diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,45619 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.30490251143384417,
+  "eval_steps": 500,
+  "global_step": 7600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 0.6082,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 0.6457,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 0.5707,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 0.9886,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 0.9204,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 1.0933,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.4468,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 1.5412,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 0.7863,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.944,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 1.3362,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 0.4761,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9e-05,
+      "loss": 1.0441,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-05,
+      "loss": 0.8828,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 0.9583,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.4751,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.1e-05,
+      "loss": 0.8238,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 0.5497,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 1.4629,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 1.0918,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 0.2344,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 0.8805,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.9e-05,
+      "loss": 0.7408,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.3671,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-05,
+      "loss": 0.5042,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-05,
+      "loss": 0.8721,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.1e-05,
+      "loss": 0.8938,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4e-05,
+      "loss": 1.0697,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 1.5208,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 1.1255,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.3e-05,
+      "loss": 0.8054,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 1.2554,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9e-05,
+      "loss": 0.339,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000102,
+      "loss": 1.6215,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 0.2207,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.7652,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 0.8505,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.3423,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000117,
+      "loss": 0.7405,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 2.8693,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 0.4785,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 0.1813,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000129,
+      "loss": 1.2773,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.4066,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000135,
+      "loss": 1.0392,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000138,
+      "loss": 0.6844,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 1.027,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.4649,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000147,
+      "loss": 0.6427,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015,
+      "loss": 0.5877,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 0.0119,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000156,
+      "loss": 0.5225,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000159,
+      "loss": 2.098,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000162,
+      "loss": 0.5874,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000165,
+      "loss": 0.3626,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000168,
+      "loss": 0.9532,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 0.5811,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 5.7897,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 0.5894,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.4869,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 0.4649,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000186,
+      "loss": 0.6504,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 1.0623,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.2459,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000195,
+      "loss": 1.4967,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000198,
+      "loss": 0.0267,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000201,
+      "loss": 1.2445,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000204,
+      "loss": 0.4555,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 0.8448,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.0172,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 0.793,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.4777,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 0.1839,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.5815,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 2.0219,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000225,
+      "loss": 2.3776,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.8549,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 0.4429,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000234,
+      "loss": 0.1514,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000237,
+      "loss": 2.9324,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.3932,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000243,
+      "loss": 0.5469,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 1.1362,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000249,
+      "loss": 0.8098,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.2951,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 0.8272,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000258,
+      "loss": 1.0487,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000261,
+      "loss": 1.3986,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 1.5231,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000267,
+      "loss": 0.9686,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00027,
+      "loss": 1.085,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 0.3078,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000276,
+      "loss": 0.9379,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000279,
+      "loss": 0.9469,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.5037,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000285,
+      "loss": 1.5305,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.5242,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 2.1693,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000294,
+      "loss": 0.534,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 1.166,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003,
+      "loss": 0.4421,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999999879898754,
+      "loss": 0.4234,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999999519595024,
+      "loss": 0.5496,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999999891908881,
+      "loss": 0.8172,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999999807838013,
+      "loss": 0.3801,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999996997468994,
+      "loss": 0.8991,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999995676355414,
+      "loss": 0.47,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999999411503942,
+      "loss": 1.6896,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999992313521027,
+      "loss": 0.3518,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999990271800263,
+      "loss": 0.734,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999987989877175,
+      "loss": 0.4002,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999985467751795,
+      "loss": 0.1489,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999998270542415,
+      "loss": 0.5509,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000299999797028943,
+      "loss": 0.8304,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999976460162286,
+      "loss": 0.6136,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999972977228157,
+      "loss": 1.107,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999969254091975,
+      "loss": 0.5259,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000299999652907538,
+      "loss": 2.1946,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999996108721369,
+      "loss": 0.3648,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999995664347172,
+      "loss": 0.6461,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999995195952794,
+      "loss": 0.5963,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999994703538246,
+      "loss": 0.5154,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999994187103533,
+      "loss": 0.8321,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999936466486646,
+      "loss": 0.9619,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999930821736495,
+      "loss": 0.9212,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999992493678495,
+      "loss": 0.8118,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999918811632136,
+      "loss": 1.0119,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999912446278124,
+      "loss": 0.526,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999990584072303,
+      "loss": 0.7132,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999989899496695,
+      "loss": 0.6287,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999891909010003,
+      "loss": 0.1722,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299998845828523,
+      "loss": 1.661,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999987701649395,
+      "loss": 0.9841,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999986920993509,
+      "loss": 0.764,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999986116317583,
+      "loss": 0.5235,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999985287621632,
+      "loss": 0.8351,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999984434905666,
+      "loss": 0.3025,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999983558169701,
+      "loss": 0.4589,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999982657413751,
+      "loss": 0.5024,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999981732637829,
+      "loss": 3.0238,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999980783841951,
+      "loss": 0.7635,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999979811026132,
+      "loss": 0.384,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999788141903876,
+      "loss": 0.4209,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999977793334734,
+      "loss": 0.8944,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999767484591866,
+      "loss": 0.7077,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999756795637625,
+      "loss": 0.5496,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999745866484794,
+      "loss": 0.6808,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999734697133543,
+      "loss": 0.3555,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999723287584055,
+      "loss": 0.55,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999971163783651,
+      "loss": 0.4174,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999699747891096,
+      "loss": 0.2511,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999687617748,
+      "loss": 0.3682,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999967524740741,
+      "loss": 0.9861,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999966263686955,
+      "loss": 0.4334,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999964978613459,
+      "loss": 0.3246,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999963669520275,
+      "loss": 0.4127,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999623364074246,
+      "loss": 1.2994,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999960979274928,
+      "loss": 0.3928,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999959598122808,
+      "loss": 0.5447,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999581929510854,
+      "loss": 1.3063,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999567637597845,
+      "loss": 0.7071,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999955310548926,
+      "loss": 0.7096,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999953833318535,
+      "loss": 0.3107,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999523320686344,
+      "loss": 1.7141,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999950806799249,
+      "loss": 0.4814,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999949257510401,
+      "loss": 0.4892,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999947684202118,
+      "loss": 0.5439,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999460868744234,
+      "loss": 0.5369,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999444655273434,
+      "loss": 0.5828,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999942820160904,
+      "loss": 0.8133,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999941150775131,
+      "loss": 0.723,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999394573700523,
+      "loss": 2.4967,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999937739945694,
+      "loss": 0.6977,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999935998502084,
+      "loss": 0.0939,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299993423303925,
+      "loss": 0.9066,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299993244355722,
+      "loss": 0.8641,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999306300560233,
+      "loss": 0.3496,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999928792535688,
+      "loss": 0.4615,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999269309962447,
+      "loss": 0.9519,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999925045437723,
+      "loss": 0.5484,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999923135860153,
+      "loss": 5.6815,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999921202263564,
+      "loss": 0.2777,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999192446479887,
+      "loss": 0.0841,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999917263013457,
+      "loss": 0.1023,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999152573600025,
+      "loss": 0.5286,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999132276876554,
+      "loss": 0.8989,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999111739964496,
+      "loss": 1.5437,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999909096286417,
+      "loss": 1.1663,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999069945575913,
+      "loss": 0.4474,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999904868810006,
+      "loss": 0.0132,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999902719043696,
+      "loss": 0.1196,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999900545258694,
+      "loss": 0.8682,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998983474550364,
+      "loss": 0.7771,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999896125632758,
+      "loss": 0.5106,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999893879791893,
+      "loss": 2.946,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299989160993248,
+      "loss": 1.9552,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999889316054553,
+      "loss": 1.8721,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299988699815815,
+      "loss": 0.0067,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998846562433076,
+      "loss": 1.5086,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999882290310064,
+      "loss": 0.6052,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999879900358456,
+      "loss": 0.0507,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999877486388522,
+      "loss": 1.2328,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998750484003017,
+      "loss": 0.8417,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998725863938333,
+      "loss": 0.3224,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999870100369156,
+      "loss": 0.5401,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998675903263104,
+      "loss": 0.804,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998650562653367,
+      "loss": 4.3887,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998624981862745,
+      "loss": 0.0483,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998599160891657,
+      "loss": 0.6673,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998573099740514,
+      "loss": 0.2473,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999854679840973,
+      "loss": 0.6671,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998520256899734,
+      "loss": 0.2406,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998493475210937,
+      "loss": 0.4182,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999846645334378,
+      "loss": 2.0499,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999843919129869,
+      "loss": 0.4257,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999841168907611,
+      "loss": 0.0019,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998383946676474,
+      "loss": 0.7959,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998355964100235,
+      "loss": 0.1089,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999832774134783,
+      "loss": 2.9715,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998299278419715,
+      "loss": 0.7391,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998270575316345,
+      "loss": 0.7319,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998241632038183,
+      "loss": 3.2964,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998212448585686,
+      "loss": 0.3525,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999818302495933,
+      "loss": 0.8063,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998153361159583,
+      "loss": 0.2223,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999812345718692,
+      "loss": 0.0074,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999809331304182,
+      "loss": 0.5793,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029998062928724757,
+      "loss": 0.5266,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999803230423623,
+      "loss": 0.5553,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999803230423623,
+      "loss": 1.6031,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999800143957672,
+      "loss": 0.7125,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999797033474673,
+      "loss": 0.6833,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999793898974675,
+      "loss": 0.5578,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997907404577295,
+      "loss": 0.521,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999787557923886,
+      "loss": 0.424,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997843513731947,
+      "loss": 0.3912,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997811208057085,
+      "loss": 1.1264,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997778662214784,
+      "loss": 1.017,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999774587620556,
+      "loss": 0.7859,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997712850029953,
+      "loss": 0.7413,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999767958368848,
+      "loss": 2.7864,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999764607718168,
+      "loss": 0.0075,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997612330510086,
+      "loss": 0.5556,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997578343674236,
+      "loss": 3.1132,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997544116674676,
+      "loss": 1.5589,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999750964951196,
+      "loss": 1.1716,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999747494218663,
+      "loss": 0.0102,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997439994699246,
+      "loss": 0.7373,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997404807050374,
+      "loss": 0.0425,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997369379240567,
+      "loss": 0.6373,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299973337112704,
+      "loss": 0.3752,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997297803140437,
+      "loss": 0.5111,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999726165485126,
+      "loss": 0.0532,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999722526640345,
+      "loss": 0.2523,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997188637797585,
+      "loss": 0.6098,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999715176903425,
+      "loss": 0.4431,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997114660114036,
+      "loss": 0.2207,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997077311037535,
+      "loss": 0.8102,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029997039721805347,
+      "loss": 0.4433,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999700189241808,
+      "loss": 0.7388,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999696382287633,
+      "loss": 1.1607,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999692551318072,
+      "loss": 0.4863,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999688696333185,
+      "loss": 0.3484,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999684817333035,
+      "loss": 0.5244,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996809143176826,
+      "loss": 0.552,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999676987287191,
+      "loss": 0.289,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999673036241623,
+      "loss": 0.4618,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996690611810425,
+      "loss": 0.2009,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996650621055127,
+      "loss": 0.452,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996610390150977,
+      "loss": 2.1568,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999656991909861,
+      "loss": 3.054,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999652920789869,
+      "loss": 0.4962,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999648825655186,
+      "loss": 0.8852,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999644706505878,
+      "loss": 0.5776,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299964056334201,
+      "loss": 0.4406,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996363961636496,
+      "loss": 0.7579,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996322049708623,
+      "loss": 1.0634,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996279897637163,
+      "loss": 0.5465,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999623750542279,
+      "loss": 0.8095,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999619487306617,
+      "loss": 0.6515,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996152000568,
+      "loss": 0.5721,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999610888792896,
+      "loss": 0.1836,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999606553514975,
+      "loss": 0.654,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029996021942231056,
+      "loss": 1.8457,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999597810917357,
+      "loss": 0.2581,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995934035978,
+      "loss": 0.3528,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995889722645054,
+      "loss": 0.4445,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995845169175444,
+      "loss": 0.5048,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995800375569873,
+      "loss": 1.1177,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999575534182907,
+      "loss": 0.489,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995710067953746,
+      "loss": 0.7172,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999566455394463,
+      "loss": 0.524,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999561879980246,
+      "loss": 0.9177,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999557280552795,
+      "loss": 0.7532,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999552657112185,
+      "loss": 0.1295,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299954800965849,
+      "loss": 0.6342,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999543338191784,
+      "loss": 0.7693,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999538642712142,
+      "loss": 0.6009,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995339232196386,
+      "loss": 0.5431,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995291797143506,
+      "loss": 0.4771,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999524412196353,
+      "loss": 2.1071,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999519620665722,
+      "loss": 0.7119,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995148051225354,
+      "loss": 0.968,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995099655668694,
+      "loss": 0.0052,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999505101998802,
+      "loss": 0.9679,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029995002144184103,
+      "loss": 0.4108,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999495302825774,
+      "loss": 3.0098,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994903672209697,
+      "loss": 2.4187,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994854076040783,
+      "loss": 0.5288,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999480423975178,
+      "loss": 1.483,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994754163343494,
+      "loss": 0.8515,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994703846816724,
+      "loss": 0.9621,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999465329017228,
+      "loss": 0.2986,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999460249341096,
+      "loss": 0.6224,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999455145653359,
+      "loss": 0.7774,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994500179540977,
+      "loss": 0.6289,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999444866243395,
+      "loss": 1.4427,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994396905213335,
+      "loss": 0.6656,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994344907879954,
+      "loss": 0.5839,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994292670434643,
+      "loss": 0.9333,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999424019287824,
+      "loss": 2.0752,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999418747521158,
+      "loss": 0.9285,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999413451743551,
+      "loss": 0.457,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029994081319550877,
+      "loss": 1.1892,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999402788155854,
+      "loss": 1.6005,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999397420345935,
+      "loss": 2.2713,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999392028525416,
+      "loss": 0.4316,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993866126943843,
+      "loss": 0.4706,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993811728529267,
+      "loss": 0.3741,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993757090011295,
+      "loss": 0.5472,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299937022113908,
+      "loss": 0.3128,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993647092668675,
+      "loss": 1.3186,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999359173384579,
+      "loss": 0.0113,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993536134923036,
+      "loss": 1.0267,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993480295901305,
+      "loss": 0.3336,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993424216781485,
+      "loss": 0.3183,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993367897564475,
+      "loss": 0.3732,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993311338251193,
+      "loss": 0.0128,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999325453884252,
+      "loss": 0.7211,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999319749933938,
+      "loss": 0.964,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993140219742686,
+      "loss": 2.0126,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993082700053355,
+      "loss": 0.5211,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029993024940272304,
+      "loss": 0.4686,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992966940400455,
+      "loss": 0.569,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992908700438747,
+      "loss": 0.6212,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992850220388105,
+      "loss": 1.2915,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999279150024947,
+      "loss": 0.8984,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999273254002378,
+      "loss": 1.6295,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999267333971198,
+      "loss": 1.1624,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992613899315016,
+      "loss": 0.3895,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999255421883384,
+      "loss": 0.5453,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992494298269407,
+      "loss": 0.4575,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992434137622684,
+      "loss": 0.0689,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992373736894624,
+      "loss": 0.7966,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299923130960862,
+      "loss": 0.4863,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992252215198386,
+      "loss": 0.1806,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999219109423215,
+      "loss": 1.1082,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999212973318847,
+      "loss": 0.598,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992068132068337,
+      "loss": 0.2244,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029992006290872734,
+      "loss": 0.6915,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029991944209602645,
+      "loss": 2.0011,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029991881888259073,
+      "loss": 0.6528,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999181932684301,
+      "loss": 0.5699,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029991756525355465,
+      "loss": 0.2,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029991693483797434,
+      "loss": 1.2829,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999163020216993,
+      "loss": 0.0038,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999156668047397,
+      "loss": 0.2096,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029991502918710574,
+      "loss": 0.2529,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999143891688075,
+      "loss": 0.0098,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999137467498554,
+      "loss": 0.3423,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999131019302595,
+      "loss": 1.5642,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029991245471003037,
+      "loss": 0.7957,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999118050891782,
+      "loss": 0.3336,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999111530677135,
+      "loss": 0.321,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999104986456466,
+      "loss": 0.783,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999098418229881,
+      "loss": 0.8318,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999091825997485,
+      "loss": 0.7515,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029990852097593827,
+      "loss": 0.6569,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999078569515681,
+      "loss": 0.0101,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999071905266486,
+      "loss": 1.3934,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029990652170119034,
+      "loss": 0.3622,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029990585047520415,
+      "loss": 0.2884,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029990517684870077,
+      "loss": 0.4,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029990450082169096,
+      "loss": 0.2324,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999038223941855,
+      "loss": 0.1775,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999031415661953,
+      "loss": 0.0027,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029990245833773127,
+      "loss": 1.7972,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999017727088043,
+      "loss": 0.3842,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999010846794255,
+      "loss": 0.0025,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029990039424960575,
+      "loss": 0.3474,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998997014193562,
+      "loss": 0.6299,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029989900618868786,
+      "loss": 0.3241,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998983085576119,
+      "loss": 0.6135,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998976085261395,
+      "loss": 0.9595,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029989690609428187,
+      "loss": 0.0035,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029989620126205023,
+      "loss": 0.8464,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998954940294559,
+      "loss": 0.5995,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029989478439651024,
+      "loss": 0.0016,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998940723632245,
+      "loss": 0.14,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029989335792961024,
+      "loss": 2.0168,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029989335792961024,
+      "loss": 0.2094,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998926410956787,
+      "loss": 0.5108,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029989192186144156,
+      "loss": 0.8928,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998912002269102,
+      "loss": 0.3447,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998904761920962,
+      "loss": 1.2306,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029988974975701127,
+      "loss": 0.6787,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029988902092166687,
+      "loss": 0.2462,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029988828968607477,
+      "loss": 3.6597,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998875560502467,
+      "loss": 0.5929,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998868200141943,
+      "loss": 1.0922,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998860815779295,
+      "loss": 0.8976,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299885340741464,
+      "loss": 0.5819,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998845975048097,
+      "loss": 0.0466,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998838518679786,
+      "loss": 1.0847,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029988310383098247,
+      "loss": 0.9847,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998823533938334,
+      "loss": 0.7482,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998816005565434,
+      "loss": 0.4283,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029988084531912447,
+      "loss": 0.7431,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029988008768158875,
+      "loss": 0.2175,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998793276439484,
+      "loss": 0.2961,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029987856520621555,
+      "loss": 0.7113,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998778003684024,
+      "loss": 0.2944,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998770331305212,
+      "loss": 0.6408,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998762634925843,
+      "loss": 0.1935,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998754914546039,
+      "loss": 0.0936,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998747170165925,
+      "loss": 0.803,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998739401785624,
+      "loss": 2.3781,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998731609405261,
+      "loss": 1.0266,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299872379302496,
+      "loss": 1.708,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998715952644848,
+      "loss": 3.5539,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998708088265048,
+      "loss": 0.385,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029987001998856873,
+      "loss": 2.2036,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029986922875068926,
+      "loss": 0.5798,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299868435112879,
+      "loss": 0.5154,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998676390751507,
+      "loss": 3.527,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029986684063751704,
+      "loss": 0.6486,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029986603979999086,
+      "loss": 0.9148,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029986523656258497,
+      "loss": 0.3921,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998644309253122,
+      "loss": 1.2489,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998636228881855,
+      "loss": 1.4371,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998628124512178,
+      "loss": 0.8653,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998619996144221,
+      "loss": 0.6369,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998611843778114,
+      "loss": 0.5806,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998603667413987,
+      "loss": 0.2169,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029985954670519715,
+      "loss": 1.8133,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029985872426921986,
+      "loss": 0.6107,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029985789943348,
+      "loss": 0.4822,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998570721979908,
+      "loss": 0.0069,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998562425627655,
+      "loss": 0.754,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998554105278174,
+      "loss": 0.3175,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029985457609315973,
+      "loss": 0.0749,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029985373925880596,
+      "loss": 1.0707,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998529000247694,
+      "loss": 1.9348,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029985205839106366,
+      "loss": 0.404,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029985121435770207,
+      "loss": 0.8832,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998503679246981,
+      "loss": 0.6141,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029984951909206543,
+      "loss": 0.3446,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998486678598176,
+      "loss": 0.6046,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998478142279683,
+      "loss": 0.6271,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029984695819653106,
+      "loss": 0.5219,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998460997655197,
+      "loss": 0.2009,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029984523893494797,
+      "loss": 0.3517,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998443757048296,
+      "loss": 0.8978,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029984351007517843,
+      "loss": 0.761,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029984264204600835,
+      "loss": 0.3242,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029984177161733325,
+      "loss": 1.0781,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299840898789167,
+      "loss": 0.4241,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998400235615237,
+      "loss": 0.2888,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983914593441726,
+      "loss": 0.9145,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983826590786184,
+      "loss": 1.1606,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998373834818714,
+      "loss": 0.3215,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983649865646014,
+      "loss": 0.3304,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998356114316423,
+      "loss": 1.107,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983472180743193,
+      "loss": 0.2646,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983382978384336,
+      "loss": 0.2637,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983293536089094,
+      "loss": 0.9527,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983203853858886,
+      "loss": 0.4542,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029983113931695155,
+      "loss": 0.7127,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998302376959935,
+      "loss": 0.1353,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982933367572897,
+      "loss": 0.3287,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982842725617253,
+      "loss": 0.5632,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982751843733875,
+      "loss": 1.5377,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982660721924206,
+      "loss": 0.2818,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982569360189715,
+      "loss": 0.2437,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982477758531864,
+      "loss": 0.5831,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982385916952107,
+      "loss": 3.5795,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982293835451935,
+      "loss": 1.0205,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029982201514032805,
+      "loss": 0.5634,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998210895269621,
+      "loss": 0.4732,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998201615144362,
+      "loss": 0.819,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998192311027653,
+      "loss": 0.697,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029981829829196424,
+      "loss": 0.4723,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029981736308204795,
+      "loss": 1.2741,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998164254730315,
+      "loss": 0.4337,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029981548546492975,
+      "loss": 1.2639,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998145430577579,
+      "loss": 1.2521,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299813598251531,
+      "loss": 0.6589,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998126510462641,
+      "loss": 0.5091,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998117014419725,
+      "loss": 0.6194,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998107494386713,
+      "loss": 2.2732,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029980979503637577,
+      "loss": 1.1198,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029980883823510127,
+      "loss": 0.426,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299807879034863,
+      "loss": 0.1394,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998069174356764,
+      "loss": 1.2332,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998059534375568,
+      "loss": 0.3686,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998049870405197,
+      "loss": 0.629,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029980401824458063,
+      "loss": 0.7355,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029980304704975495,
+      "loss": 1.0268,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998020734560583,
+      "loss": 0.9582,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002998010974635063,
+      "loss": 0.4177,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029980011907211455,
+      "loss": 0.3177,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029979913828189874,
+      "loss": 0.3716,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029979815509287446,
+      "loss": 0.6625,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997971695050576,
+      "loss": 0.6055,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029979618151846385,
+      "loss": 0.3549,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997951911331091,
+      "loss": 1.0981,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029979419834900913,
+      "loss": 0.9803,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997932031661799,
+      "loss": 0.3675,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029979220558463737,
+      "loss": 0.0049,
+      "step": 519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029979120560439743,
+      "loss": 0.7795,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997902032254762,
+      "loss": 0.5777,
+      "step": 521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997891984478896,
+      "loss": 0.3353,
+      "step": 522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997881912716538,
+      "loss": 0.1844,
+      "step": 523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029978718169678493,
+      "loss": 0.4885,
+      "step": 524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029978616972329913,
+      "loss": 0.6982,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029978515535121264,
+      "loss": 1.782,
+      "step": 526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029978413858054167,
+      "loss": 0.999,
+      "step": 527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029978311941130254,
+      "loss": 0.4867,
+      "step": 528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997820978435115,
+      "loss": 0.6495,
+      "step": 529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029978107387718496,
+      "loss": 0.3001,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029978004751233937,
+      "loss": 0.4585,
+      "step": 531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029977901874899104,
+      "loss": 0.3027,
+      "step": 532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029977798758715657,
+      "loss": 0.1915,
+      "step": 533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029977695402685233,
+      "loss": 0.3148,
+      "step": 534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299775918068095,
+      "loss": 0.5644,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997748797109012,
+      "loss": 0.3074,
+      "step": 536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997738389552874,
+      "loss": 0.6409,
+      "step": 537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997727958012704,
+      "loss": 0.6965,
+      "step": 538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997717502488668,
+      "loss": 1.0783,
+      "step": 539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997707022980934,
+      "loss": 0.4707,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029976965194896695,
+      "loss": 0.9336,
+      "step": 541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997685992015044,
+      "loss": 0.3597,
+      "step": 542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997675440557224,
+      "loss": 0.5038,
+      "step": 543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299766486511638,
+      "loss": 0.356,
+      "step": 544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997654265692681,
+      "loss": 0.6105,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997643642286296,
+      "loss": 0.7791,
+      "step": 546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029976329948973963,
+      "loss": 0.9433,
+      "step": 547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029976223235261514,
+      "loss": 0.2018,
+      "step": 548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997611628172733,
+      "loss": 1.031,
+      "step": 549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997600908837312,
+      "loss": 0.0523,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029975901655200594,
+      "loss": 0.9312,
+      "step": 551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997579398221148,
+      "loss": 0.6132,
+      "step": 552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029975686069407513,
+      "loss": 1.3225,
+      "step": 553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299755779167904,
+      "loss": 0.5417,
+      "step": 554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029975469524361876,
+      "loss": 0.559,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029975360892123693,
+      "loss": 0.3718,
+      "step": 556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997525202007757,
+      "loss": 2.4617,
+      "step": 557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997514290822527,
+      "loss": 0.5204,
+      "step": 558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029975033556568526,
+      "loss": 0.5522,
+      "step": 559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029974923965109094,
+      "loss": 0.518,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997481413384873,
+      "loss": 0.3818,
+      "step": 561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029974704062789195,
+      "loss": 0.5278,
+      "step": 562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997459375193224,
+      "loss": 0.2671,
+      "step": 563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997448320127965,
+      "loss": 0.5331,
+      "step": 564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997437241083318,
+      "loss": 2.2319,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997426138059461,
+      "loss": 0.2925,
+      "step": 566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997415011056571,
+      "loss": 0.4191,
+      "step": 567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029974038600748276,
+      "loss": 0.3273,
+      "step": 568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997392685114409,
+      "loss": 0.7598,
+      "step": 569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029973814861754933,
+      "loss": 0.4249,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299737026325826,
+      "loss": 0.6454,
+      "step": 571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299735901636289,
+      "loss": 0.5924,
+      "step": 572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997347745489562,
+      "loss": 0.7876,
+      "step": 573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029973364506384576,
+      "loss": 0.4472,
+      "step": 574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029973251318097567,
+      "loss": 0.527,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029973137890036407,
+      "loss": 0.9478,
+      "step": 576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997302422220292,
+      "loss": 0.2056,
+      "step": 577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997291031459892,
+      "loss": 0.8,
+      "step": 578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997279616722624,
+      "loss": 0.6914,
+      "step": 579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997268178008669,
+      "loss": 0.3286,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997256715318212,
+      "loss": 0.5214,
+      "step": 581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029972452286514345,
+      "loss": 0.5975,
+      "step": 582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997233718008523,
+      "loss": 0.3803,
+      "step": 583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299722218338966,
+      "loss": 0.8148,
+      "step": 584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997210624795031,
+      "loss": 1.3607,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029971990422248204,
+      "loss": 0.9083,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029971874356792146,
+      "loss": 0.3893,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029971758051583984,
+      "loss": 0.4363,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029971641506625594,
+      "loss": 0.931,
+      "step": 589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997152472191883,
+      "loss": 0.4755,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997140769746557,
+      "loss": 0.8721,
+      "step": 591
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029971290433267677,
+      "loss": 3.0693,
+      "step": 592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029971172929327044,
+      "loss": 0.6307,
+      "step": 593
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997105518564554,
+      "loss": 0.3584,
+      "step": 594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997093720222506,
+      "loss": 1.1073,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029970818979067486,
+      "loss": 1.9136,
+      "step": 596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029970700516174714,
+      "loss": 0.2816,
+      "step": 597
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997058181354865,
+      "loss": 0.0981,
+      "step": 598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997046287119117,
+      "loss": 0.3192,
+      "step": 599
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299703436891042,
+      "loss": 1.1676,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029970224267289653,
+      "loss": 0.3294,
+      "step": 601
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002997010460574942,
+      "loss": 2.1328,
+      "step": 602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996998470448544,
+      "loss": 0.4742,
+      "step": 603
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969864563499614,
+      "loss": 0.406,
+      "step": 604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969744182793874,
+      "loss": 0.4066,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969623562370144,
+      "loss": 0.3417,
+      "step": 606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969502702230366,
+      "loss": 0.3874,
+      "step": 607
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969381602376464,
+      "loss": 1.0648,
+      "step": 608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969260262810383,
+      "loss": 0.1198,
+      "step": 609
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969138683534066,
+      "loss": 0.5071,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029969016864549457,
+      "loss": 2.503,
+      "step": 611
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996889480585851,
+      "loss": 0.938,
+      "step": 612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996877250746317,
+      "loss": 0.3216,
+      "step": 613
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996864996936541,
+      "loss": 3.4333,
+      "step": 614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996852719156719,
+      "loss": 0.8181,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996840417407047,
+      "loss": 0.3687,
+      "step": 616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029968280916877214,
+      "loss": 0.3899,
+      "step": 617
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029968157419989406,
+      "loss": 0.3349,
+      "step": 618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996803368340902,
+      "loss": 0.2812,
+      "step": 619
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996790970713804,
+      "loss": 0.4847,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996778549117846,
+      "loss": 0.6741,
+      "step": 621
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029967661035532246,
+      "loss": 1.2943,
+      "step": 622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029967536340201405,
+      "loss": 0.0219,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029967411405187937,
+      "loss": 0.6679,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029967286230493835,
+      "loss": 0.7483,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029967160816121106,
+      "loss": 0.4377,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996703516207176,
+      "loss": 1.035,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996690926834781,
+      "loss": 0.4486,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996678313495127,
+      "loss": 0.1569,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996665676188416,
+      "loss": 1.0024,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299665301491485,
+      "loss": 0.5543,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996640329674633,
+      "loss": 0.5298,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996627620467966,
+      "loss": 0.4987,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996614887295055,
+      "loss": 3.366,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029966021301561024,
+      "loss": 0.865,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029965893490513124,
+      "loss": 1.4343,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299657654398089,
+      "loss": 0.0179,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996563714945041,
+      "loss": 0.8609,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996550861943969,
+      "loss": 0.2364,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029965379849778817,
+      "loss": 0.7264,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029965250840469846,
+      "loss": 0.0133,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996512159151484,
+      "loss": 0.6661,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029964992102915873,
+      "loss": 1.0984,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996486237467502,
+      "loss": 0.8999,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996473240679435,
+      "loss": 0.6427,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996460219927595,
+      "loss": 0.4339,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299644717521219,
+      "loss": 0.3558,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299643410653343,
+      "loss": 0.2202,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029964210138915237,
+      "loss": 0.6986,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029964078972866804,
+      "loss": 0.7068,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996394756719111,
+      "loss": 0.3719,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996381592189024,
+      "loss": 0.9851,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996368403696632,
+      "loss": 0.6485,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029963551912421467,
+      "loss": 0.4309,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029963419548257784,
+      "loss": 0.3915,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029963286944477386,
+      "loss": 0.5389,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996315410108241,
+      "loss": 1.9492,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996302101807498,
+      "loss": 0.3634,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029962887695457224,
+      "loss": 0.8859,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029962754133231275,
+      "loss": 1.3406,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029962620331399275,
+      "loss": 0.6705,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029962486289963365,
+      "loss": 0.9486,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996235200892569,
+      "loss": 0.2311,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996221748828841,
+      "loss": 0.2675,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029962082728053667,
+      "loss": 0.2972,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996194772822363,
+      "loss": 0.7348,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996181248880045,
+      "loss": 0.828,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029961677009786303,
+      "loss": 0.0713,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029961541291183345,
+      "loss": 0.4655,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996140533299376,
+      "loss": 1.0571,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996126913521972,
+      "loss": 0.4444,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029961132697863414,
+      "loss": 2.4133,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029960996020927017,
+      "loss": 0.0452,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029960859104412725,
+      "loss": 0.2483,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029960721948322724,
+      "loss": 0.1143,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996058455265922,
+      "loss": 0.1648,
+      "step": 676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299604469174244,
+      "loss": 0.3379,
+      "step": 677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029960309042620484,
+      "loss": 0.6503,
+      "step": 678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029960170928249665,
+      "loss": 0.0833,
+      "step": 679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002996003257431416,
+      "loss": 1.0383,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995989398081619,
+      "loss": 0.5664,
+      "step": 681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995975514775796,
+      "loss": 0.5495,
+      "step": 682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995961607514171,
+      "loss": 1.0748,
+      "step": 683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995947676296966,
+      "loss": 0.7079,
+      "step": 684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029959337211244043,
+      "loss": 0.4719,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029959197419967084,
+      "loss": 0.4273,
+      "step": 686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029959057389141033,
+      "loss": 0.6066,
+      "step": 687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029958917118768133,
+      "loss": 0.5683,
+      "step": 688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029958776608850614,
+      "loss": 0.5021,
+      "step": 689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995863585939075,
+      "loss": 0.4476,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029958494870390773,
+      "loss": 1.8491,
+      "step": 691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029958353641852957,
+      "loss": 0.6671,
+      "step": 692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995821217377955,
+      "loss": 0.7913,
+      "step": 693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995807046617283,
+      "loss": 0.2723,
