,
+      "loss": 1.1477,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002799223734920968,
+      "loss": 1.0499,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002798075541990302,
+      "loss": 1.2265,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002796927036123635,
+      "loss": 1.1505,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027957782177713086,
+      "loss": 1.042,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027946290873837916,
+      "loss": 1.15,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027934796454116725,
+      "loss": 1.1113,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002792329892305662,
+      "loss": 1.0724,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027911798285165944,
+      "loss": 0.9197,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027900294544954253,
+      "loss": 1.3029,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027888787706932316,
+      "loss": 0.8489,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027877277775612106,
+      "loss": 1.1979,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002786576475550683,
+      "loss": 1.1043,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002785424865113089,
+      "loss": 1.1103,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00027842729466999905,
+      "loss": 1.1164,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000278312072076307,
+      "loss": 1.121,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002781968187754131,
+      "loss": 1.0182,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002780815348125096,
+      "loss": 0.9005,
+      "step": 1970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027796622023280095,
+      "loss": 0.9906,
+      "step": 1971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002778508750815035,
+      "loss": 0.9942,
+      "step": 1972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002777354994038456,
+      "loss": 0.8644,
+      "step": 1973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002776200932450677,
+      "loss": 1.034,
+      "step": 1974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027750465665042184,
+      "loss": 1.1366,
+      "step": 1975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027738918966517247,
+      "loss": 1.1597,
+      "step": 1976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027727369233459566,
+      "loss": 1.1152,
+      "step": 1977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027715816470397936,
+      "loss": 0.9154,
+      "step": 1978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002770426068186236,
+      "loss": 1.1628,
+      "step": 1979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002769270187238401,
+      "loss": 1.2295,
+      "step": 1980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002768114004649525,
+      "loss": 0.9879,
+      "step": 1981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027669575208729623,
+      "loss": 0.9741,
+      "step": 1982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002765800736362186,
+      "loss": 0.9146,
+      "step": 1983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002764643651570786,
+      "loss": 1.3153,
+      "step": 1984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002763486266952471,
+      "loss": 0.9752,
+      "step": 1985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027623285829610667,
+      "loss": 0.8999,
+      "step": 1986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002761170600050517,
+      "loss": 0.9649,
+      "step": 1987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002760012318674881,
+      "loss": 0.9247,
+      "step": 1988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027588537392883377,
+      "loss": 0.9992,
+      "step": 1989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027576948623451796,
+      "loss": 1.0283,
+      "step": 1990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002756535688299819,
+      "loss": 0.8936,
+      "step": 1991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027553762176067844,
+      "loss": 1.0211,
+      "step": 1992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002754216450720718,
+      "loss": 1.3266,
+      "step": 1993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000275305638809638,
+      "loss": 1.0175,
+      "step": 1994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027518960301886467,
+      "loss": 1.3003,
+      "step": 1995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027507353774525097,
+      "loss": 0.9858,
+      "step": 1996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027495744303430764,
+      "loss": 1.2655,
+      "step": 1997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000274841318931557,
+      "loss": 1.1762,
+      "step": 1998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002747251654825327,
+      "loss": 0.9788,
+      "step": 1999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027460898273278016,
+      "loss": 1.2095,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002744927707278562,
+      "loss": 1.3003,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027437652951332913,
+      "loss": 1.3253,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002742602591347785,
+      "loss": 1.3259,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027414395963779557,
+      "loss": 1.1737,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002740276310679829,
+      "loss": 1.259,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002739112734709545,
+      "loss": 1.3894,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027379488689233567,
+      "loss": 0.9648,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002736784713777632,
+      "loss": 0.7933,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027356202697288506,
+      "loss": 1.2287,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002734455537233607,
+      "loss": 0.9479,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002733290516748609,
+      "loss": 1.2068,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027321252087306755,
+      "loss": 1.1211,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002730959613636739,
+      "loss": 0.8738,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002729793731923846,
+      "loss": 1.2323,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027286275640491534,
+      "loss": 1.1279,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002727461110469931,
+      "loss": 0.923,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027262943716435613,
+      "loss": 1.1905,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027251273480275373,
+      "loss": 1.262,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027239600400794653,
+      "loss": 1.0295,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002722792448257061,
+      "loss": 1.4338,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027216245730181545,
+      "loss": 1.2263,
+      "step": 2021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002720456414820683,
+      "loss": 0.7619,
+      "step": 2022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027192879741226984,
+      "loss": 1.241,
+      "step": 2023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002718119251382362,
+      "loss": 1.1146,
+      "step": 2024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002716950247057944,
+      "loss": 1.0139,
+      "step": 2025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027157809616078275,
+      "loss": 0.8616,
+      "step": 2026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002714611395490504,
+      "loss": 1.2791,
+      "step": 2027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002713441549164577,
+      "loss": 0.805,
+      "step": 2028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002712271423088758,
+      "loss": 1.0811,
+      "step": 2029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027111010177218695,
+      "loss": 1.0094,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027099303335228423,
+      "loss": 0.9681,
+      "step": 2031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002708759370950718,
+      "loss": 0.8405,
+      "step": 2032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002707588130464646,
+      "loss": 0.9985,
+      "step": 2033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002706416612523885,
+      "loss": 1.2022,
+      "step": 2034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002705244817587803,
+      "loss": 1.2114,
+      "step": 2035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027040727461158766,
+      "loss": 1.1754,
+      "step": 2036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000270290039856769,
+      "loss": 0.9899,
+      "step": 2037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027017277754029365,
+      "loss": 1.3421,
+      "step": 2038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027005548770814183,
+      "loss": 1.272,
+      "step": 2039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026993817040630425,
+      "loss": 1.3859,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002698208256807827,
+      "loss": 0.9226,
+      "step": 2041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026970345357758954,
+      "loss": 0.9758,
+      "step": 2042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026958605414274795,
+      "loss": 1.1357,
+      "step": 2043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026946862742229183,
+      "loss": 0.9858,
+      "step": 2044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002693511734622657,
+      "loss": 0.6967,
+      "step": 2045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002692336923087249,
+      "loss": 1.5573,
+      "step": 2046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002691161840077352,
+      "loss": 0.7966,
+      "step": 2047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002689986486053734,
+      "loss": 1.1407,
+      "step": 2048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026888108614772643,
+      "loss": 1.051,
+      "step": 2049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002687634966808923,
+      "loss": 1.0745,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026864588025097923,
+      "loss": 1.0048,
+      "step": 2051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026852823690410633,
+      "loss": 1.0496,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026841056668640295,
+      "loss": 1.2764,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026829286964400936,
+      "loss": 1.22,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026817514582307594,
+      "loss": 0.994,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002680573952697638,
+      "loss": 1.1003,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002679396180302446,
+      "loss": 1.0268,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002678218141507002,
+      "loss": 1.0004,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002677039836773232,
+      "loss": 1.1221,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002675861266563164,
+      "loss": 1.2957,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026746824313389315,
+      "loss": 1.0575,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002673503331562771,
+      "loss": 0.7385,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026723239676970244,
+      "loss": 0.9621,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002671144340204134,
+      "loss": 1.2269,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002669964449546649,
+      "loss": 1.385,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026687842961872196,
+      "loss": 0.7703,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002667603880588599,
+      "loss": 0.9737,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002666423203213644,
+      "loss": 0.849,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002665242264525315,
+      "loss": 1.0194,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026640610649866726,
+      "loss": 0.9812,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026628796050608806,
+      "loss": 1.1226,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002661697885211206,
+      "loss": 1.2003,
+      "step": 2072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026605159059010157,
+      "loss": 1.2205,
+      "step": 2073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026593336675937807,
+      "loss": 1.1378,
+      "step": 2074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002658151170753072,
+      "loss": 1.0892,
+      "step": 2075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026569684158425617,
+      "loss": 1.1819,
+      "step": 2076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026557854033260243,
+      "loss": 1.05,
+      "step": 2077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026546021336673343,
+      "loss": 1.238,
+      "step": 2078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002653418607330468,
+      "loss": 1.2561,
+      "step": 2079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026522348247795004,
+      "loss": 1.16,
+      "step": 2080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026510507864786103,
+      "loss": 1.1482,
+      "step": 2081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002649866492892073,
+      "loss": 0.8758,
+      "step": 2082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002648681944484267,
+      "loss": 1.2998,
+      "step": 2083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026474971417196684,
+      "loss": 0.9005,
+      "step": 2084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002646312085062855,
+      "loss": 1.0207,
+      "step": 2085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002645126774978503,
+      "loss": 0.9694,
+      "step": 2086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002643941211931388,
+      "loss": 1.0794,
+      "step": 2087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002642755396386385,
+      "loss": 1.2186,
+      "step": 2088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002641569328808468,
+      "loss": 1.0372,
+      "step": 2089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026403830096627097,
+      "loss": 1.156,
+      "step": 2090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002639196439414282,
+      "loss": 1.163,
+      "step": 2091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002638009618528455,
+      "loss": 0.9649,
+      "step": 2092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026368225474705955,
+      "loss": 0.8055,
+      "step": 2093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026356352267061715,
+      "loss": 1.1793,
+      "step": 2094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002634447656700746,
+      "loss": 0.9087,
+      "step": 2095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002633259837919982,
+      "loss": 0.8929,
+      "step": 2096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026320717708296386,
+      "loss": 1.0883,
+      "step": 2097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002630883455895571,
+      "loss": 1.0738,
+      "step": 2098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026296948935837364,
+      "loss": 1.1243,
+      "step": 2099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026285060843601833,
+      "loss": 0.8175,
+      "step": 2100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002627317028691061,
+      "loss": 1.0742,
+      "step": 2101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026261277270426127,
+      "loss": 0.9156,
+      "step": 2102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026249381798811803,
+      "loss": 0.9281,
+      "step": 2103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026237483876732014,
+      "loss": 0.9318,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026225583508852077,
+      "loss": 0.9941,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026213680699838305,
+      "loss": 0.852,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002620177545435793,
+      "loss": 1.0651,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026189867777079163,
+      "loss": 1.3337,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026177957672671165,
+      "loss": 1.1473,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026166045145804037,
+      "loss": 1.0604,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002615413020114885,
+      "loss": 0.8357,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026142212843377607,
+      "loss": 0.9014,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026130293077163266,
+      "loss": 1.0982,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026118370907179706,
+      "loss": 1.1886,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026106446338101794,
+      "loss": 1.3293,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026094519374605295,
+      "loss": 1.2219,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002608259002136694,
+      "loss": 1.0054,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002607065828306438,
+      "loss": 1.2241,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026058724164376213,
+      "loss": 0.7676,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002604678766998195,
+      "loss": 0.9879,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026034848804562066,
+      "loss": 0.9495,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002602290757279795,
+      "loss": 1.2082,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000260109639793719,
+      "loss": 1.094,
+      "step": 2123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002599901802896718,
+      "loss": 0.913,
+      "step": 2124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002598706972626793,
+      "loss": 0.9179,
+      "step": 2125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025975119075959254,
+      "loss": 1.2707,
+      "step": 2126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025963166082727156,
+      "loss": 1.0678,
+      "step": 2127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025951210751258574,
+      "loss": 1.1364,
+      "step": 2128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002593925308624133,
+      "loss": 1.1336,
+      "step": 2129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025927293092364205,
+      "loss": 1.1066,
+      "step": 2130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002591533077431686,
+      "loss": 0.8646,
+      "step": 2131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002590336613678988,
+      "loss": 1.0232,
+      "step": 2132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025891399184474755,
+      "loss": 0.9502,
+      "step": 2133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002587942992206389,
+      "loss": 1.2686,
+      "step": 2134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025867458354250595,
+      "loss": 0.9229,
+      "step": 2135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002585548448572907,
+      "loss": 0.9102,
+      "step": 2136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002584350832119444,
+      "loss": 1.0923,
+      "step": 2137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025831529865342703,
+      "loss": 1.0802,
+      "step": 2138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025819549122870785,
+      "loss": 1.186,
+      "step": 2139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025807566098476483,
+      "loss": 1.3048,
+      "step": 2140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002579558079685851,
+      "loss": 1.0435,
+      "step": 2141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002578359322271646,
+      "loss": 1.0595,
+      "step": 2142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000257716033807508,
+      "loss": 1.2227,
+      "step": 2143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002575961127566294,
+      "loss": 0.949,
+      "step": 2144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002574761691215511,
+      "loss": 0.7483,
+      "step": 2145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002573562029493049,
+      "loss": 0.8727,
+      "step": 2146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025723621428693085,
+      "loss": 0.7754,
+      "step": 2147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002571162031814783,
+      "loss": 1.264,
+      "step": 2148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000256996169680005,
+      "loss": 1.2148,
+      "step": 2149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002568761138295779,
+      "loss": 1.0773,
+      "step": 2150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025675603567727235,
+      "loss": 1.4511,
+      "step": 2151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002566359352701726,
+      "loss": 1.2917,
+      "step": 2152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025651581265537167,
+      "loss": 1.1503,
+      "step": 2153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002563956678799712,
+      "loss": 1.2741,
+      "step": 2154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025627550099108145,
+      "loss": 1.3428,
+      "step": 2155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002561553120358217,
+      "loss": 1.4006,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025603510106131936,
+      "loss": 1.3613,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002559148681147108,
+      "loss": 0.9244,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002557946132431411,
+      "loss": 0.6638,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002556743364937637,
+      "loss": 1.2266,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002555540379137407,
+      "loss": 0.6406,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025543371755024275,
+      "loss": 0.9717,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002553133754504491,
+      "loss": 1.0864,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002551930116615475,
+      "loss": 1.0465,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002550726262307341,
+      "loss": 1.2239,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025495221920521384,
+      "loss": 1.2375,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025483179063219964,
+      "loss": 1.0427,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025471134055891323,
+      "loss": 1.1334,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025459086903258473,
+      "loss": 0.9535,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002544703761004526,
+      "loss": 1.1023,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002543498618097637,
+      "loss": 0.9127,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002542293262077733,
+      "loss": 0.9361,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025410876934174506,
+      "loss": 0.9859,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002539881912589508,
+      "loss": 0.8568,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025386759200667097,
+      "loss": 1.181,
+      "step": 2175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025374697163219387,
+      "loss": 1.1822,
+      "step": 2176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025362633018281665,
+      "loss": 0.9788,
+      "step": 2177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025350566770584423,
+      "loss": 1.2893,
+      "step": 2178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025338498424858996,
+      "loss": 0.8091,
+      "step": 2179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025326427985837547,
+      "loss": 1.1077,
+      "step": 2180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025314355458253056,
+      "loss": 1.0921,
+      "step": 2181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025302280846839324,
+      "loss": 0.9424,
+      "step": 2182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002529020415633096,
+      "loss": 1.1646,
+      "step": 2183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000252781253914634,
+      "loss": 0.9859,
+      "step": 2184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025266044556972883,
+      "loss": 0.9365,
+      "step": 2185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002525396165759647,
+      "loss": 1.2517,
+      "step": 2186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002524187669807202,
+      "loss": 1.2585,
+      "step": 2187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025229789683138207,
+      "loss": 0.9987,
+      "step": 2188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025217700617534514,
+      "loss": 1.2451,
+      "step": 2189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002520560950600122,
+      "loss": 0.9259,
+      "step": 2190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002519351635327942,
+      "loss": 1.2296,
+      "step": 2191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002518142116411098,
+      "loss": 1.0644,
+      "step": 2192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025169323943238596,
+      "loss": 1.0655,
+      "step": 2193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002515722469540575,
+      "loss": 0.9837,
+      "step": 2194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025145123425356715,
+      "loss": 1.1543,
+      "step": 2195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025133020137836566,
+      "loss": 1.2688,
+      "step": 2196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002512091483759115,
+      "loss": 1.2254,
+      "step": 2197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025108807529367126,
+      "loss": 0.7872,
+      "step": 2198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002509669821791192,
+      "loss": 1.1694,
+      "step": 2199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002508458690797377,
+      "loss": 0.8909,
+      "step": 2200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002507247360430166,
+      "loss": 0.9912,
+      "step": 2201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025060358311645395,
+      "loss": 1.1244,
+      "step": 2202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002504824103475553,
+      "loss": 1.149,
+      "step": 2203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002503612177838342,
+      "loss": 1.0098,
+      "step": 2204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025024000547281175,
+      "loss": 0.9991,
+      "step": 2205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000250118773462017,
+      "loss": 0.9325,
+      "step": 2206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024999752179898657,
+      "loss": 1.0532,
+      "step": 2207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002498762505312649,
+      "loss": 0.8987,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000249754959706404,
+      "loss": 1.1424,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024963364937196367,
+      "loss": 0.8624,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024951231957551126,
+      "loss": 1.167,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002493909703646218,
+      "loss": 1.3615,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024926960178687797,
+      "loss": 1.0524,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002491482138898699,
+      "loss": 0.9907,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024902680672119544,
+      "loss": 1.1215,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024890538032846,
+      "loss": 1.2177,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002487839347592764,
+      "loss": 1.1604,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002486624700612652,
+      "loss": 1.0918,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024854098628205413,
+      "loss": 1.0282,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002484194834692787,
+      "loss": 1.1321,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002482979616705818,
+      "loss": 1.2481,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024817642093361367,
+      "loss": 1.0024,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024805486130603197,