+      "step": 694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029957928519035057,
+      "loss": 0.4616,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995778633236851,
+      "loss": 0.2105,
+      "step": 696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029957643906175464,
+      "loss": 0.9581,
+      "step": 697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995750124045819,
+      "loss": 0.5218,
+      "step": 698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029957358335218993,
+      "loss": 0.2372,
+      "step": 699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995721519046015,
+      "loss": 0.2521,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029957071806183946,
+      "loss": 1.9224,
+      "step": 701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995692818239269,
+      "loss": 0.7851,
+      "step": 702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029956784319088673,
+      "loss": 0.7684,
+      "step": 703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029956640216274204,
+      "loss": 0.6064,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995649587395159,
+      "loss": 0.2881,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029956351292123134,
+      "loss": 0.493,
+      "step": 706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995620647079117,
+      "loss": 1.591,
+      "step": 707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029956061409957993,
+      "loss": 0.2488,
+      "step": 708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029955916109625946,
+      "loss": 0.5169,
+      "step": 709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029955770569797347,
+      "loss": 0.2644,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029955624790474526,
+      "loss": 0.2995,
+      "step": 711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995547877165982,
+      "loss": 0.6723,
+      "step": 712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029955332513355564,
+      "loss": 0.4755,
+      "step": 713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029955186015564106,
+      "loss": 0.9366,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029955039278287794,
+      "loss": 0.8933,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029954892301528965,
+      "loss": 0.0969,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995474508528998,
+      "loss": 0.6746,
+      "step": 717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299545976295732,
+      "loss": 1.277,
+      "step": 718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995444993438098,
+      "loss": 0.4479,
+      "step": 719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029954301999715686,
+      "loss": 0.0084,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029954153825579693,
+      "loss": 0.3678,
+      "step": 721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995400541197537,
+      "loss": 0.5088,
+      "step": 722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995385675890509,
+      "loss": 0.304,
+      "step": 723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995370786637124,
+      "loss": 0.0662,
+      "step": 724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299535587343762,
+      "loss": 0.422,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995340936292235,
+      "loss": 0.7488,
+      "step": 726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029953259752012103,
+      "loss": 0.1403,
+      "step": 727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029953109901647836,
+      "loss": 0.6124,
+      "step": 728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029952959811831956,
+      "loss": 1.3099,
+      "step": 729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029952809482566866,
+      "loss": 0.7573,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029952658913854973,
+      "loss": 1.0093,
+      "step": 731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995250810569869,
+      "loss": 1.0035,
+      "step": 732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029952357058100437,
+      "loss": 0.8918,
+      "step": 733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029952205771062614,
+      "loss": 0.2706,
+      "step": 734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029952054244587663,
+      "loss": 0.4327,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029951902478678,
+      "loss": 0.9155,
+      "step": 736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995175047333606,
+      "loss": 0.4515,
+      "step": 737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995159822856427,
+      "loss": 1.0384,
+      "step": 738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029951445744365085,
+      "loss": 0.9349,
+      "step": 739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995129302074093,
+      "loss": 0.5321,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995114005769425,
+      "loss": 3.8804,
+      "step": 741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029950986855227507,
+      "loss": 0.4829,
+      "step": 742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029950833413343146,
+      "loss": 0.7533,
+      "step": 743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995067973204363,
+      "loss": 0.3983,
+      "step": 744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029950525811331413,
+      "loss": 0.353,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995037165120896,
+      "loss": 0.5139,
+      "step": 746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002995021725167874,
+      "loss": 0.2974,
+      "step": 747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029950062612743236,
+      "loss": 0.5466,
+      "step": 748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029949907734404913,
+      "loss": 0.6046,
+      "step": 749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994975261666625,
+      "loss": 0.8182,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994959725952974,
+      "loss": 0.6434,
+      "step": 751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029949441662997865,
+      "loss": 0.6042,
+      "step": 752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994928582707312,
+      "loss": 0.4315,
+      "step": 753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029949129751757995,
+      "loss": 1.0256,
+      "step": 754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029948973437054994,
+      "loss": 0.5118,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994881688296662,
+      "loss": 0.5391,
+      "step": 756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994866008949537,
+      "loss": 0.4383,
+      "step": 757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994850305664377,
+      "loss": 0.4613,
+      "step": 758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994834578441433,
+      "loss": 0.4043,
+      "step": 759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994818827280957,
+      "loss": 0.3269,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029948030521832,
+      "loss": 0.9401,
+      "step": 761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994787253148416,
+      "loss": 0.3665,
+      "step": 762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994771430176858,
+      "loss": 0.4662,
+      "step": 763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994755583268778,
+      "loss": 0.3799,
+      "step": 764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029947397124244305,
+      "loss": 0.5422,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029947238176440704,
+      "loss": 0.3896,
+      "step": 766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029947078989279514,
+      "loss": 0.0436,
+      "step": 767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029946919562763287,
+      "loss": 0.3844,
+      "step": 768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029946759896894577,
+      "loss": 0.2607,
+      "step": 769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029946599991675937,
+      "loss": 1.2097,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994643984710993,
+      "loss": 1.1048,
+      "step": 771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994627946319912,
+      "loss": 0.1256,
+      "step": 772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994611883994608,
+      "loss": 1.306,
+      "step": 773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029945957977353373,
+      "loss": 0.5848,
+      "step": 774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994579687542358,
+      "loss": 1.3041,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029945635534159277,
+      "loss": 0.3233,
+      "step": 776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994547395356306,
+      "loss": 4.6411,
+      "step": 777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299453121336375,
+      "loss": 0.2731,
+      "step": 778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029945150074385204,
+      "loss": 0.1473,
+      "step": 779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994498777580875,
+      "loss": 1.4458,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994482523791075,
+      "loss": 0.5393,
+      "step": 781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029944662460693804,
+      "loss": 0.3147,
+      "step": 782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994449944416052,
+      "loss": 0.1741,
+      "step": 783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299443361883135,
+      "loss": 0.5847,
+      "step": 784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029944172693155364,
+      "loss": 0.4571,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994400895868873,
+      "loss": 0.3589,
+      "step": 786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994384498491623,
+      "loss": 1.2126,
+      "step": 787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994368077184047,
+      "loss": 0.674,
+      "step": 788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029943516319464093,
+      "loss": 1.154,
+      "step": 789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994335162778973,
+      "loss": 0.8745,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994318669682002,
+      "loss": 0.533,
+      "step": 791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299430215265576,
+      "loss": 0.6225,
+      "step": 792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994285611700511,
+      "loss": 0.9474,
+      "step": 793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994269046816521,
+      "loss": 0.8173,
+      "step": 794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029942524580040556,
+      "loss": 0.6842,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029942358452633786,
+      "loss": 0.5623,
+      "step": 796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994219208594757,
+      "loss": 0.4162,
+      "step": 797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994202547998458,
+      "loss": 0.3911,
+      "step": 798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994185863474747,
+      "loss": 0.7209,
+      "step": 799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994169155023892,
+      "loss": 0.3378,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994152422646161,
+      "loss": 0.212,
+      "step": 801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994135666341821,
+      "loss": 0.4292,
+      "step": 802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994118886111141,
+      "loss": 2.8947,
+      "step": 803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029941020819543887,
+      "loss": 1.5582,
+      "step": 804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029940852538718344,
+      "loss": 0.9103,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994068401863747,
+      "loss": 0.558,
+      "step": 806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994051525930396,
+      "loss": 0.8993,
+      "step": 807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029940346260720526,
+      "loss": 0.1624,
+      "step": 808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002994017702288987,
+      "loss": 0.5112,
+      "step": 809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029940007545814696,
+      "loss": 1.7019,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029939837829497725,
+      "loss": 0.6312,
+      "step": 811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993966787394167,
+      "loss": 0.5053,
+      "step": 812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993949767914926,
+      "loss": 0.8625,
+      "step": 813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029939327245123216,
+      "loss": 0.4833,
+      "step": 814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029939156571866265,
+      "loss": 0.4326,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993898565938114,
+      "loss": 0.4521,
+      "step": 816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029938814507670584,
+      "loss": 0.5834,
+      "step": 817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993864311673733,
+      "loss": 0.4002,
+      "step": 818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993847148658413,
+      "loss": 0.42,
+      "step": 819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029938299617213724,
+      "loss": 1.0622,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029938127508628875,
+      "loss": 0.4922,
+      "step": 821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029937955160832326,
+      "loss": 0.3832,
+      "step": 822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993778257382685,
+      "loss": 0.3644,
+      "step": 823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299376097476152,
+      "loss": 0.3039,
+      "step": 824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993743668220015,
+      "loss": 1.1071,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029937263377584464,
+      "loss": 0.7642,
+      "step": 826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993708983377093,
+      "loss": 0.4299,
+      "step": 827
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993691605076232,
+      "loss": 0.4362,
+      "step": 828
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993674202856141,
+      "loss": 0.5481,
+      "step": 829
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029936567767171004,
+      "loss": 0.3224,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029936393266593876,
+      "loss": 0.9821,
+      "step": 831
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993621852683282,
+      "loss": 0.24,
+      "step": 832
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993604354789065,
+      "loss": 0.3447,
+      "step": 833
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993586832977015,
+      "loss": 0.7007,
+      "step": 834
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029935692872474145,
+      "loss": 0.5945,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029935517176005427,
+      "loss": 0.2171,
+      "step": 836
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029935341240366816,
+      "loss": 0.587,
+      "step": 837
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993516506556113,
+      "loss": 2.3494,
+      "step": 838
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993498865159119,
+      "loss": 0.925,
+      "step": 839
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993481199845982,
+      "loss": 0.2795,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029934635106169857,
+      "loss": 0.4914,
+      "step": 841
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993445797472412,
+      "loss": 0.6274,
+      "step": 842
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029934280604125454,
+      "loss": 0.3978,
+      "step": 843
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299341029943767,
+      "loss": 0.7608,
+      "step": 844
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029933925145480693,
+      "loss": 0.2048,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993374705744029,
+      "loss": 1.1888,
+      "step": 846
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993356873025834,
+      "loss": 0.2022,
+      "step": 847
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299333901639377,
+      "loss": 0.2959,
+      "step": 848
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993321135848123,
+      "loss": 1.2035,
+      "step": 849
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029933032313891796,
+      "loss": 0.7232,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029932853030172253,
+      "loss": 1.8751,
+      "step": 851
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993267350732548,
+      "loss": 0.8712,
+      "step": 852
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993249374535435,
+      "loss": 0.3461,
+      "step": 853
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993231374426175,
+      "loss": 0.7518,
+      "step": 854
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993213350405055,
+      "loss": 0.6311,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029931953024723646,
+      "loss": 0.3359,
+      "step": 856
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993177230628392,
+      "loss": 1.1701,
+      "step": 857
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029931591348734276,
+      "loss": 0.7335,
+      "step": 858
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299314101520776,
+      "loss": 0.2086,
+      "step": 859
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029931228716316806,
+      "loss": 0.4079,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993104704145478,
+      "loss": 0.0395,
+      "step": 861
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029930865127494454,
+      "loss": 0.266,
+      "step": 862
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993068297443873,
+      "loss": 4.7435,
+      "step": 863
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029930500582290524,
+      "loss": 1.2128,
+      "step": 864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002993031795105276,
+      "loss": 0.027,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029930135080728364,
+      "loss": 0.0161,
+      "step": 866
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002992995197132026,
+      "loss": 1.3555,
+      "step": 867
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029929768622831383,
+      "loss": 0.7428,
+      "step": 868
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029929585035264663,
+      "loss": 0.3911,
+      "step": 869
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002992940120862305,
+      "loss": 0.5341,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029929217142909476,
+      "loss": 0.7444,
+      "step": 871
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029929032838126897,
+      "loss": 0.318,
+      "step": 872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029928848294278265,
+      "loss": 0.4635,
+      "step": 873
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029928663511366533,
+      "loss": 0.1245,
+      "step": 874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992847848939466,
+      "loss": 0.2531,
+      "step": 875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992829322836561,
+      "loss": 2.031,
+      "step": 876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992810772828234,
+      "loss": 0.0135,
+      "step": 877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029927921989147835,
+      "loss": 0.8338,
+      "step": 878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992773601096506,
+      "loss": 0.2692,
+      "step": 879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029927549793737004,
+      "loss": 0.2163,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992736333746663,
+      "loss": 0.5727,
+      "step": 881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992717664215694,
+      "loss": 0.9672,
+      "step": 882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029926989707810924,
+      "loss": 0.4481,
+      "step": 883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992680253443156,
+      "loss": 0.0584,
+      "step": 884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029926615122021866,
+      "loss": 0.4092,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992642747058482,
+      "loss": 1.6374,
+      "step": 886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992623958012345,
+      "loss": 0.4533,
+      "step": 887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992605145064075,
+      "loss": 0.5229,
+      "step": 888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029925863082139745,
+      "loss": 0.0265,
+      "step": 889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992567447462343,
+      "loss": 0.0087,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992548562809485,
+      "loss": 1.8335,
+      "step": 891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029925296542557007,
+      "loss": 0.4818,
+      "step": 892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992510721801295,
+      "loss": 0.9951,
+      "step": 893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029924917654465695,
+      "loss": 1.8667,
+      "step": 894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992472785191828,
+      "loss": 0.5549,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992453781037375,
+      "loss": 0.0283,
+      "step": 896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992434752983515,
+      "loss": 0.1736,
+      "step": 897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029924157010305515,
+      "loss": 0.4723,
+      "step": 898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029923966251787907,
+      "loss": 0.5739,
+      "step": 899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992377525428538,
+      "loss": 0.0038,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029923584017800986,
+      "loss": 0.6895,
+      "step": 901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029923392542337794,
+      "loss": 0.6953,
+      "step": 902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029923200827898863,
+      "loss": 0.5351,
+      "step": 903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992300887448727,
+      "loss": 2.4263,
+      "step": 904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992281668210609,
+      "loss": 0.2575,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029922624250758393,
+      "loss": 0.9257,
+      "step": 906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029922431580447264,
+      "loss": 6.7776,
+      "step": 907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992223867117579,
+      "loss": 0.7361,
+      "step": 908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992204552294706,
+      "loss": 0.5994,
+      "step": 909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992185213576416,
+      "loss": 0.0169,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029921658509630196,
+      "loss": 0.0048,
+      "step": 911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992146464454826,
+      "loss": 1.7575,
+      "step": 912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992127054052147,
+      "loss": 1.124,
+      "step": 913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992107619755293,
+      "loss": 0.5455,
+      "step": 914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992088161564574,
+      "loss": 0.6768,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029920686794803023,
+      "loss": 0.8236,
+      "step": 916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029920491735027904,
+      "loss": 0.1598,
+      "step": 917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000299202964363235,
+      "loss": 0.0109,
+      "step": 918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002992010089869294,
+      "loss": 0.1562,
+      "step": 919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991990512213936,
+      "loss": 1.0518,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029919709106665884,
+      "loss": 0.2737,
+      "step": 921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029919512852275666,
+      "loss": 0.001,
+      "step": 922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029919316358971833,
+      "loss": 0.1649,
+      "step": 923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029919119626757546,
+      "loss": 0.3942,
+      "step": 924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029918922655635945,
+      "loss": 2.7463,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991872544561019,
+      "loss": 0.3174,
+      "step": 926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029918527996683437,
+      "loss": 0.001,
+      "step": 927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991833030885884,
+      "loss": 0.3442,
+      "step": 928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029918132382139584,
+      "loss": 2.5999,
+      "step": 929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029917934216528824,
+      "loss": 0.008,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029917735812029733,
+      "loss": 0.7515,
+      "step": 931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029917537168645494,
+      "loss": 0.5513,
+      "step": 932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991733828637928,
+      "loss": 0.3925,
+      "step": 933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029917139165234287,
+      "loss": 0.2674,
+      "step": 934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029916939805213694,
+      "loss": 0.2861,
+      "step": 935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000299167402063207,
+      "loss": 0.6527,
+      "step": 936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000299165403685585,
+      "loss": 0.401,
+      "step": 937
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029916340291930287,
+      "loss": 1.0149,
+      "step": 938
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029916139976439275,
+      "loss": 2.3198,
+      "step": 939
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991593942208867,
+      "loss": 0.4219,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991573862888167,
+      "loss": 0.0033,
+      "step": 941
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029915537596821506,
+      "loss": 0.4326,
+      "step": 942
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029915336325911395,
+      "loss": 1.0358,
+      "step": 943
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991513481615456,
+      "loss": 0.5227,
+      "step": 944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991493306755422,
+      "loss": 1.0824,
+      "step": 945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991473108011361,
+      "loss": 1.4255,
+      "step": 946
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991452885383597,
+      "loss": 0.8158,
+      "step": 947
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029914326388724537,
+      "loss": 0.2127,
+      "step": 948
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991412368478254,
+      "loss": 0.5739,
+      "step": 949
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029913920742013247,
+      "loss": 0.3536,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029913717560419887,
+      "loss": 0.8242,
+      "step": 951
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991351414000573,
+      "loss": 0.8309,
+      "step": 952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029913310480774016,
+      "loss": 0.2524,
+      "step": 953
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029913106582728026,
+      "loss": 0.2781,
+      "step": 954
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991290244587101,
+      "loss": 0.6135,
+      "step": 955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029912698070206245,
+      "loss": 0.6617,
+      "step": 956
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029912493455737,
+      "loss": 0.5783,
+      "step": 957
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029912288602466556,
+      "loss": 1.0577,
+      "step": 958
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991208351039819,
+      "loss": 0.8206,
+      "step": 959
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029911878179535184,
+      "loss": 0.2499,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029911672609880824,
+      "loss": 0.3826,
+      "step": 961
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991146680143841,
+      "loss": 6.0591,
+      "step": 962
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029911260754211235,
+      "loss": 0.7188,
+      "step": 963
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029911054468202605,
+      "loss": 0.3507,
+      "step": 964
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029910847943415803,
+      "loss": 2.4967,
+      "step": 965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991064117985416,
+      "loss": 1.0494,
+      "step": 966
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991043417752097,
+      "loss": 0.1917,
+      "step": 967
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029910226936419554,
+      "loss": 0.4005,
+      "step": 968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002991001945655323,
+      "loss": 0.5091,
+      "step": 969
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029909811737925326,
+      "loss": 1.5878,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029909603780539157,
+      "loss": 0.8375,
+      "step": 971
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990939558439807,
+      "loss": 0.4423,
+      "step": 972
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029909187149505383,
+      "loss": 0.5571,
+      "step": 973
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029908978475864435,
+      "loss": 1.1678,
+      "step": 974
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029908769563478574,
+      "loss": 0.4139,
+      "step": 975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029908560412351146,
+      "loss": 0.687,
+      "step": 976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029908351022485494,
+      "loss": 1.3629,
+      "step": 977
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990814139388498,
+      "loss": 0.7035,
+      "step": 978
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990793152655295,
+      "loss": 1.2759,
+      "step": 979
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029907721420492774,
+      "loss": 0.6908,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029907511075707814,
+      "loss": 0.5393,
+      "step": 981
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029907300492201436,
+      "loss": 0.7518,
+      "step": 982
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990708966997701,
+      "loss": 1.3136,
+      "step": 983
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990687860903792,
+      "loss": 0.794,
+      "step": 984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029906667309387535,
+      "loss": 0.2208,
+      "step": 985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029906455771029253,
+      "loss": 0.6426,
+      "step": 986
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029906243993966445,
+      "loss": 1.1011,
+      "step": 987
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029906031978202516,
+      "loss": 0.6311,
+      "step": 988
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029905819723740853,
+      "loss": 0.1482,
+      "step": 989
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990560723058486,
+      "loss": 4.3845,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990539449873794,
+      "loss": 1.5175,
+      "step": 991
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990518152820349,
+      "loss": 0.3298,
+      "step": 992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990496831898493,
+      "loss": 0.2175,
+      "step": 993
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990475487108567,
+      "loss": 0.4262,
+      "step": 994
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029904541184509135,
+      "loss": 0.2252,
+      "step": 995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029904327259258737,
+      "loss": 0.3779,
+      "step": 996
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990411309533791,
+      "loss": 0.0824,
+      "step": 997
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990389869275007,
+      "loss": 1.3027,
+      "step": 998
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029903684051498674,
+      "loss": 0.3737,
+      "step": 999
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990346917158714,
+      "loss": 0.7352,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990325405301891,
+      "loss": 0.9568,
+      "step": 1001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029903038695797433,
+      "loss": 0.6533,
+      "step": 1002
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990282309992616,
+      "loss": 0.6535,
+      "step": 1003
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990260726540854,
+      "loss": 0.1985,
+      "step": 1004
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990239119224804,
+      "loss": 0.2155,
+      "step": 1005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000299021748804481,
+      "loss": 0.0006,
+      "step": 1006
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029901958330012194,
+      "loss": 0.5097,
+      "step": 1007
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990174154094379,
+      "loss": 0.3575,
+      "step": 1008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029901524513246367,
+      "loss": 0.2923,
+      "step": 1009
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029901307246923384,
+      "loss": 0.6536,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029901089741978334,
+      "loss": 0.3554,
+      "step": 1011
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990087199841469,
+      "loss": 0.3623,
+      "step": 1012
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029900654016235955,
+      "loss": 0.3944,
+      "step": 1013
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029900435795445596,
+      "loss": 0.3554,
+      "step": 1014
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002990021733604713,
+      "loss": 0.5789,
+      "step": 1015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989999863804404,
+      "loss": 0.3797,
+      "step": 1016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989977970143984,
+      "loss": 1.0534,
+      "step": 1017
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989956052623802,
+      "loss": 0.5563,
+      "step": 1018
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989934111244211,
+      "loss": 0.5911,
+      "step": 1019
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298991214600556,
+      "loss": 0.4437,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989890156908203,
+      "loss": 1.0279,
+      "step": 1021
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989868143952491,
+      "loss": 0.6299,
+      "step": 1022
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029898461071387766,
+      "loss": 1.194,
+      "step": 1023
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029898240464674124,
+      "loss": 4.9652,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029898019619387523,
+      "loss": 0.513,
+      "step": 1025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298977985355315,
+      "loss": 0.1193,
+      "step": 1026
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029897577213109587,
+      "loss": 0.4635,
+      "step": 1027
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989735565212533,
+      "loss": 0.4456,
+      "step": 1028
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029897133852582286,
+      "loss": 0.2489,
+      "step": 1029
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029896911814484,
+      "loss": 0.6559,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989668953783403,
+      "loss": 0.8587,
+      "step": 1031
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989646702263593,
+      "loss": 4.1724,
+      "step": 1032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029896244268893266,
+      "loss": 3.4047,
+      "step": 1033
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989602127660961,
+      "loss": 0.7557,
+      "step": 1034
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029895798045788527,
+      "loss": 0.6626,
+      "step": 1035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989557457643359,
+      "loss": 0.7883,
+      "step": 1036
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989535086854839,
+      "loss": 1.3351,
+      "step": 1037
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298951269221365,
+      "loss": 0.5135,
+      "step": 1038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298949027372015,
+      "loss": 1.8054,
+      "step": 1039
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029894678313746993,
+      "loss": 0.4182,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989445365177656,
+      "loss": 0.8641,
+      "step": 1041
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989422875129381,
+      "loss": 0.3784,
+      "step": 1042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989400361230234,
+      "loss": 0.9567,
+      "step": 1043
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029893778234805753,
+      "loss": 0.533,
+      "step": 1044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029893552618807663,
+      "loss": 0.8951,
+      "step": 1045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029893326764311674,
+      "loss": 4.1852,
+      "step": 1046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029893100671321413,
+      "loss": 0.9375,
+      "step": 1047
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029892874339840496,
+      "loss": 0.8032,
+      "step": 1048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029892647769872544,
+      "loss": 0.368,
+      "step": 1049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989242096142119,
+      "loss": 0.4057,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989219391449007,
+      "loss": 0.4079,
+      "step": 1051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029891966629082804,
+      "loss": 2.0127,
+      "step": 1052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029891739105203054,
+      "loss": 0.4465,
+      "step": 1053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029891511342854446,
+      "loss": 0.5166,
+      "step": 1054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029891283342040634,
+      "loss": 0.3368,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029891055102765265,
+      "loss": 0.4449,
+      "step": 1056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029890826625032,
+      "loss": 2.3287,
+      "step": 1057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989059790884449,
+      "loss": 1.4501,
+      "step": 1058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989036895420641,
+      "loss": 0.9448,
+      "step": 1059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002989013976112141,
+      "loss": 0.4909,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029889910329593174,
+      "loss": 0.5025,
+      "step": 1061
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988968065962537,
+      "loss": 0.1797,
+      "step": 1062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988945075122168,
+      "loss": 0.2703,
+      "step": 1063
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029889220604385774,
+      "loss": 0.4091,
+      "step": 1064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029888990219121355,
+      "loss": 0.4542,
+      "step": 1065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029888759595432095,
+      "loss": 3.7607,
+      "step": 1066
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298885287333217,
+      "loss": 0.5195,
+      "step": 1067
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029888297632793857,
+      "loss": 1.3322,
+      "step": 1068
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988806629385227,
+      "loss": 1.0276,
+      "step": 1069
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988783471650066,
+      "loss": 0.1306,
+      "step": 1070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029887602900742705,
+      "loss": 1.0508,
+      "step": 1071
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988737084658214,
+      "loss": 0.7646,
+      "step": 1072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988713855402267,
+      "loss": 0.6347,
+      "step": 1073
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988690602306802,
+      "loss": 0.7973,
+      "step": 1074
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988667325372191,
+      "loss": 0.0358,
+      "step": 1075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988644024598807,
+      "loss": 2.6185,
+      "step": 1076
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029886206999870236,
+      "loss": 1.5649,
+      "step": 1077
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988597351537213,
+      "loss": 2.2712,
+      "step": 1078
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029885739792497506,
+      "loss": 0.2692,
+      "step": 1079
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298855058312501,
+      "loss": 1.6648,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029885271631633655,
+      "loss": 1.5293,
+      "step": 1081
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988503719365192,
+      "loss": 0.7427,
+      "step": 1082
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988480251730866,
+      "loss": 0.2913,
+      "step": 1083
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029884567602607627,
+      "loss": 1.6609,
+      "step": 1084
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029884332449552574,
+      "loss": 1.779,
+      "step": 1085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029884097058147286,
+      "loss": 0.0348,
+      "step": 1086
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988386142839552,
+      "loss": 1.9334,
+      "step": 1087
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988362556030104,
+      "loss": 1.3197,
+      "step": 1088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029883389453867645,
+      "loss": 0.3256,
+      "step": 1089
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298831531090991,
+      "loss": 1.4009,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029882916525999196,
+      "loss": 0.5543,
+      "step": 1091
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988267970457172,
+      "loss": 0.0235,
+      "step": 1092
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029882442644820465,
+      "loss": 1.834,
+      "step": 1093
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988220534674923,
+      "loss": 0.4342,
+      "step": 1094
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029881967810361804,
+      "loss": 0.554,
+      "step": 1095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029881730035662,
+      "loss": 1.6521,
+      "step": 1096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029881492022653625,
+      "loss": 0.407,
+      "step": 1097
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988125377134049,
+      "loss": 0.5393,
+      "step": 1098
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988101528172641,
+      "loss": 0.9902,
+      "step": 1099
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298807765538152,
+      "loss": 0.5744,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002988053758761069,
+      "loss": 1.0854,
+      "step": 1101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000298802983831167,
+      "loss": 0.6826,
+      "step": 1102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029880058940337067,
+      "loss": 0.4425,
+      "step": 1103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002987981925927562,
+      "loss": 1.3798,
+      "step": 1104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029879579339936197,
+      "loss": 0.1355,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002987933918232265,
+      "loss": 1.2349,
+      "step": 1106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029879098786438806,
+      "loss": 0.5253,
+      "step": 1107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002987885815228853,
+      "loss": 0.9236,
+      "step": 1108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029878617279875674,
+      "loss": 1.9825,
+      "step": 1109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002987837616920409,
+      "loss": 1.589,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029878134820277637,
+      "loss": 0.5732,
+      "step": 1111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002987789323310019,
+      "loss": 0.5291,
+      "step": 1112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029877651407675605,
+      "loss": 0.2494,
+      "step": 1113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029877409344007764,
+      "loss": 0.3808,
+      "step": 1114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002987716704210054,
+      "loss": 1.0081,
+      "step": 1115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029876924501957814,
+      "loss": 0.5692,
+      "step": 1116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002987668172358347,
+      "loss": 0.8389,
+      "step": 1117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029876438706981395,
+      "loss": 0.2524,
+      "step": 1118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029876195452155477,
+      "loss": 0.3396,
+      "step": 1119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029875951959109617,
+      "loss": 0.0554,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029875708227847714,
+      "loss": 0.737,
+      "step": 1121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987546425837367,
+      "loss": 0.0987,
+      "step": 1122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029875220050691386,
+      "loss": 0.1778,
+      "step": 1123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987497560480478,
+      "loss": 0.1349,
+      "step": 1124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029874730920717763,
+      "loss": 1.0875,
+      "step": 1125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029874485998434257,
+      "loss": 0.4404,
+      "step": 1126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987424083795818,
+      "loss": 1.71,
+      "step": 1127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029873995439293454,
+      "loss": 0.8929,
+      "step": 1128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987374980244402,
+      "loss": 1.1732,
+      "step": 1129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029873503927413804,
+      "loss": 0.3088,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029873257814206744,
+      "loss": 0.4324,
+      "step": 1131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987301146282678,
+      "loss": 0.0123,
+      "step": 1132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987276487327786,
+      "loss": 0.8436,
+      "step": 1133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987251804556393,
+      "loss": 0.6492,
+      "step": 1134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029872270979688947,
+      "loss": 0.4837,
+      "step": 1135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029872023675656865,
+      "loss": 0.5273,
+      "step": 1136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987177613347164,
+      "loss": 0.9841,
+      "step": 1137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987152835313724,
+      "loss": 0.0995,
+      "step": 1138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029871280334657635,
+      "loss": 0.346,
+      "step": 1139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987103207803679,
+      "loss": 1.2408,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029870783583278693,
+      "loss": 0.4761,
+      "step": 1141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987053485038731,
+      "loss": 0.7216,
+      "step": 1142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029870285879366625,
+      "loss": 0.7437,
+      "step": 1143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002987003667022063,
+      "loss": 1.1406,
+      "step": 1144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986978722295332,
+      "loss": 0.4197,
+      "step": 1145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029869537537568683,
+      "loss": 0.4418,
+      "step": 1146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986928761407071,
+      "loss": 0.558,
+      "step": 1147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029869037452463423,
+      "loss": 0.65,
+      "step": 1148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986878705275081,
+      "loss": 0.9091,
+      "step": 1149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986853641493689,
+      "loss": 0.6679,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986828553902567,
+      "loss": 0.2689,
+      "step": 1151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029868034425021173,
+      "loss": 0.3997,
+      "step": 1152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029867783072927423,
+      "loss": 0.0093,
+      "step": 1153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986753148274843,
+      "loss": 1.4716,
+      "step": 1154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986727965448824,
+      "loss": 0.7709,
+      "step": 1155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029867027588150884,
+      "loss": 0.123,
+      "step": 1156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029866775283740387,
+      "loss": 0.0037,
+      "step": 1157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298665227412608,
+      "loss": 0.0036,
+      "step": 1158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029866269960716153,
+      "loss": 0.3454,
+      "step": 1159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029866016942110515,
+      "loss": 0.3882,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029865763685447917,
+      "loss": 1.2862,
+      "step": 1161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986551019073243,
+      "loss": 0.2353,
+      "step": 1162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986525645796811,
+      "loss": 1.309,
+      "step": 1163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986500248715901,
+      "loss": 0.4305,
+      "step": 1164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029864748278309207,
+      "loss": 1.2008,
+      "step": 1165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029864493831422773,
+      "loss": 0.9319,
+      "step": 1166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029864239146503774,
+      "loss": 0.72,
+      "step": 1167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298639842235563,
+      "loss": 2.755,
+      "step": 1168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029863729062584417,
+      "loss": 0.3581,
+      "step": 1169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986347366359223,
+      "loss": 0.2871,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029863218026583813,
+      "loss": 0.8202,
+      "step": 1171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029862962151563265,
+      "loss": 0.4296,
+      "step": 1172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986270603853469,
+      "loss": 0.3077,
+      "step": 1173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029862449687502175,
+      "loss": 0.2855,
+      "step": 1174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986219309846984,
+      "loss": 0.6558,
+      "step": 1175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986193627144179,
+      "loss": 0.4507,
+      "step": 1176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029861679206422126,
+      "loss": 0.5765,
+      "step": 1177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986142190341498,
+      "loss": 2.1127,
+      "step": 1178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029861164362424467,
+      "loss": 0.5063,
+      "step": 1179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029860906583454704,
+      "loss": 0.5453,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029860648566509834,
+      "loss": 0.2578,
+      "step": 1181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029860390311593977,
+      "loss": 1.9338,
+      "step": 1182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002986013181871127,
+      "loss": 0.4206,
+      "step": 1183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985987308786585,
+      "loss": 1.0649,
+      "step": 1184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985961411906187,
+      "loss": 0.8388,
+      "step": 1185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985935491230347,
+      "loss": 0.3,
+      "step": 1186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029859095467594805,
+      "loss": 0.2407,
+      "step": 1187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029858835784940014,
+      "loss": 0.6752,
+      "step": 1188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985857586434328,
+      "loss": 0.4638,
+      "step": 1189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985831570580875,
+      "loss": 0.3337,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029858055309340593,
+      "loss": 0.6802,
+      "step": 1191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985779467494298,
+      "loss": 0.5261,
+      "step": 1192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985753380262008,
+      "loss": 0.6721,
+      "step": 1193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029857272692376076,
+      "loss": 0.7317,
+      "step": 1194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029857011344215146,
+      "loss": 0.4485,
+      "step": 1195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029856749758141484,
+      "loss": 0.484,
+      "step": 1196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985648793415926,
+      "loss": 0.4263,
+      "step": 1197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029856225872272683,
+      "loss": 0.4201,
+      "step": 1198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985596357248595,
+      "loss": 1.1119,
+      "step": 1199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029855701034803257,
+      "loss": 0.8075,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029855438259228796,
+      "loss": 0.7074,
+      "step": 1201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029855175245766793,
+      "loss": 1.0385,
+      "step": 1202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029854911994421456,
+      "loss": 0.2883,
+      "step": 1203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985464850519699,
+      "loss": 0.0522,
+      "step": 1204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985438477809763,
+      "loss": 0.3093,
+      "step": 1205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985412081312758,
+      "loss": 0.2516,
+      "step": 1206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029853856610291085,
+      "loss": 0.1572,
+      "step": 1207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029853592169592374,
+      "loss": 0.4271,
+      "step": 1208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985332749103567,
+      "loss": 1.0423,
+      "step": 1209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985306257462522,
+      "loss": 0.2668,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985279742036526,
+      "loss": 0.615,
+      "step": 1211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029852532028260047,
+      "loss": 0.6872,
+      "step": 1212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029852266398313823,
+      "loss": 0.9604,
+      "step": 1213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985200053053084,
+      "loss": 0.3855,
+      "step": 1214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985173442491536,
+      "loss": 0.5392,
+      "step": 1215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985146808147164,
+      "loss": 0.0691,
+      "step": 1216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985120150020395,
+      "loss": 0.6627,
+      "step": 1217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029850934681116553,
+      "loss": 0.0049,
+      "step": 1218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029850667624213724,
+      "loss": 0.2971,
+      "step": 1219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985040032949975,
+      "loss": 0.3956,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002985013279697889,
+      "loss": 0.3858,
+      "step": 1221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029849865026655454,
+      "loss": 0.4768,
+      "step": 1222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984959701853371,
+      "loss": 0.8525,
+      "step": 1223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984932877261795,
+      "loss": 0.0505,
+      "step": 1224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029849060288912485,
+      "loss": 0.4015,
+      "step": 1225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298487915674216,
+      "loss": 0.3356,
+      "step": 1226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298485226081496,
+      "loss": 0.4455,
+      "step": 1227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029848253411100795,
+      "loss": 0.5801,
+      "step": 1228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298479839762795,
+      "loss": 0.3539,
+      "step": 1229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029847714303690026,
+      "loss": 0.0018,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029847444393336686,
+      "loss": 0.7412,
+      "step": 1231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984717424522381,
+      "loss": 0.3175,
+      "step": 1232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984690385935572,
+      "loss": 0.6519,
+      "step": 1233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984663323573675,
+      "loss": 0.5466,
+      "step": 1234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029846362374371223,
+      "loss": 2.6799,
+      "step": 1235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029846091275263493,
+      "loss": 0.9054,
+      "step": 1236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029845819938417885,
+      "loss": 1.1795,
+      "step": 1237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029845548363838757,
+      "loss": 0.3124,
+      "step": 1238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029845276551530446,
+      "loss": 0.4187,
+      "step": 1239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029845004501497316,
+      "loss": 0.0025,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029844732213743714,
+      "loss": 0.4403,
+      "step": 1241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984445968827401,
+      "loss": 0.4305,
+      "step": 1242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984418692509255,
+      "loss": 0.4682,
+      "step": 1243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984391392420373,
+      "loss": 3.5746,
+      "step": 1244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029843640685611896,
+      "loss": 0.2206,
+      "step": 1245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984336720932144,
+      "loss": 0.2843,
+      "step": 1246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029843093495336725,
+      "loss": 1.8343,
+      "step": 1247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984281954366215,
+      "loss": 0.5017,
+      "step": 1248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298425453543021,
+      "loss": 2.1938,
+      "step": 1249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984227092726096,
+      "loss": 0.4121,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029841996262543126,
+      "loss": 1.8785,
+      "step": 1251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029841721360152995,
+      "loss": 0.1499,
+      "step": 1252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984144622009497,
+      "loss": 0.4938,
+      "step": 1253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029841170842373455,
+      "loss": 0.4759,
+      "step": 1254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029840895226992866,
+      "loss": 0.4987,
+      "step": 1255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984061937395761,
+      "loss": 1.0962,
+      "step": 1256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029840343283272114,
+      "loss": 6.4129,
+      "step": 1257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002984006695494079,
+      "loss": 0.2109,
+      "step": 1258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029839790388968066,
+      "loss": 0.0588,
+      "step": 1259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983951358535836,
+      "loss": 1.0189,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983923654411613,
+      "loss": 0.4313,
+      "step": 1261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983895926524579,
+      "loss": 0.3112,
+      "step": 1262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983868174875179,
+      "loss": 0.4846,
+      "step": 1263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029838403994638567,
+      "loss": 0.3745,
+      "step": 1264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983812600291058,
+      "loss": 0.9863,
+      "step": 1265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029837847773572264,
+      "loss": 0.7129,
+      "step": 1266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029837569306628094,
+      "loss": 0.7103,
+      "step": 1267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983729060208252,
+      "loss": 0.4657,
+      "step": 1268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029837011659939995,
+      "loss": 0.2826,
+      "step": 1269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029836732480205007,
+      "loss": 0.157,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983645306288201,
+      "loss": 0.0155,
+      "step": 1271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983617340797548,
+      "loss": 0.3915,
+      "step": 1272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298358935154899,
+      "loss": 0.5472,
+      "step": 1273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029835613385429755,
+      "loss": 0.0081,
+      "step": 1274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029835333017799525,
+      "loss": 0.5005,
+      "step": 1275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298350524126037,
+      "loss": 0.5163,
+      "step": 1276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029834771569846774,
+      "loss": 0.4016,
+      "step": 1277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983449048953325,
+      "loss": 0.5705,
+      "step": 1278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029834209171667624,
+      "loss": 0.6577,
+      "step": 1279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298339276162544,
+      "loss": 1.4493,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983364582329808,
+      "loss": 0.6448,
+      "step": 1281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983336379280319,
+      "loss": 0.667,
+      "step": 1282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983308152477425,
+      "loss": 2.2483,
+      "step": 1283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983279901921576,
+      "loss": 0.9235,
+      "step": 1284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029832516276132254,
+      "loss": 0.368,
+      "step": 1285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029832233295528266,
+      "loss": 0.1225,
+      "step": 1286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029831950077408313,
+      "loss": 0.3589,
+      "step": 1287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983166662177695,
+      "loss": 2.0767,
+      "step": 1288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298313829286387,
+      "loss": 0.0058,
+      "step": 1289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983109899799811,
+      "loss": 0.6639,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029830814829859726,
+      "loss": 0.368,
+      "step": 1291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002983053042422811,
+      "loss": 0.6515,
+      "step": 1292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029830245781107797,
+      "loss": 0.8121,
+      "step": 1293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982996090050336,
+      "loss": 0.4888,
+      "step": 1294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982967578241935,
+      "loss": 0.3614,
+      "step": 1295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029829390426860345,
+      "loss": 0.3878,
+      "step": 1296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029829104833830906,
+      "loss": 0.4714,
+      "step": 1297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982881900333561,
+      "loss": 0.3619,
+      "step": 1298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982853293537903,
+      "loss": 0.4123,
+      "step": 1299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029828246629965755,
+      "loss": 1.2433,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982796008710036,
+      "loss": 0.3795,
+      "step": 1301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029827673306787435,
+      "loss": 0.3579,
+      "step": 1302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982738628903158,
+      "loss": 0.266,
+      "step": 1303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982709903383738,
+      "loss": 1.061,
+      "step": 1304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982681154120945,
+      "loss": 0.3992,
+      "step": 1305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982652381115238,
+      "loss": 0.2739,
+      "step": 1306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982623584367078,
+      "loss": 0.5786,
+      "step": 1307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029825947638769274,
+      "loss": 0.0227,
+      "step": 1308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982565919645246,
+      "loss": 2.3752,
+      "step": 1309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029825370516724963,
+      "loss": 1.6689,
+      "step": 1310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982508159959141,
+      "loss": 0.6507,
+      "step": 1311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982479244505642,
+      "loss": 0.3078,
+      "step": 1312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029824503053124636,
+      "loss": 0.6572,
+      "step": 1313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982421342380068,
+      "loss": 0.8039,
+      "step": 1314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982392355708919,
+      "loss": 0.464,
+      "step": 1315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029823633452994813,
+      "loss": 0.4842,
+      "step": 1316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298233431115222,
+      "loss": 0.2844,
+      "step": 1317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029823052532675987,
+      "loss": 0.6501,
+      "step": 1318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029822761716460835,
+      "loss": 0.4487,
+      "step": 1319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029822470662881404,
+      "loss": 0.746,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029822179371942346,
+      "loss": 1.8234,
+      "step": 1321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982188784364833,
+      "loss": 3.481,
+      "step": 1322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002982159607800403,
+      "loss": 0.5231,
+      "step": 1323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029821304075014113,
+      "loss": 0.4446,
+      "step": 1324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029821011834683253,
+      "loss": 0.4891,
+      "step": 1325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029820719357016124,
+      "loss": 0.6432,
+      "step": 1326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029820426642017427,
+      "loss": 0.7765,
+      "step": 1327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029820133689691835,
+      "loss": 0.48,
+      "step": 1328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029819840500044047,
+      "loss": 2.749,
+      "step": 1329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029819547073078753,
+      "loss": 1.1266,
+      "step": 1330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029819253408800653,
+      "loss": 0.3576,
+      "step": 1331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029818959507214453,
+      "loss": 0.2134,
+      "step": 1332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029818665368324857,
+      "loss": 0.4379,
+      "step": 1333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981837099213657,
+      "loss": 0.4789,
+      "step": 1334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981807637865431,
+      "loss": 0.4601,
+      "step": 1335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298177815278828,
+      "loss": 4.96,
+      "step": 1336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029817486439826754,
+      "loss": 0.5905,
+      "step": 1337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029817191114490905,
+      "loss": 0.341,
+      "step": 1338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981689555187998,
+      "loss": 2.6038,
+      "step": 1339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000298165997519987,
+      "loss": 0.4313,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981630371485182,
+      "loss": 0.4402,
+      "step": 1341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029816007440444073,
+      "loss": 0.6681,
+      "step": 1342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029815710928780196,
+      "loss": 0.2891,
+      "step": 1343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029815414179864945,
+      "loss": 1.1695,
+      "step": 1344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981511719370307,
+      "loss": 0.3235,
+      "step": 1345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981481997029933,
+      "loss": 0.6936,
+      "step": 1346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029814522509658486,
+      "loss": 5.4087,
+      "step": 1347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981422481178529,
+      "loss": 0.6898,
+      "step": 1348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981392687668452,
+      "loss": 1.197,
+      "step": 1349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981362870436095,
+      "loss": 0.7382,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981333029481934,
+      "loss": 2.9387,
+      "step": 1351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029813031648064477,
+      "loss": 0.2913,
+      "step": 1352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029812732764101146,
+      "loss": 0.083,
+      "step": 1353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029812433642934134,
+      "loss": 0.5557,
+      "step": 1354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981213428456822,
+      "loss": 0.4749,
+      "step": 1355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981183468900821,
+      "loss": 0.705,
+      "step": 1356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029811534856258895,
+      "loss": 0.6068,
+      "step": 1357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981123478632508,
+      "loss": 0.8256,
+      "step": 1358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029810934479211566,
+      "loss": 0.6578,
+      "step": 1359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981063393492317,
+      "loss": 0.6963,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029810333153464693,
+      "loss": 0.0394,
+      "step": 1361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002981003213484096,
+      "loss": 1.1988,
+      "step": 1362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029809730879056787,
+      "loss": 0.5193,
+      "step": 1363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029809429386117004,
+      "loss": 0.0128,
+      "step": 1364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029809127656026433,
+      "loss": 1.0269,
+      "step": 1365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002980882568878991,
+      "loss": 0.5834,
+      "step": 1366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002980852348441227,
+      "loss": 0.7885,
+      "step": 1367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029808221042898345,
+      "loss": 0.0701,
+      "step": 1368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029807918364252987,
+      "loss": 2.0845,
+      "step": 1369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002980761544848104,
+      "loss": 0.7173,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029807312295587355,
+      "loss": 0.4817,
+      "step": 1371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029807008905576786,
+      "loss": 0.9277,
+      "step": 1372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980670527845419,
+      "loss": 0.2472,
+      "step": 1373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980640141422443,
+      "loss": 0.2713,
+      "step": 1374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029806097312892374,
+      "loss": 0.3595,
+      "step": 1375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029805792974462895,
+      "loss": 0.4002,
+      "step": 1376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980548839894086,
+      "loss": 0.6564,
+      "step": 1377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980518358633114,
+      "loss": 0.2357,
+      "step": 1378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980487853663864,
+      "loss": 0.5183,
+      "step": 1379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980457324986822,
+      "loss": 0.5629,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029804267726024776,
+      "loss": 0.1545,
+      "step": 1381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000298039619651132,
+      "loss": 0.2156,
+      "step": 1382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029803655967138395,
+      "loss": 1.6445,
+      "step": 1383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980334973210526,
+      "loss": 0.533,
+      "step": 1384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980304326001869,
+      "loss": 0.75,
+      "step": 1385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029802736550883606,
+      "loss": 0.1853,
+      "step": 1386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029802429604704905,
+      "loss": 1.4368,
+      "step": 1387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980212242148751,
+      "loss": 0.6644,
+      "step": 1388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980181500123634,
+      "loss": 0.5216,
+      "step": 1389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980150734395632,
+      "loss": 0.6819,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029801199449652375,
+      "loss": 1.3306,
+      "step": 1391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002980089131832943,
+      "loss": 0.2993,
+      "step": 1392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029800582949992426,
+      "loss": 0.2151,
+      "step": 1393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029800274344646293,
+      "loss": 0.2341,
+      "step": 1394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979996550229598,
+      "loss": 3.3148,
+      "step": 1395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979965642294644,
+      "loss": 0.789,
+      "step": 1396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029799347106602606,
+      "loss": 1.2306,
+      "step": 1397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979903755326944,
+      "loss": 0.4352,
+      "step": 1398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297987277629519,
+      "loss": 0.426,
+      "step": 1399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979841773565494,
+      "loss": 0.635,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029798107471383533,
+      "loss": 0.3757,
+      "step": 1401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979779697014264,
+      "loss": 0.2998,
+      "step": 1402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979748623193724,
+      "loss": 0.3119,
+      "step": 1403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979717525677231,
+      "loss": 0.753,
+      "step": 1404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979686404465282,
+      "loss": 0.38,
+      "step": 1405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029796552595583756,
+      "loss": 0.2268,
+      "step": 1406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029796240909570114,
+      "loss": 0.9142,
+      "step": 1407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979592898661688,
+      "loss": 0.7947,
+      "step": 1408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979561682672905,
+      "loss": 1.0531,
+      "step": 1409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979530442991162,
+      "loss": 0.8839,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979499179616959,
+      "loss": 0.1409,
+      "step": 1411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029794678925507975,
+      "loss": 1.7604,
+      "step": 1412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979436581793178,
+      "loss": 7.3418,
+      "step": 1413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029794052473446016,
+      "loss": 0.648,
+      "step": 1414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979373889205571,
+      "loss": 0.9539,
+      "step": 1415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029793425073765874,
+      "loss": 0.3229,
+      "step": 1416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979311101858154,
+      "loss": 0.8845,
+      "step": 1417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029792796726507736,
+      "loss": 0.6991,
+      "step": 1418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979248219754949,
+      "loss": 0.2007,
+      "step": 1419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979216743171184,
+      "loss": 0.6523,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029791852428999834,
+      "loss": 0.3098,
+      "step": 1421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979153718941851,
+      "loss": 0.3459,
+      "step": 1422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029791221712972913,
+      "loss": 0.4898,
+      "step": 1423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029790905999668103,
+      "loss": 0.6201,
+      "step": 1424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002979059004950913,
+      "loss": 0.1855,
+      "step": 1425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029790273862501056,
+      "loss": 0.2419,
+      "step": 1426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978995743864894,
+      "loss": 0.4038,
+      "step": 1427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029789640777957857,
+      "loss": 0.2493,
+      "step": 1428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029789323880432866,
+      "loss": 0.2204,
+      "step": 1429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978900674607905,
+      "loss": 0.4744,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029788689374901493,
+      "loss": 0.0413,
+      "step": 1431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029788371766905264,
+      "loss": 0.7745,
+      "step": 1432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029788053922095455,
+      "loss": 0.5911,
+      "step": 1433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978773584047716,
+      "loss": 0.3085,
+      "step": 1434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978741752205547,
+      "loss": 0.2638,
+      "step": 1435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978709896683548,
+      "loss": 0.4287,
+      "step": 1436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978678017482229,
+      "loss": 1.8328,
+      "step": 1437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029786461146021005,
+      "loss": 0.512,
+      "step": 1438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978614188043674,
+      "loss": 0.4144,
+      "step": 1439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029785822378074607,
+      "loss": 0.4314,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978550263893971,
+      "loss": 2.8832,
+      "step": 1441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029785182663037186,
+      "loss": 0.6054,
+      "step": 1442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978486245037215,
+      "loss": 0.2769,
+      "step": 1443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029784542000949727,
+      "loss": 0.7809,
+      "step": 1444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978422131477505,
+      "loss": 2.2076,
+      "step": 1445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029783900391853265,
+      "loss": 0.4326,
+      "step": 1446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297835792321895,
+      "loss": 0.7844,
+      "step": 1447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297832578357889,
+      "loss": 0.9925,
+      "step": 1448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978293620265661,
+      "loss": 1.0259,
+      "step": 1449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029782614332797785,
+      "loss": 0.3537,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978229222621758,
+      "loss": 0.3302,
+      "step": 1451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029781969882921146,
+      "loss": 0.0289,
+      "step": 1452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029781647302913654,
+      "loss": 0.185,
+      "step": 1453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978132448620026,
+      "loss": 0.4229,
+      "step": 1454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978100143278614,
+      "loss": 0.408,
+      "step": 1455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978067814267646,
+      "loss": 0.7042,
+      "step": 1456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029780354615876407,
+      "loss": 1.2514,
+      "step": 1457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002978003085239116,
+      "loss": 0.2353,
+      "step": 1458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977970685222589,
+      "loss": 0.2432,
+      "step": 1459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297793826153858,
+      "loss": 0.8877,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977905814187609,
+      "loss": 0.5,
+      "step": 1461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977873343170193,
+      "loss": 5.0793,
+      "step": 1462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029778408484868536,
+      "loss": 0.4218,
+      "step": 1463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977808330138111,
+      "loss": 0.6598,
+      "step": 1464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029777757881244865,
+      "loss": 0.5762,
+      "step": 1465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029777432224465,
+      "loss": 0.9055,
+      "step": 1466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029777106331046744,
+      "loss": 0.954,
+      "step": 1467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029776780200995297,
+      "loss": 0.5355,
+      "step": 1468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029776453834315895,
+      "loss": 1.0463,
+      "step": 1469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977612723101376,
+      "loss": 0.4031,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977580039109413,
+      "loss": 0.7447,
+      "step": 1471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029775473314562224,
+      "loss": 1.9116,
+      "step": 1472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029775146001423294,
+      "loss": 1.4164,
+      "step": 1473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977481845168257,
+      "loss": 0.3033,
+      "step": 1474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029774490665345307,
+      "loss": 1.2505,
+      "step": 1475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029774162642416744,
+      "loss": 0.3085,
+      "step": 1476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977383438290215,
+      "loss": 0.5353,
+      "step": 1477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977350588680676,
+      "loss": 0.4031,
+      "step": 1478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029773177154135844,
+      "loss": 1.3333,
+      "step": 1479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029772848184894674,
+      "loss": 0.4079,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977251897908851,
+      "loss": 0.0746,
+      "step": 1481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029772189536722625,
+      "loss": 0.2525,
+      "step": 1482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977185985780229,
+      "loss": 0.2004,
+      "step": 1483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977152994233279,
+      "loss": 0.7164,
+      "step": 1484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977119979031941,
+      "loss": 0.1899,
+      "step": 1485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977086940176743,
+      "loss": 2.0803,
+      "step": 1486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002977053877668215,
+      "loss": 0.61,
+      "step": 1487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029770207915068856,
+      "loss": 0.3833,
+      "step": 1488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029769876816932854,
+      "loss": 0.9496,
+      "step": 1489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029769545482279434,
+      "loss": 0.5668,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976921391111392,
+      "loss": 0.4094,
+      "step": 1491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029768882103441596,
+      "loss": 0.6348,
+      "step": 1492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297685500592678,
+      "loss": 0.9134,
+      "step": 1493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029768217778597834,
+      "loss": 0.4273,
+      "step": 1494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976788526143703,
+      "loss": 0.3059,
+      "step": 1495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297675525077907,
+      "loss": 0.356,
+      "step": 1496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029767219517664185,
+      "loss": 1.6428,
+      "step": 1497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976688629106281,
+      "loss": 0.6173,
+      "step": 1498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976655282799191,
+      "loss": 0.4395,
+      "step": 1499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976621912845683,
+      "loss": 0.6387,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976588519246291,
+      "loss": 0.6686,
+      "step": 1501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029765551020015504,
+      "loss": 0.8186,
+      "step": 1502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976521661111996,
+      "loss": 0.6635,
+      "step": 1503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029764881965781625,
+      "loss": 0.5171,
+      "step": 1504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976454708400587,
+      "loss": 1.035,
+      "step": 1505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029764211965798044,
+      "loss": 0.4761,
+      "step": 1506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976387661116353,
+      "loss": 0.4571,
+      "step": 1507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029763541020107686,
+      "loss": 0.3725,
+      "step": 1508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029763205192635887,
+      "loss": 0.9942,
+      "step": 1509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029762869128753517,
+      "loss": 0.8251,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029762532828465953,
+      "loss": 0.2106,
+      "step": 1511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976219629177858,
+      "loss": 0.5783,
+      "step": 1512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976185951869679,
+      "loss": 1.5884,
+      "step": 1513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976152250922597,
+      "loss": 0.5329,
+      "step": 1514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976118526337153,
+      "loss": 2.1877,
+      "step": 1515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029760847781138853,
+      "loss": 2.5776,
+      "step": 1516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976051006253335,
+      "loss": 0.4097,
+      "step": 1517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002976017210756044,
+      "loss": 0.6592,
+      "step": 1518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975983391622552,
+      "loss": 0.401,
+      "step": 1519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975949548853401,
+      "loss": 0.611,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975915682449133,
+      "loss": 0.2612,
+      "step": 1521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029758817924102913,
+      "loss": 0.9021,
+      "step": 1522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029758478787374166,
+      "loss": 0.7503,
+      "step": 1523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029758139414310536,
+      "loss": 0.5166,
+      "step": 1524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975779980491745,
+      "loss": 0.9672,
+      "step": 1525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029757459959200354,
+      "loss": 0.7019,
+      "step": 1526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029757119877164677,
+      "loss": 0.3943,
+      "step": 1527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029756779558815877,
+      "loss": 0.4275,
+      "step": 1528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029756439004159397,
+      "loss": 1.1522,
+      "step": 1529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975609821320069,
+      "loss": 0.5988,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975575718594522,
+      "loss": 0.3106,
+      "step": 1531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975541592239845,
+      "loss": 0.2668,
+      "step": 1532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029755074422565825,
+      "loss": 0.842,
+      "step": 1533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029754732686452834,
+      "loss": 0.5417,
+      "step": 1534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975439071406495,
+      "loss": 0.7922,
+      "step": 1535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975404850540763,
+      "loss": 0.84,
+      "step": 1536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029753706060486375,
+      "loss": 0.2458,
+      "step": 1537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029753363379306653,
+      "loss": 1.76,
+      "step": 1538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975302046187396,
+      "loss": 0.355,
+      "step": 1539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029752677308193787,
+      "loss": 0.9301,
+      "step": 1540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029752333918271626,
+      "loss": 0.6381,
+      "step": 1541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975199029211298,
+      "loss": 0.6721,
+      "step": 1542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002975164642972335,
+      "loss": 0.4546,
+      "step": 1543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029751302331108235,
+      "loss": 0.3785,
+      "step": 1544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029750957996273154,
+      "loss": 0.3738,
+      "step": 1545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029750613425223623,
+      "loss": 1.3231,
+      "step": 1546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029750268617965156,
+      "loss": 0.7197,
+      "step": 1547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029749923574503266,
+      "loss": 0.6787,
+      "step": 1548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297495782948435,
+      "loss": 0.3895,
+      "step": 1549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974923277899136,
+      "loss": 0.5227,
+      "step": 1550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297488870269524,
+      "loss": 0.1752,
+      "step": 1551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029748541038732146,
+      "loss": 1.0524,
+      "step": 1552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974819481433614,
+      "loss": 0.7697,
+      "step": 1553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974784835376993,
+      "loss": 0.8141,
+      "step": 1554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974750165703907,
+      "loss": 0.473,
+      "step": 1555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029747154724149094,
+      "loss": 0.8152,
+      "step": 1556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974680755510557,
+      "loss": 0.1941,
+      "step": 1557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974646014991406,
+      "loss": 0.6677,
+      "step": 1558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974611250858011,
+      "loss": 0.5981,
+      "step": 1559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029745764631109307,
+      "loss": 0.5608,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974541651750722,
+      "loss": 1.0788,
+      "step": 1561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029745068167779407,
+      "loss": 0.8509,
+      "step": 1562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974471958193146,
+      "loss": 0.6351,
+      "step": 1563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029744370759968955,
+      "loss": 0.2036,
+      "step": 1564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029744021701897484,
+      "loss": 0.6395,
+      "step": 1565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029743672407722634,
+      "loss": 1.5311,
+      "step": 1566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029743322877449994,
+      "loss": 0.4036,
+      "step": 1567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029742973111085164,
+      "loss": 0.7911,
+      "step": 1568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974262310863375,
+      "loss": 0.0858,
+      "step": 1569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974227287010135,
+      "loss": 0.2924,
+      "step": 1570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974192239549358,
+      "loss": 1.1187,
+      "step": 1571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974157168481604,
+      "loss": 0.4094,
+      "step": 1572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029741220738074354,
+      "loss": 0.627,
+      "step": 1573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029740869555274147,
+      "loss": 0.7465,
+      "step": 1574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974051813642103,
+      "loss": 0.6259,
+      "step": 1575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002974016648152064,
+      "loss": 0.6523,
+      "step": 1576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973981459057861,
+      "loss": 0.8543,
+      "step": 1577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029739462463600566,
+      "loss": 1.2156,
+      "step": 1578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029739110100592155,
+      "loss": 3.5162,
+      "step": 1579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029738757501559013,
+      "loss": 0.7307,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973840466650679,
+      "loss": 0.6489,
+      "step": 1581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973805159544114,
+      "loss": 1.438,
+      "step": 1582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973769828836771,
+      "loss": 0.6054,
+      "step": 1583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029737344745292153,
+      "loss": 1.0054,
+      "step": 1584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029736990966220143,
+      "loss": 0.689,
+      "step": 1585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973663695115734,
+      "loss": 0.1234,
+      "step": 1586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029736282700109414,
+      "loss": 0.7667,
+      "step": 1587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029735928213082036,
+      "loss": 0.6197,
+      "step": 1588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029735573490080884,
+      "loss": 0.8113,
+      "step": 1589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029735218531111635,
+      "loss": 0.5565,
+      "step": 1590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029734863336179976,
+      "loss": 0.5788,
+      "step": 1591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297345079052916,
+      "loss": 0.5037,
+      "step": 1592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029734152238452183,
+      "loss": 0.2083,
+      "step": 1593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973379633566744,
+      "loss": 3.3983,
+      "step": 1594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029733440196943057,
+      "loss": 0.8043,
+      "step": 1595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973308382228474,
+      "loss": 0.6331,
+      "step": 1596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297327272116982,
+      "loss": 1.8203,
+      "step": 1597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973237036518914,
+      "loss": 1.9956,
+      "step": 1598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029732013282763285,
+      "loss": 0.7871,
+      "step": 1599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973165596442634,
+      "loss": 0.1454,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029731298410184037,
+      "loss": 0.3272,
+      "step": 1601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000297309406200421,
+      "loss": 0.3979,
+      "step": 1602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973058259400626,
+      "loss": 0.3691,
+      "step": 1603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002973022433208224,
+      "loss": 0.1926,
+      "step": 1604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972986583427579,
+      "loss": 0.9778,
+      "step": 1605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972950710059264,
+      "loss": 0.6068,
+      "step": 1606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972914813103854,
+      "loss": 0.3815,
+      "step": 1607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972878892561924,
+      "loss": 0.7853,
+      "step": 1608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029728429484340494,
+      "loss": 0.7832,
+      "step": 1609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972806980720805,
+      "loss": 0.3517,
+      "step": 1610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972770989422767,
+      "loss": 0.7354,
+      "step": 1611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029727349745405123,
+      "loss": 0.2823,
+      "step": 1612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029726989360746175,
+      "loss": 0.911,
+      "step": 1613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029726628740256585,
+      "loss": 0.629,
+      "step": 1614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972626788394215,
+      "loss": 0.288,
+      "step": 1615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972590679180862,
+      "loss": 0.3261,
+      "step": 1616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029725545463861805,
+      "loss": 1.3691,
+      "step": 1617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972518390010748,
+      "loss": 0.5014,
+      "step": 1618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972482210055143,
+      "loss": 0.3119,
+      "step": 1619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002972446006519945,
+      "loss": 0.4627,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029724097794057346,
+      "loss": 1.1782,
+      "step": 1621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002972373528713091,
+      "loss": 0.7433,
+      "step": 1622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002972337254442595,
+      "loss": 0.7079,
+      "step": 1623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029723009565948283,
+      "loss": 0.4262,
+      "step": 1624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002972264635170371,
+      "loss": 0.3033,
+      "step": 1625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002972228290169805,
+      "loss": 0.4089,
+      "step": 1626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029721919215937125,
+      "loss": 0.3573,
+      "step": 1627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002972155529442675,
+      "loss": 0.739,
+      "step": 1628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029721191137172773,
+      "loss": 0.5236,
+      "step": 1629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029720826744181005,
+      "loss": 0.6916,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029720462115457295,
+      "loss": 1.0784,
+      "step": 1631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029720097251007476,
+      "loss": 1.0125,
+      "step": 1632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971973215083739,
+      "loss": 0.5124,
+      "step": 1633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971936681495288,
+      "loss": 0.9443,
+      "step": 1634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971900124335981,
+      "loss": 0.6616,
+      "step": 1635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029718635436064023,
+      "loss": 0.3269,
+      "step": 1636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971826939307138,
+      "loss": 0.8858,
+      "step": 1637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029717903114387736,
+      "loss": 0.964,
+      "step": 1638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029717536600018963,
+      "loss": 0.3771,
+      "step": 1639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971716984997093,
+      "loss": 1.756,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971680286424951,
+      "loss": 4.594,
+      "step": 1641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971643564286058,
+      "loss": 0.1376,
+      "step": 1642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971606818581002,
+      "loss": 0.2599,
+      "step": 1643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971570049310371,
+      "loss": 1.1518,
+      "step": 1644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971533256474755,
+      "loss": 0.8439,
+      "step": 1645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029714964400747416,
+      "loss": 0.3741,
+      "step": 1646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971459600110921,
+      "loss": 0.4822,
+      "step": 1647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971422736583884,
+      "loss": 1.0724,
+      "step": 1648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000297138584949422,
+      "loss": 0.4972,
+      "step": 1649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029713489388425194,
+      "loss": 0.713,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971312004629374,
+      "loss": 2.7292,
+      "step": 1651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971275046855375,
+      "loss": 0.6436,
+      "step": 1652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971238065521114,
+      "loss": 0.2636,
+      "step": 1653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029712010606271834,
+      "loss": 0.7403,
+      "step": 1654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029711640321741756,
+      "loss": 0.7713,
+      "step": 1655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971126980162684,
+      "loss": 0.6116,
+      "step": 1656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002971089904593302,
+      "loss": 0.94,
+      "step": 1657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029710528054666227,
+      "loss": 0.919,
+      "step": 1658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029710156827832406,
+      "loss": 1.0245,
+      "step": 1659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029709785365437496,
+      "loss": 0.6195,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970941366748745,
+      "loss": 0.2564,
+      "step": 1661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029709041733988223,
+      "loss": 0.5806,
+      "step": 1662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970866956494577,
+      "loss": 0.3976,
+      "step": 1663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029708297160366043,
+      "loss": 0.1976,
+      "step": 1664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029707924520255017,
+      "loss": 0.9875,
+      "step": 1665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970755164461865,
+      "loss": 0.3887,
+      "step": 1666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970717853346292,
+      "loss": 0.5115,
+      "step": 1667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970680518679379,
+      "loss": 0.4304,
+      "step": 1668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970643160461725,
+      "loss": 0.4799,
+      "step": 1669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970605778693928,
+      "loss": 0.4836,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029705683733765865,
+      "loss": 0.3456,
+      "step": 1671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029705309445103,
+      "loss": 0.329,
+      "step": 1672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029704934920956667,
+      "loss": 0.4008,
+      "step": 1673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970456016133287,
+      "loss": 0.7288,
+      "step": 1674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029704185166237615,
+      "loss": 0.4835,
+      "step": 1675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000297038099356769,
+      "loss": 0.6388,
+      "step": 1676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029703434469656735,
+      "loss": 0.6555,
+      "step": 1677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029703058768183136,
+      "loss": 0.8142,
+      "step": 1678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029702682831262117,
+      "loss": 0.2589,
+      "step": 1679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029702306658899695,
+      "loss": 0.3458,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029701930251101904,
+      "loss": 0.8724,
+      "step": 1681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970155360787476,
+      "loss": 0.6752,
+      "step": 1682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000297011767292243,
+      "loss": 0.158,
+      "step": 1683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029700799615156554,
+      "loss": 0.6774,
+      "step": 1684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970042226567757,
+      "loss": 0.1577,
+      "step": 1685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002970004468079338,
+      "loss": 1.2669,
+      "step": 1686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029699666860510043,
+      "loss": 0.4724,
+      "step": 1687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000296992888048336,
+      "loss": 0.5318,
+      "step": 1688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029698910513770106,
+      "loss": 0.5996,
+      "step": 1689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029698531987325624,
+      "loss": 0.5217,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029698153225506205,
+      "loss": 0.3173,
+      "step": 1691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029697774228317923,
+      "loss": 0.3473,
+      "step": 1692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969739499576685,
+      "loss": 2.8165,
+      "step": 1693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969701552785905,
+      "loss": 0.8546,
+      "step": 1694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000296966358246006,
+      "loss": 0.6257,
+      "step": 1695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969625588599759,
+      "loss": 0.7742,
+      "step": 1696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029695875712056096,
+      "loss": 0.2028,
+      "step": 1697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029695495302782207,
+      "loss": 0.0211,
+      "step": 1698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029695114658182014,
+      "loss": 1.6398,
+      "step": 1699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029694733778261616,
+      "loss": 0.8282,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969435266302711,
+      "loss": 0.7786,
+      "step": 1701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000296939713124846,
+      "loss": 0.8961,
+      "step": 1702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029693589726640193,
+      "loss": 0.1067,
+      "step": 1703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029693207905499993,
+      "loss": 0.6326,
+      "step": 1704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029692825849070124,
+      "loss": 0.5676,
+      "step": 1705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000296924435573567,
+      "loss": 0.9614,
+      "step": 1706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969206103036584,
+      "loss": 0.3641,
+      "step": 1707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969167826810368,
+      "loss": 0.7374,
+      "step": 1708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969129527057633,
+      "loss": 0.2661,
+      "step": 1709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969091203778994,
+      "loss": 0.4871,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969052856975065,
+      "loss": 0.3672,
+      "step": 1711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002969014486646458,
+      "loss": 0.6187,
+      "step": 1712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968976092793789,
+      "loss": 0.2748,
+      "step": 1713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968937675417673,
+      "loss": 2.0011,
+      "step": 1714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968899234518724,
+      "loss": 0.7257,
+      "step": 1715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968860770097559,
+      "loss": 0.0846,
+      "step": 1716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968822282154793,
+      "loss": 1.7246,
+      "step": 1717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968783770691042,
+      "loss": 0.0947,
+      "step": 1718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029687452357069237,
+      "loss": 0.216,
+      "step": 1719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968706677203055,
+      "loss": 0.3807,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029686680951800527,
+      "loss": 0.3842,
+      "step": 1721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968629489638535,
+      "loss": 0.1579,
+      "step": 1722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029685908605791206,
+      "loss": 0.9494,
+      "step": 1723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968552208002427,
+      "loss": 0.444,
+      "step": 1724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968513531909074,
+      "loss": 0.7181,
+      "step": 1725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029684748322996807,
+      "loss": 0.7961,
+      "step": 1726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029684361091748666,
+      "loss": 3.5834,
+      "step": 1727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029683973625352524,
+      "loss": 0.4605,
+      "step": 1728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968358592381458,
+      "loss": 0.4479,
+      "step": 1729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029683197987141047,
+      "loss": 1.0883,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968280981533813,
+      "loss": 0.3379,
+      "step": 1731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968242140841205,
+      "loss": 1.1214,
+      "step": 1732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968203276636903,
+      "loss": 3.132,
+      "step": 1733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002968164388921529,
+      "loss": 0.639,
+      "step": 1734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029681254776957056,
+      "loss": 2.1979,
+      "step": 1735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029680865429600564,
+      "loss": 0.1169,
+      "step": 1736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029680475847152036,
+      "loss": 0.5116,
+      "step": 1737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029680086029617724,
+      "loss": 0.627,
+      "step": 1738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029679695977003867,
+      "loss": 0.4306,
+      "step": 1739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967930568931671,
+      "loss": 0.5763,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029678915166562504,
+      "loss": 0.2895,
+      "step": 1741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000296785244087475,
+      "loss": 0.4716,
+      "step": 1742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029678133415877956,
+      "loss": 0.4339,
+      "step": 1743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029677742187960137,
+      "loss": 0.4464,
+      "step": 1744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029677350725000303,
+      "loss": 0.7924,
+      "step": 1745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029676959027004725,
+      "loss": 0.2881,
+      "step": 1746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029676567093979677,
+      "loss": 0.6795,
+      "step": 1747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967617492593143,
+      "loss": 1.5929,
+      "step": 1748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967578252286627,
+      "loss": 0.8956,
+      "step": 1749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029675389884790477,
+      "loss": 0.02,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029674997011710336,
+      "loss": 0.5965,
+      "step": 1751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967460390363215,
+      "loss": 1.0091,
+      "step": 1752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029674210560562197,
+      "loss": 0.596,
+      "step": 1753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967381698250679,
+      "loss": 0.4292,
+      "step": 1754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967342316947223,
+      "loss": 0.574,
+      "step": 1755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029673029121464814,
+      "loss": 0.3737,
+      "step": 1756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029672634838490863,
+      "loss": 0.1366,
+      "step": 1757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029672240320556683,
+      "loss": 0.7354,
+      "step": 1758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967184556766859,
+      "loss": 0.7025,
+      "step": 1759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967145057983291,
+      "loss": 0.998,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967105535705597,
+      "loss": 0.6016,
+      "step": 1761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029670659899344104,
+      "loss": 0.2132,
+      "step": 1762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002967026420670363,
+      "loss": 0.5612,
+      "step": 1763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966986827914089,
+      "loss": 0.2157,
+      "step": 1764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029669472116662226,
+      "loss": 1.6954,
+      "step": 1765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029669075719273983,
+      "loss": 0.5412,
+      "step": 1766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029668679086982514,
+      "loss": 1.0481,
+      "step": 1767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029668282219794156,
+      "loss": 0.1535,
+      "step": 1768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966788511771528,
+      "loss": 0.297,
+      "step": 1769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966748778075223,
+      "loss": 0.495,
+      "step": 1770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029667090208911377,
+      "loss": 0.4673,
+      "step": 1771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966669240219909,
+      "loss": 0.3394,
+      "step": 1772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966629436062173,
+      "loss": 0.2136,
+      "step": 1773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966589608418568,
+      "loss": 0.8448,
+      "step": 1774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029665497572897313,
+      "loss": 0.4301,
+      "step": 1775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029665098826763014,
+      "loss": 1.6576,
+      "step": 1776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966469984578917,
+      "loss": 0.4021,
+      "step": 1777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029664300629982164,
+      "loss": 0.3263,
+      "step": 1778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966390117934839,
+      "loss": 0.2508,
+      "step": 1779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029663501493894245,
+      "loss": 0.8907,
+      "step": 1780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966310157362613,
+      "loss": 0.0378,
+      "step": 1781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029662701418550454,
+      "loss": 0.5724,
+      "step": 1782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002966230102867362,
+      "loss": 0.6945,
+      "step": 1783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029661900404002035,
+      "loss": 2.6513,
+      "step": 1784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029661499544542125,
+      "loss": 0.2792,
+      "step": 1785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000296610984503003,
+      "loss": 0.3918,
+      "step": 1786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029660697121282987,
+      "loss": 1.9923,
+      "step": 1787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029660295557496614,
+      "loss": 0.0208,
+      "step": 1788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965989375894761,
+      "loss": 0.0119,
+      "step": 1789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029659491725642404,
+      "loss": 0.3145,
+      "step": 1790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965908945758745,
+      "loss": 0.0689,
+      "step": 1791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029658686954789164,
+      "loss": 0.5381,
+      "step": 1792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029658284217254013,
+      "loss": 0.4148,
+      "step": 1793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965788124498844,
+      "loss": 0.6205,
+      "step": 1794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029657478037998895,
+      "loss": 0.2755,
+      "step": 1795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965707459629184,
+      "loss": 0.9325,
+      "step": 1796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029656670919873734,
+      "loss": 0.3682,
+      "step": 1797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965626700875103,
+      "loss": 2.6935,
+      "step": 1798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029655862862930214,
+      "loss": 0.7072,
+      "step": 1799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965545848241775,
+      "loss": 1.4559,
+      "step": 1800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965505386722011,
+      "loss": 0.6126,
+      "step": 1801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965464901734378,
+      "loss": 0.2907,
+      "step": 1802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965424393279524,
+      "loss": 0.637,
+      "step": 1803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029653838613580974,
+      "loss": 0.0733,
+      "step": 1804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965343305970747,
+      "loss": 0.4931,
+      "step": 1805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029653027271181237,
+      "loss": 0.2105,
+      "step": 1806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029652621248008753,
+      "loss": 0.5039,
+      "step": 1807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029652214990196537,
+      "loss": 0.6053,
+      "step": 1808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965180849775109,
+      "loss": 1.0178,
+      "step": 1809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029651401770678916,
+      "loss": 0.745,
+      "step": 1810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965099480898653,
+      "loss": 0.0141,
+      "step": 1811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965058761268045,
+      "loss": 1.5108,
+      "step": 1812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029650180181767195,
+      "loss": 0.7554,
+      "step": 1813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029649772516253294,
+      "loss": 0.2272,
+      "step": 1814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029649364616145274,
+      "loss": 0.9214,
+      "step": 1815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029648956481449665,
+      "loss": 0.7316,
+      "step": 1816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029648548112173,
+      "loss": 0.469,
+      "step": 1817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964813950832182,
+      "loss": 0.1973,
+      "step": 1818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029647730669902676,
+      "loss": 0.5623,
+      "step": 1819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964732159692211,
+      "loss": 0.7657,
+      "step": 1820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029646912289386663,
+      "loss": 0.1591,
+      "step": 1821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000296465027473029,
+      "loss": 0.6613,
+      "step": 1822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964609297067738,
+      "loss": 0.5821,
+      "step": 1823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964568295951666,
+      "loss": 0.0904,
+      "step": 1824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964527271382731,
+      "loss": 0.4066,
+      "step": 1825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964486223361589,
+      "loss": 0.2005,
+      "step": 1826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029644451518888985,
+      "loss": 0.9592,
+      "step": 1827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964404056965317,
+      "loss": 0.3146,
+      "step": 1828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029643629385915014,
+      "loss": 0.7746,
+      "step": 1829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964321796768112,
+      "loss": 1.378,
+      "step": 1830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029642806314958056,
+      "loss": 0.2775,
+      "step": 1831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029642394427752437,
+      "loss": 1.0065,
+      "step": 1832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029641982306070837,
+      "loss": 0.3598,
+      "step": 1833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964156994991987,
+      "loss": 1.6378,
+      "step": 1834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029641157359306135,
+      "loss": 0.1618,
+      "step": 1835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964074453423623,
+      "loss": 0.0622,
+      "step": 1836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002964033147471678,
+      "loss": 0.5402,
+      "step": 1837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029639918180754393,
+      "loss": 0.5977,
+      "step": 1838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963950465235569,
+      "loss": 0.2563,
+      "step": 1839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963909088952729,
+      "loss": 0.3424,
+      "step": 1840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963867689227582,
+      "loss": 0.9908,
+      "step": 1841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029638262660607907,
+      "loss": 0.784,
+      "step": 1842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029637848194530184,
+      "loss": 0.2409,
+      "step": 1843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029637433494049295,
+      "loss": 0.6625,
+      "step": 1844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029637018559171876,
+      "loss": 0.2621,