+      "loss": 1.0367,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000247933282835502,
+      "loss": 1.0316,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002478116855696963,
+      "loss": 0.921,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024769006955629454,
+      "loss": 0.9671,
+      "step": 2226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024756843484298425,
+      "loss": 1.0784,
+      "step": 2227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024744678147745985,
+      "loss": 1.2268,
+      "step": 2228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002473251095074233,
+      "loss": 1.2492,
+      "step": 2229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002472034189805838,
+      "loss": 1.0791,
+      "step": 2230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002470817099446578,
+      "loss": 1.0747,
+      "step": 2231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024695998244736895,
+      "loss": 1.0206,
+      "step": 2232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002468382365364483,
+      "loss": 0.9164,
+      "step": 2233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002467164722596341,
+      "loss": 0.9939,
+      "step": 2234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024659468966467155,
+      "loss": 1.3494,
+      "step": 2235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002464728887993134,
+      "loss": 0.8876,
+      "step": 2236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002463510697113193,
+      "loss": 0.9407,
+      "step": 2237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024622923244845614,
+      "loss": 1.2143,
+      "step": 2238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024610737705849794,
+      "loss": 1.1491,
+      "step": 2239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024598550358922576,
+      "loss": 1.0158,
+      "step": 2240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002458636120884279,
+      "loss": 1.1725,
+      "step": 2241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002457417026038996,
+      "loss": 0.9844,
+      "step": 2242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002456197751834432,
+      "loss": 1.1548,
+      "step": 2243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002454978298748679,
+      "loss": 1.0112,
+      "step": 2244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002453758667259903,
+      "loss": 1.0766,
+      "step": 2245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002452538857846336,
+      "loss": 1.2333,
+      "step": 2246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002451318870986282,
+      "loss": 0.9545,
+      "step": 2247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024500987071581143,
+      "loss": 1.1814,
+      "step": 2248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002448878366840276,
+      "loss": 0.8917,
+      "step": 2249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002447657850511277,
+      "loss": 1.2634,
+      "step": 2250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002446437158649699,
+      "loss": 1.0672,
+      "step": 2251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024452162917341913,
+      "loss": 0.8912,
+      "step": 2252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024439952502434715,
+      "loss": 0.9084,
+      "step": 2253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024427740346563275,
+      "loss": 1.003,
+      "step": 2254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002441552645451613,
+      "loss": 1.0608,
+      "step": 2255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002440331083108251,
+      "loss": 1.2979,
+      "step": 2256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002439109348105232,
+      "loss": 1.1966,
+      "step": 2257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002437887440921616,
+      "loss": 0.9826,
+      "step": 2258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002436665362036527,
+      "loss": 1.2248,
+      "step": 2259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000243544311192916,
+      "loss": 1.1589,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024342206910787746,
+      "loss": 1.1668,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024329980999646984,
+      "loss": 0.8884,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024317753390663255,
+      "loss": 0.9489,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024305524088631165,
+      "loss": 1.082,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024293293098345996,
+      "loss": 1.1266,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024281060424603666,
+      "loss": 1.0658,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024268826072200778,
+      "loss": 1.397,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024256590045934577,
+      "loss": 1.255,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024244352350602982,
+      "loss": 1.3824,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002423211299100455,
+      "loss": 1.0225,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002421987197193849,
+      "loss": 1.1222,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002420762929820468,
+      "loss": 1.0661,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024195384974603625,
+      "loss": 1.3056,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002418313900593649,
+      "loss": 1.057,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024170891397005083,
+      "loss": 1.3366,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024158642152611855,
+      "loss": 0.8297,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024146391277559887,
+      "loss": 1.0523,
+      "step": 2277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002413413877665293,
+      "loss": 0.782,
+      "step": 2278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024121884654695327,
+      "loss": 1.0528,
+      "step": 2279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024109628916492098,
+      "loss": 1.0036,
+      "step": 2280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024097371566848876,
+      "loss": 1.025,
+      "step": 2281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002408511261057193,
+      "loss": 0.977,
+      "step": 2282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024072852052468168,
+      "loss": 0.9908,
+      "step": 2283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024060589897345106,
+      "loss": 1.0917,
+      "step": 2284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024048326150010897,
+      "loss": 1.1871,
+      "step": 2285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002403606081527433,
+      "loss": 1.0432,
+      "step": 2286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024023793897944793,
+      "loss": 0.8926,
+      "step": 2287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024011525402832314,
+      "loss": 1.1033,
+      "step": 2288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002399925533474753,
+      "loss": 1.3768,
+      "step": 2289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023986983698501702,
+      "loss": 0.8337,
+      "step": 2290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002397471049890669,
+      "loss": 1.3518,
+      "step": 2291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023962435740774996,
+      "loss": 1.0956,
+      "step": 2292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000239501594289197,
+      "loss": 0.8483,
+      "step": 2293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023937881568154512,
+      "loss": 1.0832,
+      "step": 2294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023925602163293747,
+      "loss": 1.0459,
+      "step": 2295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023913321219152316,
+      "loss": 0.8775,
+      "step": 2296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023901038740545744,
+      "loss": 1.1689,
+      "step": 2297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023888754732290156,
+      "loss": 1.0406,
+      "step": 2298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023876469199202274,
+      "loss": 1.386,
+      "step": 2299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002386418214609941,
+      "loss": 1.2231,
+      "step": 2300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023851893577799492,
+      "loss": 1.0372,
+      "step": 2301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023839603499121022,
+      "loss": 1.1783,
+      "step": 2302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023827311914883113,
+      "loss": 1.0659,
+      "step": 2303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023815018829905445,
+      "loss": 1.0558,
+      "step": 2304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023802724249008308,
+      "loss": 1.269,
+      "step": 2305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023790428177012567,
+      "loss": 0.6568,
+      "step": 2306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002377813061873968,
+      "loss": 1.0937,
+      "step": 2307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002376583157901168,
+      "loss": 1.1094,
+      "step": 2308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023753531062651177,
+      "loss": 1.1074,
+      "step": 2309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023741229074481375,
+      "loss": 1.0695,
+      "step": 2310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023728925619326038,
+      "loss": 1.134,
+      "step": 2311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002371662070200953,
+      "loss": 1.4381,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023704314327356753,
+      "loss": 0.9683,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023692006500193214,
+      "loss": 1.4302,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023679697225344963,
+      "loss": 1.1233,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023667386507638644,
+      "loss": 1.2956,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002365507435190144,
+      "loss": 1.1549,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023642760762961117,
+      "loss": 0.9031,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023630445745645995,
+      "loss": 1.058,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023618129304784957,
+      "loss": 0.8668,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023605811445207443,
+      "loss": 0.8558,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023593492171743443,
+      "loss": 1.1169,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002358117148922352,
+      "loss": 1.164,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023568849402478761,
+      "loss": 1.0799,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023556525916340832,
+      "loss": 0.8008,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002354420103564193,
+      "loss": 1.0951,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023531874765214813,
+      "loss": 0.9116,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023519547109892756,
+      "loss": 1.0711,
+      "step": 2328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023507218074509612,
+      "loss": 0.9732,
+      "step": 2329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023494887663899757,
+      "loss": 1.1542,
+      "step": 2330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023482555882898103,
+      "loss": 1.1763,
+      "step": 2331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002347022273634011,
+      "loss": 1.1371,
+      "step": 2332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023457888229061764,
+      "loss": 0.849,
+      "step": 2333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023445552365899594,
+      "loss": 0.9951,
+      "step": 2334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023433215151690652,
+      "loss": 1.1022,
+      "step": 2335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023420876591272521,
+      "loss": 1.0108,
+      "step": 2336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023408536689483326,
+      "loss": 0.9153,
+      "step": 2337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023396195451161686,
+      "loss": 1.0673,
+      "step": 2338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002338385288114678,
+      "loss": 1.1065,
+      "step": 2339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002337150898427829,
+      "loss": 1.1687,
+      "step": 2340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023359163765396422,
+      "loss": 0.9889,
+      "step": 2341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023346817229341897,
+      "loss": 1.2066,
+      "step": 2342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023334469380955958,
+      "loss": 0.9218,
+      "step": 2343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023322120225080357,
+      "loss": 1.1352,
+      "step": 2344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023309769766557362,
+      "loss": 1.099,
+      "step": 2345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002329741801022976,
+      "loss": 0.989,
+      "step": 2346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023285064960940828,
+      "loss": 1.2701,
+      "step": 2347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023272710623534364,
+      "loss": 0.7675,
+      "step": 2348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023260355002854675,
+      "loss": 1.1325,
+      "step": 2349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023247998103746546,
+      "loss": 0.9569,
+      "step": 2350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000232356399310553,
+      "loss": 0.9819,
+      "step": 2351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023223280489626728,
+      "loss": 1.3567,
+      "step": 2352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023210919784307134,
+      "loss": 1.0391,
+      "step": 2353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002319855781994331,
+      "loss": 0.8134,
+      "step": 2354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023186194601382566,
+      "loss": 1.2107,
+      "step": 2355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023173830133472656,
+      "loss": 0.8995,
+      "step": 2356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002316146442106187,
+      "loss": 0.9763,
+      "step": 2357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023149097468998963,
+      "loss": 1.2539,
+      "step": 2358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023136729282133183,
+      "loss": 1.1668,
+      "step": 2359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023124359865314253,
+      "loss": 1.1974,
+      "step": 2360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023111989223392399,
+      "loss": 1.0039,
+      "step": 2361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023099617361218297,
+      "loss": 1.0129,
+      "step": 2362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023087244283643128,
+      "loss": 1.0777,
+      "step": 2363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002307486999551854,
+      "loss": 1.1532,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002306249450169665,
+      "loss": 1.1833,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023050117807030062,
+      "loss": 1.2064,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023037739916371833,
+      "loss": 1.4815,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023025360834575502,
+      "loss": 1.1197,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023012980566495075,
+      "loss": 0.6957,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023000599116985014,
+      "loss": 0.9597,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022988216490900246,
+      "loss": 1.3588,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022975832693096173,
+      "loss": 1.1078,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002296344772842864,
+      "loss": 1.3443,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022951061601753954,
+      "loss": 1.0311,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022938674317928888,
+      "loss": 0.9202,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002292628588181065,
+      "loss": 1.0661,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002291389629825692,
+      "loss": 1.3191,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022901505572125804,
+      "loss": 1.2954,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00022889113708275886,
+      "loss": 1.019,
+      "step": 2379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002287672071156617,
+      "loss": 1.0772,
+      "step": 2380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002286432658685612,
+      "loss": 0.8981,
+      "step": 2381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022851931339005628,
+      "loss": 0.9573,
+      "step": 2382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002283953497287504,
+      "loss": 1.0109,
+      "step": 2383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022827137493325146,
+      "loss": 1.1061,
+      "step": 2384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022814738905217144,
+      "loss": 0.9766,
+      "step": 2385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000228023392134127,
+      "loss": 0.7921,
+      "step": 2386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022789938422773887,
+      "loss": 1.2162,
+      "step": 2387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022777536538163228,
+      "loss": 1.0376,
+      "step": 2388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022765133564443653,
+      "loss": 0.994,
+      "step": 2389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002275272950647855,
+      "loss": 0.9987,
+      "step": 2390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022740324369131698,
+      "loss": 1.0703,
+      "step": 2391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002272791815726733,
+      "loss": 1.1782,
+      "step": 2392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002271551087575008,
+      "loss": 1.1469,
+      "step": 2393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002270310252944501,
+      "loss": 1.1998,
+      "step": 2394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022690693123217587,
+      "loss": 1.0128,
+      "step": 2395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022678282661933716,
+      "loss": 1.2134,
+      "step": 2396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022665871150459702,
+      "loss": 0.9407,
+      "step": 2397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002265345859366225,
+      "loss": 1.298,
+      "step": 2398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022641044996408506,
+      "loss": 1.1144,
+      "step": 2399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022628630363565996,
+      "loss": 1.3551,
+      "step": 2400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022616214700002662,
+      "loss": 1.1346,
+      "step": 2401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022603798010586857,
+      "loss": 1.2923,
+      "step": 2402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022591380300187322,
+      "loss": 0.9676,
+      "step": 2403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002257896157367321,
+      "loss": 1.274,
+      "step": 2404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022566541835914068,
+      "loss": 0.8818,
+      "step": 2405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022554121091779842,
+      "loss": 0.9405,
+      "step": 2406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022541699346140866,
+      "loss": 1.032,
+      "step": 2407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022529276603867876,
+      "loss": 1.0623,
+      "step": 2408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022516852869831983,
+      "loss": 0.9803,
+      "step": 2409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022504428148904711,
+      "loss": 1.3313,
+      "step": 2410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022492002445957953,
+      "loss": 0.956,
+      "step": 2411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022479575765863992,
+      "loss": 1.0387,
+      "step": 2412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022467148113495492,
+      "loss": 1.1936,
+      "step": 2413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022454719493725508,
+      "loss": 1.1131,
+      "step": 2414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002244228991142745,
+      "loss": 0.7885,
+      "step": 2415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022429859371475143,
+      "loss": 0.8884,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022417427878742752,
+      "loss": 1.1895,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022404995438104832,
+      "loss": 1.0044,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022392562054436308,
+      "loss": 1.2339,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022380127732612483,
+      "loss": 1.1215,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002236769247750901,
+      "loss": 1.0418,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022355256294001922,
+      "loss": 0.8339,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022342819186967606,
+      "loss": 1.276,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022330381161282827,
+      "loss": 0.8948,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022317942221824688,
+      "loss": 1.2172,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022305502373470674,
+      "loss": 0.9743,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000222930616210986,
+      "loss": 1.1331,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022280619969586668,
+      "loss": 0.9679,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002226817742381339,
+      "loss": 0.9826,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002225573398865768,
+      "loss": 1.1312,
+      "step": 2430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022243289668998756,
+      "loss": 0.9792,
+      "step": 2431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022230844469716207,
+      "loss": 0.82,
+      "step": 2432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022218398395689967,
+      "loss": 1.438,
+      "step": 2433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022205951451800292,
+      "loss": 0.851,
+      "step": 2434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022193503642927803,
+      "loss": 1.0124,
+      "step": 2435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002218105497395345,
+      "loss": 1.1075,
+      "step": 2436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022168605449758523,
+      "loss": 0.9231,
+      "step": 2437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022156155075224644,
+      "loss": 0.746,
+      "step": 2438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022143703855233767,
+      "loss": 1.0928,
+      "step": 2439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002213125179466818,
+      "loss": 1.0086,
+      "step": 2440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022118798898410505,
+      "loss": 0.8647,
+      "step": 2441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022106345171343689,
+      "loss": 1.2227,
+      "step": 2442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022093890618350996,
+      "loss": 0.9385,
+      "step": 2443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002208143524431603,
+      "loss": 0.997,
+      "step": 2444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022068979054122703,
+      "loss": 0.758,
+      "step": 2445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022056522052655252,
+      "loss": 0.9134,
+      "step": 2446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002204406424479824,
+      "loss": 1.0113,
+      "step": 2447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022031605635436523,
+      "loss": 1.0079,
+      "step": 2448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022019146229455307,
+      "loss": 1.1407,
+      "step": 2449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022006686031740075,
+      "loss": 0.7868,
+      "step": 2450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021994225047176644,
+      "loss": 1.1032,
+      "step": 2451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021981763280651127,
+      "loss": 0.9084,
+      "step": 2452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021969300737049953,
+      "loss": 1.1133,
+      "step": 2453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021956837421259836,
+      "loss": 1.1485,
+      "step": 2454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002194437333816783,
+      "loss": 1.11,
+      "step": 2455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002193190849266125,
+      "loss": 1.2362,
+      "step": 2456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021919442889627733,
+      "loss": 0.9634,
+      "step": 2457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021906976533955205,
+      "loss": 1.1286,
+      "step": 2458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021894509430531896,
+      "loss": 0.8068,
+      "step": 2459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002188204158424632,
+      "loss": 1.0835,
+      "step": 2460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021869572999987273,
+      "loss": 1.0976,
+      "step": 2461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021857103682643874,
+      "loss": 1.0526,
+      "step": 2462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021844633637105486,
+      "loss": 1.1549,
+      "step": 2463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021832162868261793,
+      "loss": 1.036,
+      "step": 2464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002181969138100274,
+      "loss": 1.061,
+      "step": 2465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021807219180218566,
+      "loss": 0.8843,
+      "step": 2466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021794746270799794,
+      "loss": 1.1464,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021782272657637207,
+      "loss": 1.087,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021769798345621882,
+      "loss": 1.2141,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021757323339645154,
+      "loss": 1.1745,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002174484764459864,
+      "loss": 0.6856,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002173237126537423,
+      "loss": 1.1851,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002171989420686408,
+      "loss": 1.2246,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000217074164739606,
+      "loss": 1.2249,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021694938071556484,
+      "loss": 1.1122,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021682459004544668,
+      "loss": 1.172,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021669979277818367,
+      "loss": 1.0328,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021657498896271053,
+      "loss": 1.2813,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021645017864796445,
+      "loss": 1.0391,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021632536188288514,
+      "loss": 1.0117,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021620053871641498,
+      "loss": 1.064,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002160757091974988,
+      "loss": 1.0461,
+      "step": 2482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021595087337508378,
+      "loss": 1.0514,
+      "step": 2483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021582603129811987,
+      "loss": 1.1983,
+      "step": 2484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021570118301555922,
+      "loss": 1.0261,
+      "step": 2485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021557632857635648,
+      "loss": 0.7988,
+      "step": 2486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002154514680294688,