+      "step": 1845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963660338990457,
+      "loss": 0.7249,
+      "step": 1846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963618798625403,
+      "loss": 0.6126,
+      "step": 1847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029635772348226896,
+      "loss": 0.4385,
+      "step": 1848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029635356475829846,
+      "loss": 0.0089,
+      "step": 1849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963494036906952,
+      "loss": 0.6862,
+      "step": 1850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029634524027952587,
+      "loss": 0.3245,
+      "step": 1851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963410745248572,
+      "loss": 0.8761,
+      "step": 1852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029633690642675583,
+      "loss": 0.3714,
+      "step": 1853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029633273598528846,
+      "loss": 0.5311,
+      "step": 1854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029632856320052197,
+      "loss": 0.3685,
+      "step": 1855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029632438807252316,
+      "loss": 0.6596,
+      "step": 1856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963202106013589,
+      "loss": 0.3706,
+      "step": 1857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029631603078709604,
+      "loss": 0.2849,
+      "step": 1858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002963118486298015,
+      "loss": 0.6756,
+      "step": 1859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029630766412954237,
+      "loss": 2.4017,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029630347728638553,
+      "loss": 0.682,
+      "step": 1861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002962992881003981,
+      "loss": 0.7157,
+      "step": 1862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029629509657164713,
+      "loss": 0.108,
+      "step": 1863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029629090270019974,
+      "loss": 0.8233,
+      "step": 1864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002962867064861231,
+      "loss": 0.8316,
+      "step": 1865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029628250792948437,
+      "loss": 0.3882,
+      "step": 1866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002962783070303508,
+      "loss": 3.0997,
+      "step": 1867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029627410378878975,
+      "loss": 0.495,
+      "step": 1868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002962698982048684,
+      "loss": 0.7761,
+      "step": 1869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002962656902786542,
+      "loss": 0.1809,
+      "step": 1870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029626148001021444,
+      "loss": 0.5638,
+      "step": 1871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002962572673996166,
+      "loss": 0.3015,
+      "step": 1872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029625305244692805,
+      "loss": 0.564,
+      "step": 1873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029624883515221647,
+      "loss": 0.6404,
+      "step": 1874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029624461551554926,
+      "loss": 0.6773,
+      "step": 1875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000296240393536994,
+      "loss": 0.7129,
+      "step": 1876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029623616921661827,
+      "loss": 0.4507,
+      "step": 1877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002962319425544898,
+      "loss": 0.6422,
+      "step": 1878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029622771355067626,
+      "loss": 0.9085,
+      "step": 1879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002962234822052453,
+      "loss": 0.4543,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029621924851826474,
+      "loss": 0.0061,
+      "step": 1881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029621501248980233,
+      "loss": 0.1393,
+      "step": 1882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000296210774119926,
+      "loss": 0.8903,
+      "step": 1883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002962065334087035,
+      "loss": 0.7297,
+      "step": 1884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002962022903562028,
+      "loss": 0.4434,
+      "step": 1885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029619804496249186,
+      "loss": 1.4093,
+      "step": 1886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029619379722763866,
+      "loss": 0.8824,
+      "step": 1887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002961895471517112,
+      "loss": 0.0887,
+      "step": 1888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029618529473477746,
+      "loss": 1.1378,
+      "step": 1889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002961810399769057,
+      "loss": 1.5392,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002961767828781639,
+      "loss": 1.2366,
+      "step": 1891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002961725234386204,
+      "loss": 0.1585,
+      "step": 1892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029616826165834324,
+      "loss": 0.267,
+      "step": 1893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002961639975374008,
+      "loss": 0.6495,
+      "step": 1894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029615973107586125,
+      "loss": 0.6995,
+      "step": 1895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029615546227379296,
+      "loss": 0.629,
+      "step": 1896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002961511911312643,
+      "loss": 0.3995,
+      "step": 1897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029614691764834366,
+      "loss": 0.5558,
+      "step": 1898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029614264182509947,
+      "loss": 0.1971,
+      "step": 1899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029613836366160017,
+      "loss": 0.8693,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029613408315791435,
+      "loss": 0.2653,
+      "step": 1901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029612980031411046,
+      "loss": 0.797,
+      "step": 1902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029612551513025714,
+      "loss": 0.6717,
+      "step": 1903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029612122760642307,
+      "loss": 0.7522,
+      "step": 1904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029611693774267677,
+      "loss": 0.6832,
+      "step": 1905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000296112645539087,
+      "loss": 0.4061,
+      "step": 1906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029610835099572245,
+      "loss": 0.2256,
+      "step": 1907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029610405411265207,
+      "loss": 0.6422,
+      "step": 1908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960997548899444,
+      "loss": 0.4659,
+      "step": 1909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960954533276685,
+      "loss": 0.4552,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029609114942589316,
+      "loss": 1.0883,
+      "step": 1911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960868431846873,
+      "loss": 0.3115,
+      "step": 1912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960825346041199,
+      "loss": 0.4214,
+      "step": 1913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029607822368425994,
+      "loss": 1.0527,
+      "step": 1914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960739104251765,
+      "loss": 0.7575,
+      "step": 1915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029606959482693856,
+      "loss": 0.5666,
+      "step": 1916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029606527688961527,
+      "loss": 0.1651,
+      "step": 1917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960609566132758,
+      "loss": 0.3545,
+      "step": 1918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960566339979893,
+      "loss": 3.0564,
+      "step": 1919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960523090438251,
+      "loss": 0.3039,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960479817508523,
+      "loss": 0.6919,
+      "step": 1921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960436521191402,
+      "loss": 0.5549,
+      "step": 1922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029603932014875826,
+      "loss": 0.7224,
+      "step": 1923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960349858397758,
+      "loss": 0.1662,
+      "step": 1924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029603064919226215,
+      "loss": 0.5402,
+      "step": 1925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029602631020628683,
+      "loss": 0.7327,
+      "step": 1926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960219688819193,
+      "loss": 1.1573,
+      "step": 1927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002960176252192291,
+      "loss": 1.5393,
+      "step": 1928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029601327921828583,
+      "loss": 0.839,
+      "step": 1929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029600893087915897,
+      "loss": 3.5222,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029600458020191824,
+      "loss": 0.1391,
+      "step": 1931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029600022718663326,
+      "loss": 0.6844,
+      "step": 1932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029599587183337374,
+      "loss": 0.7908,
+      "step": 1933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029599151414220944,
+      "loss": 0.4733,
+      "step": 1934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959871541132102,
+      "loss": 0.1384,
+      "step": 1935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959827917464457,
+      "loss": 0.4375,
+      "step": 1936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295978427041986,
+      "loss": 0.4951,
+      "step": 1937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959740599999008,
+      "loss": 0.5396,
+      "step": 1938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959696906202601,
+      "loss": 1.1492,
+      "step": 1939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029596531890313386,
+      "loss": 0.6476,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959609448485921,
+      "loss": 0.5582,
+      "step": 1941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959565684567049,
+      "loss": 0.5573,
+      "step": 1942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959521897275423,
+      "loss": 0.2748,
+      "step": 1943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029594780866117444,
+      "loss": 0.4142,
+      "step": 1944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029594342525767137,
+      "loss": 0.3269,
+      "step": 1945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029593903951710346,
+      "loss": 1.1907,
+      "step": 1946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029593465143954085,
+      "loss": 0.1428,
+      "step": 1947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959302610250538,
+      "loss": 0.5581,
+      "step": 1948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029592586827371264,
+      "loss": 0.3081,
+      "step": 1949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959214731855877,
+      "loss": 1.2754,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959170757607493,
+      "loss": 0.6728,
+      "step": 1951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959126759992681,
+      "loss": 0.5899,
+      "step": 1952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959082739012142,
+      "loss": 0.6189,
+      "step": 1953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029590386946665837,
+      "loss": 0.6574,
+      "step": 1954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295899462695671,
+      "loss": 0.3284,
+      "step": 1955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958950535883227,
+      "loss": 1.4612,
+      "step": 1956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029589064214468404,
+      "loss": 0.5912,
+      "step": 1957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958862283648258,
+      "loss": 0.2064,
+      "step": 1958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029588181224881845,
+      "loss": 0.7779,
+      "step": 1959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029587739379673285,
+      "loss": 0.5497,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958729730086398,
+      "loss": 0.5168,
+      "step": 1961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029586854988460987,
+      "loss": 0.8094,
+      "step": 1962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029586412442471414,
+      "loss": 0.8764,
+      "step": 1963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958596966290233,
+      "loss": 0.3521,
+      "step": 1964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029585526649760836,
+      "loss": 0.6901,
+      "step": 1965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958508340305402,
+      "loss": 0.3839,
+      "step": 1966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958463992278899,
+      "loss": 1.385,
+      "step": 1967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029584196208972833,
+      "loss": 0.2363,
+      "step": 1968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958375226161266,
+      "loss": 0.5098,
+      "step": 1969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029583308080715585,
+      "loss": 1.4623,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029582863666288717,
+      "loss": 0.2541,
+      "step": 1971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029582419018339175,
+      "loss": 0.5022,
+      "step": 1972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029581974136874077,
+      "loss": 0.4048,
+      "step": 1973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958152902190054,
+      "loss": 0.6743,
+      "step": 1974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002958108367342571,
+      "loss": 0.4714,
+      "step": 1975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029580638091456705,
+      "loss": 1.0729,
+      "step": 1976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029580192276000664,
+      "loss": 0.3908,
+      "step": 1977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029579746227064726,
+      "loss": 0.5313,
+      "step": 1978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957929994465603,
+      "loss": 0.029,
+      "step": 1979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957885342878173,
+      "loss": 0.823,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957840667944897,
+      "loss": 1.6146,
+      "step": 1981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295779596966649,
+      "loss": 0.3431,
+      "step": 1982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957751248043669,
+      "loss": 0.821,
+      "step": 1983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295770650307715,
+      "loss": 0.5753,
+      "step": 1984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029576617347676484,
+      "loss": 0.0487,
+      "step": 1985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957616943115882,
+      "loss": 0.36,
+      "step": 1986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029575721281225676,
+      "loss": 1.5451,
+      "step": 1987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029575272897884233,
+      "loss": 0.6688,
+      "step": 1988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029574824281141667,
+      "loss": 0.827,
+      "step": 1989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957437543100517,
+      "loss": 0.2245,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957392634748191,
+      "loss": 0.4005,
+      "step": 1991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957347703057911,
+      "loss": 0.623,
+      "step": 1992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029573027480303934,
+      "loss": 0.3453,
+      "step": 1993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295725776966636,
+      "loss": 0.554,
+      "step": 1994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029572127679665296,
+      "loss": 0.4085,
+      "step": 1995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957167742931625,
+      "loss": 1.4576,
+      "step": 1996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029571226945623646,
+      "loss": 2.4577,
+      "step": 1997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957077622859472,
+      "loss": 0.3314,
+      "step": 1998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002957032527823668,
+      "loss": 1.243,
+      "step": 1999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029569874094556743,
+      "loss": 0.6017,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029569422677562144,
+      "loss": 0.2466,
+      "step": 2001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029568971027260096,
+      "loss": 0.3344,
+      "step": 2002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956851914365785,
+      "loss": 0.4456,
+      "step": 2003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956806702676264,
+      "loss": 2.3467,
+      "step": 2004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956761467658169,
+      "loss": 0.4629,
+      "step": 2005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029567162093122257,
+      "loss": 1.2373,
+      "step": 2006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029566709276391585,
+      "loss": 0.495,
+      "step": 2007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956625622639693,
+      "loss": 0.7967,
+      "step": 2008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956580294314554,
+      "loss": 0.3093,
+      "step": 2009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956534942664467,
+      "loss": 0.8315,
+      "step": 2010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029564895676901593,
+      "loss": 0.7545,
+      "step": 2011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956444169392357,
+      "loss": 1.1071,
+      "step": 2012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029563987477717875,
+      "loss": 0.1938,
+      "step": 2013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029563533028291775,
+      "loss": 0.7323,
+      "step": 2014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956307834565255,
+      "loss": 0.7768,
+      "step": 2015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956262342980748,
+      "loss": 1.0363,
+      "step": 2016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029562168280763854,
+      "loss": 0.2956,
+      "step": 2017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956171289852896,
+      "loss": 0.4789,
+      "step": 2018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956125728311008,
+      "loss": 0.3713,
+      "step": 2019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956080143451453,
+      "loss": 0.2789,
+      "step": 2020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002956034535274959,
+      "loss": 0.4783,
+      "step": 2021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029559889037822574,
+      "loss": 0.4352,
+      "step": 2022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002955943248974078,
+      "loss": 0.3339,
+      "step": 2023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002955897570851153,
+      "loss": 0.4524,
+      "step": 2024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029558518694142134,
+      "loss": 0.0254,
+      "step": 2025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002955806144663991,
+      "loss": 0.5248,
+      "step": 2026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029557603966012184,
+      "loss": 0.8193,
+      "step": 2027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002955714625226627,
+      "loss": 0.4201,
+      "step": 2028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029556688305409516,
+      "loss": 0.9026,
+      "step": 2029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029556230125449235,
+      "loss": 0.566,
+      "step": 2030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029555771712392784,
+      "loss": 0.4094,
+      "step": 2031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029555313066247487,
+      "loss": 0.7143,
+      "step": 2032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295548541870207,
+      "loss": 0.5933,
+      "step": 2033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029554395074719766,
+      "loss": 0.5781,
+      "step": 2034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029553935729352034,
+      "loss": 0.5477,
+      "step": 2035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002955347615092487,
+      "loss": 3.0378,
+      "step": 2036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029553016339445623,
+      "loss": 1.1007,
+      "step": 2037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029552556294921663,
+      "loss": 0.0105,
+      "step": 2038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002955209601736035,
+      "loss": 1.8098,
+      "step": 2039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029551635506769064,
+      "loss": 0.1363,
+      "step": 2040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029551174763155176,
+      "loss": 0.0137,
+      "step": 2041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002955071378652606,
+      "loss": 0.0076,
+      "step": 2042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029550252576889094,
+      "loss": 0.2702,
+      "step": 2043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954979113425168,
+      "loss": 1.4455,
+      "step": 2044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029549329458621184,
+      "loss": 0.9412,
+      "step": 2045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954886755000502,
+      "loss": 0.3426,
+      "step": 2046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954840540841058,
+      "loss": 0.957,
+      "step": 2047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954794303384526,
+      "loss": 0.3193,
+      "step": 2048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029547480426316464,
+      "loss": 1.2109,
+      "step": 2049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029547017585831596,
+      "loss": 0.7753,
+      "step": 2050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029546554512398086,
+      "loss": 0.4025,
+      "step": 2051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029546091206023327,
+      "loss": 0.7328,
+      "step": 2052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954562766671475,
+      "loss": 1.8361,
+      "step": 2053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954516389447978,
+      "loss": 0.1825,
+      "step": 2054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029544699889325835,
+      "loss": 0.9426,
+      "step": 2055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029544235651260355,
+      "loss": 1.3362,
+      "step": 2056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954377118029076,
+      "loss": 0.7484,
+      "step": 2057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295433064764245,
+      "loss": 1.1604,
+      "step": 2058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029542841539669017,
+      "loss": 2.5499,
+      "step": 2059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954237637003175,
+      "loss": 0.9694,
+      "step": 2060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029541910967520154,
+      "loss": 0.864,
+      "step": 2061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029541445332141673,
+      "loss": 0.4775,
+      "step": 2062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029540979463903773,
+      "loss": 1.1585,
+      "step": 2063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954051336281391,
+      "loss": 0.4839,
+      "step": 2064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002954004702887954,
+      "loss": 0.4499,
+      "step": 2065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029539580462108146,
+      "loss": 0.2593,
+      "step": 2066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953911366250719,
+      "loss": 0.7209,
+      "step": 2067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953864663008415,
+      "loss": 0.4966,
+      "step": 2068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295381793648465,
+      "loss": 0.4503,
+      "step": 2069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953771186680173,
+      "loss": 0.5065,
+      "step": 2070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953724413595732,
+      "loss": 0.8913,
+      "step": 2071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953677617232076,
+      "loss": 0.7994,
+      "step": 2072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953630797589955,
+      "loss": 0.1868,
+      "step": 2073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029535839546701185,
+      "loss": 0.7764,
+      "step": 2074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953537088473316,
+      "loss": 2.459,
+      "step": 2075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953490199000299,
+      "loss": 0.6697,
+      "step": 2076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029534432862518176,
+      "loss": 0.5883,
+      "step": 2077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029533963502286235,
+      "loss": 0.0612,
+      "step": 2078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029533493909314685,
+      "loss": 0.4673,
+      "step": 2079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029533024083611034,
+      "loss": 0.4739,
+      "step": 2080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953255402518281,
+      "loss": 0.5781,
+      "step": 2081
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029532083734037554,
+      "loss": 0.1359,
+      "step": 2082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953161321018278,
+      "loss": 0.3603,
+      "step": 2083
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953114245362603,
+      "loss": 0.1357,
+      "step": 2084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029530671464374843,
+      "loss": 0.1872,
+      "step": 2085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002953020024243676,
+      "loss": 0.7844,
+      "step": 2086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029529728787819326,
+      "loss": 0.5784,
+      "step": 2087
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029529257100530094,
+      "loss": 0.0629,
+      "step": 2088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029528785180576615,
+      "loss": 2.4616,
+      "step": 2089
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002952831302796644,
+      "loss": 2.2758,
+      "step": 2090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002952784064270714,
+      "loss": 0.3345,
+      "step": 2091
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029527368024806277,
+      "loss": 0.4713,
+      "step": 2092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002952689517427141,
+      "loss": 0.3064,
+      "step": 2093
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029526422091110126,
+      "loss": 0.4638,
+      "step": 2094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029525948775329996,
+      "loss": 0.0402,
+      "step": 2095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002952547522693859,
+      "loss": 0.3396,
+      "step": 2096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295250014459435,
+      "loss": 0.0052,
+      "step": 2097
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002952452743235231,
+      "loss": 0.5895,
+      "step": 2098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029524053186172615,
+      "loss": 1.0444,
+      "step": 2099
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029523578707412,
+      "loss": 3.7695,
+      "step": 2100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029523103996078074,
+      "loss": 0.0255,
+      "step": 2101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029522629052178436,
+      "loss": 0.5484,
+      "step": 2102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029522153875720684,
+      "loss": 0.9042,
+      "step": 2103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002952167846671243,
+      "loss": 0.2755,
+      "step": 2104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029521202825161295,
+      "loss": 0.007,
+      "step": 2105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029520726951074886,
+      "loss": 1.1676,
+      "step": 2106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002952025084446083,
+      "loss": 0.6597,
+      "step": 2107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002951977450532675,
+      "loss": 0.6851,
+      "step": 2108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002951929793368027,
+      "loss": 0.3638,
+      "step": 2109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029518821129529023,
+      "loss": 0.8805,
+      "step": 2110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029518344092880646,
+      "loss": 1.0173,
+      "step": 2111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002951786682374278,
+      "loss": 2.3068,
+      "step": 2112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002951738932212306,
+      "loss": 0.384,
+      "step": 2113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002951691158802914,
+      "loss": 0.4121,
+      "step": 2114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002951643362146867,
+      "loss": 0.4901,
+      "step": 2115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295159554224493,
+      "loss": 0.737,
+      "step": 2116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002951547699097869,
+      "loss": 0.4147,
+      "step": 2117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000295149983270645,
+      "loss": 0.3156,
+      "step": 2118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029514519430714396,
+      "loss": 0.3517,
+      "step": 2119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029514040301936045,
+      "loss": 1.1974,
+      "step": 2120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002951356094073712,
+      "loss": 0.4051,
+      "step": 2121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000295130813471253,
+      "loss": 0.7974,
+      "step": 2122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029512601521108266,
+      "loss": 0.2159,
+      "step": 2123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029512121462693696,
+      "loss": 0.2538,
+      "step": 2124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002951164117188928,
+      "loss": 0.1834,
+      "step": 2125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029511160648702713,
+      "loss": 0.2288,
+      "step": 2126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029510679893141685,
+      "loss": 1.6398,
+      "step": 2127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029510198905213894,
+      "loss": 0.2693,
+      "step": 2128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029509717684927044,
+      "loss": 0.3189,
+      "step": 2129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029509236232288843,
+      "loss": 0.3498,
+      "step": 2130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029508754547307,
+      "loss": 0.5925,
+      "step": 2131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950827262998923,
+      "loss": 0.6304,
+      "step": 2132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950779048034324,
+      "loss": 0.6246,
+      "step": 2133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950730809837676,
+      "loss": 0.9472,
+      "step": 2134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950682548409752,
+      "loss": 0.4369,
+      "step": 2135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950634263751324,
+      "loss": 0.3263,
+      "step": 2136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950585955863165,
+      "loss": 1.8192,
+      "step": 2137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950537624746049,
+      "loss": 0.168,
+      "step": 2138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029504892704007503,
+      "loss": 1.1371,
+      "step": 2139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029504408928280424,
+      "loss": 0.6745,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029503924920287,
+      "loss": 0.7927,
+      "step": 2141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029503440680035,
+      "loss": 0.4361,
+      "step": 2142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950295620753215,
+      "loss": 0.3524,
+      "step": 2143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950247150278623,
+      "loss": 0.3842,
+      "step": 2144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029501986565804984,
+      "loss": 0.4382,
+      "step": 2145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029501501396596194,
+      "loss": 3.7521,
+      "step": 2146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029501015995167626,
+      "loss": 0.4647,
+      "step": 2147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950053036152705,
+      "loss": 1.1955,
+      "step": 2148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002950004449568224,
+      "loss": 0.4406,
+      "step": 2149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029499558397640975,
+      "loss": 0.7835,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029499072067411047,
+      "loss": 0.4951,
+      "step": 2151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029498585505000237,
+      "loss": 0.4499,
+      "step": 2152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029498098710416346,
+      "loss": 0.3909,
+      "step": 2153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949761168366716,
+      "loss": 0.3976,
+      "step": 2154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949712442476047,
+      "loss": 0.2121,
+      "step": 2155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029496636933704106,
+      "loss": 0.2361,
+      "step": 2156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949614921050585,
+      "loss": 0.3713,
+      "step": 2157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949566125517352,
+      "loss": 1.9898,
+      "step": 2158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949517306771493,
+      "loss": 0.5532,
+      "step": 2159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294946846481379,
+      "loss": 0.4467,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949419599645025,
+      "loss": 0.7631,
+      "step": 2161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294937071126598,
+      "loss": 0.4326,
+      "step": 2162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949321799677438,
+      "loss": 0.0363,
+      "step": 2163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029492728648801826,
+      "loss": 0.6426,
+      "step": 2164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029492239068749985,
+      "loss": 0.2398,
+      "step": 2165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949174925662667,
+      "loss": 0.0255,
+      "step": 2166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029491259212439746,
+      "loss": 0.5485,
+      "step": 2167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029490768936197057,
+      "loss": 0.785,
+      "step": 2168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002949027842790645,
+      "loss": 0.0428,
+      "step": 2169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029489787687575773,
+      "loss": 0.5548,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294892967152129,
+      "loss": 0.4269,
+      "step": 2171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029488805510825687,
+      "loss": 0.6864,
+      "step": 2172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948831407442199,
+      "loss": 0.462,
+      "step": 2173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948782240600969,
+      "loss": 0.1321,
+      "step": 2174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948733050559667,
+      "loss": 0.0037,
+      "step": 2175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948683837319078,
+      "loss": 0.8147,
+      "step": 2176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029486346008799914,
+      "loss": 0.7243,
+      "step": 2177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029485853412431964,
+      "loss": 0.7363,
+      "step": 2178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948536058409481,
+      "loss": 1.7016,
+      "step": 2179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948486752379635,
+      "loss": 0.5049,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029484374231544466,
+      "loss": 0.9925,
+      "step": 2181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029483880707347073,
+      "loss": 0.3502,
+      "step": 2182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029483386951212066,
+      "loss": 0.3021,
+      "step": 2183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948289296314736,
+      "loss": 0.8339,
+      "step": 2184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029482398743160853,
+      "loss": 1.1447,
+      "step": 2185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029481904291260466,
+      "loss": 0.3248,
+      "step": 2186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029481409607454113,
+      "loss": 1.3021,
+      "step": 2187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948091469174972,
+      "loss": 0.3473,
+      "step": 2188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002948041954415521,
+      "loss": 1.0422,
+      "step": 2189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947992416467852,
+      "loss": 0.0634,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947942855332757,
+      "loss": 0.1102,
+      "step": 2191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029478932710110304,
+      "loss": 0.8585,
+      "step": 2192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947843663503466,
+      "loss": 0.9264,
+      "step": 2193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947794032810858,
+      "loss": 0.3471,
+      "step": 2194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947744378934002,
+      "loss": 0.4879,
+      "step": 2195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947694701873692,
+      "loss": 0.3766,
+      "step": 2196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947645001630724,
+      "loss": 0.5377,
+      "step": 2197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029475952782058944,
+      "loss": 0.5415,
+      "step": 2198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029475455315999984,
+      "loss": 0.3992,
+      "step": 2199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029474957618138333,
+      "loss": 0.5332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947445968848196,
+      "loss": 0.3479,
+      "step": 2201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029473961527038843,
+      "loss": 2.4514,
+      "step": 2202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947346313381695,
+      "loss": 0.2617,
+      "step": 2203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029472964508824273,
+      "loss": 0.3353,
+      "step": 2204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947246565206878,
+      "loss": 0.3675,
+      "step": 2205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947196656355847,
+      "loss": 0.4993,
+      "step": 2206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029471467243301347,
+      "loss": 0.2476,
+      "step": 2207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947096769130538,
+      "loss": 0.5226,
+      "step": 2208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002947046790757859,
+      "loss": 1.8381,
+      "step": 2209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946996789212898,
+      "loss": 0.3639,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946946764496454,
+      "loss": 0.526,
+      "step": 2211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294689671660933,
+      "loss": 0.3114,
+      "step": 2212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029468466455523254,
+      "loss": 2.0346,
+      "step": 2213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946796551326244,
+      "loss": 0.3638,
+      "step": 2214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029467464339318875,
+      "loss": 0.4863,
+      "step": 2215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029466962933700575,
+      "loss": 1.3555,
+      "step": 2216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946646129641558,
+      "loss": 0.8401,
+      "step": 2217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946595942747192,
+      "loss": 0.7897,
+      "step": 2218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946545732687762,
+      "loss": 0.5361,
+      "step": 2219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946495499464074,
+      "loss": 0.0032,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029464452430769307,
+      "loss": 0.4871,
+      "step": 2221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029463949635271384,
+      "loss": 1.6663,
+      "step": 2222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946344660815501,
+      "loss": 0.4013,
+      "step": 2223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946294334942825,
+      "loss": 0.6075,
+      "step": 2224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029462439859099154,
+      "loss": 0.4179,
+      "step": 2225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029461936137175785,
+      "loss": 0.3479,
+      "step": 2226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946143218366622,
+      "loss": 0.2047,
+      "step": 2227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946092799857852,
+      "loss": 0.5578,
+      "step": 2228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946042358192076,
+      "loss": 0.6157,
+      "step": 2229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029459918933701025,
+      "loss": 0.3897,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945941405392738,
+      "loss": 0.879,
+      "step": 2231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945890894260792,
+      "loss": 0.8428,
+      "step": 2232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029458403599750735,
+      "loss": 0.429,
+      "step": 2233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945789802536392,
+      "loss": 0.3471,
+      "step": 2234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029457392219455565,
+      "loss": 0.5923,
+      "step": 2235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945688618203377,
+      "loss": 0.3055,
+      "step": 2236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945637991310664,
+      "loss": 0.4414,
+      "step": 2237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029455873412682275,
+      "loss": 0.8274,
+      "step": 2238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294553666807688,
+      "loss": 0.6636,
+      "step": 2239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029454859717374323,
+      "loss": 0.9697,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945435252250696,
+      "loss": 0.8747,
+      "step": 2241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945384509617483,
+      "loss": 0.2815,
+      "step": 2242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029453337438386064,
+      "loss": 0.8887,
+      "step": 2243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945282954914879,
+      "loss": 1.6843,
+      "step": 2244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029452321428471143,
+      "loss": 0.1885,
+      "step": 2245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945181307636126,
+      "loss": 0.8009,
+      "step": 2246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002945130449282728,
+      "loss": 0.631,
+      "step": 2247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029450795677877346,
+      "loss": 0.5507,
+      "step": 2248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029450286631519605,
+      "loss": 1.9204,
+      "step": 2249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944977735376221,
+      "loss": 0.3173,
+      "step": 2250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944926784461332,
+      "loss": 0.3022,
+      "step": 2251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944875810408109,
+      "loss": 0.4985,
+      "step": 2252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944824813217368,
+      "loss": 0.1715,
+      "step": 2253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029447737928899265,
+      "loss": 0.1513,
+      "step": 2254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029447227494266005,
+      "loss": 0.6207,
+      "step": 2255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029446716828282086,
+      "loss": 0.468,
+      "step": 2256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944620593095567,
+      "loss": 0.4318,
+      "step": 2257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944569480229495,
+      "loss": 0.3637,
+      "step": 2258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944518344230811,
+      "loss": 0.7435,
+      "step": 2259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029444671851003334,
+      "loss": 0.2783,
+      "step": 2260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944416002838882,
+      "loss": 2.6742,
+      "step": 2261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944364797447276,
+      "loss": 0.4801,
+      "step": 2262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944313568926335,
+      "loss": 0.4221,
+      "step": 2263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294426231727688,
+      "loss": 0.3593,
+      "step": 2264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944211042499732,
+      "loss": 0.3617,
+      "step": 2265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002944159744595711,
+      "loss": 0.4354,
+      "step": 2266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029441084235656395,
+      "loss": 0.4088,
+      "step": 2267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029440570794103384,
+      "loss": 1.0703,
+      "step": 2268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029440057121306307,
+      "loss": 0.7056,
+      "step": 2269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943954321727339,
+      "loss": 0.6515,
+      "step": 2270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029439029082012856,
+      "loss": 2.5967,
+      "step": 2271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943851471553294,
+      "loss": 0.5173,
+      "step": 2272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943800011784188,
+      "loss": 0.0723,
+      "step": 2273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029437485288947917,
+      "loss": 0.8957,
+      "step": 2274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029436970228859295,
+      "loss": 0.0129,
+      "step": 2275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029436454937584264,
+      "loss": 0.4418,
+      "step": 2276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943593941513107,
+      "loss": 0.5476,
+      "step": 2277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943542366150797,
+      "loss": 0.6769,
+      "step": 2278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943490767672323,
+      "loss": 0.0081,
+      "step": 2279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294343914607851,
+      "loss": 0.9168,
+      "step": 2280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029433875013701864,
+      "loss": 1.3694,
+      "step": 2281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943335833548177,
+      "loss": 2.4886,
+      "step": 2282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029432841426133114,
+      "loss": 0.7434,
+      "step": 2283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943232428566416,
+      "loss": 0.6971,
+      "step": 2284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943180691408319,
+      "loss": 0.4448,
+      "step": 2285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029431289311398497,
+      "loss": 0.668,
+      "step": 2286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029430771477618355,
+      "loss": 0.5011,
+      "step": 2287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002943025341275107,
+      "loss": 0.4292,
+      "step": 2288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029429735116804934,
+      "loss": 0.905,
+      "step": 2289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002942921658978825,
+      "loss": 0.5756,
+      "step": 2290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029428697831709313,
+      "loss": 0.0213,
+      "step": 2291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029428178842576437,
+      "loss": 0.9554,
+      "step": 2292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029427659622397926,
+      "loss": 0.5424,
+      "step": 2293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294271401711821,
+      "loss": 0.0483,
+      "step": 2294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002942662048893728,
+      "loss": 0.8414,
+      "step": 2295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002942610057567178,
+      "loss": 0.0238,
+      "step": 2296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029425580431393934,
+      "loss": 0.523,
+      "step": 2297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029425060056112066,
+      "loss": 1.3754,
+      "step": 2298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029424539449834507,
+      "loss": 0.7274,
+      "step": 2299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294240186125696,
+      "loss": 0.6601,
+      "step": 2300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002942349754432568,
+      "loss": 0.3217,
+      "step": 2301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029422976245111095,
+      "loss": 0.1551,
+      "step": 2302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002942245471493419,
+      "loss": 0.1491,
+      "step": 2303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002942193295380332,
+      "loss": 0.6983,
+      "step": 2304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002942141096172684,
+      "loss": 0.3163,
+      "step": 2305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294208887387131,
+      "loss": 0.5731,
+      "step": 2306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029420366284770477,
+      "loss": 0.6275,
+      "step": 2307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029419843599907323,
+      "loss": 1.1048,
+      "step": 2308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941932068413202,
+      "loss": 0.5871,
+      "step": 2309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029418797537452936,
+      "loss": 0.5181,
+      "step": 2310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941827415987845,
+      "loss": 0.2913,
+      "step": 2311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941775055141694,
+      "loss": 0.5501,
+      "step": 2312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941722671207679,
+      "loss": 0.9195,
+      "step": 2313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029416702641866393,
+      "loss": 2.8206,
+      "step": 2314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941617834079414,
+      "loss": 0.5744,
+      "step": 2315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941565380886843,
+      "loss": 0.1525,
+      "step": 2316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029415129046097654,
+      "loss": 0.9401,
+      "step": 2317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029414604052490224,
+      "loss": 0.0307,
+      "step": 2318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029414078828054547,
+      "loss": 1.6692,
+      "step": 2319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029413553372799025,
+      "loss": 1.3795,
+      "step": 2320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029413027686732077,
+      "loss": 1.32,
+      "step": 2321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941250176986212,
+      "loss": 0.0076,
+      "step": 2322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029411975622197585,
+      "loss": 0.7726,
+      "step": 2323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029411449243746884,
+      "loss": 0.5174,
+      "step": 2324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029410922634518454,
+      "loss": 0.6221,
+      "step": 2325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002941039579452072,
+      "loss": 0.0207,
+      "step": 2326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029409868723762126,
+      "loss": 1.7489,
+      "step": 2327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029409341422251115,
+      "loss": 2.5337,
+      "step": 2328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029408813889996127,
+      "loss": 3.6233,
+      "step": 2329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029408286127005607,
+      "loss": 2.7544,
+      "step": 2330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029407758133288006,
+      "loss": 0.4265,
+      "step": 2331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002940722990885179,
+      "loss": 0.483,
+      "step": 2332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000294067014537054,
+      "loss": 0.6575,
+      "step": 2333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002940617276785731,
+      "loss": 0.5222,
+      "step": 2334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029405643851315987,
+      "loss": 0.0063,
+      "step": 2335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002940511470408989,
+      "loss": 1.5338,
+      "step": 2336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002940458532618751,
+      "loss": 0.3509,
+      "step": 2337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029404055717617306,
+      "loss": 0.498,
+      "step": 2338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029403525878387774,
+      "loss": 3.65,
+      "step": 2339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029402995808507386,
+      "loss": 0.6986,
+      "step": 2340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029402465507984637,
+      "loss": 0.8264,
+      "step": 2341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002940193497682802,
+      "loss": 0.6693,
+      "step": 2342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002940140421504603,
+      "loss": 0.1882,
+      "step": 2343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029400873222647163,
+      "loss": 0.5159,
+      "step": 2344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029400341999639924,
+      "loss": 0.5223,
+      "step": 2345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029399810546032815,
+      "loss": 0.3921,
+      "step": 2346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939927886183436,
+      "loss": 1.2905,
+      "step": 2347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939874694705306,
+      "loss": 0.4953,
+      "step": 2348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029398214801697436,
+      "loss": 0.2998,
+      "step": 2349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029397682425776013,
+      "loss": 0.4107,
+      "step": 2350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939714981929731,
+      "loss": 1.5027,
+      "step": 2351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029396616982269865,
+      "loss": 0.3948,
+      "step": 2352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029396083914702206,
+      "loss": 0.3528,
+      "step": 2353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939555061660286,
+      "loss": 0.2366,
+      "step": 2354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939501708798038,
+      "loss": 0.4354,
+      "step": 2355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939448332884331,
+      "loss": 0.1319,
+      "step": 2356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939394933920019,
+      "loss": 0.4216,
+      "step": 2357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939341511905957,
+      "loss": 0.5259,
+      "step": 2358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029392880668430014,
+      "loss": 0.1102,
+      "step": 2359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029392345987320074,
+      "loss": 1.4425,
+      "step": 2360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002939181107573831,
+      "loss": 0.6926,
+      "step": 2361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029391275933693286,
+      "loss": 0.0173,
+      "step": 2362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029390740561193585,
+      "loss": 0.0686,
+      "step": 2363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029390204958247763,
+      "loss": 0.9381,
+      "step": 2364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029389669124864413,
+      "loss": 0.2562,
+      "step": 2365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029389133061052107,
+      "loss": 0.006,
+      "step": 2366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029388596766819424,
+      "loss": 0.0811,
+      "step": 2367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029388060242174964,
+      "loss": 0.5287,
+      "step": 2368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938752348712731,
+      "loss": 0.5125,
+      "step": 2369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938698650168506,
+      "loss": 1.9581,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029386449285856816,
+      "loss": 0.6891,
+      "step": 2371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938591183965117,
+      "loss": 0.7247,
+      "step": 2372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938537416307675,
+      "loss": 3.7547,
+      "step": 2373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938483625614214,
+      "loss": 0.2627,
+      "step": 2374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938429811885597,
+      "loss": 0.2221,
+      "step": 2375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029383759751226853,
+      "loss": 2.346,
+      "step": 2376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029383221153263413,
+      "loss": 0.8554,
+      "step": 2377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938268232497427,
+      "loss": 0.5567,
+      "step": 2378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029382143266368054,
+      "loss": 0.8884,
+      "step": 2379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029381603977453395,
+      "loss": 0.0239,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002938106445823894,
+      "loss": 1.6237,
+      "step": 2381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029380524708733313,
+      "loss": 2.2329,
+      "step": 2382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029379984728945173,
+      "loss": 0.4629,
+      "step": 2383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029379444518883155,
+      "loss": 0.2774,
+      "step": 2384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937890407855591,
+      "loss": 2.6117,
+      "step": 2385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293783634079721,
+      "loss": 0.7404,
+      "step": 2386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937782250714038,
+      "loss": 0.547,
+      "step": 2387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937728137606941,
+      "loss": 1.207,
+      "step": 2388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029376740014767856,
+      "loss": 1.0524,
+      "step": 2389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029376198423244383,
+      "loss": 0.32,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937565660150767,
+      "loss": 0.6251,
+      "step": 2391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029375114549566394,
+      "loss": 0.6223,
+      "step": 2392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937457226742923,
+      "loss": 0.0636,
+      "step": 2393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029374029755104865,
+      "loss": 0.2682,
+      "step": 2394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937348701260199,
+      "loss": 0.0394,
+      "step": 2395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937294403992929,
+      "loss": 0.1277,
+      "step": 2396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937240083709546,
+      "loss": 0.3899,
+      "step": 2397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293718574041092,
+      "loss": 0.7485,
+      "step": 2398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002937131374097921,
+      "loss": 0.5985,
+      "step": 2399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029370769847714205,
+      "loss": 0.4614,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029370225724322885,
+      "loss": 0.8594,
+      "step": 2401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029369681370813963,
+      "loss": 0.7012,
+      "step": 2402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936913678719617,
+      "loss": 0.5537,
+      "step": 2403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029368591973478206,
+      "loss": 0.7513,
+      "step": 2404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936804692966881,
+      "loss": 0.681,
+      "step": 2405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029367501655776706,
+      "loss": 0.484,
+      "step": 2406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936695615181063,
+      "loss": 0.1413,
+      "step": 2407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936641041777931,
+      "loss": 1.6241,
+      "step": 2408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936586445369148,
+      "loss": 0.1929,
+      "step": 2409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293653182595559,
+      "loss": 0.4747,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936477183538131,
+      "loss": 0.2735,
+      "step": 2411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936422518117645,
+      "loss": 0.1057,
+      "step": 2412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936367829695008,