+      "loss": 1.1832,
+      "step": 2487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021532660142385557,
+      "loss": 1.1928,
+      "step": 2488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021520172880847875,
+      "loss": 1.3567,
+      "step": 2489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021507685023230245,
+      "loss": 0.914,
+      "step": 2490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021495196574429327,
+      "loss": 1.0438,
+      "step": 2491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021482707539342006,
+      "loss": 0.9789,
+      "step": 2492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021470217922865396,
+      "loss": 0.7321,
+      "step": 2493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021457727729896843,
+      "loss": 1.1809,
+      "step": 2494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021445236965333915,
+      "loss": 0.9301,
+      "step": 2495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021432745634074416,
+      "loss": 1.0986,
+      "step": 2496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002142025374101635,
+      "loss": 0.9936,
+      "step": 2497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021407761291057962,
+      "loss": 1.1325,
+      "step": 2498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021395268289097703,
+      "loss": 1.0194,
+      "step": 2499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002138277474003425,
+      "loss": 1.1299,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002137028064876648,
+      "loss": 1.1452,
+      "step": 2501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021357786020193502,
+      "loss": 0.9838,
+      "step": 2502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021345290859214622,
+      "loss": 1.0987,
+      "step": 2503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002133279517072935,
+      "loss": 0.9274,
+      "step": 2504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021320298959637426,
+      "loss": 1.2765,
+      "step": 2505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002130780223083877,
+      "loss": 0.7157,
+      "step": 2506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021295304989233516,
+      "loss": 0.9498,
+      "step": 2507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021282807239722003,
+      "loss": 1.0345,
+      "step": 2508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002127030898720476,
+      "loss": 1.1004,
+      "step": 2509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021257810236582515,
+      "loss": 0.9387,
+      "step": 2510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021245310992756204,
+      "loss": 0.9655,
+      "step": 2511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021232811260626932,
+      "loss": 1.1096,
+      "step": 2512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021220311045096017,
+      "loss": 0.9403,
+      "step": 2513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002120781035106497,
+      "loss": 0.9095,
+      "step": 2514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002119530918343546,
+      "loss": 1.0219,
+      "step": 2515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021182807547109374,
+      "loss": 1.0341,
+      "step": 2516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002117030544698876,
+      "loss": 0.9877,
+      "step": 2517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002115780288797587,
+      "loss": 1.0319,
+      "step": 2518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021145299874973102,
+      "loss": 1.3868,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021132796412883079,
+      "loss": 1.1743,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021120292506608554,
+      "loss": 1.1115,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021107788161052487,
+      "loss": 1.2227,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021095283381117986,
+      "loss": 0.9228,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021082778171708353,
+      "loss": 0.8227,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021070272537727043,
+      "loss": 1.1453,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002105776648407767,
+      "loss": 1.0917,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021045260015664045,
+      "loss": 1.0859,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021032753137390094,
+      "loss": 1.1659,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021020245854159954,
+      "loss": 0.8206,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002100773817087788,
+      "loss": 1.2238,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020995230092448307,
+      "loss": 1.1536,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002098272162377582,
+      "loss": 1.0671,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002097021276976515,
+      "loss": 1.1933,
+      "step": 2533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002095770353532119,
+      "loss": 1.3679,
+      "step": 2534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020945193925348972,
+      "loss": 1.0052,
+      "step": 2535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002093268394475369,
+      "loss": 1.0543,
+      "step": 2536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020920173598440656,
+      "loss": 0.9233,
+      "step": 2537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002090766289131536,
+      "loss": 0.8987,
+      "step": 2538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020895151828283405,
+      "loss": 0.8496,
+      "step": 2539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020882640414250552,
+      "loss": 0.7962,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020870128654122683,
+      "loss": 1.0423,
+      "step": 2541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002085761655280583,
+      "loss": 1.1098,
+      "step": 2542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020845104115206162,
+      "loss": 0.9147,
+      "step": 2543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002083259134622996,
+      "loss": 1.1105,
+      "step": 2544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002082007825078365,
+      "loss": 1.2088,
+      "step": 2545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020807564833773788,
+      "loss": 0.8453,
+      "step": 2546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020795051100107044,
+      "loss": 1.0888,
+      "step": 2547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020782537054690216,
+      "loss": 1.0989,
+      "step": 2548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020770022702430239,
+      "loss": 1.1718,
+      "step": 2549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020757508048234143,
+      "loss": 1.4114,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020744993097009101,
+      "loss": 1.0248,
+      "step": 2551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020732477853662392,
+      "loss": 1.0107,
+      "step": 2552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020719962323101392,
+      "loss": 1.2108,
+      "step": 2553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020707446510233627,
+      "loss": 0.904,
+      "step": 2554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020694930419966703,
+      "loss": 1.0979,
+      "step": 2555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020682414057208344,
+      "loss": 1.4337,
+      "step": 2556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020669897426866385,
+      "loss": 1.0177,
+      "step": 2557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020657380533848767,
+      "loss": 0.689,
+      "step": 2558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020644863383063514,
+      "loss": 1.4163,
+      "step": 2559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020632345979418784,
+      "loss": 0.9922,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020619828327822815,
+      "loss": 1.041,
+      "step": 2561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020607310433183931,
+      "loss": 1.008,
+      "step": 2562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002059479230041058,
+      "loss": 1.1688,
+      "step": 2563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020582273934411277,
+      "loss": 0.8632,
+      "step": 2564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020569755340094648,
+      "loss": 0.983,
+      "step": 2565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020557236522369387,
+      "loss": 0.9046,
+      "step": 2566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020544717486144305,
+      "loss": 0.9726,
+      "step": 2567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020532198236328271,
+      "loss": 0.763,
+      "step": 2568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002051967877783025,
+      "loss": 1.1688,
+      "step": 2569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002050715911555929,
+      "loss": 0.8871,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020494639254424514,
+      "loss": 1.3052,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020482119199335133,
+      "loss": 0.7519,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002046959895520041,
+      "loss": 0.8914,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002045707852692972,
+      "loss": 1.0223,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020444557919432473,
+      "loss": 1.0231,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020432037137618173,
+      "loss": 1.0298,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020419516186396383,
+      "loss": 1.3097,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002040699507067673,
+      "loss": 0.9631,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002039447379536892,
+      "loss": 1.1985,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020381952365382695,
+      "loss": 0.9621,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002036943078562789,
+      "loss": 0.9897,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020356909061014373,
+      "loss": 1.1119,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002034438719645208,
+      "loss": 1.0112,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020331865196851002,
+      "loss": 1.2338,
+      "step": 2584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002031934306712118,
+      "loss": 1.0586,
+      "step": 2585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020306820812172703,
+      "loss": 0.9987,
+      "step": 2586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020294298436915721,
+      "loss": 1.1103,
+      "step": 2587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020281775946260418,
+      "loss": 0.8503,
+      "step": 2588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002026925334511703,
+      "loss": 0.985,
+      "step": 2589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020256730638395837,
+      "loss": 0.8247,
+      "step": 2590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020244207831007156,
+      "loss": 0.9893,
+      "step": 2591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020231684927861346,
+      "loss": 1.0251,
+      "step": 2592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000202191619338688,
+      "loss": 0.7957,
+      "step": 2593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002020663885393996,
+      "loss": 0.959,
+      "step": 2594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002019411569298528,
+      "loss": 1.197,
+      "step": 2595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020181592455915267,
+      "loss": 0.9681,
+      "step": 2596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020169069147640438,
+      "loss": 1.1268,
+      "step": 2597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020156545773071356,
+      "loss": 1.1385,
+      "step": 2598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020144022337118602,
+      "loss": 1.0843,
+      "step": 2599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002013149884469278,
+      "loss": 0.8827,
+      "step": 2600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002011897530070452,
+      "loss": 0.9899,
+      "step": 2601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020106451710064462,
+      "loss": 0.9271,
+      "step": 2602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002009392807768328,
+      "loss": 0.7408,
+      "step": 2603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020081404408471651,
+      "loss": 0.9497,
+      "step": 2604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020068880707340277,
+      "loss": 0.8931,
+      "step": 2605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020056356979199862,
+      "loss": 0.9711,
+      "step": 2606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002004383322896113,
+      "loss": 0.932,
+      "step": 2607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020031309461534806,
+      "loss": 0.8938,
+      "step": 2608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020018785681831627,
+      "loss": 1.0464,
+      "step": 2609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020006261894762333,
+      "loss": 0.8854,
+      "step": 2610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001999373810523767,
+      "loss": 1.214,
+      "step": 2611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019981214318168374,
+      "loss": 1.0089,
+      "step": 2612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019968690538465201,
+      "loss": 1.1469,
+      "step": 2613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019956166771038878,
+      "loss": 0.9504,
+      "step": 2614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019943643020800143,
+      "loss": 1.0496,
+      "step": 2615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019931119292659725,
+      "loss": 0.9934,
+      "step": 2616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019918595591528348,
+      "loss": 1.2315,
+      "step": 2617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019906071922316728,
+      "loss": 1.0322,
+      "step": 2618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019893548289935545,
+      "loss": 1.1078,
+      "step": 2619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019881024699295485,
+      "loss": 0.982,
+      "step": 2620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001986850115530722,
+      "loss": 0.9865,
+      "step": 2621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019855977662881405,
+      "loss": 1.0762,
+      "step": 2622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001984345422692865,
+      "loss": 0.9865,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019830930852359567,
+      "loss": 1.1148,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019818407544084737,
+      "loss": 1.0254,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001980588430701472,
+      "loss": 1.0396,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019793361146060049,
+      "loss": 0.951,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019780838066131204,
+      "loss": 1.1584,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019768315072138655,
+      "loss": 0.8997,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001975579216899285,
+      "loss": 1.144,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019743269361604168,
+      "loss": 1.0206,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019730746654882973,
+      "loss": 1.0708,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019718224053739587,
+      "loss": 1.1143,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019705701563084283,
+      "loss": 0.7961,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019693179187827296,
+      "loss": 1.0246,
+      "step": 2635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001968065693287883,
+      "loss": 0.7893,
+      "step": 2636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019668134803149003,
+      "loss": 1.2321,
+      "step": 2637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019655612803547924,
+      "loss": 0.8603,
+      "step": 2638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019643090938985627,
+      "loss": 0.8044,
+      "step": 2639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019630569214372116,
+      "loss": 0.8703,
+      "step": 2640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019618047634617307,
+      "loss": 1.0229,
+      "step": 2641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019605526204631086,
+      "loss": 1.005,
+      "step": 2642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001959300492932327,
+      "loss": 0.9907,
+      "step": 2643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019580483813603616,
+      "loss": 1.0598,
+      "step": 2644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019567962862381832,
+      "loss": 1.1644,
+      "step": 2645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001955544208056753,
+      "loss": 0.9835,
+      "step": 2646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019542921473070284,
+      "loss": 0.9931,
+      "step": 2647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001953040104479959,
+      "loss": 1.1263,
+      "step": 2648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001951788080066488,
+      "loss": 1.0825,
+      "step": 2649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001950536074557549,
+      "loss": 0.9895,
+      "step": 2650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019492840884440714,
+      "loss": 1.2983,
+      "step": 2651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019480321222169756,
+      "loss": 0.965,
+      "step": 2652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019467801763671733,
+      "loss": 1.0286,
+      "step": 2653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000194552825138557,
+      "loss": 0.9459,
+      "step": 2654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019442763477630617,
+      "loss": 0.9936,
+      "step": 2655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001943024465990536,
+      "loss": 0.96,
+      "step": 2656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019417726065588725,
+      "loss": 0.7675,
+      "step": 2657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019405207699589428,
+      "loss": 0.9456,
+      "step": 2658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019392689566816073,
+      "loss": 1.1433,
+      "step": 2659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019380171672177192,
+      "loss": 0.9955,
+      "step": 2660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019367654020581215,
+      "loss": 1.0678,
+      "step": 2661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019355136616936483,
+      "loss": 0.9411,
+      "step": 2662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019342619466151243,
+      "loss": 0.9527,
+      "step": 2663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001933010257313362,
+      "loss": 0.9859,
+      "step": 2664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019317585942791658,
+      "loss": 1.2407,
+      "step": 2665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000193050695800333,
+      "loss": 1.0149,
+      "step": 2666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001929255348976638,
+      "loss": 1.1105,
+      "step": 2667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019280037676898612,
+      "loss": 1.1362,
+      "step": 2668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019267522146337616,
+      "loss": 0.9875,
+      "step": 2669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000192550069029909,
+      "loss": 1.0704,
+      "step": 2670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019242491951765864,
+      "loss": 0.9149,
+      "step": 2671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001922997729756977,
+      "loss": 0.9228,
+      "step": 2672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001921746294530979,
+      "loss": 0.939,
+      "step": 2673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001920494889989296,
+      "loss": 1.2366,
+      "step": 2674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019192435166226216,
+      "loss": 0.9275,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019179921749216353,
+      "loss": 1.3046,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019167408653770042,
+      "loss": 1.1112,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019154895884793843,
+      "loss": 0.6505,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001914238344719417,
+      "loss": 1.2205,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019129871345877322,
+      "loss": 0.8804,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019117359585749458,
+      "loss": 0.8298,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000191048481717166,
+      "loss": 0.8238,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019092337108684643,
+      "loss": 1.2675,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019079826401559343,
+      "loss": 1.0072,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001906731605524632,
+      "loss": 1.266,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001905480607465103,
+      "loss": 0.725,
+      "step": 2686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019042296464678814,
+      "loss": 1.0718,
+      "step": 2687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019029787230234852,
+      "loss": 0.9819,
+      "step": 2688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001901727837622419,
+      "loss": 1.2234,
+      "step": 2689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000190047699075517,
+      "loss": 0.8758,
+      "step": 2690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018992261829122126,
+      "loss": 1.2579,
+      "step": 2691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001897975414584005,
+      "loss": 1.123,
+      "step": 2692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018967246862609905,
+      "loss": 1.2403,
+      "step": 2693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018954739984335968,
+      "loss": 1.2298,
+      "step": 2694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001894223351592233,
+      "loss": 1.0908,
+      "step": 2695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018929727462272964,
+      "loss": 0.9907,
+      "step": 2696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018917221828291652,
+      "loss": 0.8568,
+      "step": 2697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001890471661888202,
+      "loss": 0.809,
+      "step": 2698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001889221183894752,
+      "loss": 1.0874,
+      "step": 2699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018879707493391453,
+      "loss": 0.9671,
+      "step": 2700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018867203587116926,
+      "loss": 1.2625,
+      "step": 2701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018854700125026897,
+      "loss": 0.7871,
+      "step": 2702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018842197112024144,
+      "loss": 0.9994,
+      "step": 2703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018829694553011245,
+      "loss": 1.2119,
+      "step": 2704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001881719245289063,
+      "loss": 1.3324,
+      "step": 2705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001880469081656454,
+      "loss": 1.3412,
+      "step": 2706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018792189648935038,
+      "loss": 1.0707,
+      "step": 2707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018779688954903985,
+      "loss": 1.0899,
+      "step": 2708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018767188739373072,
+      "loss": 1.2452,
+      "step": 2709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018754689007243803,
+      "loss": 1.2617,
+      "step": 2710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018742189763417487,
+      "loss": 1.1084,
+      "step": 2711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018729691012795249,
+      "loss": 0.9204,
+      "step": 2712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018717192760278004,
+      "loss": 1.1042,
+      "step": 2713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018704695010766486,
+      "loss": 1.465,
+      "step": 2714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018692197769161232,
+      "loss": 1.279,
+      "step": 2715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018679701040362582,
+      "loss": 0.9906,
+      "step": 2716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018667204829270654,
+      "loss": 0.8899,
+      "step": 2717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018654709140785385,
+      "loss": 1.0926,
+      "step": 2718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018642213979806502,
+      "loss": 1.2343,
+      "step": 2719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001862971935123352,
+      "loss": 0.976,
+      "step": 2720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018617225259965757,
+      "loss": 1.0711,
+      "step": 2721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018604731710902304,
+      "loss": 0.8428,
+      "step": 2722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018592238708942043,
+      "loss": 1.1541,
+      "step": 2723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001857974625898365,
+      "loss": 0.9392,
+      "step": 2724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001856725436592559,
+      "loss": 0.9172,
+      "step": 2725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018554763034666087,
+      "loss": 1.2735,
+      "step": 2726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018542272270103162,
+      "loss": 1.1998,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018529782077134608,
+      "loss": 1.3102,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018517292460657996,
+      "loss": 0.9294,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001850480342557068,
+      "loss": 0.8987,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018492314976769757,
+      "loss": 0.9117,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001847982711915213,
+      "loss": 1.1942,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018467339857614442,
+      "loss": 0.8924,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018454853197053127,
+      "loss": 0.9767,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018442367142364356,
+      "loss": 1.1391,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018429881698444082,
+      "loss": 1.1651,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018417396870188014,
+      "loss": 0.9439,
+      "step": 2737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018404912662491624,
+      "loss": 1.0174,
+      "step": 2738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001839242908025013,
+      "loss": 1.0259,
+      "step": 2739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018379946128358507,
+      "loss": 1.0512,
+      "step": 2740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001836746381171149,
+      "loss": 1.3568,
+      "step": 2741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001835498213520356,
+      "loss": 0.7263,
+      "step": 2742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018342501103728952,
+      "loss": 1.1328,
+      "step": 2743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018330020722181635,
+      "loss": 0.93,
+      "step": 2744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018317540995455337,
+      "loss": 1.0267,
+      "step": 2745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018305061928443523,
+      "loss": 1.0039,
+      "step": 2746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000182925835260394,
+      "loss": 1.3492,
+      "step": 2747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018280105793135932,
+      "loss": 1.023,