+      "loss": 0.5853,
+      "step": 2413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936313118271097,
+      "loss": 0.3527,
+      "step": 2414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029362583838467857,
+      "loss": 0.7464,
+      "step": 2415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936203626422953,
+      "loss": 0.0236,
+      "step": 2416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936148846000474,
+      "loss": 0.0429,
+      "step": 2417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002936094042580227,
+      "loss": 0.8147,
+      "step": 2418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029360392161630885,
+      "loss": 2.0524,
+      "step": 2419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029359843667499375,
+      "loss": 0.5595,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029359294943416526,
+      "loss": 0.0111,
+      "step": 2421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029358745989391114,
+      "loss": 0.2972,
+      "step": 2422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029358196805431933,
+      "loss": 0.019,
+      "step": 2423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935764739154778,
+      "loss": 2.4233,
+      "step": 2424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935709774774745,
+      "loss": 0.1753,
+      "step": 2425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935654787403975,
+      "loss": 0.3079,
+      "step": 2426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935599777043348,
+      "loss": 0.6071,
+      "step": 2427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029355447436937447,
+      "loss": 0.027,
+      "step": 2428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029354896873560473,
+      "loss": 0.5005,
+      "step": 2429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935434608031137,
+      "loss": 0.0087,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935379505719896,
+      "loss": 1.4845,
+      "step": 2431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029353243804232056,
+      "loss": 0.6282,
+      "step": 2432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029352692321419493,
+      "loss": 1.3454,
+      "step": 2433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935214060877011,
+      "loss": 0.3686,
+      "step": 2434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935158866629273,
+      "loss": 0.4234,
+      "step": 2435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293510364939962,
+      "loss": 0.7052,
+      "step": 2436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002935048409188936,
+      "loss": 2.1507,
+      "step": 2437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029349931459981047,
+      "loss": 0.596,
+      "step": 2438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002934937859828012,
+      "loss": 0.2778,
+      "step": 2439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002934882550679544,
+      "loss": 0.3016,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029348272185535844,
+      "loss": 0.4963,
+      "step": 2441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293477186345102,
+      "loss": 2.7574,
+      "step": 2442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029347164853727386,
+      "loss": 0.0079,
+      "step": 2443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002934661084319625,
+      "loss": 0.6956,
+      "step": 2444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002934605660292568,
+      "loss": 0.4086,
+      "step": 2445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029345502132924536,
+      "loss": 0.3194,
+      "step": 2446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029344947433201707,
+      "loss": 0.6537,
+      "step": 2447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029344392503766073,
+      "loss": 1.6508,
+      "step": 2448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029343837344626526,
+      "loss": 0.4832,
+      "step": 2449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029343281955791946,
+      "loss": 0.0017,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029342726337271233,
+      "loss": 0.0233,
+      "step": 2451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029342170489073284,
+      "loss": 0.4756,
+      "step": 2452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029341614411207,
+      "loss": 0.7188,
+      "step": 2453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002934105810368128,
+      "loss": 0.3446,
+      "step": 2454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002934050156650505,
+      "loss": 0.707,
+      "step": 2455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293399447996872,
+      "loss": 0.8833,
+      "step": 2456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933938780323665,
+      "loss": 0.9731,
+      "step": 2457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933883057716233,
+      "loss": 1.1877,
+      "step": 2458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029338273121473163,
+      "loss": 0.514,
+      "step": 2459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933771543617806,
+      "loss": 0.4467,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933715752128597,
+      "loss": 0.5766,
+      "step": 2461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933659937680582,
+      "loss": 0.694,
+      "step": 2462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933604100274655,
+      "loss": 0.3897,
+      "step": 2463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933548239911709,
+      "loss": 0.4439,
+      "step": 2464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029334923565926395,
+      "loss": 3.49,
+      "step": 2465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933436450318342,
+      "loss": 0.7767,
+      "step": 2466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029333805210897103,
+      "loss": 1.42,
+      "step": 2467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933324568907641,
+      "loss": 1.0078,
+      "step": 2468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293326859377303,
+      "loss": 0.4147,
+      "step": 2469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029332125956867735,
+      "loss": 0.3582,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002933156574649768,
+      "loss": 0.2734,
+      "step": 2471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029331005306629113,
+      "loss": 0.354,
+      "step": 2472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029330444637271,
+      "loss": 0.0819,
+      "step": 2473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029329883738432323,
+      "loss": 0.674,
+      "step": 2474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932932261012206,
+      "loss": 0.2126,
+      "step": 2475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932876125234921,
+      "loss": 0.814,
+      "step": 2476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932819966512275,
+      "loss": 1.0446,
+      "step": 2477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932763784845167,
+      "loss": 0.775,
+      "step": 2478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029327075802344977,
+      "loss": 0.6052,
+      "step": 2479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029326513526811666,
+      "loss": 0.6949,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029325951021860747,
+      "loss": 0.3504,
+      "step": 2481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029325951021860747,
+      "loss": 1.5343,
+      "step": 2482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932538828750122,
+      "loss": 0.3815,
+      "step": 2483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029324825323742096,
+      "loss": 0.7963,
+      "step": 2484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932426213059239,
+      "loss": 0.7522,
+      "step": 2485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029323698708061125,
+      "loss": 0.5938,
+      "step": 2486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932313505615733,
+      "loss": 0.4582,
+      "step": 2487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029322571174890015,
+      "loss": 0.8109,
+      "step": 2488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932200706426822,
+      "loss": 0.7002,
+      "step": 2489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029321442724300977,
+      "loss": 0.3347,
+      "step": 2490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002932087815499733,
+      "loss": 0.7794,
+      "step": 2491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293203133563663,
+      "loss": 0.3165,
+      "step": 2492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931974832841695,
+      "loss": 0.2842,
+      "step": 2493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029319183071158317,
+      "loss": 0.3621,
+      "step": 2494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931861758459946,
+      "loss": 0.4394,
+      "step": 2495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931805186874943,
+      "loss": 2.907,
+      "step": 2496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931748592361729,
+      "loss": 2.5243,
+      "step": 2497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293169197492121,
+      "loss": 0.3047,
+      "step": 2498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029316353345542923,
+      "loss": 1.1268,
+      "step": 2499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931578671261884,
+      "loss": 0.7788,
+      "step": 2500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029315219850448913,
+      "loss": 0.2955,
+      "step": 2501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029314652759042224,
+      "loss": 1.03,
+      "step": 2502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029314085438407857,
+      "loss": 0.3138,
+      "step": 2503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029313517888554893,
+      "loss": 0.7221,
+      "step": 2504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931295010949242,
+      "loss": 0.5437,
+      "step": 2505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931238210122953,
+      "loss": 0.8696,
+      "step": 2506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931181386377533,
+      "loss": 0.3989,
+      "step": 2507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000293112453971389,
+      "loss": 0.3588,
+      "step": 2508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002931067670132936,
+      "loss": 0.5692,
+      "step": 2509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029310107776355806,
+      "loss": 0.3786,
+      "step": 2510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930953862222735,
+      "loss": 1.2952,
+      "step": 2511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930896923895311,
+      "loss": 0.8107,
+      "step": 2512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930839962654221,
+      "loss": 0.4119,
+      "step": 2513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029307829785003755,
+      "loss": 0.6514,
+      "step": 2514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930725971434688,
+      "loss": 0.9037,
+      "step": 2515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930668941458072,
+      "loss": 0.3681,
+      "step": 2516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930611888571439,
+      "loss": 0.321,
+      "step": 2517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029305548127757044,
+      "loss": 0.9423,
+      "step": 2518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930497714071781,
+      "loss": 0.0062,
+      "step": 2519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029304405924605837,
+      "loss": 0.7918,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930383447943027,
+      "loss": 0.858,
+      "step": 2521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029303262805200264,
+      "loss": 0.4526,
+      "step": 2522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029302690901924973,
+      "loss": 0.4959,
+      "step": 2523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029302118769613546,
+      "loss": 0.6132,
+      "step": 2524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029301546408275154,
+      "loss": 0.6183,
+      "step": 2525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002930097381791896,
+      "loss": 0.6585,
+      "step": 2526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029300400998554134,
+      "loss": 1.7988,
+      "step": 2527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929982795018985,
+      "loss": 0.5585,
+      "step": 2528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029299254672835275,
+      "loss": 0.0661,
+      "step": 2529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029298681166499603,
+      "loss": 0.7228,
+      "step": 2530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929810743119201,
+      "loss": 0.1031,
+      "step": 2531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929753346692169,
+      "loss": 1.8304,
+      "step": 2532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029296959273697817,
+      "loss": 0.2459,
+      "step": 2533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029296384851529604,
+      "loss": 0.4619,
+      "step": 2534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929581020042624,
+      "loss": 0.7704,
+      "step": 2535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029295235320396935,
+      "loss": 0.3514,
+      "step": 2536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929466021145089,
+      "loss": 0.6979,
+      "step": 2537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929408487359732,
+      "loss": 0.3962,
+      "step": 2538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929350930684543,
+      "loss": 0.0136,
+      "step": 2539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029292933511204433,
+      "loss": 1.1373,
+      "step": 2540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029292357486683567,
+      "loss": 0.2949,
+      "step": 2541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029291781233292035,
+      "loss": 1.2874,
+      "step": 2542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029291204751039085,
+      "loss": 0.4398,
+      "step": 2543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029290628039933933,
+      "loss": 0.0291,
+      "step": 2544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002929005109998582,
+      "loss": 1.0888,
+      "step": 2545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928947393120399,
+      "loss": 0.439,
+      "step": 2546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928889653359768,
+      "loss": 0.7187,
+      "step": 2547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928831890717613,
+      "loss": 0.6237,
+      "step": 2548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000292877410519486,
+      "loss": 0.8481,
+      "step": 2549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928716296792434,
+      "loss": 0.5366,
+      "step": 2550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928658465511261,
+      "loss": 0.023,
+      "step": 2551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029286006113522664,
+      "loss": 0.9813,
+      "step": 2552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029285427343163774,
+      "loss": 0.9033,
+      "step": 2553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000292848483440452,
+      "loss": 0.5136,
+      "step": 2554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928426911617622,
+      "loss": 0.4399,
+      "step": 2555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029283689659566115,
+      "loss": 0.0583,
+      "step": 2556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029283109974224145,
+      "loss": 0.6765,
+      "step": 2557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928253006015961,
+      "loss": 0.0725,
+      "step": 2558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928194991738179,
+      "loss": 0.8215,
+      "step": 2559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002928136954589998,
+      "loss": 0.4982,
+      "step": 2560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029280788945723466,
+      "loss": 0.4659,
+      "step": 2561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029280208116861555,
+      "loss": 0.0044,
+      "step": 2562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927962705932354,
+      "loss": 0.0135,
+      "step": 2563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029279045773118723,
+      "loss": 0.4199,
+      "step": 2564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029278464258256423,
+      "loss": 0.0103,
+      "step": 2565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029277882514745947,
+      "loss": 0.4628,
+      "step": 2566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029277300542596607,
+      "loss": 0.0025,
+      "step": 2567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927671834181773,
+      "loss": 0.5404,
+      "step": 2568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029276135912418633,
+      "loss": 0.4926,
+      "step": 2569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927555325440864,
+      "loss": 0.2918,
+      "step": 2570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000292749703677971,
+      "loss": 0.7254,
+      "step": 2571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927438725259332,
+      "loss": 0.2525,
+      "step": 2572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029273803908806657,
+      "loss": 0.9592,
+      "step": 2573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029273220336446446,
+      "loss": 0.367,
+      "step": 2574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029272636535522035,
+      "loss": 1.0793,
+      "step": 2575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029272052506042765,
+      "loss": 0.424,
+      "step": 2576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927146824801799,
+      "loss": 2.5424,
+      "step": 2577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927088376145708,
+      "loss": 0.1345,
+      "step": 2578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927029904636938,
+      "loss": 0.6754,
+      "step": 2579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029269714102764255,
+      "loss": 0.5494,
+      "step": 2580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029269128930651075,
+      "loss": 0.3204,
+      "step": 2581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029268543530039213,
+      "loss": 0.4086,
+      "step": 2582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926795790093804,
+      "loss": 0.4667,
+      "step": 2583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926737204335693,
+      "loss": 0.5598,
+      "step": 2584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029266785957305275,
+      "loss": 0.5391,
+      "step": 2585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926619964279245,
+      "loss": 2.0431,
+      "step": 2586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926561309982785,
+      "loss": 0.4481,
+      "step": 2587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926502632842086,
+      "loss": 0.5743,
+      "step": 2588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926443932858089,
+      "loss": 0.1951,
+      "step": 2589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926385210031733,
+      "loss": 0.3339,
+      "step": 2590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029263264643639587,
+      "loss": 0.5305,
+      "step": 2591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029262676958557066,
+      "loss": 0.5555,
+      "step": 2592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029262089045079177,
+      "loss": 0.912,
+      "step": 2593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029261500903215336,
+      "loss": 0.7105,
+      "step": 2594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029260912532974966,
+      "loss": 0.0008,
+      "step": 2595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002926032393436748,
+      "loss": 0.0005,
+      "step": 2596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925973510740231,
+      "loss": 0.0014,
+      "step": 2597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925914605208888,
+      "loss": 2.6329,
+      "step": 2598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925855676843663,
+      "loss": 0.001,
+      "step": 2599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925796725645499,
+      "loss": 0.079,
+      "step": 2600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029257377516153405,
+      "loss": 0.7009,
+      "step": 2601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925678754754131,
+      "loss": 0.6578,
+      "step": 2602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925619735062817,
+      "loss": 0.0046,
+      "step": 2603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925560692542342,
+      "loss": 0.6126,
+      "step": 2604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925501627193652,
+      "loss": 0.0085,
+      "step": 2605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029254425390176923,
+      "loss": 0.3498,
+      "step": 2606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000292538342801541,
+      "loss": 1.5423,
+      "step": 2607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029253242941877514,
+      "loss": 0.3087,
+      "step": 2608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925265137535663,
+      "loss": 0.6949,
+      "step": 2609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925205958060093,
+      "loss": 0.4507,
+      "step": 2610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029251467557619884,
+      "loss": 0.597,
+      "step": 2611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002925087530642297,
+      "loss": 0.0088,
+      "step": 2612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029250282827019677,
+      "loss": 0.4552,
+      "step": 2613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002924969011941949,
+      "loss": 0.0164,
+      "step": 2614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000292490971836319,
+      "loss": 0.4816,
+      "step": 2615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002924850401966641,
+      "loss": 0.6562,
+      "step": 2616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002924791062753251,
+      "loss": 1.0109,
+      "step": 2617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000292473170072397,
+      "loss": 0.6601,
+      "step": 2618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924672315879749,
+      "loss": 0.0311,
+      "step": 2619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029246129082215395,
+      "loss": 0.8509,
+      "step": 2620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924553477750292,
+      "loss": 0.2096,
+      "step": 2621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924494024466959,
+      "loss": 0.3817,
+      "step": 2622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029244345483724917,
+      "loss": 0.455,
+      "step": 2623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924375049467843,
+      "loss": 0.9321,
+      "step": 2624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029243155277539653,
+      "loss": 0.3475,
+      "step": 2625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924255983231812,
+      "loss": 0.5622,
+      "step": 2626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924196415902337,
+      "loss": 0.4503,
+      "step": 2627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924136825766494,
+      "loss": 1.2062,
+      "step": 2628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002924077212825236,
+      "loss": 0.6234,
+      "step": 2629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000292401757707952,
+      "loss": 0.4917,
+      "step": 2630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002923957918530299,
+      "loss": 0.7625,
+      "step": 2631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002923898237178529,
+      "loss": 0.303,
+      "step": 2632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002923838533025166,
+      "loss": 2.6312,
+      "step": 2633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029237788060711655,
+      "loss": 0.3118,
+      "step": 2634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002923719056317484,
+      "loss": 0.5686,
+      "step": 2635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029236592837650783,
+      "loss": 0.2564,
+      "step": 2636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029235994884149063,
+      "loss": 0.994,
+      "step": 2637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029235396702679254,
+      "loss": 0.3185,
+      "step": 2638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029234798293250924,
+      "loss": 0.9648,
+      "step": 2639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002923419965587367,
+      "loss": 0.4886,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029233600790557065,
+      "loss": 0.3284,
+      "step": 2641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002923300169731071,
+      "loss": 0.2507,
+      "step": 2642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029232402376144187,
+      "loss": 1.321,
+      "step": 2643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029231802827067103,
+      "loss": 0.5593,
+      "step": 2644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002923120305008905,
+      "loss": 1.5628,
+      "step": 2645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029230603045219647,
+      "loss": 0.2047,
+      "step": 2646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029230002812468493,
+      "loss": 0.446,
+      "step": 2647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029229402351845196,
+      "loss": 0.2008,
+      "step": 2648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029228801663359373,
+      "loss": 1.16,
+      "step": 2649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002922820074702065,
+      "loss": 0.3847,
+      "step": 2650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002922759960283865,
+      "loss": 0.7801,
+      "step": 2651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002922699823082299,
+      "loss": 5.334,
+      "step": 2652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029226396630983306,
+      "loss": 0.6861,
+      "step": 2653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029225794803329225,
+      "loss": 0.3213,
+      "step": 2654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029225192747870396,
+      "loss": 2.008,
+      "step": 2655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029224590464616457,
+      "loss": 0.4802,
+      "step": 2656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029223987953577046,
+      "loss": 0.3488,
+      "step": 2657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002922338521476181,
+      "loss": 0.6615,
+      "step": 2658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002922278224818041,
+      "loss": 0.2752,
+      "step": 2659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029222179053842504,
+      "loss": 0.2712,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029221575631757743,
+      "loss": 0.0553,
+      "step": 2661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029220971981935785,
+      "loss": 0.8702,
+      "step": 2662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002922036810438631,
+      "loss": 0.9495,
+      "step": 2663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029219763999118977,
+      "loss": 1.0555,
+      "step": 2664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921915966614347,
+      "loss": 0.0646,
+      "step": 2665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029218555105469453,
+      "loss": 0.8341,
+      "step": 2666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921795031710662,
+      "loss": 0.0125,
+      "step": 2667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029217345301064653,
+      "loss": 0.6714,
+      "step": 2668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921674005735324,
+      "loss": 0.6335,
+      "step": 2669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029216134585982064,
+      "loss": 0.8972,
+      "step": 2670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921552888696083,
+      "loss": 0.2506,
+      "step": 2671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029214922960299234,
+      "loss": 1.3595,
+      "step": 2672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029214316806006985,
+      "loss": 0.396,
+      "step": 2673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029213710424093786,
+      "loss": 1.505,
+      "step": 2674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029213103814569336,
+      "loss": 0.4928,
+      "step": 2675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921249697744337,
+      "loss": 0.6055,
+      "step": 2676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921188991272559,
+      "loss": 0.9607,
+      "step": 2677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921128262042572,
+      "loss": 0.0079,
+      "step": 2678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921067510055349,
+      "loss": 0.006,
+      "step": 2679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002921006735311863,
+      "loss": 1.0296,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029209459378130864,
+      "loss": 0.2374,
+      "step": 2681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029208851175599926,
+      "loss": 1.7443,
+      "step": 2682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002920824274553557,
+      "loss": 0.198,
+      "step": 2683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029207634087947525,
+      "loss": 0.578,
+      "step": 2684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029207025202845547,
+      "loss": 0.8216,
+      "step": 2685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029206416090239376,
+      "loss": 0.008,
+      "step": 2686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029205806750138777,
+      "loss": 1.1571,
+      "step": 2687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029205197182553496,
+      "loss": 0.8046,
+      "step": 2688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002920458738749331,
+      "loss": 1.0991,
+      "step": 2689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002920397736496797,
+      "loss": 0.0614,
+      "step": 2690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002920336711498725,
+      "loss": 0.0156,
+      "step": 2691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029202756637560927,
+      "loss": 0.0434,
+      "step": 2692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029202145932698763,
+      "loss": 0.0111,
+      "step": 2693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002920153500041055,
+      "loss": 0.0893,
+      "step": 2694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002920092384070607,
+      "loss": 0.0237,
+      "step": 2695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029200312453595104,
+      "loss": 0.0285,
+      "step": 2696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002919970083908745,
+      "loss": 0.9549,
+      "step": 2697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029199088997192894,
+      "loss": 0.2691,
+      "step": 2698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002919847692792124,
+      "loss": 0.5661,
+      "step": 2699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029197864631282284,
+      "loss": 0.8415,
+      "step": 2700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029197252107285835,
+      "loss": 0.4173,
+      "step": 2701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029196639355941705,
+      "loss": 0.5952,
+      "step": 2702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029196026377259693,
+      "loss": 0.3228,
+      "step": 2703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029195413171249635,
+      "loss": 0.3265,
+      "step": 2704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002919479973792133,
+      "loss": 1.1885,
+      "step": 2705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002919418607728461,
+      "loss": 1.156,
+      "step": 2706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029193572189349307,
+      "loss": 0.9065,
+      "step": 2707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029192958074125247,
+      "loss": 0.0123,
+      "step": 2708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029192343731622265,
+      "loss": 0.9571,
+      "step": 2709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029191729161850196,
+      "loss": 0.2603,
+      "step": 2710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029191114364818883,
+      "loss": 0.9116,
+      "step": 2711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002919049934053817,
+      "loss": 0.3535,
+      "step": 2712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918988408901791,
+      "loss": 0.1799,
+      "step": 2713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029189268610267953,
+      "loss": 0.0012,
+      "step": 2714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918865290429815,
+      "loss": 0.0018,
+      "step": 2715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918803697111837,
+      "loss": 0.7749,
+      "step": 2716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918742081073847,
+      "loss": 0.6578,
+      "step": 2717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918680442316832,
+      "loss": 0.5879,
+      "step": 2718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918618780841778,
+      "loss": 0.6032,
+      "step": 2719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029185570966496735,
+      "loss": 0.4864,
+      "step": 2720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029184953897415063,
+      "loss": 0.5627,
+      "step": 2721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918433660118265,
+      "loss": 0.4295,
+      "step": 2722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029183719077809366,
+      "loss": 0.3666,
+      "step": 2723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029183101327305105,
+      "loss": 0.08,
+      "step": 2724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029182483349679767,
+      "loss": 1.0876,
+      "step": 2725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918186514494325,
+      "loss": 0.0015,
+      "step": 2726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029181246713105434,
+      "loss": 0.4418,
+      "step": 2727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918062805417624,
+      "loss": 1.1223,
+      "step": 2728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002918000916816557,
+      "loss": 0.5687,
+      "step": 2729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029179390055083333,
+      "loss": 1.6339,
+      "step": 2730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002917877071493945,
+      "loss": 0.0204,
+      "step": 2731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002917815114774383,
+      "loss": 0.5665,
+      "step": 2732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029177531353506394,
+      "loss": 0.5404,
+      "step": 2733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029176911332237074,
+      "loss": 0.4842,
+      "step": 2734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029176291083945794,
+      "loss": 0.3993,
+      "step": 2735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002917567060864249,
+      "loss": 0.3167,
+      "step": 2736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291750499063371,
+      "loss": 5.222,
+      "step": 2737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029174428977039554,
+      "loss": 0.7891,
+      "step": 2738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291738078207598,
+      "loss": 0.8394,
+      "step": 2739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029173186437507785,
+      "loss": 0.9859,
+      "step": 2740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029172564827293463,
+      "loss": 1.1328,
+      "step": 2741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029171942990126787,
+      "loss": 0.8112,
+      "step": 2742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002917132092601771,
+      "loss": 0.4057,
+      "step": 2743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291706986349762,
+      "loss": 0.3766,
+      "step": 2744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029170076117012215,
+      "loss": 0.0013,
+      "step": 2745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029169453372135726,
+      "loss": 0.3128,
+      "step": 2746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916883040035671,
+      "loss": 2.0938,
+      "step": 2747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916820720168514,
+      "loss": 0.0011,
+      "step": 2748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916758377613099,
+      "loss": 0.5517,
+      "step": 2749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916696012370425,
+      "loss": 0.0005,
+      "step": 2750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029166336244414913,
+      "loss": 1.2712,
+      "step": 2751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029165712138272957,
+      "loss": 0.2364,
+      "step": 2752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916508780528838,
+      "loss": 0.004,
+      "step": 2753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916446324547118,
+      "loss": 0.9027,
+      "step": 2754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029163838458831355,
+      "loss": 0.4729,
+      "step": 2755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916321344537892,
+      "loss": 1.2828,
+      "step": 2756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916258820512388,
+      "loss": 0.816,
+      "step": 2757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916196273807624,
+      "loss": 0.6651,
+      "step": 2758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916133704424602,
+      "loss": 1.2438,
+      "step": 2759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029160711123643243,
+      "loss": 0.4163,
+      "step": 2760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029160084976277933,
+      "loss": 3.6647,
+      "step": 2761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029159458602160104,
+      "loss": 0.4261,
+      "step": 2762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029158832001299805,
+      "loss": 0.4375,
+      "step": 2763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029158205173707055,
+      "loss": 2.5335,
+      "step": 2764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291575781193919,
+      "loss": 0.6596,
+      "step": 2765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029156950838364376,
+      "loss": 0.3776,
+      "step": 2766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002915632333063454,
+      "loss": 0.3683,
+      "step": 2767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002915569559621242,
+      "loss": 0.5658,
+      "step": 2768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029155067635108084,
+      "loss": 0.2712,
+      "step": 2769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029154439447331585,
+      "loss": 0.0091,
+      "step": 2770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029153811032892984,
+      "loss": 1.256,
+      "step": 2771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029153182391802336,
+      "loss": 0.6606,
+      "step": 2772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002915255352406972,
+      "loss": 0.1948,
+      "step": 2773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002915192442970519,
+      "loss": 0.5809,
+      "step": 2774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029151295108718835,
+      "loss": 4.0452,
+      "step": 2775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002915066556112073,
+      "loss": 0.4954,
+      "step": 2776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029150035786920946,
+      "loss": 0.2425,
+      "step": 2777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914940578612958,
+      "loss": 1.2172,
+      "step": 2778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914877555875671,
+      "loss": 0.4861,
+      "step": 2779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029148145104812437,
+      "loss": 2.7737,
+      "step": 2780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914751442430685,
+      "loss": 0.8931,
+      "step": 2781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029146883517250057,
+      "loss": 0.256,
+      "step": 2782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029146252383652156,
+      "loss": 1.1459,
+      "step": 2783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029145621023523254,
+      "loss": 0.6312,
+      "step": 2784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029144989436873456,
+      "loss": 0.233,
+      "step": 2785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914435762371289,
+      "loss": 0.1031,
+      "step": 2786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029143725584051655,
+      "loss": 0.2966,
+      "step": 2787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914309331789989,
+      "loss": 0.555,
+      "step": 2788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029142460825267705,
+      "loss": 0.9459,
+      "step": 2789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914182810616523,
+      "loss": 0.5671,
+      "step": 2790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914119516060261,
+      "loss": 0.2625,
+      "step": 2791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002914056198858997,
+      "loss": 0.3098,
+      "step": 2792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029139928590137453,
+      "loss": 0.6231,
+      "step": 2793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291392949652552,
+      "loss": 0.408,
+      "step": 2794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002913866111395336,
+      "loss": 0.0048,
+      "step": 2795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029138027036242076,
+      "loss": 0.3608,
+      "step": 2796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029137392732131516,
+      "loss": 0.2772,
+      "step": 2797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002913675820163182,
+      "loss": 0.3357,
+      "step": 2798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002913612344475316,
+      "loss": 0.4085,
+      "step": 2799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291354884615057,
+      "loss": 0.1726,
+      "step": 2800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002913485325189961,
+      "loss": 0.4893,
+      "step": 2801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029134217815945056,
+      "loss": 0.5994,
+      "step": 2802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002913358215365221,
+      "loss": 1.0739,
+      "step": 2803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029132946265031265,
+      "loss": 0.4155,
+      "step": 2804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002913231015009239,
+      "loss": 0.9725,
+      "step": 2805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002913167380884578,
+      "loss": 1.2451,
+      "step": 2806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029131037241301625,
+      "loss": 5.6612,
+      "step": 2807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029130400447470114,
+      "loss": 0.5568,
+      "step": 2808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029129763427361447,
+      "loss": 0.1197,
+      "step": 2809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029129126180985824,
+      "loss": 0.0079,
+      "step": 2810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002912848870835345,
+      "loss": 0.6988,
+      "step": 2811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029127851009474537,
+      "loss": 0.525,
+      "step": 2812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029127213084359293,
+      "loss": 0.7679,
+      "step": 2813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029126574933017926,
+      "loss": 1.0434,
+      "step": 2814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002912593655546067,
+      "loss": 0.583,
+      "step": 2815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002912529795169774,
+      "loss": 0.746,
+      "step": 2816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029124659121739356,
+      "loss": 0.3334,
+      "step": 2817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029124020065595757,
+      "loss": 0.7069,
+      "step": 2818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029123380783277177,
+      "loss": 0.1506,
+      "step": 2819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002912274127479385,
+      "loss": 0.3845,
+      "step": 2820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002912210154015601,
+      "loss": 0.3683,
+      "step": 2821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029121461579373917,
+      "loss": 0.297,
+      "step": 2822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291208213924578,
+      "loss": 0.3974,
+      "step": 2823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029120180979417935,
+      "loss": 0.0225,
+      "step": 2824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029119540340264555,
+      "loss": 0.7232,
+      "step": 2825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029118899475007924,
+      "loss": 0.322,
+      "step": 2826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002911825838365831,
+      "loss": 0.1299,
+      "step": 2827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002911761706622598,
+      "loss": 0.1465,
+      "step": 2828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002911697552272119,
+      "loss": 0.8805,
+      "step": 2829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029116333753154237,
+      "loss": 0.6294,
+      "step": 2830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029115691757535374,
+      "loss": 0.3931,
+      "step": 2831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291150495358749,
+      "loss": 1.1673,
+      "step": 2832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029114407088183086,
+      "loss": 0.3674,
+      "step": 2833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002911376441447022,
+      "loss": 0.56,
+      "step": 2834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002911312151474661,
+      "loss": 0.3035,
+      "step": 2835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002911247838902254,
+      "loss": 0.3793,
+      "step": 2836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291118350373083,
+      "loss": 0.4079,
+      "step": 2837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000291111914596142,
+      "loss": 0.5314,
+      "step": 2838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029110547655950556,
+      "loss": 0.7743,
+      "step": 2839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029109903626327663,
+      "loss": 1.8775,
+      "step": 2840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029109259370755844,
+      "loss": 0.0161,
+      "step": 2841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029108614889245405,
+      "loss": 0.7485,
+      "step": 2842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029107970181806674,
+      "loss": 0.5147,
+      "step": 2843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029107325248449976,
+      "loss": 0.3187,
+      "step": 2844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910668008918563,
+      "loss": 0.2234,
+      "step": 2845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910603470402398,
+      "loss": 0.0054,
+      "step": 2846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910538909297536,
+      "loss": 1.0557,
+      "step": 2847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910474325605009,
+      "loss": 1.633,
+      "step": 2848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910409719325853,
+      "loss": 0.4538,
+      "step": 2849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910345090461102,
+      "loss": 0.4249,
+      "step": 2850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910280439011791,
+      "loss": 0.2575,
+      "step": 2851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029102157649789557,
+      "loss": 0.7226,
+      "step": 2852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910151068363631,
+      "loss": 0.9339,
+      "step": 2853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029100863491668535,
+      "loss": 0.2573,
+      "step": 2854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002910021607389659,
+      "loss": 0.4022,
+      "step": 2855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029099568430330846,
+      "loss": 0.528,
+      "step": 2856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002909892056098168,
+      "loss": 0.581,
+      "step": 2857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002909827246585946,
+      "loss": 0.0095,
+      "step": 2858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002909762414497456,
+      "loss": 1.0283,
+      "step": 2859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002909697559833737,
+      "loss": 0.9226,
+      "step": 2860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002909632682595827,
+      "loss": 0.3038,
+      "step": 2861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029095677827847653,
+      "loss": 0.7273,
+      "step": 2862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002909502860401591,
+      "loss": 0.7953,
+      "step": 2863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029094379154473443,
+      "loss": 0.4912,
+      "step": 2864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002909372947923064,
+      "loss": 0.6562,
+      "step": 2865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029093079578297907,
+      "loss": 0.7922,
+      "step": 2866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002909242945168566,
+      "loss": 1.1633,
+      "step": 2867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029091779099404307,
+      "loss": 0.7039,
+      "step": 2868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029091128521464263,
+      "loss": 0.7093,
+      "step": 2869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002909047771787594,
+      "loss": 0.9817,
+      "step": 2870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029089826688649765,
+      "loss": 0.7708,
+      "step": 2871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002908917543379616,
+      "loss": 0.692,
+      "step": 2872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029088523953325554,
+      "loss": 0.0064,
+      "step": 2873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002908787224724838,
+      "loss": 0.6329,
+      "step": 2874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029087220315575076,
+      "loss": 0.5718,
+      "step": 2875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029086568158316086,
+      "loss": 0.8213,
+      "step": 2876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002908591577548184,
+      "loss": 0.0086,
+      "step": 2877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290852631670828,
+      "loss": 0.7785,
+      "step": 2878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029084610333129403,
+      "loss": 0.3129,
+      "step": 2879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029083957273632114,
+      "loss": 0.5478,
+      "step": 2880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002908330398860138,
+      "loss": 0.5136,
+      "step": 2881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029082650478047675,
+      "loss": 0.514,
+      "step": 2882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029081996741981455,
+      "loss": 0.1883,
+      "step": 2883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002908134278041319,
+      "loss": 0.8436,
+      "step": 2884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029080688593353357,
+      "loss": 0.0077,
+      "step": 2885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029080034180812425,
+      "loss": 0.2907,
+      "step": 2886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002907937954280088,
+      "loss": 0.3741,
+      "step": 2887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029078724679329205,
+      "loss": 0.4401,
+      "step": 2888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029078069590407875,
+      "loss": 0.8538,
+      "step": 2889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029077414276047395,
+      "loss": 0.4022,
+      "step": 2890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002907675873625825,
+      "loss": 0.9429,
+      "step": 2891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002907610297105094,
+      "loss": 0.2936,
+      "step": 2892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029075446980435964,
+      "loss": 0.823,
+      "step": 2893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029074790764423836,
+      "loss": 0.598,
+      "step": 2894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002907413432302505,
+      "loss": 1.1347,
+      "step": 2895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029073477656250127,
+      "loss": 1.4007,
+      "step": 2896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029072820764109586,
+      "loss": 0.0052,
+      "step": 2897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002907216364661393,
+      "loss": 0.5646,
+      "step": 2898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290715063037737,
+      "loss": 0.1755,
+      "step": 2899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029070848735599414,
+      "loss": 0.1118,
+      "step": 2900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290701909421016,
+      "loss": 0.5721,
+      "step": 2901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290695329232908,
+      "loss": 1.3955,
+      "step": 2902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029068874679177544,
+      "loss": 0.898,
+      "step": 2903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002906821620977237,
+      "loss": 0.3702,
+      "step": 2904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029067557515085837,
+      "loss": 0.2938,
+      "step": 2905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029066898595128476,
+      "loss": 0.0051,
+      "step": 2906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002906623944991084,
+      "loss": 0.5489,
+      "step": 2907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029065580079443496,
+      "loss": 0.3766,
+      "step": 2908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029064920483736997,
+      "loss": 0.0956,
+      "step": 2909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029064260662801906,
+      "loss": 1.4928,
+      "step": 2910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002906360061664878,
+      "loss": 0.2605,
+      "step": 2911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029062940345288204,
+      "loss": 0.5212,
+      "step": 2912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029062279848730745,
+      "loss": 0.5279,
+      "step": 2913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029061619126986973,
+      "loss": 0.7536,
+      "step": 2914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002906095818006748,
+      "loss": 0.0069,
+      "step": 2915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029060297007982836,
+      "loss": 0.5163,
+      "step": 2916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029059635610743646,
+      "loss": 0.6662,
+      "step": 2917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029058973988360486,
+      "loss": 1.7453,
+      "step": 2918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002905831214084396,
+      "loss": 0.0071,
+      "step": 2919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002905765006820466,
+      "loss": 0.3121,
+      "step": 2920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290569877704532,
+      "loss": 0.6735,
+      "step": 2921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002905632524760017,
+      "loss": 0.2238,
+      "step": 2922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002905566249965619,
+      "loss": 0.5351,
+      "step": 2923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002905499952663187,
+      "loss": 0.3849,
+      "step": 2924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002905433632853783,
+      "loss": 0.7099,
+      "step": 2925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029053672905384674,
+      "loss": 1.0698,
+      "step": 2926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029053009257183046,
+      "loss": 0.5514,
+      "step": 2927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029052345383943565,
+      "loss": 0.3332,
+      "step": 2928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002905168128567686,
+      "loss": 0.6902,
+      "step": 2929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029051016962393573,
+      "loss": 0.6623,
+      "step": 2930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029050352414104335,
+      "loss": 0.3268,
+      "step": 2931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029049687640819786,
+      "loss": 0.5317,
+      "step": 2932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029049022642550575,
+      "loss": 0.1906,
+      "step": 2933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904835741930736,
+      "loss": 0.2501,
+      "step": 2934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904769197110077,
+      "loss": 0.5151,
+      "step": 2935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029047026297941486,
+      "loss": 0.0083,
+      "step": 2936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904636039984016,
+      "loss": 1.0653,
+      "step": 2937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029045694276807446,
+      "loss": 0.1498,
+      "step": 2938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904502792885402,
+      "loss": 0.6291,
+      "step": 2939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029044361355990556,
+      "loss": 0.8857,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904369455822771,
+      "loss": 0.673,
+      "step": 2941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029043027535576184,
+      "loss": 0.0244,
+      "step": 2942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904236028804664,
+      "loss": 0.9737,
+      "step": 2943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904169281564977,
+      "loss": 0.5094,
+      "step": 2944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002904102511839627,
+      "loss": 0.015,
+      "step": 2945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029040357196296826,
+      "loss": 0.2281,
+      "step": 2946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029039689049362123,
+      "loss": 0.968,
+      "step": 2947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903902067760288,
+      "loss": 0.0025,
+      "step": 2948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903835208102979,
+      "loss": 0.2425,
+      "step": 2949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903768325965355,
+      "loss": 0.0125,
+      "step": 2950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903701421348489,
+      "loss": 0.3842,
+      "step": 2951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903634494253451,
+      "loss": 0.4831,
+      "step": 2952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029035675446813133,
+      "loss": 3.0789,
+      "step": 2953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903500572633148,
+      "loss": 0.2437,
+      "step": 2954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903433578110027,
+      "loss": 0.5762,
+      "step": 2955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029033665611130237,
+      "loss": 0.4102,
+      "step": 2956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903299521643211,
+      "loss": 0.7511,
+      "step": 2957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029032324597016626,
+      "loss": 2.2648,
+      "step": 2958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903165375289452,
+      "loss": 0.3693,
+      "step": 2959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029030982684076544,
+      "loss": 0.2735,
+      "step": 2960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002903031139057344,
+      "loss": 1.9174,
+      "step": 2961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029029639872395945,
+      "loss": 0.3768,
+      "step": 2962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902896812955483,
+      "loss": 0.4183,
+      "step": 2963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902829616206084,
+      "loss": 0.4495,
+      "step": 2964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902762396992475,
+      "loss": 1.867,
+      "step": 2965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029026951553157306,
+      "loss": 0.0752,
+      "step": 2966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029026278911769285,
+      "loss": 1.1424,
+      "step": 2967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902560604577146,
+      "loss": 0.3862,
+      "step": 2968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290249329551746,
+      "loss": 0.3463,
+      "step": 2969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029024259639989493,
+      "loss": 0.2845,
+      "step": 2970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902358610022691,
+      "loss": 0.5284,
+      "step": 2971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902291233589765,
+      "loss": 0.2335,
+      "step": 2972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902223834701249,
+      "loss": 0.3516,
+      "step": 2973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902156413358222,
+      "loss": 2.339,
+      "step": 2974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002902088969561765,
+      "loss": 0.3963,
+      "step": 2975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029020215033129574,