+      "step": 2748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018267628734625773,
+      "loss": 0.893,
+      "step": 2749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018255152355401363,
+      "loss": 0.8069,
+      "step": 2750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018242676660354848,
+      "loss": 1.0222,
+      "step": 2751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018230201654378128,
+      "loss": 0.9205,
+      "step": 2752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018217727342362798,
+      "loss": 1.0005,
+      "step": 2753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018205253729200208,
+      "loss": 1.0799,
+      "step": 2754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018192780819781433,
+      "loss": 0.8965,
+      "step": 2755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001818030861899726,
+      "loss": 0.8739,
+      "step": 2756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018167837131738217,
+      "loss": 1.0533,
+      "step": 2757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018155366362894521,
+      "loss": 1.2201,
+      "step": 2758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018142896317356133,
+      "loss": 1.1733,
+      "step": 2759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001813042700001273,
+      "loss": 0.9648,
+      "step": 2760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018117958415753689,
+      "loss": 0.8978,
+      "step": 2761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018105490569468106,
+      "loss": 1.2115,
+      "step": 2762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018093023466044796,
+      "loss": 1.3232,
+      "step": 2763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001808055711037227,
+      "loss": 1.245,
+      "step": 2764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018068091507338752,
+      "loss": 0.8209,
+      "step": 2765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018055626661832178,
+      "loss": 1.0283,
+      "step": 2766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018043162578740166,
+      "loss": 1.5123,
+      "step": 2767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018030699262950052,
+      "loss": 0.8847,
+      "step": 2768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018018236719348873,
+      "loss": 1.2561,
+      "step": 2769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018005774952823363,
+      "loss": 1.1892,
+      "step": 2770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001799331396825993,
+      "loss": 1.1055,
+      "step": 2771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017980853770544698,
+      "loss": 0.704,
+      "step": 2772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017968394364563476,
+      "loss": 1.1632,
+      "step": 2773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017955935755201762,
+      "loss": 0.9931,
+      "step": 2774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017943477947344753,
+      "loss": 1.2135,
+      "step": 2775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017931020945877304,
+      "loss": 1.1065,
+      "step": 2776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017918564755683972,
+      "loss": 1.2995,
+      "step": 2777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017906109381649006,
+      "loss": 1.1099,
+      "step": 2778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017893654828656321,
+      "loss": 1.176,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000178812011015895,
+      "loss": 1.0736,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017868748205331823,
+      "loss": 1.0752,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001785629614476624,
+      "loss": 0.9646,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017843844924775364,
+      "loss": 1.059,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001783139455024148,
+      "loss": 1.0433,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017818945026046553,
+      "loss": 1.2775,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017806496357072199,
+      "loss": 0.8468,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001779404854819971,
+      "loss": 1.1264,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017781601604310043,
+      "loss": 0.9377,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017769155530283795,
+      "loss": 1.2013,
+      "step": 2789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017756710331001246,
+      "loss": 0.9687,
+      "step": 2790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017744266011342322,
+      "loss": 1.1787,
+      "step": 2791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017731822576186614,
+      "loss": 0.7932,
+      "step": 2792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017719380030413345,
+      "loss": 0.715,
+      "step": 2793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017706938378901402,
+      "loss": 1.2248,
+      "step": 2794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017694497626529333,
+      "loss": 0.9283,
+      "step": 2795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001768205777817531,
+      "loss": 1.0168,
+      "step": 2796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001766961883871718,
+      "loss": 1.0966,
+      "step": 2797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017657180813032396,
+      "loss": 1.3058,
+      "step": 2798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001764474370599808,
+      "loss": 1.0294,
+      "step": 2799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017632307522490992,
+      "loss": 0.9426,
+      "step": 2800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017619872267387525,
+      "loss": 1.0865,
+      "step": 2801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017607437945563696,
+      "loss": 0.8702,
+      "step": 2802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017595004561895173,
+      "loss": 1.2601,
+      "step": 2803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017582572121257252,
+      "loss": 0.8321,
+      "step": 2804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017570140628524864,
+      "loss": 0.9203,
+      "step": 2805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017557710088572553,
+      "loss": 1.0673,
+      "step": 2806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017545280506274497,
+      "loss": 0.8355,
+      "step": 2807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001753285188650451,
+      "loss": 0.9479,
+      "step": 2808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001752042423413601,
+      "loss": 1.0729,
+      "step": 2809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001750799755404205,
+      "loss": 0.9827,
+      "step": 2810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017495571851095293,
+      "loss": 1.0893,
+      "step": 2811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001748314713016802,
+      "loss": 0.7748,
+      "step": 2812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017470723396132128,
+      "loss": 0.9544,
+      "step": 2813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017458300653859136,
+      "loss": 0.8269,
+      "step": 2814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017445878908220165,
+      "loss": 1.0924,
+      "step": 2815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017433458164085934,
+      "loss": 1.1993,
+      "step": 2816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001742103842632679,
+      "loss": 1.4035,
+      "step": 2817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001740861969981268,
+      "loss": 0.9586,
+      "step": 2818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001739620198941315,
+      "loss": 1.2477,
+      "step": 2819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017383785299997343,
+      "loss": 0.8785,
+      "step": 2820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001737136963643401,
+      "loss": 1.3223,
+      "step": 2821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017358955003591496,
+      "loss": 1.1367,
+      "step": 2822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017346541406337748,
+      "loss": 0.9133,
+      "step": 2823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001733412884954031,
+      "loss": 1.0605,
+      "step": 2824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017321717338066291,
+      "loss": 0.89,
+      "step": 2825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017309306876782415,
+      "loss": 0.7132,
+      "step": 2826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017296897470554998,
+      "loss": 0.7991,
+      "step": 2827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017284489124249925,
+      "loss": 1.0109,
+      "step": 2828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017272081842732674,
+      "loss": 0.8709,
+      "step": 2829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017259675630868304,
+      "loss": 1.0873,
+      "step": 2830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017247270493521454,
+      "loss": 0.9207,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017234866435556347,
+      "loss": 0.7426,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017222463461836785,
+      "loss": 0.6954,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001721006157722612,
+      "loss": 0.9429,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017197660786587304,
+      "loss": 0.9327,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017185261094782858,
+      "loss": 1.2141,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017172862506674864,
+      "loss": 1.0554,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017160465027124964,
+      "loss": 0.925,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017148068660994376,
+      "loss": 0.7818,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017135673413143886,
+      "loss": 1.2254,
+      "step": 2840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017123279288433832,
+      "loss": 1.0106,
+      "step": 2841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017110886291724122,
+      "loss": 1.0157,
+      "step": 2842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000170984944278742,
+      "loss": 0.901,
+      "step": 2843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017086103701743088,
+      "loss": 1.0331,
+      "step": 2844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017073714118189353,
+      "loss": 0.9967,
+      "step": 2845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001706132568207112,
+      "loss": 1.1053,
+      "step": 2846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017048938398246048,
+      "loss": 0.8311,
+      "step": 2847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001703655227157136,
+      "loss": 0.8747,
+      "step": 2848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017024167306903831,
+      "loss": 1.2573,
+      "step": 2849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017011783509099756,
+      "loss": 1.1544,
+      "step": 2850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016999400883014993,
+      "loss": 1.276,
+      "step": 2851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016987019433504933,
+      "loss": 0.953,
+      "step": 2852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000169746391654245,
+      "loss": 0.9302,
+      "step": 2853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016962260083628168,
+      "loss": 0.9962,
+      "step": 2854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016949882192969945,
+      "loss": 1.2855,
+      "step": 2855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016937505498303354,
+      "loss": 1.0814,
+      "step": 2856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016925130004481463,
+      "loss": 0.8348,
+      "step": 2857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016912755716356874,
+      "loss": 1.0059,
+      "step": 2858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016900382638781702,
+      "loss": 1.4271,
+      "step": 2859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001688801077660761,
+      "loss": 0.9983,
+      "step": 2860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001687564013468575,
+      "loss": 1.1033,
+      "step": 2861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016863270717866818,
+      "loss": 0.8629,
+      "step": 2862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016850902531001036,
+      "loss": 1.0467,
+      "step": 2863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016838535578938134,
+      "loss": 1.1881,
+      "step": 2864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016826169866527348,
+      "loss": 0.9229,
+      "step": 2865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001681380539861744,
+      "loss": 0.6564,
+      "step": 2866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016801442180056685,
+      "loss": 1.1616,
+      "step": 2867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001678908021569287,
+      "loss": 0.8301,
+      "step": 2868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001677671951037328,
+      "loss": 1.1243,
+      "step": 2869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016764360068944706,
+      "loss": 0.923,
+      "step": 2870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001675200189625346,
+      "loss": 0.9882,
+      "step": 2871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016739644997145332,
+      "loss": 1.2219,
+      "step": 2872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016727289376465638,
+      "loss": 0.9741,
+      "step": 2873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001671493503905918,
+      "loss": 0.9826,
+      "step": 2874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016702581989770245,
+      "loss": 1.1095,
+      "step": 2875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016690230233442637,
+      "loss": 1.0853,
+      "step": 2876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016677879774919645,
+      "loss": 0.9557,
+      "step": 2877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001666553061904405,
+      "loss": 1.2,
+      "step": 2878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001665318277065811,
+      "loss": 1.0695,
+      "step": 2879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001664083623460358,
+      "loss": 1.0711,
+      "step": 2880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001662849101572171,
+      "loss": 0.9324,
+      "step": 2881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016616147118853224,
+      "loss": 1.047,
+      "step": 2882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001660380454883832,
+      "loss": 0.9173,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016591463310516681,
+      "loss": 1.03,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001657912340872748,
+      "loss": 1.1142,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016566784848309358,
+      "loss": 1.2219,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016554447634100413,
+      "loss": 1.2724,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001654211177093824,
+      "loss": 1.006,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016529777263659895,
+      "loss": 1.1013,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016517444117101902,
+      "loss": 1.1861,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001650511233610025,
+      "loss": 0.9663,
+      "step": 2891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001649278192549039,
+      "loss": 1.025,
+      "step": 2892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001648045289010725,
+      "loss": 1.2763,
+      "step": 2893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016468125234785194,
+      "loss": 1.0701,
+      "step": 2894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016455798964358072,
+      "loss": 0.9603,
+      "step": 2895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016443474083659173,
+      "loss": 1.0712,
+      "step": 2896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016431150597521243,
+      "loss": 0.9918,
+      "step": 2897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016418828510776485,
+      "loss": 1.0558,
+      "step": 2898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016406507828256557,
+      "loss": 1.1244,
+      "step": 2899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016394188554792565,
+      "loss": 1.0349,
+      "step": 2900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001638187069521505,
+      "loss": 0.9296,
+      "step": 2901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016369554254354007,
+      "loss": 0.7947,
+      "step": 2902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016357239237038885,
+      "loss": 1.0931,
+      "step": 2903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001634492564809857,
+      "loss": 0.9371,
+      "step": 2904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016332613492361366,
+      "loss": 0.8704,
+      "step": 2905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001632030277465504,
+      "loss": 1.0644,
+      "step": 2906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016307993499806793,
+      "loss": 1.3722,
+      "step": 2907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001629568567264325,
+      "loss": 1.1338,
+      "step": 2908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001628337929799048,
+      "loss": 0.9017,
+      "step": 2909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016271074380673964,
+      "loss": 0.9228,
+      "step": 2910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001625877092551863,
+      "loss": 1.2793,
+      "step": 2911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016246468937348827,
+      "loss": 1.129,
+      "step": 2912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016234168420988329,
+      "loss": 1.2524,
+      "step": 2913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016221869381260323,
+      "loss": 0.8357,
+      "step": 2914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016209571822987438,
+      "loss": 0.7584,
+      "step": 2915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016197275750991694,
+      "loss": 1.068,
+      "step": 2916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016184981170094554,
+      "loss": 0.9633,
+      "step": 2917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016172688085116897,
+      "loss": 1.067,
+      "step": 2918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001616039650087898,
+      "loss": 0.9703,
+      "step": 2919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001614810642220051,
+      "loss": 1.0671,
+      "step": 2920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001613581785390059,
+      "loss": 1.0535,
+      "step": 2921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016123530800797738,
+      "loss": 1.1358,
+      "step": 2922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016111245267709848,
+      "loss": 1.0232,
+      "step": 2923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016098961259454258,
+      "loss": 1.0825,
+      "step": 2924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016086678780847686,
+      "loss": 1.0402,
+      "step": 2925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016074397836706255,
+      "loss": 1.027,
+      "step": 2926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016062118431845495,
+      "loss": 0.9565,
+      "step": 2927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016049840571080308,
+      "loss": 0.8537,
+      "step": 2928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016037564259225012,
+      "loss": 0.9841,
+      "step": 2929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001602528950109331,
+      "loss": 1.1743,
+      "step": 2930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001601301630149831,
+      "loss": 1.1207,
+      "step": 2931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016000744665252477,
+      "loss": 1.1122,
+      "step": 2932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001598847459716769,
+      "loss": 0.7865,
+      "step": 2933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001597620610205521,
+      "loss": 1.0784,
+      "step": 2934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015963939184725676,
+      "loss": 0.9389,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015951673849989105,
+      "loss": 1.1864,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015939410102654898,
+      "loss": 0.7166,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015927147947531836,
+      "loss": 1.1808,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015914887389428069,
+      "loss": 0.9087,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015902628433151126,
+      "loss": 0.9285,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001589037108350791,
+      "loss": 0.8553,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00015878115345304677,
+      "loss": 1.0755,
+      "step": 2942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001586586122334708,
+      "loss": 1.1512,
+      "step": 2943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001585360872244011,
+      "loss": 1.1286,
+      "step": 2944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015841357847388155,
+      "loss": 1.1539,
+      "step": 2945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015829108602994922,
+      "loss": 0.7901,
+      "step": 2946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015816860994063512,
+      "loss": 1.0795,
+      "step": 2947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015804615025396377,
+      "loss": 0.8576,
+      "step": 2948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001579237070179533,
+      "loss": 0.9472,
+      "step": 2949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015780128028061513,
+      "loss": 0.9853,
+      "step": 2950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015767887008995456,
+      "loss": 1.1827,
+      "step": 2951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001575564764939702,
+      "loss": 1.0469,
+      "step": 2952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015743409954065423,
+      "loss": 0.9642,
+      "step": 2953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015731173927799232,
+      "loss": 1.1827,
+      "step": 2954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015718939575396342,
+      "loss": 1.1697,
+      "step": 2955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015706706901654012,
+      "loss": 1.0016,
+      "step": 2956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015694475911368837,
+      "loss": 1.1447,
+      "step": 2957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015682246609336753,
+      "loss": 1.0456,
+      "step": 2958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015670019000353023,
+      "loss": 0.852,
+      "step": 2959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001565779308921226,
+      "loss": 0.9818,
+      "step": 2960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015645568880708402,
+      "loss": 0.9564,
+      "step": 2961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015633346379634728,
+      "loss": 1.0723,
+      "step": 2962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001562112559078385,
+      "loss": 0.8555,
+      "step": 2963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015608906518947683,
+      "loss": 0.9941,
+      "step": 2964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015596689168917495,
+      "loss": 1.0984,
+      "step": 2965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015584473545483872,
+      "loss": 0.8777,
+      "step": 2966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015572259653436732,
+      "loss": 0.8086,
+      "step": 2967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015560047497565287,
+      "loss": 0.9905,
+      "step": 2968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001554783708265809,
+      "loss": 1.0542,
+      "step": 2969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015535628413503011,
+      "loss": 0.8854,
+      "step": 2970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001552342149488723,
+      "loss": 1.0461,
+      "step": 2971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015511216331597248,
+      "loss": 0.9855,
+      "step": 2972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015499012928418856,
+      "loss": 0.8561,
+      "step": 2973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001548681129013718,
+      "loss": 1.2183,
+      "step": 2974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015474611421536642,
+      "loss": 1.2871,
+      "step": 2975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015462413327400982,
+      "loss": 1.0896,
+      "step": 2976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015450217012513215,
+      "loss": 1.1243,
+      "step": 2977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001543802248165569,
+      "loss": 0.8697,
+      "step": 2978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015425829739610045,
+      "loss": 1.2102,
+      "step": 2979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001541363879115721,
+      "loss": 0.9103,
+      "step": 2980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015401449641077423,
+      "loss": 1.1511,
+      "step": 2981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015389262294150213,
+      "loss": 1.0304,
+      "step": 2982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001537707675515439,
+      "loss": 1.1983,
+      "step": 2983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015364893028868071,
+      "loss": 0.8537,
+      "step": 2984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001535271112006867,
+      "loss": 1.3009,
+      "step": 2985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001534053103353285,
+      "loss": 1.0596,
+      "step": 2986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000153283527740366,
+      "loss": 1.0425,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015316176346355168,
+      "loss": 1.222,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015304001755263105,
+      "loss": 0.8566,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001529182900553423,
+      "loss": 0.9579,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015279658101941625,
+      "loss": 0.8945,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015267489049257667,
+      "loss": 1.1821,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001525532185225401,
+      "loss": 1.1149,
+      "step": 2993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001524315651570158,
+      "loss": 1.0833,
+      "step": 2994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015230993044370545,
+      "loss": 0.8274,
+      "step": 2995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015218831443030378,
+      "loss": 0.8175,
+      "step": 2996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000152066717164498,
+      "loss": 1.0981,
+      "step": 2997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001519451386939681,
+      "loss": 1.0015,
+      "step": 2998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015182357906638643,
+      "loss": 0.9063,
+      "step": 2999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015170203832941829,
+      "loss": 1.1121,
+      "step": 3000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015158051653072133,
+      "loss": 0.954,
+      "step": 3001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001514590137179459,
+      "loss": 1.2108,
+      "step": 3002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015133752993873488,
+      "loss": 1.2399,
+      "step": 3003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001512160652407236,
+      "loss": 0.9343,
+      "step": 3004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015109461967154002,
+      "loss": 0.8705,
+      "step": 3005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015097319327880455,
+      "loss": 1.0407,
+      "step": 3006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001508517861101302,
+      "loss": 0.9649,
+      "step": 3007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015073039821312213,
+      "loss": 1.1848,
+      "step": 3008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015060902963537825,
+      "loss": 1.0608,
+      "step": 3009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015048768042448876,
+      "loss": 1.063,
+      "step": 3010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015036635062803635,
+      "loss": 0.9505,
+      "step": 3011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015024504029359606,
+      "loss": 0.9222,
+      "step": 3012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015012374946873515,
+      "loss": 1.0593,
+      "step": 3013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015000247820101345,
+      "loss": 0.8194,
+      "step": 3014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014988122653798302,
+      "loss": 1.2492,
+      "step": 3015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014975999452718835,
+      "loss": 1.3209,
+      "step": 3016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001496387822161659,
+      "loss": 0.987,
+      "step": 3017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014951758965244474,
+      "loss": 0.8455,
+      "step": 3018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001493964168835461,
+      "loss": 0.814,
+      "step": 3019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014927526395698342,
+      "loss": 1.1008,
+      "step": 3020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014915413092026243,
+      "loss": 0.9167,
+      "step": 3021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014903301782088083,
+      "loss": 0.932,
+      "step": 3022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001489119247063288,
+      "loss": 1.1901,
+      "step": 3023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001487908516240885,
+      "loss": 1.2311,
+      "step": 3024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001486697986216344,
+      "loss": 0.9778,
+      "step": 3025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014854876574643287,
+      "loss": 0.9337,
+      "step": 3026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014842775304594255,
+      "loss": 1.3383,
+      "step": 3027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014830676056761406,
+      "loss": 1.0491,
+      "step": 3028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014818578835889022,
+      "loss": 1.2018,
+      "step": 3029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014806483646720594,
+      "loss": 0.9717,
+      "step": 3030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014794390493998782,
+      "loss": 0.8534,
+      "step": 3031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014782299382465488,
+      "loss": 1.0423,
+      "step": 3032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014770210316861793,
+      "loss": 1.0839,
+      "step": 3033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014758123301927988,
+      "loss": 0.8899,
+      "step": 3034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001474603834240354,
+      "loss": 0.9132,
+      "step": 3035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001473395544302712,
+      "loss": 0.7831,
+      "step": 3036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000147218746085366,
+      "loss": 1.0145,
+      "step": 3037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001470979584366904,
+      "loss": 0.991,
+      "step": 3038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014697719153160683,
+      "loss": 0.9806,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014685644541746946,
+      "loss": 0.9522,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014673572014162455,
+      "loss": 0.9853,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001466150157514101,
+      "loss": 1.3674,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014649433229415587,
+      "loss": 0.7342,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014637366981718342,
+      "loss": 1.1415,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014625302836780612,
+      "loss": 1.1691,
+      "step": 3045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001461324079933291,
+      "loss": 0.6774,
+      "step": 3046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014601180874104915,
+      "loss": 1.0077,
+      "step": 3047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014589123065825502,
+      "loss": 0.8146,
+      "step": 3048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014577067379222672,
+      "loss": 1.1019,
+      "step": 3049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001456501381902363,
+      "loss": 1.1855,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014552962389954743,
+      "loss": 0.7086,
+      "step": 3051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014540913096741534,
+      "loss": 0.9378,
+      "step": 3052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014528865944108684,
+      "loss": 0.8318,
+      "step": 3053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014516820936780046,
+      "loss": 1.087,
+      "step": 3054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014504778079478623,
+      "loss": 0.9683,
+      "step": 3055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014492737376926583,
+      "loss": 0.9119,
+      "step": 3056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014480698833845254,
+      "loss": 1.0182,
+      "step": 3057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014468662454955093,
+      "loss": 1.2262,
+      "step": 3058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014456628244975724,
+      "loss": 0.901,
+      "step": 3059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001444459620862593,
+      "loss": 1.0507,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001443256635062364,
+      "loss": 0.8988,
+      "step": 3061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014420538675685894,
+      "loss": 0.9787,
+      "step": 3062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014408513188528922,
+      "loss": 0.9828,
+      "step": 3063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001439648989386807,
+      "loss": 1.1664,
+      "step": 3064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014384468796417837,
+      "loss": 0.9303,
+      "step": 3065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014372449900891855,
+      "loss": 1.061,
+      "step": 3066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014360433212002883,
+      "loss": 1.1171,
+      "step": 3067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014348418734462837,
+      "loss": 1.2714,
+      "step": 3068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014336406472982737,
+      "loss": 1.3951,
+      "step": 3069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001432439643227277,
+      "loss": 1.2219,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014312388617042214,
+      "loss": 1.1252,
+      "step": 3071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014300383031999497,
+      "loss": 0.7713,
+      "step": 3072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014288379681852176,
+      "loss": 1.0504,
+      "step": 3073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014276378571306914,
+      "loss": 0.7852,
+      "step": 3074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001426437970506952,
+      "loss": 1.1158,
+      "step": 3075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001425238308784489,
+      "loss": 0.9408,
+      "step": 3076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014240388724337066,
+      "loss": 1.1094,
+      "step": 3077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014228396619249195,
+      "loss": 1.2207,
+      "step": 3078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014216406777283554,
+      "loss": 1.0464,
+      "step": 3079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014204419203141494,
+      "loss": 1.0326,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014192433901523516,
+      "loss": 1.2287,
+      "step": 3081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014180450877129217,
+      "loss": 1.2048,
+      "step": 3082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014168470134657293,
+      "loss": 1.2048,
+      "step": 3083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014156491678805568,
+      "loss": 0.8277,
+      "step": 3084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014144515514270933,
+      "loss": 1.0071,
+      "step": 3085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014132541645749407,
+      "loss": 1.1189,
+      "step": 3086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001412057007793611,
+      "loss": 1.1853,
+      "step": 3087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014108600815525247,
+      "loss": 1.0689,
+      "step": 3088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014096633863210125,
+      "loss": 0.8509,
+      "step": 3089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014084669225683145,
+      "loss": 1.1803,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014072706907635797,
+      "loss": 0.8923,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014060746913758667,
+      "loss": 1.3492,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014048789248741436,
+      "loss": 0.9138,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014036833917272846,
+      "loss": 1.1932,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014024880924040748,
+      "loss": 1.222,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001401293027373207,
+      "loss": 1.1706,
+      "step": 3096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014000981971032834,
+      "loss": 1.1534,
+      "step": 3097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013989036020628103,
+      "loss": 0.6753,
+      "step": 3098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013977092427202057,
+      "loss": 1.0592,
+      "step": 3099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001396515119543793,
+      "loss": 1.0274,
+      "step": 3100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013953212330018047,
+      "loss": 1.1365,
+      "step": 3101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000139412758356238,
+      "loss": 0.7869,
+      "step": 3102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013929341716935624,
+      "loss": 1.2653,
+      "step": 3103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001391740997863306,
+      "loss": 0.8377,
+      "step": 3104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013905480625394704,
+      "loss": 0.9682,
+      "step": 3105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013893553661898213,
+      "loss": 0.8843,
+      "step": 3106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013881629092820296,
+      "loss": 1.1116,
+      "step": 3107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013869706922836741,
+      "loss": 1.1337,
+      "step": 3108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013857787156622398,
+      "loss": 1.3313,
+      "step": 3109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013845869798851155,
+      "loss": 0.8172,
+      "step": 3110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013833954854195965,
+      "loss": 0.8984,
+      "step": 3111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013822042327328842,
+      "loss": 1.2053,
+      "step": 3112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013810132222920842,
+      "loss": 0.8523,
+      "step": 3113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013798224545642074,
+      "loss": 1.1962,
+      "step": 3114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013786319300161705,
+      "loss": 0.9966,
+      "step": 3115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013774416491147925,
+      "loss": 0.9391,
+      "step": 3116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013762516123267996,
+      "loss": 0.8127,
+      "step": 3117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013750618201188196,
+      "loss": 1.226,
+      "step": 3118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013738722729573883,
+      "loss": 0.8606,
+      "step": 3119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013726829713089401,
+      "loss": 0.7914,
+      "step": 3120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013714939156398172,
+      "loss": 0.9711,
+      "step": 3121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013703051064162643,
+      "loss": 1.2499,
+      "step": 3122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013691165441044286,
+      "loss": 0.8214,
+      "step": 3123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013679282291703627,
+      "loss": 1.3807,
+      "step": 3124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013667401620800186,
+      "loss": 0.9644,
+      "step": 3125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001365552343299254,
+      "loss": 1.043,
+      "step": 3126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013643647732938287,
+      "loss": 0.9616,
+      "step": 3127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001363177452529405,
+      "loss": 1.0714,
+      "step": 3128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013619903814715459,
+      "loss": 0.8736,
+      "step": 3129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001360803560585718,
+      "loss": 1.2906,
+      "step": 3130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013596169903372905,
+      "loss": 0.8195,
+      "step": 3131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013584306711915322,
+      "loss": 1.1516,
+      "step": 3132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013572446036136154,
+      "loss": 1.0472,
+      "step": 3133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013560587880686125,
+      "loss": 1.2446,
+      "step": 3134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013548732250214974,
+      "loss": 1.0728,
+      "step": 3135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001353687914937145,
+      "loss": 1.0064,
+      "step": 3136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013525028582803323,
+      "loss": 1.0462,
+      "step": 3137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001351318055515734,
+      "loss": 1.1653,
+      "step": 3138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013501335071079275,
+      "loss": 0.9444,
+      "step": 3139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013489492135213902,
+      "loss": 0.9918,
+      "step": 3140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013477651752204995,
+      "loss": 1.0757,
+      "step": 3141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013465813926695329,
+      "loss": 0.9934,
+      "step": 3142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013453978663326662,
+      "loss": 0.9139,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013442145966739762,
+      "loss": 1.0604,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013430315841574383,
+      "loss": 1.0721,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013418488292469286,
+      "loss": 1.0676,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013406663324062195,
+      "loss": 0.7782,
+      "step": 3147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013394840940989845,
+      "loss": 0.8856,
+      "step": 3148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013383021147887948,
+      "loss": 1.1991,
+      "step": 3149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000133712039493912,
+      "loss": 0.9819,
+      "step": 3150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013359389350133284,
+      "loss": 0.7893,
+      "step": 3151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013347577354746857,
+      "loss": 1.0013,
+      "step": 3152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013335767967863563,
+      "loss": 0.8091,
+      "step": 3153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013323961194114015,
+      "loss": 1.111,
+      "step": 3154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013312157038127814,
+      "loss": 1.1182,
+      "step": 3155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013300355504533516,
+      "loss": 0.8265,
+      "step": 3156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013288556597958662,
+      "loss": 0.9828,
+      "step": 3157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013276760323029758,
+      "loss": 1.3656,
+      "step": 3158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013264966684372283,
+      "loss": 1.1568,
+      "step": 3159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001325317568661069,
+      "loss": 1.2966,
+      "step": 3160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013241387334368362,
+      "loss": 1.0437,
+      "step": 3161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013229601632267683,
+      "loss": 1.1958,
+      "step": 3162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013217818584929978,
+      "loss": 1.1285,
+      "step": 3163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001320603819697555,
+      "loss": 1.0317,
+      "step": 3164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013194260473023626,
+      "loss": 0.973,
+      "step": 3165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013182485417692413,
+      "loss": 1.1093,
+      "step": 3166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013170713035599068,
+      "loss": 0.9099,
+      "step": 3167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013158943331359702,
+      "loss": 0.9187,
+      "step": 3168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013147176309589374,
+      "loss": 0.8995,
+      "step": 3169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001313541197490208,
+      "loss": 1.0806,
+      "step": 3170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013123650331910773,
+      "loss": 0.9539,
+      "step": 3171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013111891385227359,
+      "loss": 0.757,
+      "step": 3172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013100135139462667,
+      "loss": 0.9526,
+      "step": 3173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001308838159922648,
+      "loss": 0.5828,
+      "step": 3174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013076630769127514,
+      "loss": 1.2666,
+      "step": 3175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001306488265377343,
+      "loss": 1.2904,
+      "step": 3176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001305313725777082,
+      "loss": 0.9816,
+      "step": 3177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013041394585725207,
+      "loss": 1.0256,
+      "step": 3178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013029654642241053,
+      "loss": 0.9573,
+      "step": 3179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013017917431921736,
+      "loss": 1.238,
+      "step": 3180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013006182959369577,
+      "loss": 0.918,
+      "step": 3181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012994451229185824,
+      "loss": 0.9694,
+      "step": 3182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012982722245970637,
+      "loss": 0.9151,
+      "step": 3183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000129709960143231,
+      "loss": 1.0053,
+      "step": 3184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012959272538841238,
+      "loss": 0.9684,
+      "step": 3185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001294755182412197,
+      "loss": 1.0096,
+      "step": 3186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012935833874761158,
+      "loss": 1.0058,
+      "step": 3187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001292411869535355,
+      "loss": 0.9808,
+      "step": 3188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012912406290492825,
+      "loss": 0.9292,
+      "step": 3189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012900696664771576,
+      "loss": 0.8724,
+      "step": 3190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012888989822781312,
+      "loss": 0.8565,
+      "step": 3191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012877285769112424,
+      "loss": 0.9684,
+      "step": 3192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012865584508354232,
+      "loss": 0.995,
+      "step": 3193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001285388604509496,
+      "loss": 1.082,
+      "step": 3194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012842190383921732,
+      "loss": 0.8888,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012830497529420568,
+      "loss": 1.1036,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012818807486176384,
+      "loss": 1.0302,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00012807120258773015,
+      "loss": 1.1906,
+      "step": 3198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001279543585179317,
+      "loss": 1.1123,
+      "step": 3199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001278375426981846,
+      "loss": 0.8629,
+      "step": 3200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012772075517429392,
+      "loss": 0.9924,
+      "step": 3201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012760399599205354,
+      "loss": 0.8751,
+      "step": 3202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001274872651972463,
+      "loss": 0.8736,
+      "step": 3203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001273705628356439,
+      "loss": 1.2284,
+      "step": 3204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012725388895300696,
+      "loss": 1.1491,
+      "step": 3205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012713724359508474,
+      "loss": 0.9369,
+      "step": 3206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012702062680761546,
+      "loss": 0.9331,
+      "step": 3207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001269040386363261,
+      "loss": 1.2371,
+      "step": 3208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012678747912693255,
+      "loss": 0.9346,
+      "step": 3209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001266709483251392,
+      "loss": 0.9808,
+      "step": 3210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001265544462766393,
+      "loss": 1.1349,
+      "step": 3211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012643797302711499,
+      "loss": 1.1878,
+      "step": 3212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012632152862223693,
+      "loss": 0.7917,
+      "step": 3213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012620511310766438,
+      "loss": 0.7926,
+      "step": 3214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012608872652904556,
+      "loss": 1.0182,
+      "step": 3215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012597236893201712,
+      "loss": 0.8684,
+      "step": 3216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012585604036220447,
+      "loss": 1.078,
+      "step": 3217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012573974086522158,
+      "loss": 0.9148,
+      "step": 3218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012562347048667097,
+      "loss": 0.7219,
+      "step": 3219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012550722927214382,
+      "loss": 1.0616,
+      "step": 3220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012539101726721983,
+      "loss": 1.1549,
+      "step": 3221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012527483451746734,
+      "loss": 1.2237,
+      "step": 3222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012515868106844312,
+      "loss": 0.8863,
+      "step": 3223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001250425569656924,
+      "loss": 1.0366,
+      "step": 3224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012492646225474908,
+      "loss": 0.9441,
+      "step": 3225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012481039698113535,
+      "loss": 1.4199,
+      "step": 3226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012469436119036205,
+      "loss": 1.0512,
+      "step": 3227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012457835492792827,
+      "loss": 1.0611,
+      "step": 3228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012446237823932158,
+      "loss": 0.8887,
+      "step": 3229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012434643117001804,
+      "loss": 0.9814,
+      "step": 3230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012423051376548208,
+      "loss": 0.8981,
+      "step": 3231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012411462607116636,
+      "loss": 0.8696,
+      "step": 3232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012399876813251197,
+      "loss": 1.1643,
+      "step": 3233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012388293999494838,
+      "loss": 0.9942,
+      "step": 3234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012376714170389335,
+      "loss": 1.255,
+      "step": 3235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012365137330475298,
+      "loss": 0.7744,
+      "step": 3236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012353563484292145,
+      "loss": 1.0378,
+      "step": 3237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012341992636378147,
+      "loss": 1.0753,
+      "step": 3238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012330424791270382,
+      "loss": 1.2339,
+      "step": 3239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012318859953504754,
+      "loss": 1.1096,
+      "step": 3240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012307298127615994,
+      "loss": 0.6476,
+      "step": 3241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012295739318137644,
+      "loss": 0.9137,
+      "step": 3242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012284183529602064,
+      "loss": 1.0019,
+      "step": 3243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012272630766540436,
+      "loss": 0.9772,
+      "step": 3244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012261081033482758,
+      "loss": 1.1382,
+      "step": 3245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012249534334957818,
+      "loss": 0.9524,
+      "step": 3246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012237990675493236,
+      "loss": 1.1515,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012226450059615437,
+      "loss": 0.8202,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012214912491849655,
+      "loss": 1.3248,
+      "step": 3249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001220337797671991,
+      "loss": 0.9438,
+      "step": 3250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012191846518749044,
+      "loss": 1.0807,
+      "step": 3251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012180318122458696,
+      "loss": 1.0408,
+      "step": 3252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012168792792369302,
+      "loss": 0.8005,
+      "step": 3253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012157270533000102,
+      "loss": 0.9047,
+      "step": 3254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012145751348869117,
+      "loss": 0.8157,
+      "step": 3255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012134235244493175,
+      "loss": 0.9695,
+      "step": 3256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012122722224387895,
+      "loss": 0.9844,
+      "step": 3257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012111212293067692,
+      "loss": 1.018,
+      "step": 3258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012099705455045752,
+      "loss": 0.8854,