+      "loss": 0.4147,
+      "step": 2976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029019540146128797,
+      "loss": 0.7152,
+      "step": 2977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901886503462612,
+      "loss": 1.9053,
+      "step": 2978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901818969863237,
+      "loss": 0.389,
+      "step": 2979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901751413815834,
+      "loss": 0.6216,
+      "step": 2980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029016838353214863,
+      "loss": 0.147,
+      "step": 2981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901616234381275,
+      "loss": 1.0092,
+      "step": 2982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901548610996284,
+      "loss": 0.7278,
+      "step": 2983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901480965167595,
+      "loss": 0.711,
+      "step": 2984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901413296896292,
+      "loss": 0.4619,
+      "step": 2985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029013456061834574,
+      "loss": 0.4105,
+      "step": 2986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029012778930301774,
+      "loss": 0.3418,
+      "step": 2987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901210157437534,
+      "loss": 0.5282,
+      "step": 2988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901142399406613,
+      "loss": 0.8139,
+      "step": 2989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029010746189384994,
+      "loss": 0.4609,
+      "step": 2990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002901006816034279,
+      "loss": 0.6258,
+      "step": 2991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900938990695036,
+      "loss": 0.1288,
+      "step": 2992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900871142921858,
+      "loss": 0.3169,
+      "step": 2993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029008032727158317,
+      "loss": 0.1637,
+      "step": 2994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900735380078043,
+      "loss": 0.4503,
+      "step": 2995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900667465009579,
+      "loss": 1.3429,
+      "step": 2996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900599527511528,
+      "loss": 0.3106,
+      "step": 2997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900531567584977,
+      "loss": 0.0013,
+      "step": 2998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029004635852310153,
+      "loss": 0.6926,
+      "step": 2999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029003955804507306,
+      "loss": 0.2336,
+      "step": 3000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029003275532452137,
+      "loss": 1.184,
+      "step": 3001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029002595036155515,
+      "loss": 0.335,
+      "step": 3002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900191431562835,
+      "loss": 0.3167,
+      "step": 3003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002900123337088154,
+      "loss": 0.4289,
+      "step": 3004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029000552201925993,
+      "loss": 2.2371,
+      "step": 3005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899987080877262,
+      "loss": 1.559,
+      "step": 3006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028999189191432313,
+      "loss": 0.4815,
+      "step": 3007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028998507349916014,
+      "loss": 0.1133,
+      "step": 3008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899782528423462,
+      "loss": 0.5336,
+      "step": 3009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899714299439907,
+      "loss": 0.884,
+      "step": 3010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028996460480420274,
+      "loss": 0.0013,
+      "step": 3011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028995777742309175,
+      "loss": 0.2799,
+      "step": 3012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028995094780076693,
+      "loss": 0.742,
+      "step": 3013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899441159373378,
+      "loss": 0.3783,
+      "step": 3014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899372818329136,
+      "loss": 0.8425,
+      "step": 3015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899304454876039,
+      "loss": 2.6509,
+      "step": 3016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028992360690151817,
+      "loss": 0.6559,
+      "step": 3017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899167660747657,
+      "loss": 1.2047,
+      "step": 3018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899099230074564,
+      "loss": 0.2862,
+      "step": 3019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899030776996996,
+      "loss": 1.0497,
+      "step": 3020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898962301516049,
+      "loss": 2.3888,
+      "step": 3021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898893803632821,
+      "loss": 0.5187,
+      "step": 3022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898825283348408,
+      "loss": 0.0009,
+      "step": 3023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028987567406639074,
+      "loss": 1.049,
+      "step": 3024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898688175580417,
+      "loss": 0.0268,
+      "step": 3025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898619588099034,
+      "loss": 0.9113,
+      "step": 3026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898550978220858,
+      "loss": 0.5525,
+      "step": 3027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898482345946987,
+      "loss": 0.9018,
+      "step": 3028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000289841369127852,
+      "loss": 0.4968,
+      "step": 3029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028983450142165565,
+      "loss": 0.1195,
+      "step": 3030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898276314762196,
+      "loss": 0.5656,
+      "step": 3031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898207592916539,
+      "loss": 0.4641,
+      "step": 3032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028981388486806856,
+      "loss": 0.3192,
+      "step": 3033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898070082055737,
+      "loss": 0.0203,
+      "step": 3034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002898001293042794,
+      "loss": 1.1226,
+      "step": 3035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028979324816429586,
+      "loss": 0.5605,
+      "step": 3036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028978636478573324,
+      "loss": 0.3611,
+      "step": 3037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028977947916870183,
+      "loss": 0.9629,
+      "step": 3038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002897725913133118,
+      "loss": 0.6379,
+      "step": 3039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028976570121967346,
+      "loss": 1.027,
+      "step": 3040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028975880888789724,
+      "loss": 0.3077,
+      "step": 3041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002897519143180934,
+      "loss": 0.5043,
+      "step": 3042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002897450175103724,
+      "loss": 1.3095,
+      "step": 3043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028973811846484465,
+      "loss": 1.034,
+      "step": 3044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028973121718162066,
+      "loss": 0.8137,
+      "step": 3045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028972431366081097,
+      "loss": 0.6275,
+      "step": 3046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028971740790252604,
+      "loss": 0.1889,
+      "step": 3047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028971049990687657,
+      "loss": 1.1624,
+      "step": 3048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002897035896739731,
+      "loss": 0.1554,
+      "step": 3049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002896966772039263,
+      "loss": 0.2187,
+      "step": 3050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028968976249684684,
+      "loss": 6.6514,
+      "step": 3051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028968284555284555,
+      "loss": 0.3444,
+      "step": 3052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028967592637203305,
+      "loss": 0.2976,
+      "step": 3053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028966900495452024,
+      "loss": 0.6929,
+      "step": 3054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028966208130041794,
+      "loss": 0.4317,
+      "step": 3055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028965515540983695,
+      "loss": 0.2994,
+      "step": 3056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002896482272828883,
+      "loss": 2.25,
+      "step": 3057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002896412969196829,
+      "loss": 0.3307,
+      "step": 3058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028963436432033165,
+      "loss": 0.5437,
+      "step": 3059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028962742948494564,
+      "loss": 0.0202,
+      "step": 3060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002896204924136359,
+      "loss": 0.001,
+      "step": 3061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002896135531065135,
+      "loss": 0.4305,
+      "step": 3062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002896066115636896,
+      "loss": 1.1666,
+      "step": 3063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895996677852753,
+      "loss": 0.4174,
+      "step": 3064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028959272177138184,
+      "loss": 0.3312,
+      "step": 3065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028958577352212046,
+      "loss": 0.7167,
+      "step": 3066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895788230376024,
+      "loss": 1.0622,
+      "step": 3067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028957187031793894,
+      "loss": 3.5701,
+      "step": 3068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895649153632415,
+      "loss": 0.8698,
+      "step": 3069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028955795817362136,
+      "loss": 0.3839,
+      "step": 3070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895509987491899,
+      "loss": 0.0012,
+      "step": 3071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028954403709005876,
+      "loss": 0.4878,
+      "step": 3072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895370731963392,
+      "loss": 1.0188,
+      "step": 3073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895301070681429,
+      "loss": 0.5564,
+      "step": 3074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895231387055813,
+      "loss": 0.5411,
+      "step": 3075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028951616810876603,
+      "loss": 0.4348,
+      "step": 3076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002895091952778087,
+      "loss": 0.408,
+      "step": 3077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028950222021282104,
+      "loss": 0.6566,
+      "step": 3078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028949524291391464,
+      "loss": 0.9367,
+      "step": 3079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002894882633812013,
+      "loss": 5.1926,
+      "step": 3080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028948128161479275,
+      "loss": 0.2961,
+      "step": 3081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028947429761480076,
+      "loss": 0.531,
+      "step": 3082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002894673113813373,
+      "loss": 0.6398,
+      "step": 3083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002894603229145141,
+      "loss": 0.0075,
+      "step": 3084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028945333221444314,
+      "loss": 0.4314,
+      "step": 3085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028944633928123636,
+      "loss": 2.7501,
+      "step": 3086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028943934411500576,
+      "loss": 1.1642,
+      "step": 3087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002894323467158633,
+      "loss": 1.1582,
+      "step": 3088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028942534708392107,
+      "loss": 2.1948,
+      "step": 3089
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002894183452192912,
+      "loss": 0.8927,
+      "step": 3090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002894113411220857,
+      "loss": 0.3329,
+      "step": 3091
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002894043347924168,
+      "loss": 1.1188,
+      "step": 3092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028939732623039675,
+      "loss": 0.7112,
+      "step": 3093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893903154361377,
+      "loss": 0.761,
+      "step": 3094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028938330240975196,
+      "loss": 0.1155,
+      "step": 3095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028937628715135184,
+      "loss": 0.1082,
+      "step": 3096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893692696610496,
+      "loss": 0.0623,
+      "step": 3097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893622499389577,
+      "loss": 0.0157,
+      "step": 3098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893552279851885,
+      "loss": 0.6298,
+      "step": 3099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028934820379985447,
+      "loss": 0.6629,
+      "step": 3100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028934117738306807,
+      "loss": 2.0375,
+      "step": 3101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893341487349419,
+      "loss": 0.7951,
+      "step": 3102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028932711785558844,
+      "loss": 0.4666,
+      "step": 3103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893200847451202,
+      "loss": 0.0292,
+      "step": 3104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893130494036499,
+      "loss": 2.7496,
+      "step": 3105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002893060118312903,
+      "loss": 0.3631,
+      "step": 3106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028929897202815394,
+      "loss": 0.8596,
+      "step": 3107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002892919299943536,
+      "loss": 0.2715,
+      "step": 3108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028928488573000204,
+      "loss": 0.5081,
+      "step": 3109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002892778392352121,
+      "loss": 0.6111,
+      "step": 3110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028927079051009656,
+      "loss": 0.0088,
+      "step": 3111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028926373955476834,
+      "loss": 0.5952,
+      "step": 3112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002892566863693403,
+      "loss": 0.1215,
+      "step": 3113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028924963095392544,
+      "loss": 0.7766,
+      "step": 3114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028924257330863683,
+      "loss": 0.9469,
+      "step": 3115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028923551343358725,
+      "loss": 0.061,
+      "step": 3116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028922845132888994,
+      "loss": 0.4277,
+      "step": 3117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002892213869946579,
+      "loss": 0.5685,
+      "step": 3118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002892143204310043,
+      "loss": 0.8966,
+      "step": 3119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002892072516380423,
+      "loss": 0.0575,
+      "step": 3120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028920018061588513,
+      "loss": 0.0124,
+      "step": 3121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891931073646459,
+      "loss": 0.1106,
+      "step": 3122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000289186031884438,
+      "loss": 0.6484,
+      "step": 3123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891789541753747,
+      "loss": 0.7255,
+      "step": 3124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891718742375693,
+      "loss": 0.4977,
+      "step": 3125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891647920711352,
+      "loss": 0.2894,
+      "step": 3126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891577076761858,
+      "loss": 2.1231,
+      "step": 3127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891506210528346,
+      "loss": 0.0032,
+      "step": 3128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000289143532201195,
+      "loss": 0.1613,
+      "step": 3129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891364411213806,
+      "loss": 1.4024,
+      "step": 3130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891293478135049,
+      "loss": 0.2116,
+      "step": 3131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028912225227768144,
+      "loss": 0.4568,
+      "step": 3132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028911515451402397,
+      "loss": 0.7067,
+      "step": 3133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028910805452264605,
+      "loss": 0.4694,
+      "step": 3134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002891009523036614,
+      "loss": 0.7889,
+      "step": 3135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002890938478571838,
+      "loss": 0.1517,
+      "step": 3136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028908674118332693,
+      "loss": 0.5504,
+      "step": 3137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002890796322822047,
+      "loss": 0.5454,
+      "step": 3138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028907252115393086,
+      "loss": 0.1451,
+      "step": 3139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002890654077986193,
+      "loss": 0.3295,
+      "step": 3140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002890582922163839,
+      "loss": 0.4959,
+      "step": 3141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028905117440733874,
+      "loss": 0.3678,
+      "step": 3142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028904405437159763,
+      "loss": 0.8208,
+      "step": 3143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002890369321092747,
+      "loss": 0.6408,
+      "step": 3144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000289029807620484,
+      "loss": 0.4105,
+      "step": 3145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002890226809053395,
+      "loss": 0.2085,
+      "step": 3146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002890155519639555,
+      "loss": 0.7684,
+      "step": 3147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000289008420796446,
+      "loss": 0.9252,
+      "step": 3148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028900128740292534,
+      "loss": 0.8446,
+      "step": 3149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002889941517835076,
+      "loss": 0.7054,
+      "step": 3150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028898701393830715,
+      "loss": 0.4193,
+      "step": 3151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028897987386743824,
+      "loss": 0.5485,
+      "step": 3152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028897273157101527,
+      "loss": 0.0086,
+      "step": 3153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028896558704915253,
+      "loss": 0.7933,
+      "step": 3154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002889584403019645,
+      "loss": 0.5166,
+      "step": 3155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002889512913295656,
+      "loss": 0.1461,
+      "step": 3156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002889441401320703,
+      "loss": 0.7665,
+      "step": 3157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002889369867095931,
+      "loss": 0.02,
+      "step": 3158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002889298310622486,
+      "loss": 1.2706,
+      "step": 3159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028892267319015135,
+      "loss": 1.4492,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000288915513093416,
+      "loss": 0.3956,
+      "step": 3161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002889083507721572,
+      "loss": 0.4451,
+      "step": 3162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028890118622648957,
+      "loss": 0.4761,
+      "step": 3163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028889401945652796,
+      "loss": 0.4134,
+      "step": 3164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000288886850462387,
+      "loss": 0.3646,
+      "step": 3165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028887967924418165,
+      "loss": 0.2121,
+      "step": 3166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002888725058020267,
+      "loss": 0.4993,
+      "step": 3167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028886533013603685,
+      "loss": 0.5327,
+      "step": 3168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028885815224632727,
+      "loss": 0.9227,
+      "step": 3169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028885097213301276,
+      "loss": 0.2107,
+      "step": 3170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028884378979620825,
+      "loss": 0.3218,
+      "step": 3171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002888366052360289,
+      "loss": 0.01,
+      "step": 3172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028882941845258963,
+      "loss": 0.3596,
+      "step": 3173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002888222294460056,
+      "loss": 0.1847,
+      "step": 3174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002888150382163919,
+      "loss": 0.4068,
+      "step": 3175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002888078447638637,
+      "loss": 2.1745,
+      "step": 3176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002888006490885362,
+      "loss": 0.3407,
+      "step": 3177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028879345119052464,
+      "loss": 0.4212,
+      "step": 3178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887862510699442,
+      "loss": 0.429,
+      "step": 3179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887790487269102,
+      "loss": 0.4952,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028877184416153815,
+      "loss": 0.7496,
+      "step": 3181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887646373739432,
+      "loss": 0.5347,
+      "step": 3182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028875742836424083,
+      "loss": 0.4059,
+      "step": 3183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887502171325465,
+      "loss": 0.1724,
+      "step": 3184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887430036789757,
+      "loss": 0.4237,
+      "step": 3185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887357880036439,
+      "loss": 0.5653,
+      "step": 3186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028872857010666666,
+      "loss": 0.4832,
+      "step": 3187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028872134998815963,
+      "loss": 1.4477,
+      "step": 3188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887141276482383,
+      "loss": 0.6316,
+      "step": 3189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002887069030870184,
+      "loss": 0.2408,
+      "step": 3190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028869967630461567,
+      "loss": 0.5575,
+      "step": 3191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002886924473011458,
+      "loss": 0.3371,
+      "step": 3192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028868521607672444,
+      "loss": 0.1188,
+      "step": 3193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028867798263146756,
+      "loss": 0.9227,
+      "step": 3194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002886707469654908,
+      "loss": 0.5838,
+      "step": 3195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028866350907891025,
+      "loss": 0.1321,
+      "step": 3196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002886562689718417,
+      "loss": 0.4928,
+      "step": 3197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028864902664440113,
+      "loss": 0.5588,
+      "step": 3198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028864178209670436,
+      "loss": 0.002,
+      "step": 3199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002886345353288676,
+      "loss": 3.4513,
+      "step": 3200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028862728634100685,
+      "loss": 0.5147,
+      "step": 3201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002886200351332381,
+      "loss": 0.2316,
+      "step": 3202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028861278170567756,
+      "loss": 0.0005,
+      "step": 3203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002886055260584413,
+      "loss": 0.2747,
+      "step": 3204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028859826819164564,
+      "loss": 0.0126,
+      "step": 3205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028859100810540663,
+      "loss": 3.129,
+      "step": 3206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002885837457998407,
+      "loss": 0.7262,
+      "step": 3207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000288576481275064,
+      "loss": 2.1168,
+      "step": 3208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028856921453119296,
+      "loss": 2.2462,
+      "step": 3209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028856194556834395,
+      "loss": 0.2795,
+      "step": 3210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002885546743866333,
+      "loss": 0.3482,
+      "step": 3211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002885474009861775,
+      "loss": 1.6671,
+      "step": 3212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028854012536709295,
+      "loss": 1.7521,
+      "step": 3213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002885328475294962,
+      "loss": 0.2476,
+      "step": 3214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002885255674735039,
+      "loss": 0.2298,
+      "step": 3215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002885182851992325,
+      "loss": 0.502,
+      "step": 3216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002885110007067986,
+      "loss": 0.8081,
+      "step": 3217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028850371399631895,
+      "loss": 0.0252,
+      "step": 3218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028849642506791013,
+      "loss": 0.481,
+      "step": 3219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028848913392168896,
+      "loss": 0.9902,
+      "step": 3220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884818405577722,
+      "loss": 1.0947,
+      "step": 3221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884745449762765,
+      "loss": 0.3267,
+      "step": 3222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028846724717731886,
+      "loss": 0.8801,
+      "step": 3223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028845994716101604,
+      "loss": 0.2168,
+      "step": 3224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884526449274849,
+      "loss": 0.7358,
+      "step": 3225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028844534047684254,
+      "loss": 1.3145,
+      "step": 3226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884380338092058,
+      "loss": 0.9678,
+      "step": 3227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028843072492469173,
+      "loss": 0.4736,
+      "step": 3228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884234138234173,
+      "loss": 0.2361,
+      "step": 3229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884161005054997,
+      "loss": 0.4887,
+      "step": 3230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884087849710559,
+      "loss": 0.2936,
+      "step": 3231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884014672202032,
+      "loss": 0.767,
+      "step": 3232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028839414725305874,
+      "loss": 0.6175,
+      "step": 3233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028838682506973965,
+      "loss": 0.3445,
+      "step": 3234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002883795006703633,
+      "loss": 0.1338,
+      "step": 3235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028837217405504686,
+      "loss": 0.5931,
+      "step": 3236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028836484522390773,
+      "loss": 0.6628,
+      "step": 3237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002883575141770633,
+      "loss": 0.4431,
+      "step": 3238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002883501809146309,
+      "loss": 0.9977,
+      "step": 3239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028834284543672803,
+      "loss": 1.4095,
+      "step": 3240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002883355077434721,
+      "loss": 0.1119,
+      "step": 3241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002883281678349806,
+      "loss": 1.3921,
+      "step": 3242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002883208257113711,
+      "loss": 2.2476,
+      "step": 3243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028831348137276115,
+      "loss": 0.5313,
+      "step": 3244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002883061348192684,
+      "loss": 0.8864,
+      "step": 3245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028829878605101045,
+      "loss": 0.1949,
+      "step": 3246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000288291435068105,
+      "loss": 0.3381,
+      "step": 3247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002882840818706698,
+      "loss": 0.7211,
+      "step": 3248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028827672645882253,
+      "loss": 0.2927,
+      "step": 3249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028826936883268105,
+      "loss": 0.5008,
+      "step": 3250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028826200899236313,
+      "loss": 0.3796,
+      "step": 3251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028825464693798663,
+      "loss": 0.5459,
+      "step": 3252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002882472826696694,
+      "loss": 0.3055,
+      "step": 3253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028823991618752946,
+      "loss": 0.5042,
+      "step": 3254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002882325474916847,
+      "loss": 0.9706,
+      "step": 3255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002882251765822532,
+      "loss": 0.6708,
+      "step": 3256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028821780345935293,
+      "loss": 0.564,
+      "step": 3257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028821042812310197,
+      "loss": 0.2427,
+      "step": 3258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028820305057361843,
+      "loss": 0.3675,
+      "step": 3259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028819567081102043,
+      "loss": 0.3778,
+      "step": 3260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028818828883542616,
+      "loss": 0.4025,
+      "step": 3261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881809046469539,
+      "loss": 0.5742,
+      "step": 3262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028817351824572175,
+      "loss": 0.519,
+      "step": 3263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028816612963184814,
+      "loss": 0.3622,
+      "step": 3264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881587388054513,
+      "loss": 0.1245,
+      "step": 3265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881513457666496,
+      "loss": 0.4583,
+      "step": 3266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881439505155614,
+      "loss": 0.6157,
+      "step": 3267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881365530523052,
+      "loss": 1.5013,
+      "step": 3268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028812915337699945,
+      "loss": 0.4184,
+      "step": 3269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881217514897625,
+      "loss": 0.064,
+      "step": 3270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881143473907131,
+      "loss": 0.166,
+      "step": 3271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002881069410799697,
+      "loss": 0.4742,
+      "step": 3272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028809953255765086,
+      "loss": 0.4448,
+      "step": 3273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028809212182387525,
+      "loss": 1.1866,
+      "step": 3274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002880847088787616,
+      "loss": 0.55,
+      "step": 3275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028807729372242857,
+      "loss": 1.667,
+      "step": 3276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002880698763549949,
+      "loss": 0.669,
+      "step": 3277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028806245677657935,
+      "loss": 0.6921,
+      "step": 3278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028805503498730077,
+      "loss": 0.4814,
+      "step": 3279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028804761098727807,
+      "loss": 0.067,
+      "step": 3280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028804018477662996,
+      "loss": 0.2714,
+      "step": 3281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002880327563554755,
+      "loss": 0.0265,
+      "step": 3282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002880253257239336,
+      "loss": 0.3783,
+      "step": 3283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002880178928821232,
+      "loss": 0.0872,
+      "step": 3284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002880104578301635,
+      "loss": 0.8803,
+      "step": 3285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028800302056817336,
+      "loss": 0.561,
+      "step": 3286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028799558109627194,
+      "loss": 0.5218,
+      "step": 3287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028798813941457844,
+      "loss": 0.2249,
+      "step": 3288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287980695523212,
+      "loss": 0.0717,
+      "step": 3289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028797324942229176,
+      "loss": 2.4117,
+      "step": 3290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287965801111937,
+      "loss": 0.3477,
+      "step": 3291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287958350592267,
+      "loss": 1.4316,
+      "step": 3292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028795089786340107,
+      "loss": 0.2983,
+      "step": 3293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028794344292545855,
+      "loss": 0.2698,
+      "step": 3294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002879359857785588,
+      "loss": 0.0191,
+      "step": 3295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002879285264228212,
+      "loss": 0.0139,
+      "step": 3296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002879210648583654,
+      "loss": 0.7425,
+      "step": 3297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002879136010853106,
+      "loss": 0.0274,
+      "step": 3298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002879061351037765,
+      "loss": 0.23,
+      "step": 3299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878986669138826,
+      "loss": 0.7008,
+      "step": 3300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878911965157485,
+      "loss": 0.3203,
+      "step": 3301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878837239094939,
+      "loss": 0.8218,
+      "step": 3302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878762490952383,
+      "loss": 0.477,
+      "step": 3303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028786877207310157,
+      "loss": 1.3941,
+      "step": 3304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028786129284320334,
+      "loss": 0.6598,
+      "step": 3305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028785381140566344,
+      "loss": 0.2398,
+      "step": 3306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878463277606016,
+      "loss": 0.5775,
+      "step": 3307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878388419081377,
+      "loss": 2.4318,
+      "step": 3308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878313538483916,
+      "loss": 0.5031,
+      "step": 3309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028782386358148327,
+      "loss": 0.7223,
+      "step": 3310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878163711075326,
+      "loss": 0.306,
+      "step": 3311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028780887642665956,
+      "loss": 0.8527,
+      "step": 3312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002878013795389842,
+      "loss": 0.4233,
+      "step": 3313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028779388044462655,
+      "loss": 0.6732,
+      "step": 3314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002877863791437067,
+      "loss": 0.2712,
+      "step": 3315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028777887563634477,
+      "loss": 0.5593,
+      "step": 3316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028777136992266095,
+      "loss": 0.929,
+      "step": 3317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002877638620027754,
+      "loss": 0.8708,
+      "step": 3318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002877563518768083,
+      "loss": 0.2016,
+      "step": 3319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028774883954488006,
+      "loss": 0.0155,
+      "step": 3320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028774132500711085,
+      "loss": 0.2918,
+      "step": 3321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287733808263621,
+      "loss": 0.332,
+      "step": 3322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028772628931453097,
+      "loss": 0.0269,
+      "step": 3323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002877187681599611,
+      "loss": 0.5706,
+      "step": 3324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002877112448000318,
+      "loss": 0.4927,
+      "step": 3325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002877037192348636,
+      "loss": 0.0095,
+      "step": 3326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876961914645771,
+      "loss": 0.0144,
+      "step": 3327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876886614892927,
+      "loss": 1.3695,
+      "step": 3328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876811293091309,
+      "loss": 0.5251,
+      "step": 3329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876735949242126,
+      "loss": 0.5272,
+      "step": 3330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876660583346583,
+      "loss": 0.8878,
+      "step": 3331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876585195405886,
+      "loss": 0.4554,
+      "step": 3332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876509785421243,
+      "loss": 0.3405,
+      "step": 3333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028764343533938625,
+      "loss": 0.1924,
+      "step": 3334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028763588993249505,
+      "loss": 0.0155,
+      "step": 3335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028762834232157176,
+      "loss": 0.3429,
+      "step": 3336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028762079250673703,
+      "loss": 0.7408,
+      "step": 3337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028761324048811183,
+      "loss": 0.0024,
+      "step": 3338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876056862658172,
+      "loss": 1.6212,
+      "step": 3339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028759812983997397,
+      "loss": 0.3311,
+      "step": 3340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002875905712107032,
+      "loss": 0.5793,
+      "step": 3341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028758301037812593,
+      "loss": 0.3985,
+      "step": 3342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002875754473423632,
+      "loss": 0.5095,
+      "step": 3343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028756788210353623,
+      "loss": 0.578,
+      "step": 3344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287560314661766,
+      "loss": 2.0177,
+      "step": 3345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002875527450171738,
+      "loss": 1.614,
+      "step": 3346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002875451731698809,
+      "loss": 0.2675,
+      "step": 3347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002875375991200084,
+      "loss": 1.0604,
+      "step": 3348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002875300228676777,
+      "loss": 0.1997,
+      "step": 3349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028752244441301004,
+      "loss": 0.248,
+      "step": 3350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028751486375612684,
+      "loss": 0.4078,
+      "step": 3351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028750728089714956,
+      "loss": 0.0068,
+      "step": 3352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028749969583619946,
+      "loss": 0.8588,
+      "step": 3353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002874921085733981,
+      "loss": 0.0444,
+      "step": 3354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287484519108867,
+      "loss": 0.1252,
+      "step": 3355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028747692744272765,
+      "loss": 0.6066,
+      "step": 3356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002874693335751016,
+      "loss": 0.0231,
+      "step": 3357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002874617375061105,
+      "loss": 1.9157,
+      "step": 3358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287454139235876,
+      "loss": 1.3243,
+      "step": 3359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002874465387645197,
+      "loss": 0.5614,
+      "step": 3360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028743893609216347,
+      "loss": 0.012,
+      "step": 3361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028743133121892885,
+      "loss": 0.733,
+      "step": 3362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028742372414493773,
+      "loss": 2.0274,
+      "step": 3363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002874161148703119,
+      "loss": 4.109,
+      "step": 3364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002874085033951733,
+      "loss": 0.3453,
+      "step": 3365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028740088971964366,
+      "loss": 0.0025,
+      "step": 3366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000287393273843845,
+      "loss": 0.5288,
+      "step": 3367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028738565576789927,
+      "loss": 0.4556,
+      "step": 3368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028737803549192843,
+      "loss": 1.181,
+      "step": 3369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002873704130160545,
+      "loss": 0.0452,
+      "step": 3370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002873627883403996,
+      "loss": 0.2214,
+      "step": 3371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002873551614650858,
+      "loss": 0.5592,
+      "step": 3372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002873475323902353,
+      "loss": 0.0044,
+      "step": 3373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028733990111597015,
+      "loss": 2.9592,
+      "step": 3374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002873322676424125,
+      "loss": 0.2757,
+      "step": 3375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028732463196968485,
+      "loss": 0.4214,
+      "step": 3376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002873169940979092,
+      "loss": 0.7574,
+      "step": 3377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028730935402720807,
+      "loss": 0.4462,
+      "step": 3378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002873017117577036,
+      "loss": 0.3293,
+      "step": 3379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028729406728951836,
+      "loss": 0.4389,
+      "step": 3380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872864206227746,
+      "loss": 0.6759,
+      "step": 3381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872787717575949,
+      "loss": 0.9859,
+      "step": 3382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872711206941017,
+      "loss": 0.0189,
+      "step": 3383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872634674324176,
+      "loss": 0.2962,
+      "step": 3384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872558119726649,
+      "loss": 0.6803,
+      "step": 3385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872481543149665,
+      "loss": 0.8611,
+      "step": 3386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872404944594448,
+      "loss": 0.4107,
+      "step": 3387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872328324062226,
+      "loss": 0.2605,
+      "step": 3388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872251681554226,
+      "loss": 0.9027,
+      "step": 3389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872175017071674,
+      "loss": 0.0042,
+      "step": 3390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002872098330615798,
+      "loss": 0.0042,
+      "step": 3391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028720216221878275,
+      "loss": 0.3481,
+      "step": 3392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028719448917889894,
+      "loss": 1.3039,
+      "step": 3393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028718681394205133,
+      "loss": 0.5327,
+      "step": 3394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002871791365083627,
+      "loss": 0.0213,
+      "step": 3395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002871714568779561,
+      "loss": 0.2536,
+      "step": 3396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002871637750509545,
+      "loss": 0.7808,
+      "step": 3397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002871560910274809,
+      "loss": 0.0023,
+      "step": 3398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028714840480765835,
+      "loss": 0.1971,
+      "step": 3399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028714071639160995,
+      "loss": 0.6374,
+      "step": 3400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028713302577945877,
+      "loss": 0.3377,
+      "step": 3401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028712533297132797,
+      "loss": 0.0112,
+      "step": 3402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002871176379673408,
+      "loss": 0.3465,
+      "step": 3403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002871099407676204,
+      "loss": 0.0117,
+      "step": 3404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002871022413722901,
+      "loss": 2.3986,
+      "step": 3405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028709453978147315,
+      "loss": 0.0197,
+      "step": 3406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002870868359952929,
+      "loss": 1.1365,
+      "step": 3407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028707913001387275,
+      "loss": 0.254,
+      "step": 3408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000287071421837336,
+      "loss": 0.009,
+      "step": 3409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028706371146580614,
+      "loss": 0.6086,
+      "step": 3410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002870559988994067,
+      "loss": 0.3411,
+      "step": 3411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002870482841382611,
+      "loss": 0.2377,
+      "step": 3412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002870405671824929,
+      "loss": 3.1681,
+      "step": 3413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002870328480322257,
+      "loss": 0.4563,
+      "step": 3414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028702512668758306,
+      "loss": 0.2908,
+      "step": 3415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028701740314868873,
+      "loss": 0.4455,
+      "step": 3416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002870096774156663,
+      "loss": 0.3911,
+      "step": 3417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028700194948863946,
+      "loss": 0.336,
+      "step": 3418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002869942193677321,
+      "loss": 0.3945,
+      "step": 3419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028698648705306783,
+      "loss": 0.8353,
+      "step": 3420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028697875254477055,
+      "loss": 0.5819,
+      "step": 3421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002869710158429642,
+      "loss": 0.336,
+      "step": 3422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028696327694777255,
+      "loss": 0.8992,
+      "step": 3423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028695553585931957,
+      "loss": 0.5391,
+      "step": 3424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028694779257772927,
+      "loss": 0.549,
+      "step": 3425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028694004710312556,
+      "loss": 0.5265,
+      "step": 3426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028693229943563254,
+      "loss": 0.4261,
+      "step": 3427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028692454957537423,
+      "loss": 0.3091,
+      "step": 3428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002869167975224747,
+      "loss": 0.789,
+      "step": 3429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002869090432770583,
+      "loss": 0.2855,
+      "step": 3430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002869012868392489,
+      "loss": 2.3675,
+      "step": 3431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028689352820917093,
+      "loss": 0.1859,
+      "step": 3432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002868857673869485,
+      "loss": 0.3175,
+      "step": 3433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028687800437270593,
+      "loss": 2.3094,
+      "step": 3434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028687023916656757,
+      "loss": 0.0125,
+      "step": 3435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028686247176865776,
+      "loss": 0.9028,
+      "step": 3436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028685470217910085,
+      "loss": 0.2157,
+      "step": 3437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028684693039802126,
+      "loss": 0.0076,
+      "step": 3438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002868391564255435,
+      "loss": 0.4915,
+      "step": 3439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028683138026179203,
+      "loss": 0.3953,
+      "step": 3440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002868236019068913,
+      "loss": 0.5489,
+      "step": 3441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028681582136096595,
+      "loss": 0.7711,
+      "step": 3442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028680803862414056,
+      "loss": 0.5793,
+      "step": 3443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002868002536965397,
+      "loss": 0.7162,
+      "step": 3444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028679246657828817,
+      "loss": 1.7203,
+      "step": 3445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028678467726951054,
+      "loss": 1.0123,
+      "step": 3446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028677688577033163,
+      "loss": 0.6052,
+      "step": 3447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028676909208087616,
+      "loss": 1.3425,
+      "step": 3448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002867612962012689,
+      "loss": 0.3425,
+      "step": 3449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028675349813163473,
+      "loss": 0.825,
+      "step": 3450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002867456978720985,
+      "loss": 0.7634,
+      "step": 3451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028673789542278525,
+      "loss": 0.3686,
+      "step": 3452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028673009078381977,
+      "loss": 1.0143,
+      "step": 3453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000286722283955327,
+      "loss": 0.2612,
+      "step": 3454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002867144749374322,
+      "loss": 0.7032,
+      "step": 3455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028670666373026017,
+      "loss": 0.5906,
+      "step": 3456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002866988503339361,
+      "loss": 0.3826,
+      "step": 3457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002866910347485851,
+      "loss": 0.3805,
+      "step": 3458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002866832169743323,
+      "loss": 0.6455,
+      "step": 3459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028667539701130297,
+      "loss": 3.9404,
+      "step": 3460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028666757485962225,
+      "loss": 1.531,
+      "step": 3461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002866597505194154,
+      "loss": 0.2142,
+      "step": 3462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028665192399080777,
+      "loss": 0.7215,
+      "step": 3463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028664409527392467,
+      "loss": 0.493,
+      "step": 3464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028663626436889147,
+      "loss": 0.3323,
+      "step": 3465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002866284312758335,
+      "loss": 0.0069,
+      "step": 3466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002866205959948763,
+      "loss": 0.5897,
+      "step": 3467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028661275852614527,
+      "loss": 0.4736,
+      "step": 3468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028660491886976596,
+      "loss": 0.4896,
+      "step": 3469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002865970770258639,
+      "loss": 1.2272,
+      "step": 3470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028658923299456463,
+      "loss": 0.3456,
+      "step": 3471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002865813867759938,
+      "loss": 0.7517,
+      "step": 3472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028657353837027705,
+      "loss": 0.6229,
+      "step": 3473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028656568777754,
+      "loss": 0.5297,
+      "step": 3474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028655783499790844,
+      "loss": 0.4587,
+      "step": 3475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028654998003150813,
+      "loss": 1.8559,
+      "step": 3476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002865421228784648,
+      "loss": 0.6183,
+      "step": 3477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002865342635389043,
+      "loss": 0.912,
+      "step": 3478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002865264020129525,
+      "loss": 0.4947,
+      "step": 3479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028651853830073524,
+      "loss": 0.8186,
+      "step": 3480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002865106724023784,
+      "loss": 0.3694,
+      "step": 3481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028650280431800814,
+      "loss": 0.2892,
+      "step": 3482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864949340477503,
+      "loss": 4.106,
+      "step": 3483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028648706159173094,
+      "loss": 0.3842,
+      "step": 3484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864791869500761,
+      "loss": 0.2068,
+      "step": 3485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864713101229119,
+      "loss": 0.4897,
+      "step": 3486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864634311103645,
+      "loss": 3.6938,
+      "step": 3487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028645554991256006,
+      "loss": 0.0163,
+      "step": 3488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864476665296248,
+      "loss": 1.2725,
+      "step": 3489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864397809616849,
+      "loss": 0.6327,
+      "step": 3490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864318932088667,
+      "loss": 1.254,
+      "step": 3491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028642400327129645,
+      "loss": 0.2348,
+      "step": 3492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028641611114910056,
+      "loss": 0.4009,
+      "step": 3493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864082168424054,
+      "loss": 0.2209,
+      "step": 3494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028640032035133733,
+      "loss": 0.3194,
+      "step": 3495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002863924216760229,
+      "loss": 0.8821,
+      "step": 3496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028638452081658847,
+      "loss": 0.3167,
+      "step": 3497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002863766177731606,
+      "loss": 1.3509,
+      "step": 3498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028636871254586596,
+      "loss": 1.2844,
+      "step": 3499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000286360805134831,
+      "loss": 0.8201,
+      "step": 3500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028635289554018235,
+      "loss": 0.3622,
+      "step": 3501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002863449837620468,
+      "loss": 0.2047,
+      "step": 3502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002863370698005509,
+      "loss": 1.0408,
+      "step": 3503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028632915365582145,
+      "loss": 0.6501,
+      "step": 3504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028632123532798526,
+      "loss": 0.0761,
+      "step": 3505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000286313314817169,
+      "loss": 0.3956,
+      "step": 3506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002863053921234996,
+      "loss": 0.3704,
+      "step": 3507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028629746724710394,
+      "loss": 0.4581,
+      "step": 3508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002862895401881089,
+      "loss": 0.2292,
+      "step": 3509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028628161094664136,
+      "loss": 0.4931,
+      "step": 3510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002862736795228284,
+      "loss": 0.5352,
+      "step": 3511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000286265745916797,
+      "loss": 0.019,
+      "step": 3512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028625781012867415,
+      "loss": 0.9202,
+      "step": 3513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028624987215858697,
+      "loss": 0.6549,
+      "step": 3514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028624193200666256,
+      "loss": 3.2373,
+      "step": 3515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002862339896730281,
+      "loss": 0.8273,
+      "step": 3516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002862260451578107,
+      "loss": 0.538,
+      "step": 3517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002862180984611377,
+      "loss": 0.1792,
+      "step": 3518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028621014958313625,
+      "loss": 1.946,
+      "step": 3519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028620219852393367,
+      "loss": 1.4633,
+      "step": 3520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002861942452836573,
+      "loss": 0.5146,
+      "step": 3521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002861862898624345,
+      "loss": 1.1917,
+      "step": 3522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002861783322603927,
+      "loss": 0.63,
+      "step": 3523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002861703724776592,
+      "loss": 0.126,
+      "step": 3524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002861624105143616,
+      "loss": 0.7556,
+      "step": 3525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002861544463706273,
+      "loss": 0.4045,
+      "step": 3526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028614648004658395,
+      "loss": 0.2249,
+      "step": 3527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028613851154235905,
+      "loss": 0.5386,
+      "step": 3528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028613054085808016,
+      "loss": 0.7869,
+      "step": 3529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028612256799387495,
+      "loss": 0.5024,
+      "step": 3530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028611459294987113,
+      "loss": 0.8106,
+      "step": 3531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002861066157261964,
+      "loss": 0.6646,
+      "step": 3532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028609863632297845,
+      "loss": 0.5082,
+      "step": 3533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028609065474034513,
+      "loss": 0.6754,
+      "step": 3534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028608267097842414,
+      "loss": 0.5459,
+      "step": 3535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002860746850373435,
+      "loss": 0.2801,