+      "step": 3259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012088201714834055,
+      "loss": 0.7734,
+      "step": 3260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012076701076943387,
+      "loss": 1.0172,
+      "step": 3261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012065203545883283,
+      "loss": 1.1837,
+      "step": 3262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012053709126162092,
+      "loss": 0.8939,
+      "step": 3263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012042217822286921,
+      "loss": 0.8524,
+      "step": 3264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012030729638763659,
+      "loss": 1.0289,
+      "step": 3265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012019244580096978,
+      "loss": 0.9551,
+      "step": 3266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012007762650790333,
+      "loss": 1.3134,
+      "step": 3267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001199628385534592,
+      "loss": 1.1058,
+      "step": 3268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001198480819826474,
+      "loss": 1.1699,
+      "step": 3269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011973335684046549,
+      "loss": 1.1176,
+      "step": 3270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001196186631718987,
+      "loss": 1.006,
+      "step": 3271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011950400102192003,
+      "loss": 1.1753,
+      "step": 3272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011938937043548985,
+      "loss": 1.1261,
+      "step": 3273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011927477145755643,
+      "loss": 0.7857,
+      "step": 3274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001191602041330556,
+      "loss": 0.8999,
+      "step": 3275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001190456685069107,
+      "loss": 1.0291,
+      "step": 3276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001189311646240326,
+      "loss": 1.2228,
+      "step": 3277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011881669252931981,
+      "loss": 0.7543,
+      "step": 3278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011870225226765841,
+      "loss": 1.0251,
+      "step": 3279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011858784388392195,
+      "loss": 0.857,
+      "step": 3280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011847346742297141,
+      "loss": 1.0772,
+      "step": 3281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011835912292965532,
+      "loss": 1.0821,
+      "step": 3282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011824481044880975,
+      "loss": 0.8548,
+      "step": 3283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011813053002525804,
+      "loss": 0.9728,
+      "step": 3284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011801628170381116,
+      "loss": 1.0698,
+      "step": 3285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011790206552926735,
+      "loss": 0.9279,
+      "step": 3286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001177878815464122,
+      "loss": 1.089,
+      "step": 3287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011767372980001888,
+      "loss": 0.8308,
+      "step": 3288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011755961033484773,
+      "loss": 0.973,
+      "step": 3289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011744552319564664,
+      "loss": 1.3826,
+      "step": 3290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011733146842715048,
+      "loss": 1.1286,
+      "step": 3291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011721744607408178,
+      "loss": 1.0348,
+      "step": 3292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011710345618115017,
+      "loss": 1.0361,
+      "step": 3293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011698949879305269,
+      "loss": 0.8303,
+      "step": 3294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001168755739544734,
+      "loss": 0.9164,
+      "step": 3295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011676168171008384,
+      "loss": 1.0884,
+      "step": 3296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011664782210454252,
+      "loss": 0.8603,
+      "step": 3297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011653399518249552,
+      "loss": 1.0265,
+      "step": 3298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001164202009885758,
+      "loss": 0.9984,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011630643956740352,
+      "loss": 0.9157,
+      "step": 3300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011619271096358602,
+      "loss": 0.6374,
+      "step": 3301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011607901522171791,
+      "loss": 0.9159,
+      "step": 3302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011596535238638075,
+      "loss": 0.7942,
+      "step": 3303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001158517225021432,
+      "loss": 1.153,
+      "step": 3304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011573812561356104,
+      "loss": 0.8853,
+      "step": 3305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001156245617651772,
+      "loss": 0.7964,
+      "step": 3306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011551103100152146,
+      "loss": 0.9552,
+      "step": 3307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011539753336711094,
+      "loss": 1.1677,
+      "step": 3308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011528406890644929,
+      "loss": 1.1359,
+      "step": 3309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011517063766402763,
+      "loss": 1.0039,
+      "step": 3310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011505723968432373,
+      "loss": 0.7512,
+      "step": 3311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011494387501180264,
+      "loss": 0.9202,
+      "step": 3312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011483054369091588,
+      "loss": 0.9603,
+      "step": 3313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011471724576610237,
+      "loss": 1.039,
+      "step": 3314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011460398128178764,
+      "loss": 0.8717,
+      "step": 3315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011449075028238414,
+      "loss": 0.6817,
+      "step": 3316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001143775528122914,
+      "loss": 0.7594,
+      "step": 3317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001142643889158956,
+      "loss": 0.8002,
+      "step": 3318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011415125863756972,
+      "loss": 0.9101,
+      "step": 3319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011403816202167363,
+      "loss": 1.2155,
+      "step": 3320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011392509911255427,
+      "loss": 1.2757,
+      "step": 3321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011381206995454472,
+      "loss": 1.0397,
+      "step": 3322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001136990745919655,
+      "loss": 0.8254,
+      "step": 3323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001135861130691234,
+      "loss": 0.6969,
+      "step": 3324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011347318543031242,
+      "loss": 0.9766,
+      "step": 3325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011336029171981261,
+      "loss": 1.1854,
+      "step": 3326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011324743198189133,
+      "loss": 0.8588,
+      "step": 3327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011313460626080228,
+      "loss": 0.8785,
+      "step": 3328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011302181460078602,
+      "loss": 1.0264,
+      "step": 3329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011290905704606962,
+      "loss": 1.128,
+      "step": 3330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011279633364086677,
+      "loss": 1.0353,
+      "step": 3331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011268364442937776,
+      "loss": 0.8803,
+      "step": 3332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011257098945578968,
+      "loss": 0.8694,
+      "step": 3333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011245836876427594,
+      "loss": 0.8354,
+      "step": 3334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011234578239899662,
+      "loss": 1.0786,
+      "step": 3335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011223323040409827,
+      "loss": 1.0172,
+      "step": 3336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011212071282371415,
+      "loss": 0.9025,
+      "step": 3337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011200822970196383,
+      "loss": 1.1874,
+      "step": 3338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011189578108295343,
+      "loss": 1.1647,
+      "step": 3339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011178336701077551,
+      "loss": 1.0723,
+      "step": 3340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011167098752950911,
+      "loss": 1.0903,
+      "step": 3341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001115586426832198,
+      "loss": 1.0695,
+      "step": 3342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011144633251595946,
+      "loss": 1.003,
+      "step": 3343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011133405707176638,
+      "loss": 1.0042,
+      "step": 3344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011122181639466518,
+      "loss": 1.0454,
+      "step": 3345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011110961052866705,
+      "loss": 1.0271,
+      "step": 3346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011099743951776922,
+      "loss": 1.4653,
+      "step": 3347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011088530340595573,
+      "loss": 1.0575,
+      "step": 3348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001107732022371963,
+      "loss": 0.8416,
+      "step": 3349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011066113605544748,
+      "loss": 0.8078,
+      "step": 3350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011054910490465178,
+      "loss": 0.7039,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011043710882873837,
+      "loss": 1.1496,
+      "step": 3352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011032514787162202,
+      "loss": 1.0392,
+      "step": 3353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011021322207720439,
+      "loss": 0.977,
+      "step": 3354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011010133148937289,
+      "loss": 1.0411,
+      "step": 3355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010998947615200146,
+      "loss": 0.8493,
+      "step": 3356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010987765610894996,
+      "loss": 1.0663,
+      "step": 3357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010976587140406454,
+      "loss": 0.8133,
+      "step": 3358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010965412208117749,
+      "loss": 1.0947,
+      "step": 3359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010954240818410705,
+      "loss": 0.7462,
+      "step": 3360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010943072975665796,
+      "loss": 0.904,
+      "step": 3361
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010931908684262069,
+      "loss": 1.0075,
+      "step": 3362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010920747948577187,
+      "loss": 1.4955,
+      "step": 3363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010909590772987424,
+      "loss": 1.211,
+      "step": 3364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010898437161867665,
+      "loss": 0.9643,
+      "step": 3365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001088728711959138,
+      "loss": 1.2,
+      "step": 3366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010876140650530655,
+      "loss": 0.9528,
+      "step": 3367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010864997759056157,
+      "loss": 0.9302,
+      "step": 3368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010853858449537174,
+      "loss": 1.0183,
+      "step": 3369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010842722726341574,
+      "loss": 1.134,
+      "step": 3370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010831590593835818,
+      "loss": 0.9657,
+      "step": 3371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010820462056384958,
+      "loss": 1.0304,
+      "step": 3372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010809337118352658,
+      "loss": 1.0691,
+      "step": 3373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010798215784101131,
+      "loss": 1.2141,
+      "step": 3374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001078709805799123,
+      "loss": 0.9695,
+      "step": 3375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010775983944382331,
+      "loss": 0.8179,
+      "step": 3376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010764873447632444,
+      "loss": 0.8598,
+      "step": 3377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010753766572098132,
+      "loss": 0.899,
+      "step": 3378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010742663322134568,
+      "loss": 0.7452,
+      "step": 3379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010731563702095453,
+      "loss": 1.3534,
+      "step": 3380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010720467716333119,
+      "loss": 0.8045,
+      "step": 3381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001070937536919844,
+      "loss": 0.8195,
+      "step": 3382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010698286665040869,
+      "loss": 0.814,
+      "step": 3383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010687201608208447,
+      "loss": 1.0117,
+      "step": 3384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001067612020304776,
+      "loss": 1.1912,
+      "step": 3385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010665042453903979,
+      "loss": 0.9393,
+      "step": 3386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010653968365120825,
+      "loss": 1.1808,
+      "step": 3387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010642897941040624,
+      "loss": 1.0836,
+      "step": 3388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000106318311860042,
+      "loss": 1.0947,
+      "step": 3389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010620768104350998,
+      "loss": 1.088,
+      "step": 3390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010609708700418986,
+      "loss": 0.8042,
+      "step": 3391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010598652978544716,
+      "loss": 0.7544,
+      "step": 3392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010587600943063276,
+      "loss": 0.978,
+      "step": 3393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010576552598308315,
+      "loss": 0.7911,
+      "step": 3394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010565507948612033,
+      "loss": 1.1573,
+      "step": 3395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001055446699830519,
+      "loss": 0.8062,
+      "step": 3396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010543429751717089,
+      "loss": 0.8204,
+      "step": 3397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010532396213175574,
+      "loss": 1.0148,
+      "step": 3398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010521366387007043,
+      "loss": 0.9903,
+      "step": 3399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010510340277536444,
+      "loss": 0.7426,
+      "step": 3400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010499317889087259,
+      "loss": 0.8665,
+      "step": 3401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010488299225981509,
+      "loss": 1.174,
+      "step": 3402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010477284292539751,
+      "loss": 1.093,
+      "step": 3403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010466273093081103,
+      "loss": 0.8985,
+      "step": 3404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010455265631923192,
+      "loss": 0.8333,
+      "step": 3405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010444261913382192,
+      "loss": 0.9395,
+      "step": 3406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010433261941772808,
+      "loss": 0.9156,
+      "step": 3407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010422265721408261,
+      "loss": 1.1133,
+      "step": 3408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001041127325660034,
+      "loss": 1.14,
+      "step": 3409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010400284551659311,
+      "loss": 0.9336,
+      "step": 3410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001038929961089402,
+      "loss": 1.1366,
+      "step": 3411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010378318438611776,
+      "loss": 1.1837,
+      "step": 3412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010367341039118464,
+      "loss": 1.0012,
+      "step": 3413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010356367416718453,
+      "loss": 1.0074,
+      "step": 3414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001034539757571467,
+      "loss": 0.9042,
+      "step": 3415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010334431520408503,
+      "loss": 0.9978,
+      "step": 3416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010323469255099914,
+      "loss": 1.1612,
+      "step": 3417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001031251078408733,
+      "loss": 1.1623,
+      "step": 3418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010301556111667736,
+      "loss": 0.8954,
+      "step": 3419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010290605242136593,
+      "loss": 0.7587,
+      "step": 3420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010279658179787883,
+      "loss": 0.8462,
+      "step": 3421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010268714928914085,
+      "loss": 0.766,
+      "step": 3422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010257775493806207,
+      "loss": 0.8408,
+      "step": 3423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010246839878753744,
+      "loss": 0.8709,
+      "step": 3424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010235908088044694,
+      "loss": 0.9921,
+      "step": 3425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010224980125965555,
+      "loss": 1.0126,
+      "step": 3426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001021405599680132,
+      "loss": 1.1956,
+      "step": 3427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010203135704835495,
+      "loss": 0.9367,
+      "step": 3428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010192219254350073,
+      "loss": 0.946,
+      "step": 3429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010181306649625533,
+      "loss": 1.2204,
+      "step": 3430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010170397894940845,
+      "loss": 0.9124,
+      "step": 3431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001015949299457349,
+      "loss": 1.0326,
+      "step": 3432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001014859195279942,
+      "loss": 1.0713,
+      "step": 3433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010137694773893076,
+      "loss": 0.9821,
+      "step": 3434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001012680146212738,
+      "loss": 1.1261,
+      "step": 3435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010115912021773755,
+      "loss": 0.8565,
+      "step": 3436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010105026457102096,
+      "loss": 0.9687,
+      "step": 3437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010094144772380768,
+      "loss": 0.709,
+      "step": 3438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010083266971876622,
+      "loss": 0.9643,
+      "step": 3439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010072393059854999,
+      "loss": 0.8092,
+      "step": 3440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010061523040579692,
+      "loss": 0.892,
+      "step": 3441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010050656918313003,
+      "loss": 1.0906,
+      "step": 3442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010039794697315651,
+      "loss": 1.1741,
+      "step": 3443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001002893638184688,
+      "loss": 1.3078,
+      "step": 3444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001001808197616437,
+      "loss": 1.2342,
+      "step": 3445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001000723148452428,
+      "loss": 0.9675,
+      "step": 3446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.996384911181223e-05,
+      "loss": 1.0039,
+      "step": 3447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.985542260388297e-05,
+      "loss": 1.017,
+      "step": 3448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.974703536397045e-05,
+      "loss": 0.8987,
+      "step": 3449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.963868743457465e-05,
+      "loss": 1.0335,
+      "step": 3450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.953037885818048e-05,
+      "loss": 1.0698,
+      "step": 3451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.942210967725684e-05,
+      "loss": 0.9456,
+      "step": 3452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.931387993425776e-05,
+      "loss": 1.0617,
+      "step": 3453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.920568967162138e-05,
+      "loss": 0.9336,
+      "step": 3454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.909753893177078e-05,
+      "loss": 0.9514,
+      "step": 3455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.898942775711301e-05,
+      "loss": 0.9384,
+      "step": 3456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.888135619004011e-05,
+      "loss": 0.8041,
+      "step": 3457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.877332427292821e-05,
+      "loss": 0.9841,
+      "step": 3458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.866533204813819e-05,
+      "loss": 1.0165,
+      "step": 3459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.855737955801522e-05,
+      "loss": 0.8853,
+      "step": 3460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.844946684488887e-05,
+      "loss": 0.9138,
+      "step": 3461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.834159395107306e-05,
+      "loss": 0.8032,
+      "step": 3462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.823376091886633e-05,
+      "loss": 1.1633,
+      "step": 3463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.81259677905514e-05,
+      "loss": 0.8654,
+      "step": 3464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.801821460839535e-05,
+      "loss": 0.9963,
+      "step": 3465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.791050141464959e-05,
+      "loss": 0.6803,
+      "step": 3466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.780282825155001e-05,
+      "loss": 0.9827,
+      "step": 3467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.769519516131662e-05,
+      "loss": 1.0636,
+      "step": 3468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.758760218615377e-05,
+      "loss": 0.6689,
+      "step": 3469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.748004936825012e-05,
+      "loss": 0.865,
+      "step": 3470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.737253674977847e-05,
+      "loss": 0.7251,
+      "step": 3471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.726506437289609e-05,
+      "loss": 1.1566,
+      "step": 3472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.715763227974423e-05,
+      "loss": 0.8538,
+      "step": 3473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.705024051244849e-05,
+      "loss": 1.0407,
+      "step": 3474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.694288911311848e-05,
+      "loss": 1.129,
+      "step": 3475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.683557812384829e-05,
+      "loss": 0.9294,
+      "step": 3476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.672830758671581e-05,
+      "loss": 1.1199,
+      "step": 3477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.662107754378347e-05,
+      "loss": 0.7938,
+      "step": 3478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.651388803709728e-05,
+      "loss": 1.126,
+      "step": 3479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.640673910868793e-05,
+      "loss": 1.0172,
+      "step": 3480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.629963080056976e-05,
+      "loss": 1.059,
+      "step": 3481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.619256315474157e-05,
+      "loss": 1.0924,
+      "step": 3482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.608553621318572e-05,
+      "loss": 0.7618,
+      "step": 3483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.597855001786914e-05,
+      "loss": 0.803,
+      "step": 3484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.587160461074234e-05,
+      "loss": 0.7599,
+      "step": 3485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.57647000337402e-05,
+      "loss": 1.2173,
+      "step": 3486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.565783632878138e-05,
+      "loss": 1.193,
+      "step": 3487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.555101353776852e-05,
+      "loss": 0.9229,
+      "step": 3488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.544423170258816e-05,
+      "loss": 0.947,
+      "step": 3489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.533749086511108e-05,
+      "loss": 0.9984,
+      "step": 3490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.523079106719166e-05,
+      "loss": 1.2517,
+      "step": 3491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.51241323506683e-05,
+      "loss": 1.0602,
+      "step": 3492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.501751475736334e-05,
+      "loss": 1.0577,
+      "step": 3493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.491093832908285e-05,
+      "loss": 0.7559,
+      "step": 3494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.4804403107617e-05,
+      "loss": 1.0556,
+      "step": 3495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.46979091347396e-05,
+      "loss": 1.2032,
+      "step": 3496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.459145645220835e-05,
+      "loss": 1.1177,
+      "step": 3497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.448504510176468e-05,
+      "loss": 0.942,
+      "step": 3498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.437867512513406e-05,
+      "loss": 0.9988,
+      "step": 3499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.427234656402546e-05,
+      "loss": 1.1135,
+      "step": 3500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.416605946013174e-05,
+      "loss": 1.0623,
+      "step": 3501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.405981385512941e-05,
+      "loss": 1.3011,
+      "step": 3502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.395360979067893e-05,
+      "loss": 0.8327,
+      "step": 3503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.384744730842416e-05,
+      "loss": 1.0632,
+      "step": 3504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.374132644999305e-05,
+      "loss": 1.0911,
+      "step": 3505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.363524725699668e-05,
+      "loss": 1.272,
+      "step": 3506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.352920977103037e-05,
+      "loss": 0.8568,
+      "step": 3507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.342321403367265e-05,
+      "loss": 1.1242,
+      "step": 3508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.331726008648607e-05,
+      "loss": 0.8583,
+      "step": 3509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32113479710163e-05,
+      "loss": 1.039,
+      "step": 3510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.310547772879307e-05,
+      "loss": 1.0221,
+      "step": 3511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.299964940132949e-05,
+      "loss": 1.1984,
+      "step": 3512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.289386303012213e-05,
+      "loss": 1.0497,
+      "step": 3513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.27881186566514e-05,
+      "loss": 0.962,
+      "step": 3514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.268241632238099e-05,
+      "loss": 1.1217,
+      "step": 3515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.25767560687582e-05,
+      "loss": 0.6396,
+      "step": 3516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.24711379372137e-05,
+      "loss": 1.1024,
+      "step": 3517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.236556196916204e-05,
+      "loss": 1.0498,
+      "step": 3518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.226002820600062e-05,
+      "loss": 0.8975,
+      "step": 3519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.215453668911085e-05,
+      "loss": 1.0646,
+      "step": 3520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.204908745985723e-05,
+      "loss": 1.1271,
+      "step": 3521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.194368055958793e-05,
+      "loss": 0.8359,
+      "step": 3522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.183831602963433e-05,
+      "loss": 1.0702,
+      "step": 3523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.173299391131125e-05,
+      "loss": 1.2644,
+      "step": 3524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.16277142459168e-05,
+      "loss": 0.7704,
+      "step": 3525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.152247707473274e-05,
+      "loss": 1.0516,
+      "step": 3526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.141728243902384e-05,
+      "loss": 1.0002,
+      "step": 3527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.131213038003829e-05,
+      "loss": 0.8267,
+      "step": 3528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.120702093900759e-05,
+      "loss": 0.7761,
+      "step": 3529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.110195415714664e-05,
+      "loss": 0.9499,
+      "step": 3530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.099693007565349e-05,
+      "loss": 0.7918,
+      "step": 3531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.089194873570943e-05,
+      "loss": 0.8086,
+      "step": 3532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.078701017847901e-05,
+      "loss": 1.0549,
+      "step": 3533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.068211444511015e-05,
+      "loss": 1.0361,
+      "step": 3534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.057726157673378e-05,
+      "loss": 0.9833,
+      "step": 3535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.047245161446411e-05,
+      "loss": 0.9817,
+      "step": 3536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.03676845993985e-05,
+      "loss": 0.9825,
+      "step": 3537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.026296057261744e-05,
+      "loss": 0.8748,
+      "step": 3538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.015827957518474e-05,
+      "loss": 0.9269,
+      "step": 3539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.005364164814713e-05,
+      "loss": 1.4107,
+      "step": 3540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.994904683253453e-05,
+      "loss": 1.1382,
+      "step": 3541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.98444951693599e-05,
+      "loss": 0.9709,
+      "step": 3542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.973998669961947e-05,
+      "loss": 1.0446,
+      "step": 3543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.963552146429224e-05,
+      "loss": 1.0556,
+      "step": 3544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.953109950434062e-05,
+      "loss": 1.1315,
+      "step": 3545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.94267208607096e-05,
+      "loss": 0.6106,
+      "step": 3546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.932238557432759e-05,
+      "loss": 0.8919,
+      "step": 3547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.921809368610575e-05,
+      "loss": 0.8968,
+      "step": 3548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.91138452369385e-05,
+      "loss": 1.0631,
+      "step": 3549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.900964026770275e-05,
+      "loss": 0.9992,
+      "step": 3550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.890547881925888e-05,
+      "loss": 0.8539,
+      "step": 3551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.880136093244981e-05,
+      "loss": 1.2034,
+      "step": 3552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.869728664810171e-05,
+      "loss": 1.1328,
+      "step": 3553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.85932560070234e-05,
+      "loss": 1.2112,
+      "step": 3554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.848926905000668e-05,
+      "loss": 1.0242,
+      "step": 3555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.838532581782621e-05,
+      "loss": 0.8828,
+      "step": 3556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.828142635123943e-05,
+      "loss": 0.7675,
+      "step": 3557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.817757069098689e-05,
+      "loss": 0.9747,
+      "step": 3558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.807375887779163e-05,
+      "loss": 0.8931,
+      "step": 3559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.796999095235969e-05,
+      "loss": 1.103,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.786626695537976e-05,
+      "loss": 0.968,
+      "step": 3561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.776258692752355e-05,
+      "loss": 0.9085,
+      "step": 3562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.765895090944527e-05,
+      "loss": 1.1328,
+      "step": 3563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.755535894178199e-05,
+      "loss": 0.9681,
+      "step": 3564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.745181106515346e-05,
+      "loss": 1.0025,
+      "step": 3565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.734830732016226e-05,
+      "loss": 0.6238,
+      "step": 3566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.724484774739354e-05,
+      "loss": 1.1524,
+      "step": 3567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.714143238741519e-05,
+      "loss": 0.7211,
+      "step": 3568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.703806128077763e-05,
+      "loss": 1.0204,
+      "step": 3569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.693473446801422e-05,
+      "loss": 1.1308,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.68314519896406e-05,
+      "loss": 0.9018,
+      "step": 3571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.672821388615542e-05,
+      "loss": 1.1139,
+      "step": 3572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.662502019803944e-05,
+      "loss": 0.7766,
+      "step": 3573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.65218709657565e-05,
+      "loss": 0.7504,
+      "step": 3574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.641876622975269e-05,
+      "loss": 1.044,
+      "step": 3575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.631570603045673e-05,
+      "loss": 1.0522,
+      "step": 3576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.621269040827984e-05,
+      "loss": 0.791,
+      "step": 3577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.610971940361594e-05,
+      "loss": 1.0462,
+      "step": 3578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.600679305684127e-05,
+      "loss": 0.8567,
+      "step": 3579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.590391140831454e-05,
+      "loss": 0.898,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.580107449837722e-05,
+      "loss": 1.3029,
+      "step": 3581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.569828236735273e-05,
+      "loss": 0.7477,
+      "step": 3582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.559553505554747e-05,
+      "loss": 1.2503,
+      "step": 3583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.549283260324985e-05,
+      "loss": 1.0758,
+      "step": 3584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.53901750507311e-05,
+      "loss": 1.0889,
+      "step": 3585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.528756243824429e-05,
+      "loss": 1.266,
+      "step": 3586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.518499480602547e-05,
+      "loss": 0.9642,
+      "step": 3587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.508247219429253e-05,
+      "loss": 1.2279,
+      "step": 3588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.497999464324616e-05,
+      "loss": 0.8894,
+      "step": 3589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.487756219306908e-05,
+      "loss": 1.0761,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.477517488392639e-05,
+      "loss": 0.8548,
+      "step": 3591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.467283275596544e-05,
+      "loss": 1.3351,
+      "step": 3592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.457053584931613e-05,
+      "loss": 1.0107,
+      "step": 3593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.44682842040903e-05,
+      "loss": 0.8851,
+      "step": 3594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.43660778603822e-05,
+      "loss": 1.0255,
+      "step": 3595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.426391685826824e-05,
+      "loss": 0.8754,
+      "step": 3596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.416180123780722e-05,
+      "loss": 1.1676,
+      "step": 3597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.405973103903997e-05,
+      "loss": 0.8865,
+      "step": 3598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.395770630198953e-05,
+      "loss": 1.0105,
+      "step": 3599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.385572706666123e-05,
+      "loss": 1.1266,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.375379337304235e-05,
+      "loss": 1.0286,
+      "step": 3601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.36519052611026e-05,
+      "loss": 1.0135,
+      "step": 3602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.35500627707936e-05,
+      "loss": 1.2555,
+      "step": 3603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.344826594204911e-05,
+      "loss": 0.9714,
+      "step": 3604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.3346514814785e-05,
+      "loss": 1.2906,
+      "step": 3605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.324480942889932e-05,
+      "loss": 0.925,
+      "step": 3606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.314314982427198e-05,
+      "loss": 0.8268,
+      "step": 3607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.304153604076528e-05,
+      "loss": 1.1914,
+      "step": 3608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.293996811822304e-05,
+      "loss": 0.9956,
+      "step": 3609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.28384460964716e-05,
+      "loss": 1.0338,
+      "step": 3610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.273697001531897e-05,
+      "loss": 1.0054,
+      "step": 3611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.263553991455546e-05,
+      "loss": 1.1798,
+      "step": 3612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.253415583395286e-05,
+      "loss": 0.9824,
+      "step": 3613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.243281781326546e-05,
+      "loss": 0.9368,
+      "step": 3614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.233152589222908e-05,
+      "loss": 1.1793,
+      "step": 3615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.223028011056176e-05,
+      "loss": 0.8378,
+      "step": 3616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.212908050796326e-05,
+      "loss": 1.0781,
+      "step": 3617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.202792712411527e-05,
+      "loss": 1.0611,
+      "step": 3618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.192681999868133e-05,
+      "loss": 0.7624,
+      "step": 3619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.182575917130699e-05,
+      "loss": 0.8801,
+      "step": 3620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.172474468161948e-05,
+      "loss": 0.8787,
+      "step": 3621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.162377656922795e-05,
+      "loss": 1.0574,
+      "step": 3622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.152285487372331e-05,
+      "loss": 1.0797,
+      "step": 3623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.142197963467822e-05,
+      "loss": 0.8296,
+      "step": 3624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.132115089164734e-05,
+      "loss": 0.7731,
+      "step": 3625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.12203686841669e-05,
+      "loss": 0.9405,
+      "step": 3626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.111963305175494e-05,
+      "loss": 0.9874,
+      "step": 3627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.101894403391115e-05,
+      "loss": 1.0385,
+      "step": 3628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.091830167011718e-05,
+      "loss": 0.8501,
+      "step": 3629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.081770599983618e-05,
+      "loss": 1.0924,
+      "step": 3630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.071715706251304e-05,
+      "loss": 1.1283,
+      "step": 3631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.061665489757421e-05,
+      "loss": 1.1032,
+      "step": 3632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.051619954442815e-05,
+      "loss": 0.9404,
+      "step": 3633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.041579104246453e-05,
+      "loss": 1.0379,
+      "step": 3634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.031542943105511e-05,
+      "loss": 0.8437,
+      "step": 3635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.021511474955268e-05,
+      "loss": 0.9199,
+      "step": 3636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.01148470372922e-05,
+      "loss": 1.1248,
+      "step": 3637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.001462633358981e-05,
+      "loss": 0.8483,
+      "step": 3638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.99144526777436e-05,
+      "loss": 1.1014,
+      "step": 3639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.981432610903273e-05,
+      "loss": 1.1491,
+      "step": 3640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.971424666671832e-05,
+      "loss": 1.1648,
+      "step": 3641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.96142143900428e-05,
+      "loss": 0.897,
+      "step": 3642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.951422931823009e-05,
+      "loss": 1.1075,
+      "step": 3643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.941429149048578e-05,
+      "loss": 1.142,
+      "step": 3644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.931440094599677e-05,
+      "loss": 1.014,
+      "step": 3645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.921455772393145e-05,
+      "loss": 0.9838,
+      "step": 3646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.911476186343962e-05,
+      "loss": 0.8105,
+      "step": 3647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.901501340365276e-05,
+      "loss": 1.0193,
+      "step": 3648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.891531238368328e-05,
+      "loss": 0.9384,
+      "step": 3649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.881565884262547e-05,
+      "loss": 1.0874,
+      "step": 3650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.87160528195547e-05,
+      "loss": 1.0233,
+      "step": 3651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.8616494353528e-05,
+      "loss": 1.0843,
+      "step": 3652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.851698348358327e-05,
+      "loss": 0.9888,
+      "step": 3653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.841752024874029e-05,
+      "loss": 1.2995,
+      "step": 3654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.831810468799974e-05,
+      "loss": 0.854,
+      "step": 3655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.821873684034394e-05,
+      "loss": 1.1173,
+      "step": 3656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.811941674473624e-05,
+      "loss": 0.9192,
+      "step": 3657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.802014444012142e-05,
+      "loss": 1.2384,
+      "step": 3658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.792091996542534e-05,
+      "loss": 1.0687,
+      "step": 3659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.782174335955541e-05,
+      "loss": 1.043,
+      "step": 3660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.772261466140001e-05,
+      "loss": 0.844,
+      "step": 3661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.762353390982882e-05,
+      "loss": 0.8825,
+      "step": 3662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.752450114369262e-05,
+      "loss": 0.7919,
+      "step": 3663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.742551640182367e-05,
+      "loss": 1.0134,
+      "step": 3664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.732657972303508e-05,
+      "loss": 0.7302,
+      "step": 3665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.722769114612123e-05,
+      "loss": 1.1932,
+      "step": 3666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.712885070985769e-05,
+      "loss": 1.0045,
+      "step": 3667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.703005845300097e-05,
+      "loss": 0.9467,
+      "step": 3668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.693131441428902e-05,
+      "loss": 1.0056,
+      "step": 3669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.683261863244063e-05,
+      "loss": 0.9301,
+      "step": 3670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.673397114615564e-05,
+      "loss": 1.4037,
+      "step": 3671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.663537199411504e-05,
+      "loss": 0.9793,
+      "step": 3672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.653682121498099e-05,
+      "loss": 0.8361,
+      "step": 3673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.64383188473964e-05,
+      "loss": 1.2425,
+      "step": 3674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.633986492998557e-05,
+      "loss": 1.1341,
+      "step": 3675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.624145950135333e-05,
+      "loss": 0.9567,
+      "step": 3676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.614310260008594e-05,
+      "loss": 0.9809,
+      "step": 3677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.60447942647503e-05,
+      "loss": 0.7337,
+      "step": 3678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.594653453389464e-05,
+      "loss": 0.9649,
+      "step": 3679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.584832344604762e-05,
+      "loss": 0.9035,
+      "step": 3680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.575016103971933e-05,
+      "loss": 0.9586,
+      "step": 3681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.565204735340036e-05,
+      "loss": 1.1162,
+      "step": 3682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.55539824255626e-05,
+      "loss": 0.9139,
+      "step": 3683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.54559662946585e-05,
+      "loss": 0.8866,
+      "step": 3684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.535799899912145e-05,
+      "loss": 0.6938,
+      "step": 3685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.52600805773658e-05,
+      "loss": 0.8513,
+      "step": 3686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.516221106778656e-05,
+      "loss": 0.8315,
+      "step": 3687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.506439050875978e-05,
+      "loss": 0.9135,
+      "step": 3688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.496661893864215e-05,
+      "loss": 0.9496,
+      "step": 3689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.48688963957712e-05,
+      "loss": 0.9519,
+      "step": 3690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.477122291846515e-05,
+      "loss": 0.8601,
+      "step": 3691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.46735985450232e-05,
+      "loss": 0.9256,
+      "step": 3692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.457602331372514e-05,
+      "loss": 1.0453,
+      "step": 3693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.447849726283147e-05,
+      "loss": 0.9557,
+      "step": 3694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.438102043058336e-05,
+      "loss": 0.8458,
+      "step": 3695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.428359285520297e-05,
+      "loss": 1.0933,
+      "step": 3696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.418621457489283e-05,
+      "loss": 1.2727,
+      "step": 3697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.408888562783624e-05,
+      "loss": 1.1454,
+      "step": 3698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.399160605219715e-05,
+      "loss": 1.1143,
+      "step": 3699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.389437588612029e-05,
+      "loss": 1.1086,
+      "step": 3700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.379719516773074e-05,
+      "loss": 0.8956,
+      "step": 3701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.370006393513457e-05,
+      "loss": 1.1748,
+      "step": 3702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.360298222641798e-05,
+      "loss": 1.1452,
+      "step": 3703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.350595007964813e-05,
+      "loss": 0.9435,
+      "step": 3704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.340896753287261e-05,
+      "loss": 1.0379,
+      "step": 3705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.331203462411951e-05,
+      "loss": 1.0239,
+      "step": 3706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.32151513913975e-05,
+      "loss": 1.0141,
+      "step": 3707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.311831787269586e-05,
+      "loss": 0.9176,
+      "step": 3708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.302153410598424e-05,
+      "loss": 0.8949,
+      "step": 3709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.292480012921281e-05,
+      "loss": 1.1127,
+      "step": 3710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.28281159803124e-05,
+      "loss": 0.7791,
+      "step": 3711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.27314816971939e-05,
+      "loss": 0.9663,
+      "step": 3712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.263489731774911e-05,
+      "loss": 0.9087,
+      "step": 3713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.253836287984987e-05,
+      "loss": 0.6508,
+      "step": 3714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.244187842134886e-05,
+      "loss": 1.2082,
+      "step": 3715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.234544398007863e-05,
+      "loss": 1.316,
+      "step": 3716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.224905959385258e-05,
+      "loss": 1.0667,
+      "step": 3717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.21527253004642e-05,
+      "loss": 0.8578,
+      "step": 3718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.205644113768757e-05,
+      "loss": 1.0265,
+      "step": 3719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.196020714327692e-05,
+      "loss": 0.9806,
+      "step": 3720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.186402335496687e-05,
+      "loss": 1.0956,
+      "step": 3721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.176788981047231e-05,
+      "loss": 0.9998,
+      "step": 3722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.16718065474886e-05,
+      "loss": 0.7369,
+      "step": 3723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.157577360369116e-05,
+      "loss": 1.1141,
+      "step": 3724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.147979101673585e-05,
+      "loss": 0.7825,
+      "step": 3725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.13838588242586e-05,
+      "loss": 1.1145,
+      "step": 3726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.128797706387585e-05,
+      "loss": 0.8878,
+      "step": 3727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.119214577318403e-05,
+      "loss": 0.9914,
+      "step": 3728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.109636498975985e-05,
+      "loss": 1.1366,
+      "step": 3729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.100063475116026e-05,
+      "loss": 0.7963,
+      "step": 3730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.090495509492223e-05,
+      "loss": 1.1307,
+      "step": 3731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.080932605856323e-05,
+      "loss": 0.9249,
+      "step": 3732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.071374767958055e-05,
+      "loss": 0.6992,
+      "step": 3733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.061821999545177e-05,
+      "loss": 0.856,
+      "step": 3734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.052274304363449e-05,
+      "loss": 0.7369,
+      "step": 3735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.042731686156659e-05,
+      "loss": 1.3029,
+      "step": 3736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.033194148666587e-05,
+      "loss": 1.2308,
+      "step": 3737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.023661695633042e-05,
+      "loss": 0.9905,
+      "step": 3738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0141343307938e-05,
+      "loss": 0.8462,
+      "step": 3739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.00461205788469e-05,
+      "loss": 0.8372,
+      "step": 3740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.995094880639506e-05,
+      "loss": 0.7861,
+      "step": 3741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.985582802790078e-05,
+      "loss": 0.9877,
+      "step": 3742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.976075828066196e-05,
+      "loss": 1.0601,
+      "step": 3743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.966573960195683e-05,
+      "loss": 0.9032,
+      "step": 3744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.957077202904342e-05,
+      "loss": 0.8447,
+      "step": 3745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.947585559915988e-05,
+      "loss": 0.6936,
+      "step": 3746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.93809903495241e-05,
+      "loss": 1.0104,
+      "step": 3747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.928617631733405e-05,
+      "loss": 0.852,
+      "step": 3748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.919141353976748e-05,
+      "loss": 0.7798,
+      "step": 3749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.909670205398227e-05,
+      "loss": 0.8654,
+      "step": 3750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.900204189711599e-05,
+      "loss": 1.195,
+      "step": 3751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.890743310628611e-05,
+      "loss": 0.8512,
+      "step": 3752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.881287571859003e-05,
+      "loss": 1.2082,
+      "step": 3753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.871836977110489e-05,
+      "loss": 0.9298,
+      "step": 3754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.862391530088781e-05,
+      "loss": 1.2007,
+      "step": 3755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.852951234497562e-05,