+      "step": 3536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002860666969172309,
+      "loss": 0.3741,
+      "step": 3537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028605870661821447,
+      "loss": 0.4803,
+      "step": 3538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028605071414042197,
+      "loss": 0.3651,
+      "step": 3539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028604271948398153,
+      "loss": 0.8864,
+      "step": 3540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028603472264902103,
+      "loss": 0.342,
+      "step": 3541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002860267236356687,
+      "loss": 0.5878,
+      "step": 3542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028601872244405247,
+      "loss": 0.5592,
+      "step": 3543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028601071907430055,
+      "loss": 0.4243,
+      "step": 3544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028600271352654106,
+      "loss": 4.0689,
+      "step": 3545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859947058009023,
+      "loss": 0.8625,
+      "step": 3546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028598669589751237,
+      "loss": 0.3842,
+      "step": 3547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859786838164996,
+      "loss": 0.2096,
+      "step": 3548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028597066955799225,
+      "loss": 0.4136,
+      "step": 3549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028596265312211874,
+      "loss": 0.6745,
+      "step": 3550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859546345090074,
+      "loss": 0.1574,
+      "step": 3551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859466137187866,
+      "loss": 0.6172,
+      "step": 3552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859385907515848,
+      "loss": 0.5135,
+      "step": 3553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028593056560753053,
+      "loss": 0.7359,
+      "step": 3554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859225382867522,
+      "loss": 0.3809,
+      "step": 3555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859145087893784,
+      "loss": 0.4328,
+      "step": 3556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002859064771155378,
+      "loss": 1.2179,
+      "step": 3557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028589844326535886,
+      "loss": 0.9037,
+      "step": 3558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028589040723897037,
+      "loss": 0.3571,
+      "step": 3559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002858823690365009,
+      "loss": 0.2284,
+      "step": 3560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028587432865807925,
+      "loss": 0.938,
+      "step": 3561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002858662861038341,
+      "loss": 0.1851,
+      "step": 3562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002858582413738944,
+      "loss": 2.1646,
+      "step": 3563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028585019446838875,
+      "loss": 2.0679,
+      "step": 3564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002858421453874462,
+      "loss": 0.6205,
+      "step": 3565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028583409413119546,
+      "loss": 0.9392,
+      "step": 3566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002858260406997656,
+      "loss": 0.1014,
+      "step": 3567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028581798509328554,
+      "loss": 0.4444,
+      "step": 3568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028580992731188433,
+      "loss": 0.1879,
+      "step": 3569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028580186735569093,
+      "loss": 0.8931,
+      "step": 3570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857938052248344,
+      "loss": 0.0751,
+      "step": 3571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857857409194439,
+      "loss": 0.6503,
+      "step": 3572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857776744396486,
+      "loss": 0.9782,
+      "step": 3573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857696057855775,
+      "loss": 0.0689,
+      "step": 3574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028576153495736003,
+      "loss": 0.2574,
+      "step": 3575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857534619551253,
+      "loss": 0.8925,
+      "step": 3576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857453867790026,
+      "loss": 0.3454,
+      "step": 3577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028573730942912126,
+      "loss": 0.3887,
+      "step": 3578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028572922990561064,
+      "loss": 0.4318,
+      "step": 3579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857211482086001,
+      "loss": 0.9379,
+      "step": 3580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000285713064338219,
+      "loss": 0.9601,
+      "step": 3581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028570497829459693,
+      "loss": 0.5525,
+      "step": 3582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002856968900778633,
+      "loss": 0.7546,
+      "step": 3583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028568879968814755,
+      "loss": 2.0454,
+      "step": 3584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028568070712557935,
+      "loss": 0.4037,
+      "step": 3585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028567261239028827,
+      "loss": 0.5389,
+      "step": 3586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002856645154824039,
+      "loss": 0.506,
+      "step": 3587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002856564164020559,
+      "loss": 0.4321,
+      "step": 3588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028564831514937405,
+      "loss": 0.6231,
+      "step": 3589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028564021172448797,
+      "loss": 0.9701,
+      "step": 3590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028563210612752747,
+      "loss": 0.4565,
+      "step": 3591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002856239983586224,
+      "loss": 0.1245,
+      "step": 3592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002856158884179024,
+      "loss": 0.377,
+      "step": 3593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002856077763054976,
+      "loss": 0.9277,
+      "step": 3594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028559966202153773,
+      "loss": 0.4197,
+      "step": 3595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002855915455661528,
+      "loss": 0.6178,
+      "step": 3596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028558342693947277,
+      "loss": 0.008,
+      "step": 3597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002855753061416276,
+      "loss": 1.759,
+      "step": 3598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002855671831727474,
+      "loss": 0.4555,
+      "step": 3599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028555905803296223,
+      "loss": 1.057,
+      "step": 3600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028555093072240217,
+      "loss": 0.4822,
+      "step": 3601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002855428012411974,
+      "loss": 0.3562,
+      "step": 3602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028553466958947815,
+      "loss": 0.3969,
+      "step": 3603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028552653576737445,
+      "loss": 1.1281,
+      "step": 3604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028551839977501676,
+      "loss": 0.0119,
+      "step": 3605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002855102616125353,
+      "loss": 0.4535,
+      "step": 3606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002855021212800603,
+      "loss": 0.6482,
+      "step": 3607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002854939787777222,
+      "loss": 0.2514,
+      "step": 3608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028548583410565145,
+      "loss": 0.4342,
+      "step": 3609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028547768726397833,
+      "loss": 0.2888,
+      "step": 3610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028546953825283336,
+      "loss": 0.0227,
+      "step": 3611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002854613870723471,
+      "loss": 0.2697,
+      "step": 3612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028545323372265,
+      "loss": 0.2405,
+      "step": 3613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028544507820387263,
+      "loss": 0.6486,
+      "step": 3614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028543692051614564,
+      "loss": 0.8433,
+      "step": 3615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028542876065959966,
+      "loss": 0.3777,
+      "step": 3616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028542059863436527,
+      "loss": 0.0132,
+      "step": 3617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002854124344405732,
+      "loss": 0.522,
+      "step": 3618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028540426807835434,
+      "loss": 0.7425,
+      "step": 3619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028539609954783925,
+      "loss": 0.367,
+      "step": 3620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028538792884915884,
+      "loss": 0.6717,
+      "step": 3621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028537975598244396,
+      "loss": 0.5022,
+      "step": 3622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028537158094782547,
+      "loss": 1.7339,
+      "step": 3623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028536340374543423,
+      "loss": 1.2409,
+      "step": 3624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028535522437540123,
+      "loss": 0.6108,
+      "step": 3625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002853470428378575,
+      "loss": 0.3327,
+      "step": 3626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028533885913293395,
+      "loss": 0.4287,
+      "step": 3627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028533067326076174,
+      "loss": 0.306,
+      "step": 3628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028532248522147185,
+      "loss": 0.6526,
+      "step": 3629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002853142950151955,
+      "loss": 0.9243,
+      "step": 3630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002853061026420638,
+      "loss": 1.0839,
+      "step": 3631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028529790810220785,
+      "loss": 0.3178,
+      "step": 3632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000285289711395759,
+      "loss": 0.4686,
+      "step": 3633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002852815125228485,
+      "loss": 0.3977,
+      "step": 3634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002852733114836076,
+      "loss": 0.6751,
+      "step": 3635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002852651082781676,
+      "loss": 0.2635,
+      "step": 3636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028525690290665993,
+      "loss": 0.0178,
+      "step": 3637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002852486953692159,
+      "loss": 0.4116,
+      "step": 3638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028524048566596705,
+      "loss": 0.0023,
+      "step": 3639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002852322737970448,
+      "loss": 0.2307,
+      "step": 3640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028522405976258056,
+      "loss": 0.3647,
+      "step": 3641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000285215843562706,
+      "loss": 0.206,
+      "step": 3642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028520762519755263,
+      "loss": 2.0139,
+      "step": 3643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028519940466725205,
+      "loss": 0.513,
+      "step": 3644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028519118197193596,
+      "loss": 0.6198,
+      "step": 3645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002851829571117359,
+      "loss": 2.4242,
+      "step": 3646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002851747300867837,
+      "loss": 0.2906,
+      "step": 3647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002851665008972111,
+      "loss": 0.4993,
+      "step": 3648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028515826954314977,
+      "loss": 0.4166,
+      "step": 3649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028515003602473167,
+      "loss": 0.0241,
+      "step": 3650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028514180034208854,
+      "loss": 0.3588,
+      "step": 3651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002851335624953523,
+      "loss": 0.4358,
+      "step": 3652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028512532248465487,
+      "loss": 0.9162,
+      "step": 3653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002851170803101282,
+      "loss": 0.7644,
+      "step": 3654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028510883597190423,
+      "loss": 0.4105,
+      "step": 3655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028510058947011507,
+      "loss": 0.1203,
+      "step": 3656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028509234080489274,
+      "loss": 0.0325,
+      "step": 3657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028508408997636925,
+      "loss": 0.6934,
+      "step": 3658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002850758369846769,
+      "loss": 0.6503,
+      "step": 3659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002850675818299476,
+      "loss": 0.7976,
+      "step": 3660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002850593245123138,
+      "loss": 0.4508,
+      "step": 3661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028505106503190757,
+      "loss": 0.0024,
+      "step": 3662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002850428033888612,
+      "loss": 0.4479,
+      "step": 3663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000285034539583307,
+      "loss": 0.1874,
+      "step": 3664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028502627361537734,
+      "loss": 0.01,
+      "step": 3665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002850180054852046,
+      "loss": 0.2426,
+      "step": 3666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028500973519292105,
+      "loss": 0.004,
+      "step": 3667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028500146273865924,
+      "loss": 0.5028,
+      "step": 3668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028499318812255165,
+      "loss": 0.3234,
+      "step": 3669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028498491134473065,
+      "loss": 0.2297,
+      "step": 3670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000284976632405329,
+      "loss": 0.5333,
+      "step": 3671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002849683513044791,
+      "loss": 0.5568,
+      "step": 3672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002849600680423136,
+      "loss": 0.962,
+      "step": 3673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028495178261896516,
+      "loss": 0.9128,
+      "step": 3674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028494349503456647,
+      "loss": 0.4102,
+      "step": 3675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002849352052892502,
+      "loss": 0.3746,
+      "step": 3676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028492691338314913,
+      "loss": 0.5391,
+      "step": 3677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028491861931639607,
+      "loss": 2.8051,
+      "step": 3678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028491032308912385,
+      "loss": 4.7542,
+      "step": 3679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002849020247014652,
+      "loss": 1.0177,
+      "step": 3680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002848937241535531,
+      "loss": 0.4718,
+      "step": 3681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028488542144552043,
+      "loss": 0.7812,
+      "step": 3682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002848771165775002,
+      "loss": 1.125,
+      "step": 3683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002848688095496254,
+      "loss": 0.322,
+      "step": 3684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000284860500362029,
+      "loss": 0.3227,
+      "step": 3685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002848521890148441,
+      "loss": 0.363,
+      "step": 3686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028484387550820376,
+      "loss": 1.6935,
+      "step": 3687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002848355598422411,
+      "loss": 0.9436,
+      "step": 3688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028482724201708934,
+      "loss": 2.2755,
+      "step": 3689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028481892203288163,
+      "loss": 0.0053,
+      "step": 3690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028481059988975124,
+      "loss": 0.0056,
+      "step": 3691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002848022755878314,
+      "loss": 1.4552,
+      "step": 3692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002847939491272554,
+      "loss": 0.4262,
+      "step": 3693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028478562050815665,
+      "loss": 0.488,
+      "step": 3694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028477728973066844,
+      "loss": 0.1671,
+      "step": 3695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002847689567949242,
+      "loss": 0.2465,
+      "step": 3696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028476062170105745,
+      "loss": 0.3162,
+      "step": 3697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028475228444920154,
+      "loss": 0.3165,
+      "step": 3698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028474394503949003,
+      "loss": 0.2957,
+      "step": 3699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028473560347205644,
+      "loss": 0.8295,
+      "step": 3700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002847272597470344,
+      "loss": 0.7106,
+      "step": 3701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002847189138645575,
+      "loss": 0.7689,
+      "step": 3702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002847105658247593,
+      "loss": 0.5874,
+      "step": 3703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002847022156277736,
+      "loss": 0.6538,
+      "step": 3704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028469386327373416,
+      "loss": 0.5386,
+      "step": 3705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028468550876277455,
+      "loss": 0.5167,
+      "step": 3706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002846771520950287,
+      "loss": 0.7555,
+      "step": 3707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028466879327063036,
+      "loss": 0.1263,
+      "step": 3708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002846604322897134,
+      "loss": 0.7866,
+      "step": 3709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028465206915241175,
+      "loss": 1.6987,
+      "step": 3710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002846437038588593,
+      "loss": 0.4501,
+      "step": 3711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028463533640918994,
+      "loss": 0.12,
+      "step": 3712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002846269668035378,
+      "loss": 0.9733,
+      "step": 3713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002846185950420368,
+      "loss": 0.4929,
+      "step": 3714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000284610221124821,
+      "loss": 0.4276,
+      "step": 3715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028460184505202455,
+      "loss": 0.0078,
+      "step": 3716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002845934668237816,
+      "loss": 1.5462,
+      "step": 3717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002845850864402263,
+      "loss": 0.3844,
+      "step": 3718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028457670390149274,
+      "loss": 0.461,
+      "step": 3719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028456831920771524,
+      "loss": 0.831,
+      "step": 3720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028455993235902813,
+      "loss": 0.7782,
+      "step": 3721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028455154335556564,
+      "loss": 0.2744,
+      "step": 3722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028454315219746205,
+      "loss": 0.7242,
+      "step": 3723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028453475888485185,
+      "loss": 0.6372,
+      "step": 3724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002845263634178694,
+      "loss": 0.4257,
+      "step": 3725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028451796579664917,
+      "loss": 0.0071,
+      "step": 3726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002845095660213256,
+      "loss": 0.2734,
+      "step": 3727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002845011640920331,
+      "loss": 1.7249,
+      "step": 3728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028449276000890635,
+      "loss": 0.0042,
+      "step": 3729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028448435377207993,
+      "loss": 0.3845,
+      "step": 3730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028447594538168837,
+      "loss": 0.4642,
+      "step": 3731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028446753483786643,
+      "loss": 0.5409,
+      "step": 3732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002844591221407487,
+      "loss": 0.4596,
+      "step": 3733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002844507072904699,
+      "loss": 0.0179,
+      "step": 3734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002844422902871648,
+      "loss": 0.427,
+      "step": 3735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028443387113096826,
+      "loss": 0.5443,
+      "step": 3736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028442544982201497,
+      "loss": 0.8151,
+      "step": 3737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028441702636043984,
+      "loss": 0.0022,
+      "step": 3738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028440860074637777,
+      "loss": 0.992,
+      "step": 3739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002844001729799637,
+      "loss": 1.0488,
+      "step": 3740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843917430613326,
+      "loss": 0.4679,
+      "step": 3741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028438331099061936,
+      "loss": 0.4786,
+      "step": 3742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843748767679591,
+      "loss": 0.6736,
+      "step": 3743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843664403934869,
+      "loss": 0.6597,
+      "step": 3744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843580018673378,
+      "loss": 0.432,
+      "step": 3745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843495611896469,
+      "loss": 0.3286,
+      "step": 3746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843411183605494,
+      "loss": 0.5017,
+      "step": 3747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028433267338018054,
+      "loss": 2.6801,
+      "step": 3748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843242262486755,
+      "loss": 0.5735,
+      "step": 3749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002843157769661696,
+      "loss": 0.2785,
+      "step": 3750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028430732553279807,
+      "loss": 0.0078,
+      "step": 3751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028429887194869635,
+      "loss": 0.4213,
+      "step": 3752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842904162139997,
+      "loss": 0.2119,
+      "step": 3753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842819583288436,
+      "loss": 3.2309,
+      "step": 3754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842734982933634,
+      "loss": 0.1309,
+      "step": 3755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842650361076947,
+      "loss": 0.4571,
+      "step": 3756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028425657177197296,
+      "loss": 0.9511,
+      "step": 3757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842481052863337,
+      "loss": 0.2949,
+      "step": 3758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842396366509126,
+      "loss": 0.4608,
+      "step": 3759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028423116586584506,
+      "loss": 5.6621,
+      "step": 3760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842226929312669,
+      "loss": 0.7883,
+      "step": 3761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028421421784731373,
+      "loss": 0.9131,
+      "step": 3762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002842057406141213,
+      "loss": 0.0833,
+      "step": 3763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841972612318253,
+      "loss": 0.3698,
+      "step": 3764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841887797005616,
+      "loss": 0.3154,
+      "step": 3765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028418029602046604,
+      "loss": 0.7328,
+      "step": 3766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028417181019167433,
+      "loss": 0.3897,
+      "step": 3767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841633222143225,
+      "loss": 0.8632,
+      "step": 3768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841548320885464,
+      "loss": 0.063,
+      "step": 3769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028414633981448204,
+      "loss": 0.9287,
+      "step": 3770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841378453922653,
+      "loss": 0.5314,
+      "step": 3771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841293488220323,
+      "loss": 2.0727,
+      "step": 3772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841208501039191,
+      "loss": 0.2057,
+      "step": 3773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002841123492380618,
+      "loss": 0.4533,
+      "step": 3774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028410384622459643,
+      "loss": 0.0874,
+      "step": 3775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840953410636593,
+      "loss": 1.344,
+      "step": 3776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028408683375538647,
+      "loss": 0.3745,
+      "step": 3777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028407832429991423,
+      "loss": 0.4969,
+      "step": 3778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840698126973789,
+      "loss": 0.5038,
+      "step": 3779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840612989479167,
+      "loss": 0.2629,
+      "step": 3780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000284052783051664,
+      "loss": 0.8141,
+      "step": 3781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840442650087572,
+      "loss": 0.1384,
+      "step": 3782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840357448193326,
+      "loss": 0.3164,
+      "step": 3783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028402722248352673,
+      "loss": 0.6467,
+      "step": 3784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840186980014761,
+      "loss": 0.5706,
+      "step": 3785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840101713733171,
+      "loss": 2.5658,
+      "step": 3786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002840016425991863,
+      "loss": 0.2649,
+      "step": 3787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002839931116792203,
+      "loss": 2.2538,
+      "step": 3788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028398457861355576,
+      "loss": 0.6807,
+      "step": 3789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002839760434023293,
+      "loss": 0.3224,
+      "step": 3790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002839675060456775,
+      "loss": 0.4197,
+      "step": 3791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002839589665437372,
+      "loss": 0.4672,
+      "step": 3792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002839504248966451,
+      "loss": 0.0993,
+      "step": 3793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000283941881104538,
+      "loss": 2.5867,
+      "step": 3794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028393333516755256,
+      "loss": 0.2776,
+      "step": 3795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002839247870858259,
+      "loss": 1.7821,
+      "step": 3796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002839162368594947,
+      "loss": 1.0648,
+      "step": 3797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028390768448869594,
+      "loss": 0.5082,
+      "step": 3798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028389912997356663,
+      "loss": 0.5488,
+      "step": 3799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002838905733142437,
+      "loss": 0.0054,
+      "step": 3800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028388201451086414,
+      "loss": 0.0097,
+      "step": 3801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002838734535635651,
+      "loss": 0.5075,
+      "step": 3802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028386489047248356,
+      "loss": 0.4991,
+      "step": 3803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028385632523775673,
+      "loss": 0.0192,
+      "step": 3804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002838477578595217,
+      "loss": 0.9071,
+      "step": 3805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028383918833791577,
+      "loss": 0.2727,
+      "step": 3806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002838306166730761,
+      "loss": 0.5166,
+      "step": 3807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028382204286513997,
+      "loss": 0.1999,
+      "step": 3808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002838134669142446,
+      "loss": 1.0827,
+      "step": 3809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002838048888205274,
+      "loss": 2.9902,
+      "step": 3810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002837963085841257,
+      "loss": 0.4825,
+      "step": 3811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028378772620517696,
+      "loss": 1.1124,
+      "step": 3812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028377914168381855,
+      "loss": 0.4816,
+      "step": 3813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000283770555020188,
+      "loss": 0.0068,
+      "step": 3814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002837619662144227,
+      "loss": 0.3639,
+      "step": 3815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028375337526666034,
+      "loss": 0.5187,
+      "step": 3816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028374478217703836,
+      "loss": 0.3376,
+      "step": 3817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002837361869456944,
+      "loss": 0.4811,
+      "step": 3818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028372758957276615,
+      "loss": 0.3213,
+      "step": 3819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002837189900583913,
+      "loss": 0.6542,
+      "step": 3820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002837103884027074,
+      "loss": 0.4749,
+      "step": 3821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002837017846058524,
+      "loss": 4.2183,
+      "step": 3822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002836931786679639,
+      "loss": 0.5738,
+      "step": 3823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028368457058917983,
+      "loss": 2.4494,
+      "step": 3824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000283675960369638,
+      "loss": 0.4284,
+      "step": 3825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028366734800947625,
+      "loss": 0.0136,
+      "step": 3826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028365873350883256,
+      "loss": 0.3473,
+      "step": 3827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028365011686784487,
+      "loss": 0.0087,
+      "step": 3828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002836414980866511,
+      "loss": 0.348,
+      "step": 3829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002836328771653893,
+      "loss": 0.5328,
+      "step": 3830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028362425410419754,
+      "loss": 0.2252,
+      "step": 3831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002836156289032139,
+      "loss": 0.3036,
+      "step": 3832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002836070015625764,
+      "loss": 0.0098,
+      "step": 3833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002835983720824234,
+      "loss": 0.8061,
+      "step": 3834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028358974046289293,
+      "loss": 0.4004,
+      "step": 3835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002835811067041233,
+      "loss": 2.1369,
+      "step": 3836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002835724708062527,
+      "loss": 0.2569,
+      "step": 3837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002835638327694194,
+      "loss": 1.0764,
+      "step": 3838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028355519259376185,
+      "loss": 0.5194,
+      "step": 3839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028354655027941825,
+      "loss": 0.5982,
+      "step": 3840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002835379058265271,
+      "loss": 0.138,
+      "step": 3841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002835292592352268,
+      "loss": 0.361,
+      "step": 3842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028352061050565585,
+      "loss": 0.308,
+      "step": 3843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028351195963795267,
+      "loss": 1.3702,
+      "step": 3844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028350330663225583,
+      "loss": 0.7233,
+      "step": 3845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002834946514887039,
+      "loss": 0.2216,
+      "step": 3846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002834859942074355,
+      "loss": 0.0434,
+      "step": 3847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028347733478858924,
+      "loss": 0.5975,
+      "step": 3848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002834686732323037,
+      "loss": 0.6725,
+      "step": 3849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028346000953871777,
+      "loss": 2.4808,
+      "step": 3850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028345134370797,
+      "loss": 0.0543,
+      "step": 3851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028344267574019927,
+      "loss": 0.9902,
+      "step": 3852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002834340056355444,
+      "loss": 0.8356,
+      "step": 3853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002834253333941441,
+      "loss": 0.7317,
+      "step": 3854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028341665901613744,
+      "loss": 0.3115,
+      "step": 3855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028340798250166315,
+      "loss": 0.3436,
+      "step": 3856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028339930385086026,
+      "loss": 0.7363,
+      "step": 3857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002833906230638677,
+      "loss": 1.0451,
+      "step": 3858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028338194014082454,
+      "loss": 0.5825,
+      "step": 3859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002833732550818697,
+      "loss": 0.3183,
+      "step": 3860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028336456788714245,
+      "loss": 1.8364,
+      "step": 3861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028335587855678174,
+      "loss": 0.5371,
+      "step": 3862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002833471870909268,
+      "loss": 0.205,
+      "step": 3863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028333849348971676,
+      "loss": 0.0951,
+      "step": 3864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002833297977532909,
+      "loss": 0.4156,
+      "step": 3865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002833210998817884,
+      "loss": 0.5349,
+      "step": 3866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002833123998753486,
+      "loss": 0.4445,
+      "step": 3867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002833036977341107,
+      "loss": 0.0353,
+      "step": 3868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002832949934582142,
+      "loss": 0.5954,
+      "step": 3869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028328628704779845,
+      "loss": 1.0516,
+      "step": 3870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028327757850300284,
+      "loss": 0.6938,
+      "step": 3871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002832688678239668,
+      "loss": 0.3589,
+      "step": 3872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002832601550108299,
+      "loss": 0.7331,
+      "step": 3873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002832514400637316,
+      "loss": 0.39,
+      "step": 3874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028324272298281143,
+      "loss": 0.5919,
+      "step": 3875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028323400376820904,
+      "loss": 0.5673,
+      "step": 3876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002832252824200641,
+      "loss": 0.3725,
+      "step": 3877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002832165589385161,
+      "loss": 0.5343,
+      "step": 3878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028320783332370485,
+      "loss": 0.4679,
+      "step": 3879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028319910557577017,
+      "loss": 0.3979,
+      "step": 3880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028319037569485164,
+      "loss": 0.657,
+      "step": 3881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028318164368108914,
+      "loss": 1.5665,
+      "step": 3882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028317290953462247,
+      "loss": 0.9722,
+      "step": 3883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028316417325559154,
+      "loss": 0.6615,
+      "step": 3884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002831554348441362,
+      "loss": 0.0097,
+      "step": 3885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002831466943003965,
+      "loss": 0.1128,
+      "step": 3886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028313795162451225,
+      "loss": 0.3375,
+      "step": 3887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002831292068166235,
+      "loss": 0.3425,
+      "step": 3888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028312045987687035,
+      "loss": 0.3479,
+      "step": 3889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002831117108053928,
+      "loss": 0.0494,
+      "step": 3890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000283102959602331,
+      "loss": 0.9703,
+      "step": 3891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028309420626782504,
+      "loss": 0.4379,
+      "step": 3892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028308545080201505,
+      "loss": 0.0061,
+      "step": 3893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002830766932050414,
+      "loss": 0.6393,
+      "step": 3894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028306793347704416,
+      "loss": 0.5325,
+      "step": 3895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002830591716181637,
+      "loss": 0.0383,
+      "step": 3896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002830504076285403,
+      "loss": 0.3364,
+      "step": 3897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028304164150831434,
+      "loss": 4.015,
+      "step": 3898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028303287325762605,
+      "loss": 0.2847,
+      "step": 3899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000283024102876616,
+      "loss": 0.9237,
+      "step": 3900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028301533036542465,
+      "loss": 0.2906,
+      "step": 3901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028300655572419234,
+      "loss": 0.4824,
+      "step": 3902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028299777895305966,
+      "loss": 0.362,
+      "step": 3903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028298900005216724,
+      "loss": 2.0821,
+      "step": 3904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028298021902165547,
+      "loss": 0.7065,
+      "step": 3905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002829714358616651,
+      "loss": 0.5206,
+      "step": 3906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002829626505723367,
+      "loss": 0.8235,
+      "step": 3907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002829538631538111,
+      "loss": 1.0255,
+      "step": 3908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028294507360622885,
+      "loss": 0.4224,
+      "step": 3909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028293628192973077,
+      "loss": 0.0421,
+      "step": 3910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028292748812445767,
+      "loss": 0.5057,
+      "step": 3911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002829186921905503,
+      "loss": 0.5649,
+      "step": 3912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028290989412814964,
+      "loss": 0.2942,
+      "step": 3913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028290109393739644,
+      "loss": 0.313,
+      "step": 3914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002828922916184317,
+      "loss": 0.1747,
+      "step": 3915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002828834871713963,
+      "loss": 0.6228,
+      "step": 3916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028287468059643133,
+      "loss": 0.1012,
+      "step": 3917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028286587189367774,
+      "loss": 0.0353,
+      "step": 3918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028285706106327663,
+      "loss": 0.1336,
+      "step": 3919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002828482481053691,
+      "loss": 1.1515,
+      "step": 3920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028283943302009624,
+      "loss": 0.7156,
+      "step": 3921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028283061580759917,
+      "loss": 0.2488,
+      "step": 3922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002828217964680192,
+      "loss": 0.7509,
+      "step": 3923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028281297500149747,
+      "loss": 0.2181,
+      "step": 3924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002828041514081753,
+      "loss": 1.0415,
+      "step": 3925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028279532568819394,
+      "loss": 0.4923,
+      "step": 3926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002827864978416948,
+      "loss": 1.7824,
+      "step": 3927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028277766786881907,
+      "loss": 0.718,
+      "step": 3928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002827688357697083,
+      "loss": 0.7834,
+      "step": 3929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028276000154450385,
+      "loss": 0.3372,
+      "step": 3930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028275116519334726,
+      "loss": 0.4229,
+      "step": 3931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028274232671638,
+      "loss": 0.146,
+      "step": 3932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002827334861137436,
+      "loss": 0.3976,
+      "step": 3933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028272464338557966,
+      "loss": 1.7805,
+      "step": 3934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002827157985320297,
+      "loss": 0.3198,
+      "step": 3935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002827069515532354,
+      "loss": 0.421,
+      "step": 3936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826981024493384,
+      "loss": 0.2223,
+      "step": 3937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826892512204805,
+      "loss": 0.414,
+      "step": 3938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826803978668034,
+      "loss": 0.1488,
+      "step": 3939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826715423884488,
+      "loss": 0.3334,
+      "step": 3940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028266268478555854,
+      "loss": 0.6445,
+      "step": 3941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028265382505827453,
+      "loss": 0.2507,
+      "step": 3942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826449632067386,
+      "loss": 0.5217,
+      "step": 3943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826360992310927,
+      "loss": 0.3384,
+      "step": 3944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826272331314786,
+      "loss": 0.261,
+      "step": 3945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826183649080385,
+      "loss": 1.1914,
+      "step": 3946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002826094945609143,
+      "loss": 0.9391,
+      "step": 3947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028260062209024806,
+      "loss": 0.4602,
+      "step": 3948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002825917474961819,
+      "loss": 0.018,
+      "step": 3949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028258287077885785,
+      "loss": 2.0218,
+      "step": 3950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028257399193841806,
+      "loss": 0.2938,
+      "step": 3951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002825651109750048,
+      "loss": 0.1649,
+      "step": 3952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002825562278887603,
+      "loss": 0.6509,
+      "step": 3953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002825473426798267,
+      "loss": 0.5051,
+      "step": 3954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002825384553483463,
+      "loss": 0.6537,
+      "step": 3955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002825295658944615,
+      "loss": 0.3921,
+      "step": 3956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002825206743183146,
+      "loss": 0.7091,
+      "step": 3957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282511780620048,
+      "loss": 1.6545,
+      "step": 3958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282502884799804,
+      "loss": 0.3621,
+      "step": 3959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028249398685772524,
+      "loss": 0.4944,
+      "step": 3960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028248508679395415,
+      "loss": 0.0146,
+      "step": 3961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002824761846086332,
+      "loss": 0.5072,
+      "step": 3962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028246728030190496,
+      "loss": 0.6659,
+      "step": 3963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028245837387391205,
+      "loss": 0.9645,
+      "step": 3964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002824494653247971,
+      "loss": 1.3156,
+      "step": 3965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028244055465470266,
+      "loss": 0.8181,
+      "step": 3966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028243164186377153,
+      "loss": 0.4426,
+      "step": 3967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002824227269521465,
+      "loss": 0.4215,
+      "step": 3968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028241380991997014,
+      "loss": 1.3217,
+      "step": 3969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028240489076738535,
+      "loss": 0.1371,
+      "step": 3970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028239596949453495,
+      "loss": 0.6227,
+      "step": 3971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002823870461015618,
+      "loss": 0.0106,
+      "step": 3972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002823781205886088,
+      "loss": 1.1805,
+      "step": 3973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002823691929558189,
+      "loss": 0.9174,
+      "step": 3974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282360263203335,
+      "loss": 0.5624,
+      "step": 3975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028235133133130015,
+      "loss": 0.2993,
+      "step": 3976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002823423973398573,
+      "loss": 0.3535,
+      "step": 3977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002823334612291496,
+      "loss": 0.586,
+      "step": 3978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028232452299932015,
+      "loss": 0.2663,
+      "step": 3979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028231558265051204,
+      "loss": 0.5788,
+      "step": 3980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002823066401828685,
+      "loss": 0.6351,
+      "step": 3981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002822976955965326,
+      "loss": 0.2307,
+      "step": 3982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002822887488916477,
+      "loss": 0.2432,
+      "step": 3983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282279800068357,
+      "loss": 0.5875,
+      "step": 3984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028227084912680386,
+      "loss": 0.4632,
+      "step": 3985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002822618960671316,
+      "loss": 0.8313,
+      "step": 3986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028225294088948355,
+      "loss": 0.0039,
+      "step": 3987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002822439835940031,
+      "loss": 0.8477,
+      "step": 3988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028223502418083375,
+      "loss": 0.363,
+      "step": 3989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028222606265011893,
+      "loss": 0.5953,
+      "step": 3990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028221709900200215,
+      "loss": 0.2107,
+      "step": 3991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282208133236627,
+      "loss": 0.2315,
+      "step": 3992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282199165354137,
+      "loss": 0.9206,
+      "step": 3993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002821901953546758,
+      "loss": 0.2808,
+      "step": 3994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282181223238387,
+      "loss": 0.3459,
+      "step": 3995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028217224900541424,
+      "loss": 0.312,
+      "step": 3996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028216327265590126,
+      "loss": 0.4644,
+      "step": 3997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002821542941899919,
+      "loss": 0.415,
+      "step": 3998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002821453136078298,
+      "loss": 0.5242,
+      "step": 3999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002821363309095588,
+      "loss": 0.0152,
+      "step": 4000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028212734609532283,
+      "loss": 0.1088,
+      "step": 4001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028211835916526566,
+      "loss": 0.4315,
+      "step": 4002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002821093701195313,
+      "loss": 2.5104,
+      "step": 4003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028210037895826356,
+      "loss": 0.4073,
+      "step": 4004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002820913856816066,
+      "loss": 0.5876,
+      "step": 4005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002820823902897043,
+      "loss": 0.5914,
+      "step": 4006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028207339278270077,
+      "loss": 0.3061,
+      "step": 4007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028206439316074,
+      "loss": 0.3871,
+      "step": 4008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002820553914239663,
+      "loss": 2.0868,
+      "step": 4009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002820463875725236,
+      "loss": 0.1274,
+      "step": 4010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002820373816065562,
+      "loss": 0.3229,
+      "step": 4011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028202837352620826,
+      "loss": 0.4397,
+      "step": 4012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002820193633316241,
+      "loss": 0.1604,
+      "step": 4013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000282010351022948,
+      "loss": 0.4313,
+      "step": 4014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002820013366003242,
+      "loss": 1.9944,
+      "step": 4015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002819923200638971,
+      "loss": 0.4672,
+      "step": 4016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002819833014138111,
+      "loss": 0.0356,
+      "step": 4017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002819742806502106,
+      "loss": 0.1579,
+      "step": 4018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002819652577732401,
+      "loss": 0.0066,
+      "step": 4019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028195623278304396,
+      "loss": 0.4014,
+      "step": 4020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028194720567976684,
+      "loss": 0.9764,
+      "step": 4021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028193817646355327,
+      "loss": 0.5534,
+      "step": 4022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002819291451345478,
+      "loss": 0.7895,
+      "step": 4023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028192011169289507,
+      "loss": 0.6791,
+      "step": 4024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028191107613873977,
+      "loss": 0.8852,
+      "step": 4025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002819020384722265,
+      "loss": 0.9089,
+      "step": 4026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002818929986935,
+      "loss": 0.4116,
+      "step": 4027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002818839568027052,
+      "loss": 0.8005,
+      "step": 4028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002818749127999867,
+      "loss": 0.0153,
+      "step": 4029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028186586668548934,
+      "loss": 0.5332,
+      "step": 4030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028185681845935805,
+      "loss": 0.7026,
+      "step": 4031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028184776812173777,
+      "loss": 1.1567,
+      "step": 4032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002818387156727733,
+      "loss": 2.2191,
+      "step": 4033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028182966111260963,
+      "loss": 1.0056,
+      "step": 4034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028182060444139186,
+      "loss": 0.3547,
+      "step": 4035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028181154565926487,
+      "loss": 0.2662,
+      "step": 4036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028180248476637385,
+      "loss": 0.563,
+      "step": 4037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028179342176286385,
+      "loss": 0.346,
+      "step": 4038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028178435664887996,
+      "loss": 0.4959,
+      "step": 4039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002817752894245674,
+      "loss": 0.1405,
+      "step": 4040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002817662200900713,
+      "loss": 0.4412,
+      "step": 4041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028175714864553706,
+      "loss": 0.1607,
+      "step": 4042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002817480750911097,
+      "loss": 0.8594,
+      "step": 4043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028173899942693473,
+      "loss": 0.4573,
+      "step": 4044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002817299216531573,
+      "loss": 2.1642,
+      "step": 4045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000281720841769923,
+      "loss": 0.949,
+      "step": 4046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000281711759777377,
+      "loss": 0.4012,
+      "step": 4047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002817026756756649,
+      "loss": 0.3865,
+      "step": 4048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000281693589464932,
+      "loss": 0.8176,
+      "step": 4049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000281684501145324,
+      "loss": 0.6412,
+      "step": 4050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028167541071698636,
+      "loss": 0.2613,
+      "step": 4051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002816663181800646,
+      "loss": 0.4028,
+      "step": 4052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002816572235347044,
+      "loss": 0.5256,
+      "step": 4053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002816481267810513,
+      "loss": 0.3726,
+      "step": 4054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028163902791925104,
+      "loss": 0.6552,
+      "step": 4055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002816299269494493,
+      "loss": 0.9967,
+      "step": 4056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028162082387179186,
+      "loss": 0.1381,
+      "step": 4057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028161171868642447,
+      "loss": 0.2301,
+      "step": 4058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002816026113934929,
+      "loss": 0.5813,
+      "step": 4059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028159350199314303,
+      "loss": 0.2552,
+      "step": 4060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002815843904855207,
+      "loss": 0.682,
+      "step": 4061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002815752768707718,
+      "loss": 0.4241,
+      "step": 4062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002815661611490424,
+      "loss": 2.4526,
+      "step": 4063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002815570433204783,
+      "loss": 0.1253,
+      "step": 4064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028154792338522564,
+      "loss": 0.4189,
+      "step": 4065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028153880134343043,
+      "loss": 0.2804,
+      "step": 4066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028152967719523867,
+      "loss": 0.6276,
+      "step": 4067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028152055094079657,
+      "loss": 0.4402,
+      "step": 4068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002815114225802502,
+      "loss": 0.8173,
+      "step": 4069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002815022921137458,
+      "loss": 0.2763,
+      "step": 4070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814931595414295,
+      "loss": 0.7075,
+      "step": 4071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814840248634477,
+      "loss": 0.1557,
+      "step": 4072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814748880799464,
+      "loss": 0.0104,
+      "step": 4073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028146574919107225,
+      "loss": 0.0287,
+      "step": 4074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814566081969713,
+      "loss": 0.2974,
+      "step": 4075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028144746509779013,
+      "loss": 0.4801,
+      "step": 4076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814383198936751,
+      "loss": 0.6304,
+      "step": 4077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814291725847726,
+      "loss": 0.695,
+      "step": 4078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028142002317122915,
+      "loss": 0.032,
+      "step": 4079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814108716531913,
+      "loss": 0.4559,
+      "step": 4080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002814017180308055,
+      "loss": 1.0522,
+      "step": 4081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002813925623042185,
+      "loss": 0.0771,
+      "step": 4082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028138340447357666,
+      "loss": 2.0958,
+      "step": 4083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028137424453902687,
+      "loss": 0.1076,
+      "step": 4084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028136508250071573,
+      "loss": 0.3002,
+      "step": 4085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002813559183587899,
+      "loss": 0.6258,
+      "step": 4086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028134675211339617,
+      "loss": 0.6258,
+      "step": 4087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002813375837646814,
+      "loss": 0.2202,
+      "step": 4088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028132841331279226,
+      "loss": 0.5469,
+      "step": 4089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028131924075787564,
+      "loss": 0.1395,
+      "step": 4090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002813100661000786,
+      "loss": 0.4671,
+      "step": 4091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002813008893395478,
+      "loss": 0.4417,
+      "step": 4092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002812917104764304,
+      "loss": 0.3437,
+      "step": 4093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002812825295108732,
+      "loss": 0.51,