+      "loss": 0.8172,
+      "step": 3756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.843516094038498e-05,
+      "loss": 0.8038,
+      "step": 3757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.834086112411222e-05,
+      "loss": 0.8543,
+      "step": 3758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.824661293313367e-05,
+      "loss": 0.9038,
+      "step": 3759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.815241640440531e-05,
+      "loss": 1.153,
+      "step": 3760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.805827157486278e-05,
+      "loss": 1.219,
+      "step": 3761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.796417848142144e-05,
+      "loss": 1.052,
+      "step": 3762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.787013716097661e-05,
+      "loss": 0.8572,
+      "step": 3763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.777614765040308e-05,
+      "loss": 1.16,
+      "step": 3764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.768220998655535e-05,
+      "loss": 1.0191,
+      "step": 3765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.758832420626758e-05,
+      "loss": 1.1438,
+      "step": 3766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.749449034635373e-05,
+      "loss": 1.0139,
+      "step": 3767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.74007084436072e-05,
+      "loss": 0.9695,
+      "step": 3768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.73069785348013e-05,
+      "loss": 1.1478,
+      "step": 3769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.721330065668853e-05,
+      "loss": 1.0586,
+      "step": 3770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.71196748460014e-05,
+      "loss": 0.7982,
+      "step": 3771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.702610113945179e-05,
+      "loss": 0.8852,
+      "step": 3772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.693257957373116e-05,
+      "loss": 0.9663,
+      "step": 3773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.683911018551052e-05,
+      "loss": 1.0539,
+      "step": 3774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.674569301144056e-05,
+      "loss": 1.0815,
+      "step": 3775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.665232808815136e-05,
+      "loss": 1.032,
+      "step": 3776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.655901545225244e-05,
+      "loss": 0.8661,
+      "step": 3777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.646575514033315e-05,
+      "loss": 0.8869,
+      "step": 3778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.637254718896182e-05,
+      "loss": 0.7618,
+      "step": 3779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.627939163468672e-05,
+      "loss": 1.2375,
+      "step": 3780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.618628851403526e-05,
+      "loss": 0.9853,
+      "step": 3781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.609323786351458e-05,
+      "loss": 1.1127,
+      "step": 3782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.600023971961085e-05,
+      "loss": 0.9989,
+      "step": 3783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.590729411879002e-05,
+      "loss": 1.0279,
+      "step": 3784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.581440109749719e-05,
+      "loss": 0.7989,
+      "step": 3785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.57215606921571e-05,
+      "loss": 1.0877,
+      "step": 3786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.56287729391736e-05,
+      "loss": 0.8807,
+      "step": 3787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.553603787493002e-05,
+      "loss": 0.8585,
+      "step": 3788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.544335553578891e-05,
+      "loss": 0.8323,
+      "step": 3789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.535072595809242e-05,
+      "loss": 0.8297,
+      "step": 3790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.525814917816176e-05,
+      "loss": 0.9362,
+      "step": 3791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.516562523229748e-05,
+      "loss": 0.9984,
+      "step": 3792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.507315415677942e-05,
+      "loss": 1.205,
+      "step": 3793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.498073598786682e-05,
+      "loss": 1.1961,
+      "step": 3794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4888370761798e-05,
+      "loss": 1.126,
+      "step": 3795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.479605851479065e-05,
+      "loss": 0.9271,
+      "step": 3796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.470379928304156e-05,
+      "loss": 1.0838,
+      "step": 3797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.461159310272673e-05,
+      "loss": 0.904,
+      "step": 3798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.45194400100016e-05,
+      "loss": 1.0287,
+      "step": 3799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.442734004100055e-05,
+      "loss": 0.9719,
+      "step": 3800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.433529323183715e-05,
+      "loss": 0.9536,
+      "step": 3801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.424329961860415e-05,
+      "loss": 1.0559,
+      "step": 3802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.415135923737362e-05,
+      "loss": 0.685,
+      "step": 3803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.40594721241964e-05,
+      "loss": 1.1417,
+      "step": 3804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.39676383151029e-05,
+      "loss": 1.0539,
+      "step": 3805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.38758578461021e-05,
+      "loss": 0.8861,
+      "step": 3806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.378413075318253e-05,
+      "loss": 1.0804,
+      "step": 3807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.369245707231144e-05,
+      "loss": 1.2182,
+      "step": 3808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.360083683943554e-05,
+      "loss": 0.9635,
+      "step": 3809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.350927009048007e-05,
+      "loss": 1.1038,
+      "step": 3810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.341775686134974e-05,
+      "loss": 0.791,
+      "step": 3811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.332629718792798e-05,
+      "loss": 1.031,
+      "step": 3812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.323489110607745e-05,
+      "loss": 0.8553,
+      "step": 3813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.314353865163966e-05,
+      "loss": 1.0315,
+      "step": 3814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.305223986043511e-05,
+      "loss": 1.0831,
+      "step": 3815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.296099476826327e-05,
+      "loss": 1.0033,
+      "step": 3816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.286980341090245e-05,
+      "loss": 1.1202,
+      "step": 3817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.277866582411014e-05,
+      "loss": 0.65,
+      "step": 3818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.268758204362255e-05,
+      "loss": 1.1257,
+      "step": 3819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.25965521051548e-05,
+      "loss": 1.0086,
+      "step": 3820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.250557604440093e-05,
+      "loss": 1.1591,
+      "step": 3821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.241465389703393e-05,
+      "loss": 1.2009,
+      "step": 3822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.232378569870554e-05,
+      "loss": 1.114,
+      "step": 3823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.223297148504639e-05,
+      "loss": 1.4721,
+      "step": 3824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.214221129166586e-05,
+      "loss": 0.7438,
+      "step": 3825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.205150515415237e-05,
+      "loss": 0.9147,
+      "step": 3826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.196085310807294e-05,
+      "loss": 0.9571,
+      "step": 3827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.18702551889734e-05,
+      "loss": 1.0536,
+      "step": 3828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.177971143237835e-05,
+      "loss": 0.8341,
+      "step": 3829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.168922187379136e-05,
+      "loss": 1.1392,
+      "step": 3830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.159878654869442e-05,
+      "loss": 1.0676,
+      "step": 3831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.150840549254862e-05,
+      "loss": 1.0152,
+      "step": 3832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.14180787407933e-05,
+      "loss": 0.9525,
+      "step": 3833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1327806328847e-05,
+      "loss": 1.0634,
+      "step": 3834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.123758829210666e-05,
+      "loss": 0.973,
+      "step": 3835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.114742466594796e-05,
+      "loss": 0.776,
+      "step": 3836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.10573154857252e-05,
+      "loss": 0.9274,
+      "step": 3837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0967260786771485e-05,
+      "loss": 1.1305,
+      "step": 3838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.08772606043984e-05,
+      "loss": 0.9177,
+      "step": 3839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.078731497389618e-05,
+      "loss": 0.6796,
+      "step": 3840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.069742393053384e-05,
+      "loss": 1.154,
+      "step": 3841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.060758750955864e-05,
+      "loss": 1.1916,
+      "step": 3842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.051780574619681e-05,
+      "loss": 1.0458,
+      "step": 3843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.042807867565283e-05,
+      "loss": 0.8816,
+      "step": 3844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.033840633311005e-05,
+      "loss": 1.0665,
+      "step": 3845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.024878875372999e-05,
+      "loss": 0.8605,
+      "step": 3846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.015922597265302e-05,
+      "loss": 1.1478,
+      "step": 3847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0069718024997814e-05,
+      "loss": 1.0069,
+      "step": 3848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.998026494586171e-05,
+      "loss": 0.9647,
+      "step": 3849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9890866770320405e-05,
+      "loss": 1.1356,
+      "step": 3850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9801523533428116e-05,
+      "loss": 0.9472,
+      "step": 3851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.971223527021743e-05,
+      "loss": 0.967,
+      "step": 3852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9623002015699593e-05,
+      "loss": 0.8707,
+      "step": 3853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9533823804864095e-05,
+      "loss": 1.2539,
+      "step": 3854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.944470067267887e-05,
+      "loss": 1.3119,
+      "step": 3855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.935563265409025e-05,
+      "loss": 0.8034,
+      "step": 3856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.926661978402308e-05,
+      "loss": 0.991,
+      "step": 3857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.917766209738045e-05,
+      "loss": 1.0007,
+      "step": 3858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.908875962904383e-05,
+      "loss": 1.063,
+      "step": 3859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.899991241387306e-05,
+      "loss": 1.0316,
+      "step": 3860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.8911120486706237e-05,
+      "loss": 0.8768,
+      "step": 3861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.882238388236001e-05,
+      "loss": 1.0428,
+      "step": 3862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.873370263562912e-05,
+      "loss": 0.9335,
+      "step": 3863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.864507678128661e-05,
+      "loss": 1.0952,
+      "step": 3864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8556506354083805e-05,
+      "loss": 1.1125,
+      "step": 3865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.846799138875052e-05,
+      "loss": 0.7687,
+      "step": 3866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8379531919994526e-05,
+      "loss": 0.9197,
+      "step": 3867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.829112798250198e-05,
+      "loss": 0.523,
+      "step": 3868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8202779610937164e-05,
+      "loss": 1.086,
+      "step": 3869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.811448683994278e-05,
+      "loss": 0.9947,
+      "step": 3870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8026249704139456e-05,
+      "loss": 0.9043,
+      "step": 3871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.793806823812635e-05,
+      "loss": 1.1352,
+      "step": 3872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7849942476480324e-05,
+      "loss": 0.9973,
+      "step": 3873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.776187245375684e-05,
+      "loss": 1.0288,
+      "step": 3874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7673858204489205e-05,
+      "loss": 0.8714,
+      "step": 3875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.758589976318918e-05,
+      "loss": 1.1538,
+      "step": 3876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.749799716434614e-05,
+      "loss": 1.1456,
+      "step": 3877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.741015044242808e-05,
+      "loss": 0.7869,
+      "step": 3878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.732235963188073e-05,
+      "loss": 0.851,
+      "step": 3879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.723462476712815e-05,
+      "loss": 0.91,
+      "step": 3880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.71469458825723e-05,
+      "loss": 0.9833,
+      "step": 3881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7059323012593205e-05,
+      "loss": 0.967,
+      "step": 3882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6971756191548995e-05,
+      "loss": 1.0288,
+      "step": 3883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6884245453775685e-05,
+      "loss": 0.8033,
+      "step": 3884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.679679083358751e-05,
+      "loss": 0.9906,
+      "step": 3885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.670939236527657e-05,
+      "loss": 1.0285,
+      "step": 3886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6622050083112944e-05,
+      "loss": 1.1295,
+      "step": 3887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.653476402134461e-05,
+      "loss": 0.9197,
+      "step": 3888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.644753421419773e-05,
+      "loss": 1.0887,
+      "step": 3889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6360360695876225e-05,
+      "loss": 1.1151,
+      "step": 3890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.627324350056198e-05,
+      "loss": 1.167,
+      "step": 3891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6186182662414735e-05,
+      "loss": 1.0553,
+      "step": 3892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.609917821557231e-05,
+      "loss": 0.9562,
+      "step": 3893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.601223019415025e-05,
+      "loss": 1.2249,
+      "step": 3894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.592533863224203e-05,
+      "loss": 1.1945,
+      "step": 3895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.583850356391893e-05,
+      "loss": 0.9755,
+      "step": 3896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.575172502323025e-05,
+      "loss": 0.7047,
+      "step": 3897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.566500304420286e-05,
+      "loss": 0.9752,
+      "step": 3898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5578337660841814e-05,
+      "loss": 1.131,
+      "step": 3899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.54917289071295e-05,
+      "loss": 0.936,
+      "step": 3900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.540517681702655e-05,
+      "loss": 0.9278,
+      "step": 3901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.531868142447112e-05,
+      "loss": 1.1244,
+      "step": 3902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.523224276337919e-05,
+      "loss": 0.6069,
+      "step": 3903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.514586086764446e-05,
+      "loss": 0.9169,
+      "step": 3904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.505953577113854e-05,
+      "loss": 1.3792,
+      "step": 3905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4973267507710546e-05,
+      "loss": 0.9719,
+      "step": 3906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4887056111187404e-05,
+      "loss": 0.8986,
+      "step": 3907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4800901615373876e-05,
+      "loss": 1.1967,
+      "step": 3908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.471480405405209e-05,
+      "loss": 1.2089,
+      "step": 3909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.462876346098218e-05,
+      "loss": 0.8369,
+      "step": 3910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.45427798699017e-05,
+      "loss": 0.959,
+      "step": 3911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.445685331452617e-05,
+      "loss": 1.0444,
+      "step": 3912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4370983828548213e-05,
+      "loss": 0.7501,
+      "step": 3913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.428517144563865e-05,
+      "loss": 1.05,
+      "step": 3914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.419941619944549e-05,
+      "loss": 0.8504,
+      "step": 3915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.411371812359465e-05,
+      "loss": 1.0192,
+      "step": 3916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.402807725168939e-05,
+      "loss": 1.1557,
+      "step": 3917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3942493617310654e-05,
+      "loss": 0.977,
+      "step": 3918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.385696725401685e-05,
+      "loss": 1.2873,
+      "step": 3919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.377149819534413e-05,
+      "loss": 1.1183,
+      "step": 3920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3686086474805975e-05,
+      "loss": 0.9059,
+      "step": 3921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.360073212589345e-05,
+      "loss": 0.9563,
+      "step": 3922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3515435182075094e-05,
+      "loss": 1.1424,
+      "step": 3923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.343019567679706e-05,
+      "loss": 1.1084,
+      "step": 3924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.334501364348283e-05,
+      "loss": 0.9887,
+      "step": 3925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3259889115533415e-05,
+      "loss": 1.1131,
+      "step": 3926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.317482212632727e-05,
+      "loss": 0.9454,
+      "step": 3927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.308981270922027e-05,
+      "loss": 1.0716,
+      "step": 3928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.300486089754577e-05,
+      "loss": 0.8237,
+      "step": 3929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.291996672461452e-05,
+      "loss": 1.0026,
+      "step": 3930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.28351302237146e-05,
+      "loss": 1.0834,
+      "step": 3931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.275035142811149e-05,
+      "loss": 0.9548,
+      "step": 3932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.266563037104821e-05,
+      "loss": 1.1453,
+      "step": 3933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.258096708574487e-05,
+      "loss": 0.9101,
+      "step": 3934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2496361605399256e-05,
+      "loss": 1.2423,
+      "step": 3935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.241181396318604e-05,
+      "loss": 0.6983,
+      "step": 3936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.23273241922577e-05,
+      "loss": 1.1268,
+      "step": 3937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.224289232574362e-05,
+      "loss": 0.8865,
+      "step": 3938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.215851839675088e-05,
+      "loss": 0.9077,
+      "step": 3939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.207420243836334e-05,
+      "loss": 1.1332,
+      "step": 3940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.198994448364261e-05,
+      "loss": 0.9379,
+      "step": 3941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.19057445656272e-05,
+      "loss": 0.6698,
+      "step": 3942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1821602717333186e-05,
+      "loss": 0.9815,
+      "step": 3943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.173751897175361e-05,
+      "loss": 1.1755,
+      "step": 3944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.165349336185881e-05,
+      "loss": 0.9008,
+      "step": 3945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.156952592059636e-05,
+      "loss": 1.1585,
+      "step": 3946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1485616680890956e-05,
+      "loss": 1.1158,
+      "step": 3947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.140176567564463e-05,
+      "loss": 1.1494,
+      "step": 3948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1317972937736436e-05,
+      "loss": 0.8812,
+      "step": 3949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.123423850002258e-05,
+      "loss": 0.9504,
+      "step": 3950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.11505623953364e-05,
+      "loss": 0.8357,
+      "step": 3951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.106694465648853e-05,
+      "loss": 0.5476,
+      "step": 3952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.098338531626654e-05,
+      "loss": 1.142,
+      "step": 3953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.089988440743516e-05,
+      "loss": 0.9295,
+      "step": 3954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.081644196273612e-05,
+      "loss": 1.0321,
+      "step": 3955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0733058014888436e-05,
+      "loss": 1.4322,
+      "step": 3956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.064973259658803e-05,
+      "loss": 0.7571,
+      "step": 3957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.056646574050785e-05,
+      "loss": 0.6699,
+      "step": 3958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.048325747929792e-05,
+      "loss": 1.0966,
+      "step": 3959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.04001078455854e-05,
+      "loss": 1.0317,
+      "step": 3960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0317016871974256e-05,
+      "loss": 0.9864,
+      "step": 3961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0233984591045737e-05,
+      "loss": 0.9563,
+      "step": 3962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.015101103535764e-05,
+      "loss": 0.8567,
+      "step": 3963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0068096237445214e-05,
+      "loss": 0.9572,
+      "step": 3964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.998524022982039e-05,
+      "loss": 0.9765,
+      "step": 3965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9902443044972045e-05,
+      "loss": 0.8893,
+      "step": 3966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.981970471536606e-05,
+      "loss": 0.9461,
+      "step": 3967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9737025273445326e-05,
+      "loss": 0.9323,
+      "step": 3968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.96544047516295e-05,
+      "loss": 1.0907,
+      "step": 3969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.957184318231512e-05,
+      "loss": 0.98,
+      "step": 3970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9489340597875824e-05,
+      "loss": 1.1301,
+      "step": 3971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.940689703066177e-05,
+      "loss": 1.1273,
+      "step": 3972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.932451251300032e-05,
+      "loss": 0.8798,
+      "step": 3973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.924218707719546e-05,
+      "loss": 0.9905,
+      "step": 3974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.91599207555282e-05,
+      "loss": 0.7881,
+      "step": 3975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.907771358025608e-05,
+      "loss": 1.381,
+      "step": 3976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.899556558361378e-05,
+      "loss": 0.9313,
+      "step": 3977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8913476797812486e-05,
+      "loss": 0.8101,
+      "step": 3978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.883144725504047e-05,
+      "loss": 0.8775,
+      "step": 3979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.874947698746242e-05,
+      "loss": 1.219,
+      "step": 3980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.866756602722009e-05,
+      "loss": 0.6577,
+      "step": 3981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8585714406431734e-05,
+      "loss": 0.6963,
+      "step": 3982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.850392215719261e-05,
+      "loss": 0.9485,
+      "step": 3983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.842218931157447e-05,
+      "loss": 1.1303,
+      "step": 3984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.834051590162585e-05,
+      "loss": 0.8343,
+      "step": 3985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.825890195937191e-05,
+      "loss": 1.149,
+      "step": 3986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8177347516814665e-05,
+      "loss": 0.6877,
+      "step": 3987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.809585260593266e-05,
+      "loss": 1.2502,
+      "step": 3988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.801441725868112e-05,
+      "loss": 0.8111,
+      "step": 3989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793304150699192e-05,
+      "loss": 0.7562,
+      "step": 3990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7851725382773514e-05,
+      "loss": 0.8836,
+      "step": 3991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.777046891791117e-05,
+      "loss": 0.9518,
+      "step": 3992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.768927214426653e-05,
+      "loss": 0.7618,
+      "step": 3993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7608135093677966e-05,
+      "loss": 0.8272,
+      "step": 3994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7527057797960293e-05,
+      "loss": 0.9495,
+      "step": 3995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.744604028890511e-05,
+      "loss": 1.1262,
+      "step": 3996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7365082598280454e-05,
+      "loss": 0.7091,
+      "step": 3997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.728418475783085e-05,
+      "loss": 0.8026,
+      "step": 3998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.72033467992774e-05,
+      "loss": 0.9071,
+      "step": 3999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.712256875431782e-05,
+      "loss": 0.9876,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 5117,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 4.026612723923026e+17,
+  "trial_name": null,
+  "trial_params": null
+}