+      "step": 4094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002812733464430234,
+      "loss": 1.2999,
+      "step": 4095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028126416127302795,
+      "loss": 0.6424,
+      "step": 4096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028125497400103395,
+      "loss": 3.9963,
+      "step": 4097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002812457846271885,
+      "loss": 0.0157,
+      "step": 4098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028123659315163884,
+      "loss": 0.8515,
+      "step": 4099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028122739957453204,
+      "loss": 0.6472,
+      "step": 4100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002812182038960154,
+      "loss": 0.2547,
+      "step": 4101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028120900611623614,
+      "loss": 0.3136,
+      "step": 4102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002811998062353416,
+      "loss": 2.047,
+      "step": 4103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002811906042534791,
+      "loss": 0.5325,
+      "step": 4104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002811814001707959,
+      "loss": 0.6103,
+      "step": 4105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028117219398743944,
+      "loss": 0.0539,
+      "step": 4106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002811629857035572,
+      "loss": 0.5971,
+      "step": 4107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002811537753192966,
+      "loss": 1.6261,
+      "step": 4108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028114456283480507,
+      "loss": 0.0148,
+      "step": 4109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002811353482502302,
+      "loss": 0.8907,
+      "step": 4110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028112613156571956,
+      "loss": 0.1884,
+      "step": 4111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002811169127814207,
+      "loss": 0.6855,
+      "step": 4112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028110769189748126,
+      "loss": 0.0039,
+      "step": 4113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002810984689140489,
+      "loss": 0.1323,
+      "step": 4114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002810892438312713,
+      "loss": 1.2032,
+      "step": 4115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002810800166492962,
+      "loss": 0.0307,
+      "step": 4116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002810707873682714,
+      "loss": 0.3828,
+      "step": 4117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028106155598834465,
+      "loss": 0.9292,
+      "step": 4118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028105232250966373,
+      "loss": 0.4576,
+      "step": 4119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002810430869323765,
+      "loss": 0.5067,
+      "step": 4120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000281033849256631,
+      "loss": 0.4815,
+      "step": 4121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028102460948257503,
+      "loss": 0.7411,
+      "step": 4122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002810153676103566,
+      "loss": 0.6507,
+      "step": 4123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002810061236401236,
+      "loss": 0.8208,
+      "step": 4124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002809968775720242,
+      "loss": 0.5499,
+      "step": 4125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002809876294062064,
+      "loss": 0.2598,
+      "step": 4126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028097837914281836,
+      "loss": 0.1875,
+      "step": 4127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028096912678200807,
+      "loss": 0.0199,
+      "step": 4128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002809598723239238,
+      "loss": 0.481,
+      "step": 4129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002809506157687137,
+      "loss": 0.229,
+      "step": 4130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280941357116526,
+      "loss": 0.1214,
+      "step": 4131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028093209636750904,
+      "loss": 0.3319,
+      "step": 4132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028092283352181105,
+      "loss": 0.3965,
+      "step": 4133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002809135685795803,
+      "loss": 0.0541,
+      "step": 4134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002809043015409653,
+      "loss": 0.6846,
+      "step": 4135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002808950324061144,
+      "loss": 0.8124,
+      "step": 4136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002808857611751759,
+      "loss": 0.3368,
+      "step": 4137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028087648784829843,
+      "loss": 0.654,
+      "step": 4138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028086721242563033,
+      "loss": 0.7277,
+      "step": 4139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028085793490732037,
+      "loss": 0.413,
+      "step": 4140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028084865529351687,
+      "loss": 0.5271,
+      "step": 4141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002808393735843686,
+      "loss": 0.9577,
+      "step": 4142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028083008978002405,
+      "loss": 0.2275,
+      "step": 4143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280820803880632,
+      "loss": 0.619,
+      "step": 4144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028081151588634114,
+      "loss": 0.5106,
+      "step": 4145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028080222579730015,
+      "loss": 1.0352,
+      "step": 4146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002807929336136578,
+      "loss": 0.6676,
+      "step": 4147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028078363933556295,
+      "loss": 0.2838,
+      "step": 4148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002807743429631644,
+      "loss": 0.76,
+      "step": 4149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002807650444966109,
+      "loss": 0.8321,
+      "step": 4150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002807557439360516,
+      "loss": 0.6694,
+      "step": 4151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002807464412816352,
+      "loss": 0.2467,
+      "step": 4152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002807371365335108,
+      "loss": 0.8754,
+      "step": 4153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028072782969182743,
+      "loss": 0.0797,
+      "step": 4154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280718520756734,
+      "loss": 0.0201,
+      "step": 4155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002807092097283797,
+      "loss": 0.5242,
+      "step": 4156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806998966069135,
+      "loss": 0.6695,
+      "step": 4157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806905813924846,
+      "loss": 0.8619,
+      "step": 4158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028068126408524225,
+      "loss": 0.5803,
+      "step": 4159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806719446853355,
+      "loss": 0.2317,
+      "step": 4160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806626231929137,
+      "loss": 0.7624,
+      "step": 4161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806532996081262,
+      "loss": 0.6095,
+      "step": 4162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280643973931122,
+      "loss": 0.4621,
+      "step": 4163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806346461620507,
+      "loss": 0.611,
+      "step": 4164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806253163010616,
+      "loss": 0.3934,
+      "step": 4165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028061598434830416,
+      "loss": 0.0561,
+      "step": 4166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002806066503039277,
+      "loss": 1.0086,
+      "step": 4167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805973141680818,
+      "loss": 0.0764,
+      "step": 4168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028058797594091587,
+      "loss": 0.28,
+      "step": 4169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805786356225795,
+      "loss": 0.7364,
+      "step": 4170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805692932132222,
+      "loss": 0.5019,
+      "step": 4171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805599487129937,
+      "loss": 0.3854,
+      "step": 4172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805506021220435,
+      "loss": 0.3332,
+      "step": 4173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805412534405214,
+      "loss": 1.1816,
+      "step": 4174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280531902668577,
+      "loss": 1.8419,
+      "step": 4175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028052254980636006,
+      "loss": 0.3354,
+      "step": 4176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805131948540204,
+      "loss": 0.2983,
+      "step": 4177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002805038378117078,
+      "loss": 0.7225,
+      "step": 4178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002804944786795721,
+      "loss": 0.2588,
+      "step": 4179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028048511745776316,
+      "loss": 0.5814,
+      "step": 4180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028047575414643093,
+      "loss": 0.3001,
+      "step": 4181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002804663887457253,
+      "loss": 1.9747,
+      "step": 4182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028045702125579624,
+      "loss": 0.4426,
+      "step": 4183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002804476516767937,
+      "loss": 0.9451,
+      "step": 4184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002804382800088679,
+      "loss": 0.6191,
+      "step": 4185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002804289062521688,
+      "loss": 0.4526,
+      "step": 4186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002804195304068465,
+      "loss": 0.5799,
+      "step": 4187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028041015247305115,
+      "loss": 0.0662,
+      "step": 4188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002804007724509329,
+      "loss": 3.295,
+      "step": 4189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280391390340642,
+      "loss": 0.2687,
+      "step": 4190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028038200614232864,
+      "loss": 0.752,
+      "step": 4191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028037261985614317,
+      "loss": 0.9116,
+      "step": 4192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028036323148223583,
+      "loss": 0.4969,
+      "step": 4193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280353841020757,
+      "loss": 0.0517,
+      "step": 4194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280344448471857,
+      "loss": 4.5277,
+      "step": 4195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028033505383568633,
+      "loss": 0.3933,
+      "step": 4196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002803256571123954,
+      "loss": 0.5507,
+      "step": 4197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002803162583021346,
+      "loss": 0.2749,
+      "step": 4198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028030685740505453,
+      "loss": 0.6076,
+      "step": 4199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802974544213057,
+      "loss": 0.7901,
+      "step": 4200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028028804935103873,
+      "loss": 0.3741,
+      "step": 4201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802786421944041,
+      "loss": 0.2838,
+      "step": 4202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802692329515526,
+      "loss": 0.1035,
+      "step": 4203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802598216226348,
+      "loss": 0.5401,
+      "step": 4204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802504082078015,
+      "loss": 0.2434,
+      "step": 4205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028024099270720335,
+      "loss": 0.3238,
+      "step": 4206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802315751209912,
+      "loss": 0.7792,
+      "step": 4207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802221554493158,
+      "loss": 0.5513,
+      "step": 4208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280212733692328,
+      "loss": 0.4654,
+      "step": 4209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002802033098501787,
+      "loss": 0.9078,
+      "step": 4210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002801938839230188,
+      "loss": 0.6671,
+      "step": 4211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002801844559109992,
+      "loss": 0.0216,
+      "step": 4212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000280175025814271,
+      "loss": 0.6659,
+      "step": 4213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002801655936329851,
+      "loss": 0.6683,
+      "step": 4214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002801561593672926,
+      "loss": 0.3009,
+      "step": 4215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028014672301734445,
+      "loss": 0.9509,
+      "step": 4216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028013728458329195,
+      "loss": 2.0034,
+      "step": 4217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002801278440652861,
+      "loss": 0.108,
+      "step": 4218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002801184014634782,
+      "loss": 0.5128,
+      "step": 4219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002801089567780192,
+      "loss": 0.4548,
+      "step": 4220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028009951000906066,
+      "loss": 0.2766,
+      "step": 4221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002800900611567537,
+      "loss": 0.4293,
+      "step": 4222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028008061022124966,
+      "loss": 0.6536,
+      "step": 4223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028007115720269987,
+      "loss": 0.3032,
+      "step": 4224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002800617021012557,
+      "loss": 0.5988,
+      "step": 4225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028005224491706854,
+      "loss": 0.4227,
+      "step": 4226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002800427856502899,
+      "loss": 0.7656,
+      "step": 4227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002800333243010712,
+      "loss": 0.5999,
+      "step": 4228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002800238608695639,
+      "loss": 0.6409,
+      "step": 4229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002800143953559197,
+      "loss": 0.3997,
+      "step": 4230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028000492776029,
+      "loss": 0.3835,
+      "step": 4231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027999545808282656,
+      "loss": 1.8859,
+      "step": 4232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027998598632368094,
+      "loss": 0.5593,
+      "step": 4233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027997651248300485,
+      "loss": 0.2204,
+      "step": 4234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002799670365609499,
+      "loss": 0.3413,
+      "step": 4235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000279957558557668,
+      "loss": 0.5179,
+      "step": 4236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027994807847331077,
+      "loss": 0.5735,
+      "step": 4237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002799385963080301,
+      "loss": 1.3478,
+      "step": 4238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027992911206197786,
+      "loss": 0.3329,
+      "step": 4239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002799196257353059,
+      "loss": 0.0861,
+      "step": 4240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002799101373281661,
+      "loss": 1.1619,
+      "step": 4241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027990064684071036,
+      "loss": 0.0601,
+      "step": 4242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027989115427309077,
+      "loss": 0.4092,
+      "step": 4243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027988165962545923,
+      "loss": 0.7797,
+      "step": 4244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002798721628979679,
+      "loss": 0.8289,
+      "step": 4245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002798626640907688,
+      "loss": 1.6118,
+      "step": 4246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000279853163204014,
+      "loss": 0.497,
+      "step": 4247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027984366023785565,
+      "loss": 0.2213,
+      "step": 4248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000279834155192446,
+      "loss": 0.461,
+      "step": 4249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002798246480679372,
+      "loss": 0.5236,
+      "step": 4250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027981513886448143,
+      "loss": 0.5033,
+      "step": 4251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027980562758223113,
+      "loss": 0.264,
+      "step": 4252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027979611422133846,
+      "loss": 0.0229,
+      "step": 4253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027978659878195584,
+      "loss": 2.088,
+      "step": 4254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002797770812642356,
+      "loss": 0.26,
+      "step": 4255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002797675616683302,
+      "loss": 0.0154,
+      "step": 4256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027975803999439204,
+      "loss": 0.7586,
+      "step": 4257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002797485162425736,
+      "loss": 0.3637,
+      "step": 4258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002797389904130274,
+      "loss": 1.2241,
+      "step": 4259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027972946250590596,
+      "loss": 0.5586,
+      "step": 4260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027971993252136194,
+      "loss": 0.5215,
+      "step": 4261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002797104004595478,
+      "loss": 0.5738,
+      "step": 4262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027970086632061634,
+      "loss": 0.5211,
+      "step": 4263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027969133010472013,
+      "loss": 0.3055,
+      "step": 4264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002796817918120119,
+      "loss": 0.567,
+      "step": 4265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002796722514426444,
+      "loss": 0.9663,
+      "step": 4266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027966270899677044,
+      "loss": 0.3425,
+      "step": 4267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027965316447454275,
+      "loss": 0.4018,
+      "step": 4268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002796436178761142,
+      "loss": 0.0241,
+      "step": 4269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027963406920163776,
+      "loss": 0.0209,
+      "step": 4270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027962451845126615,
+      "loss": 0.5274,
+      "step": 4271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002796149656251525,
+      "loss": 0.2309,
+      "step": 4272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027960541072344965,
+      "loss": 0.7831,
+      "step": 4273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795958537463106,
+      "loss": 0.0416,
+      "step": 4274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027958629469388856,
+      "loss": 0.0158,
+      "step": 4275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795767335663365,
+      "loss": 0.7389,
+      "step": 4276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795671703638074,
+      "loss": 0.0448,
+      "step": 4277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795576050864546,
+      "loss": 0.5461,
+      "step": 4278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795480377344312,
+      "loss": 0.4247,
+      "step": 4279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027953846830789037,
+      "loss": 1.6354,
+      "step": 4280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795288968069854,
+      "loss": 0.7513,
+      "step": 4281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027951932323186954,
+      "loss": 0.6022,
+      "step": 4282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795097475826961,
+      "loss": 0.4227,
+      "step": 4283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002795001698596184,
+      "loss": 1.1749,
+      "step": 4284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002794905900627899,
+      "loss": 0.4575,
+      "step": 4285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002794810081923639,
+      "loss": 0.8736,
+      "step": 4286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027947142424849386,
+      "loss": 0.0147,
+      "step": 4287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027946183823133325,
+      "loss": 0.3215,
+      "step": 4288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002794522501410356,
+      "loss": 0.3259,
+      "step": 4289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002794426599777545,
+      "loss": 0.9876,
+      "step": 4290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002794330677416434,
+      "loss": 0.0068,
+      "step": 4291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000279423473432856,
+      "loss": 0.5751,
+      "step": 4292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002794138770515459,
+      "loss": 0.4189,
+      "step": 4293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002794042785978668,
+      "loss": 0.0084,
+      "step": 4294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027939467807197235,
+      "loss": 1.9854,
+      "step": 4295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002793850754740164,
+      "loss": 1.2459,
+      "step": 4296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002793754708041526,
+      "loss": 1.2858,
+      "step": 4297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002793658640625348,
+      "loss": 0.2324,
+      "step": 4298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002793562552493168,
+      "loss": 0.9893,
+      "step": 4299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002793466443646526,
+      "loss": 0.3138,
+      "step": 4300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027933703140869596,
+      "loss": 0.0068,
+      "step": 4301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002793274163816009,
+      "loss": 1.1611,
+      "step": 4302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002793177992835213,
+      "loss": 0.7265,
+      "step": 4303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027930818011461124,
+      "loss": 0.3686,
+      "step": 4304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027929855887502474,
+      "loss": 0.035,
+      "step": 4305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027928893556491594,
+      "loss": 0.2936,
+      "step": 4306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002792793101844388,
+      "loss": 0.2819,
+      "step": 4307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027926968273374753,
+      "loss": 0.3809,
+      "step": 4308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002792600532129963,
+      "loss": 0.003,
+      "step": 4309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027925042162233935,
+      "loss": 0.3725,
+      "step": 4310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002792407879619309,
+      "loss": 1.1052,
+      "step": 4311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027923115223192516,
+      "loss": 0.8955,
+      "step": 4312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002792215144324764,
+      "loss": 0.527,
+      "step": 4313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027921187456373905,
+      "loss": 0.4939,
+      "step": 4314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002792022326258675,
+      "loss": 0.4732,
+      "step": 4315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002791925886190161,
+      "loss": 0.371,
+      "step": 4316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002791829425433392,
+      "loss": 0.734,
+      "step": 4317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002791732943989914,
+      "loss": 0.7179,
+      "step": 4318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002791636441861272,
+      "loss": 0.1081,
+      "step": 4319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000279153991904901,
+      "loss": 0.8209,
+      "step": 4320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027914433755546757,
+      "loss": 0.0046,
+      "step": 4321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027913468113798125,
+      "loss": 0.3426,
+      "step": 4322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002791250226525969,
+      "loss": 0.7745,
+      "step": 4323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002791153620994691,
+      "loss": 0.0237,
+      "step": 4324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002791056994787525,
+      "loss": 0.0078,
+      "step": 4325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027909603479060193,
+      "loss": 0.4542,
+      "step": 4326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002790863680351721,
+      "loss": 1.8731,
+      "step": 4327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027907669921261783,
+      "loss": 0.629,
+      "step": 4328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027906702832309395,
+      "loss": 0.8727,
+      "step": 4329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027905735536675526,
+      "loss": 1.1504,
+      "step": 4330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027904768034375674,
+      "loss": 1.3006,
+      "step": 4331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002790380032542533,
+      "loss": 1.0055,
+      "step": 4332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002790283240983999,
+      "loss": 1.8324,
+      "step": 4333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002790186428763515,
+      "loss": 0.2731,
+      "step": 4334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002790089595882632,
+      "loss": 0.2351,
+      "step": 4335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027899927423429005,
+      "loss": 0.5169,
+      "step": 4336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002789895868145871,
+      "loss": 0.2773,
+      "step": 4337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002789798973293095,
+      "loss": 0.4029,
+      "step": 4338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002789702057786124,
+      "loss": 2.1134,
+      "step": 4339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027896051216265097,
+      "loss": 0.5441,
+      "step": 4340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027895081648158056,
+      "loss": 0.005,
+      "step": 4341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027894111873555627,
+      "loss": 0.3038,
+      "step": 4342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027893141892473356,
+      "loss": 0.3177,
+      "step": 4343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002789217170492677,
+      "loss": 0.0039,
+      "step": 4344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027891201310931396,
+      "loss": 0.2755,
+      "step": 4345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002789023071050278,
+      "loss": 0.4741,
+      "step": 4346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027889259903656464,
+      "loss": 0.2919,
+      "step": 4347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027888288890407995,
+      "loss": 0.4332,
+      "step": 4348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027887317670772924,
+      "loss": 0.0037,
+      "step": 4349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000278863462447668,
+      "loss": 0.6553,
+      "step": 4350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002788537461240519,
+      "loss": 0.0056,
+      "step": 4351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002788440277370363,
+      "loss": 0.3714,
+      "step": 4352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000278834307286777,
+      "loss": 0.419,
+      "step": 4353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027882458477342966,
+      "loss": 0.0437,
+      "step": 4354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002788148601971499,
+      "loss": 0.3629,
+      "step": 4355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027880513355809354,
+      "loss": 0.1398,
+      "step": 4356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002787954048564162,
+      "loss": 0.5873,
+      "step": 4357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027878567409227373,
+      "loss": 0.4231,
+      "step": 4358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027877594126582205,
+      "loss": 0.4535,
+      "step": 4359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002787662063772169,
+      "loss": 0.5649,
+      "step": 4360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002787564694266142,
+      "loss": 0.8206,
+      "step": 4361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002787467304141699,
+      "loss": 0.5902,
+      "step": 4362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027873698934003993,
+      "loss": 2.0392,
+      "step": 4363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002787272462043802,
+      "loss": 0.1958,
+      "step": 4364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027871750100734693,
+      "loss": 0.163,
+      "step": 4365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000278707753749096,
+      "loss": 0.3599,
+      "step": 4366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002786980044297836,
+      "loss": 0.7988,
+      "step": 4367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002786882530495658,
+      "loss": 0.8592,
+      "step": 4368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002786784996085987,
+      "loss": 0.0057,
+      "step": 4369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002786687441070386,
+      "loss": 0.452,
+      "step": 4370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027865898654504166,
+      "loss": 0.9084,
+      "step": 4371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027864922692276417,
+      "loss": 0.3247,
+      "step": 4372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027863946524036236,
+      "loss": 0.35,
+      "step": 4373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002786297014979925,
+      "loss": 0.5794,
+      "step": 4374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027861993569581117,
+      "loss": 0.405,
+      "step": 4375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027861016783397446,
+      "loss": 0.0104,
+      "step": 4376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027860039791263903,
+      "loss": 0.0164,
+      "step": 4377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027859062593196116,
+      "loss": 3.3765,
+      "step": 4378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027858085189209745,
+      "loss": 0.7989,
+      "step": 4379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002785710757932043,
+      "loss": 1.2049,
+      "step": 4380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027856129763543846,
+      "loss": 0.006,
+      "step": 4381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027855151741895624,
+      "loss": 0.6369,
+      "step": 4382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002785417351439145,
+      "loss": 3.0782,
+      "step": 4383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002785319508104697,
+      "loss": 2.4709,
+      "step": 4384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002785221644187787,
+      "loss": 0.6792,
+      "step": 4385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027851237596899803,
+      "loss": 0.7062,
+      "step": 4386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002785025854612846,
+      "loss": 0.1333,
+      "step": 4387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027849279289579506,
+      "loss": 0.4848,
+      "step": 4388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027848299827268634,
+      "loss": 0.6412,
+      "step": 4389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002784732015921152,
+      "loss": 0.4653,
+      "step": 4390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027846340285423853,
+      "loss": 0.2497,
+      "step": 4391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002784536020592133,
+      "loss": 0.3355,
+      "step": 4392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027844379920719637,
+      "loss": 0.8142,
+      "step": 4393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027843399429834485,
+      "loss": 0.9562,
+      "step": 4394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002784241873328156,
+      "loss": 0.6317,
+      "step": 4395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002784143783107657,
+      "loss": 0.6153,
+      "step": 4396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002784045672323523,
+      "loss": 0.0747,
+      "step": 4397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027839475409773245,
+      "loss": 0.2239,
+      "step": 4398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027838493890706327,
+      "loss": 0.5388,
+      "step": 4399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000278375121660502,
+      "loss": 0.8335,
+      "step": 4400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002783653023582059,
+      "loss": 0.2119,
+      "step": 4401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000278355481000332,
+      "loss": 0.0697,
+      "step": 4402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027834565758703777,
+      "loss": 0.2172,
+      "step": 4403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027833583211848045,
+      "loss": 0.7571,
+      "step": 4404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002783260045948174,
+      "loss": 0.1159,
+      "step": 4405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002783161750162059,
+      "loss": 0.6592,
+      "step": 4406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027830634338280356,
+      "loss": 0.0361,
+      "step": 4407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782965096947676,
+      "loss": 0.3402,
+      "step": 4408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782866739522556,
+      "loss": 0.2689,
+      "step": 4409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027827683615542505,
+      "loss": 0.34,
+      "step": 4410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782669963044335,
+      "loss": 0.3058,
+      "step": 4411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782571543994385,
+      "loss": 3.178,
+      "step": 4412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782473104405977,
+      "loss": 2.172,
+      "step": 4413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782374644280686,
+      "loss": 0.3023,
+      "step": 4414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027822761636200904,
+      "loss": 3.9289,
+      "step": 4415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782177662425766,
+      "loss": 1.3095,
+      "step": 4416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002782079140699291,
+      "loss": 1.1827,
+      "step": 4417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027819805984422426,
+      "loss": 0.1429,
+      "step": 4418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002781882035656199,
+      "loss": 0.7578,
+      "step": 4419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002781783452342738,
+      "loss": 0.2782,
+      "step": 4420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002781684848503439,
+      "loss": 1.9076,
+      "step": 4421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027815862241398805,
+      "loss": 1.1187,
+      "step": 4422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002781487579253642,
+      "loss": 0.6677,
+      "step": 4423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002781388913846304,
+      "loss": 0.2719,
+      "step": 4424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027812902279194444,
+      "loss": 0.5352,
+      "step": 4425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027811915214746456,
+      "loss": 0.4238,
+      "step": 4426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002781092794513487,
+      "loss": 0.5649,
+      "step": 4427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000278099404703755,
+      "loss": 0.1175,
+      "step": 4428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002780895279048416,
+      "loss": 2.7913,
+      "step": 4429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027807964905476667,
+      "loss": 0.9143,
+      "step": 4430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002780697681536883,
+      "loss": 0.1744,
+      "step": 4431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027805988520176485,
+      "loss": 0.8424,
+      "step": 4432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002780500001991545,
+      "loss": 0.6612,
+      "step": 4433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027804011314601563,
+      "loss": 0.4019,
+      "step": 4434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027803022404250646,
+      "loss": 0.5821,
+      "step": 4435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002780203328887854,
+      "loss": 0.0955,
+      "step": 4436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027801043968501087,
+      "loss": 0.0732,
+      "step": 4437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002780005444313413,
+      "loss": 0.7397,
+      "step": 4438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027799064712793505,
+      "loss": 0.9572,
+      "step": 4439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002779807477749507,
+      "loss": 0.9484,
+      "step": 4440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002779708463725467,
+      "loss": 1.2755,
+      "step": 4441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002779609429208817,
+      "loss": 0.1855,
+      "step": 4442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027795103742011423,
+      "loss": 0.5347,
+      "step": 4443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002779411298704029,
+      "loss": 0.3326,
+      "step": 4444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002779312202719064,
+      "loss": 0.364,
+      "step": 4445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027792130862478343,
+      "loss": 0.4792,
+      "step": 4446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002779113949291927,
+      "loss": 0.864,
+      "step": 4447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027790147918529284,
+      "loss": 0.5777,
+      "step": 4448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002778915613932428,
+      "loss": 0.0255,
+      "step": 4449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027788164155320146,
+      "loss": 0.4516,
+      "step": 4450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002778717196653274,
+      "loss": 0.3654,
+      "step": 4451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027786179572977977,
+      "loss": 0.5517,
+      "step": 4452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002778518697467173,
+      "loss": 0.1329,
+      "step": 4453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027784194171629904,
+      "loss": 0.0712,
+      "step": 4454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027783201163868393,
+      "loss": 0.4304,
+      "step": 4455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027782207951403104,
+      "loss": 0.0506,
+      "step": 4456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002778121453424993,
+      "loss": 0.6375,
+      "step": 4457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027780220912424793,
+      "loss": 0.6125,
+      "step": 4458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000277792270859436,
+      "loss": 0.692,
+      "step": 4459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027778233054822265,
+      "loss": 0.5271,
+      "step": 4460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027777238819076704,
+      "loss": 0.4906,
+      "step": 4461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027776244378722835,
+      "loss": 0.0133,
+      "step": 4462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002777524973377659,
+      "loss": 0.3871,
+      "step": 4463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027774254884253895,
+      "loss": 0.1396,
+      "step": 4464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002777325983017068,
+      "loss": 0.2706,
+      "step": 4465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002777226457154287,
+      "loss": 1.4036,
+      "step": 4466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027771269108386423,
+      "loss": 2.9023,
+      "step": 4467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002777027344071726,
+      "loss": 0.3598,
+      "step": 4468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002776927756855134,
+      "loss": 0.2591,
+      "step": 4469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027768281491904594,
+      "loss": 0.9047,
+      "step": 4470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002776728521079299,
+      "loss": 0.6496,
+      "step": 4471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027766288725232475,
+      "loss": 0.5866,
+      "step": 4472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027765292035239003,
+      "loss": 0.6965,
+      "step": 4473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027764295140828536,
+      "loss": 2.0685,
+      "step": 4474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027763298042017044,
+      "loss": 1.172,
+      "step": 4475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027762300738820486,
+      "loss": 0.3979,
+      "step": 4476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002776130323125483,
+      "loss": 0.4309,
+      "step": 4477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002776030551933606,
+      "loss": 0.3802,
+      "step": 4478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027759307603080147,
+      "loss": 0.007,
+      "step": 4479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002775830948250307,
+      "loss": 0.3232,
+      "step": 4480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027757311157620816,
+      "loss": 0.451,
+      "step": 4481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027756312628449375,
+      "loss": 0.4759,
+      "step": 4482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027755313895004726,
+      "loss": 0.5381,
+      "step": 4483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027754314957302865,
+      "loss": 0.8698,
+      "step": 4484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027753315815359797,
+      "loss": 0.4446,
+      "step": 4485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002775231646919151,
+      "loss": 0.3708,
+      "step": 4486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002775131691881402,
+      "loss": 0.9157,
+      "step": 4487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027750317164243325,
+      "loss": 0.5528,
+      "step": 4488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027749317205495434,
+      "loss": 2.487,
+      "step": 4489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002774831704258636,
+      "loss": 0.9498,
+      "step": 4490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002774731667553213,
+      "loss": 0.5479,
+      "step": 4491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002774631610434875,
+      "loss": 0.2179,
+      "step": 4492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002774531532905224,
+      "loss": 0.4739,
+      "step": 4493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002774431434965864,
+      "loss": 0.3543,
+      "step": 4494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027743313166183974,
+      "loss": 0.5958,
+      "step": 4495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027742311778644266,
+      "loss": 0.4422,
+      "step": 4496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002774131018705557,
+      "loss": 0.4815,
+      "step": 4497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027740308391433903,
+      "loss": 0.3579,
+      "step": 4498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002773930639179532,
+      "loss": 1.0003,
+      "step": 4499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027738304188155863,
+      "loss": 0.9886,
+      "step": 4500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027737301780531584,
+      "loss": 2.2835,
+      "step": 4501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002773629916893853,
+      "loss": 0.5661,
+      "step": 4502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002773529635339277,
+      "loss": 0.6287,
+      "step": 4503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002773429333391034,
+      "loss": 0.6207,
+      "step": 4504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002773329011050732,
+      "loss": 0.751,
+      "step": 4505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002773228668319976,
+      "loss": 0.3909,
+      "step": 4506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027731283052003744,
+      "loss": 0.9652,
+      "step": 4507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002773027921693534,
+      "loss": 0.5336,
+      "step": 4508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027729275178010616,
+      "loss": 0.4778,
+      "step": 4509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027728270935245655,
+      "loss": 0.4051,
+      "step": 4510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027727266488656536,
+      "loss": 0.3682,
+      "step": 4511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002772626183825934,
+      "loss": 0.425,
+      "step": 4512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002772525698407017,
+      "loss": 0.7613,
+      "step": 4513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027724251926105107,
+      "loss": 0.3795,
+      "step": 4514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002772324666438024,
+      "loss": 0.5998,
+      "step": 4515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027722241198911674,
+      "loss": 0.7604,
+      "step": 4516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002772123552971551,
+      "loss": 0.2375,
+      "step": 4517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002772022965680785,
+      "loss": 0.7477,
+      "step": 4518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000277192235802048,
+      "loss": 0.1486,
+      "step": 4519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027718217299922476,
+      "loss": 0.4627,
+      "step": 4520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002771721081597699,
+      "loss": 0.2302,
+      "step": 4521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002771620412838446,
+      "loss": 0.0929,
+      "step": 4522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027715197237161005,
+      "loss": 0.3909,
+      "step": 4523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002771419014232274,
+      "loss": 0.6392,
+      "step": 4524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027713182843885816,
+      "loss": 0.7922,
+      "step": 4525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027712175341866336,
+      "loss": 0.4876,
+      "step": 4526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027711167636280455,
+      "loss": 0.2769,
+      "step": 4527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027710159727144294,
+      "loss": 0.0404,
+      "step": 4528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027709151614474005,
+      "loss": 1.3165,
+      "step": 4529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027708143298285724,
+      "loss": 0.3936,
+      "step": 4530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000277071347785956,
+      "loss": 0.1908,
+      "step": 4531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027706126055419784,
+      "loss": 0.5606,
+      "step": 4532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027705117128774435,
+      "loss": 0.2655,
+      "step": 4533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002770410799867569,
+      "loss": 1.4874,
+      "step": 4534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027703098665139735,
+      "loss": 0.1534,
+      "step": 4535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002770208912818271,
+      "loss": 0.2581,
+      "step": 4536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027701079387820797,
+      "loss": 1.2315,
+      "step": 4537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027700069444070157,
+      "loss": 1.0477,
+      "step": 4538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027699059296946966,
+      "loss": 0.3013,
+      "step": 4539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002769804894646739,
+      "loss": 0.2818,
+      "step": 4540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027697038392647625,
+      "loss": 0.5636,
+      "step": 4541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002769602763550385,
+      "loss": 0.1991,
+      "step": 4542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002769501667505224,
+      "loss": 0.0335,
+      "step": 4543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027694005511308997,
+      "loss": 0.4355,
+      "step": 4544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000276929941442903,
+      "loss": 0.9795,
+      "step": 4545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002769198257401235,
+      "loss": 0.4668,
+      "step": 4546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027690970800491353,
+      "loss": 1.2355,
+      "step": 4547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000276899588237435,
+      "loss": 0.4232,
+      "step": 4548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027688946643785004,
+      "loss": 0.7761,
+      "step": 4549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002768793426063207,
+      "loss": 0.0287,
+      "step": 4550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027686921674300913,
+      "loss": 2.5073,
+      "step": 4551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027685908884807745,
+      "loss": 0.4831,
+      "step": 4552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002768489589216878,
+      "loss": 0.5816,
+      "step": 4553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027683882696400255,
+      "loss": 0.479,
+      "step": 4554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027682869297518376,
+      "loss": 0.3167,
+      "step": 4555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002768185569553938,
+      "loss": 0.2036,
+      "step": 4556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027680841890479504,
+      "loss": 0.3015,
+      "step": 4557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002767982788235497,
+      "loss": 0.1785,
+      "step": 4558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027678813671182023,
+      "loss": 0.592,
+      "step": 4559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002767779925697691,
+      "loss": 2.0463,
+      "step": 4560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027676784639755857,
+      "loss": 1.2711,
+      "step": 4561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027675769819535134,
+      "loss": 0.209,
+      "step": 4562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027674754796330975,
+      "loss": 0.3874,
+      "step": 4563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002767373957015964,
+      "loss": 0.5306,
+      "step": 4564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002767272414103739,
+      "loss": 0.0265,
+      "step": 4565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002767170850898048,
+      "loss": 0.1205,
+      "step": 4566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027670692674005177,
+      "loss": 0.1229,
+      "step": 4567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027669676636127744,
+      "loss": 0.2017,
+      "step": 4568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027668660395364453,
+      "loss": 0.0966,
+      "step": 4569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002766764395173159,
+      "loss": 0.3506,
+      "step": 4570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002766662730524541,
+      "loss": 1.0515,
+      "step": 4571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027665610455922203,
+      "loss": 0.3124,
+      "step": 4572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027664593403778256,
+      "loss": 1.0887,
+      "step": 4573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027663576148829853,
+      "loss": 0.0447,
+      "step": 4574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027662558691093285,
+      "loss": 0.2446,
+      "step": 4575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002766154103058484,
+      "loss": 0.8983,
+      "step": 4576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027660523167320815,
+      "loss": 0.4363,
+      "step": 4577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002765950510131752,
+      "loss": 0.5447,
+      "step": 4578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027658486832591244,
+      "loss": 0.3466,
+      "step": 4579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000276574683611583,
+      "loss": 0.8632,
+      "step": 4580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027656449687034996,
+      "loss": 2.6544,
+      "step": 4581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002765543081023765,
+      "loss": 0.0464,
+      "step": 4582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002765441173078257,
+      "loss": 0.0081,
+      "step": 4583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002765339244868608,
+      "loss": 0.8443,
+      "step": 4584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027652372963964496,
+      "loss": 2.3009,
+      "step": 4585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027651353276634145,
+      "loss": 0.954,
+      "step": 4586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027650333386711363,
+      "loss": 0.9177,
+      "step": 4587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027649313294212475,
+      "loss": 1.4751,
+      "step": 4588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027648292999153825,
+      "loss": 0.0168,
+      "step": 4589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002764727250155174,
+      "loss": 0.614,
+      "step": 4590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027646251801422564,
+      "loss": 0.5569,
+      "step": 4591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027645230898782643,
+      "loss": 1.1219,
+      "step": 4592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027644209793648337,
+      "loss": 0.1616,
+      "step": 4593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002764318848603598,
+      "loss": 0.5348,
+      "step": 4594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027642166975961935,
+      "loss": 0.4356,
+      "step": 4595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002764114526344256,
+      "loss": 0.6636,
+      "step": 4596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027640123348494217,
+      "loss": 0.0339,
+      "step": 4597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002763910123113327,
+      "loss": 0.5001,
+      "step": 4598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002763807891137608,
+      "loss": 0.5303,
+      "step": 4599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027637056389239026,
+      "loss": 1.057,
+      "step": 4600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002763603366473848,
+      "loss": 0.7144,
+      "step": 4601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027635010737890814,
+      "loss": 0.37,
+      "step": 4602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027633987608712424,
+      "loss": 0.1531,
+      "step": 4603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027632964277219676,
+      "loss": 0.3102,
+      "step": 4604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002763194074342896,
+      "loss": 0.8325,
+      "step": 4605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002763091700735668,
+      "loss": 0.9156,
+      "step": 4606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002762989306901922,
+      "loss": 0.4611,
+      "step": 4607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002762886892843297,
+      "loss": 0.6333,
+      "step": 4608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027627844585614343,
+      "loss": 0.5614,
+      "step": 4609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027626820040579735,
+      "loss": 0.6592,
+      "step": 4610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002762579529334556,
+      "loss": 0.7703,
+      "step": 4611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002762477034392821,
+      "loss": 0.4122,
+      "step": 4612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002762374519234412,
+      "loss": 0.5301,
+      "step": 4613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002762271983860969,
+      "loss": 1.3707,
+      "step": 4614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002762169428274135,
+      "loss": 0.8262,
+      "step": 4615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002762066852475552,
+      "loss": 0.1937,
+      "step": 4616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027619642564668623,
+      "loss": 0.3326,
+      "step": 4617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002761861640249709,
+      "loss": 0.6488,
+      "step": 4618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002761759003825735,
+      "loss": 0.4868,
+      "step": 4619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002761656347196585,
+      "loss": 0.8142,
+      "step": 4620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002761553670363901,
+      "loss": 0.5672,
+      "step": 4621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002761450973329329,
+      "loss": 0.1039,
+      "step": 4622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027613482560945123,
+      "loss": 1.6641,
+      "step": 4623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027612455186610966,
+      "loss": 0.0757,
+      "step": 4624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002761142761030727,
+      "loss": 0.6269,
+      "step": 4625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002761039983205048,
+      "loss": 0.7405,
+      "step": 4626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027609371851857075,
+      "loss": 0.4267,
+      "step": 4627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027608343669743493,
+      "loss": 0.5557,
+      "step": 4628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027607315285726216,
+      "loss": 0.5575,
+      "step": 4629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000276062866998217,
+      "loss": 0.3985,
+      "step": 4630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002760525791204643,
+      "loss": 1.904,
+      "step": 4631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027604228922416865,
+      "loss": 0.5421,
+      "step": 4632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027603199730949496,
+      "loss": 1.7813,
+      "step": 4633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002760217033766079,
+      "loss": 0.447,
+      "step": 4634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002760114074256724,
+      "loss": 0.7645,
+      "step": 4635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002760011094568534,
+      "loss": 0.6218,
+      "step": 4636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027599080947031575,
+      "loss": 1.9143,
+      "step": 4637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027598050746622426,
+      "loss": 0.7057,
+      "step": 4638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002759702034447441,
+      "loss": 0.6541,
+      "step": 4639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027595989740604015,
+      "loss": 0.4559,
+      "step": 4640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002759495893502775,
+      "loss": 0.752,
+      "step": 4641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027593927927762116,
+      "loss": 1.313,
+      "step": 4642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027592896718823634,
+      "loss": 0.4693,
+      "step": 4643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027591865308228806,
+      "loss": 0.3868,
+      "step": 4644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027590833695994156,
+      "loss": 0.5622,
+      "step": 4645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027589801882136196,
+      "loss": 0.2214,
+      "step": 4646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027588769866671454,
+      "loss": 0.2573,
+      "step": 4647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002758773764961646,
+      "loss": 0.5488,
+      "step": 4648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027586705230987736,
+      "loss": 0.2722,
+      "step": 4649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027585672610801816,
+      "loss": 0.7463,
+      "step": 4650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002758463978907524,
+      "loss": 0.362,
+      "step": 4651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002758360676582455,
+      "loss": 0.2928,
+      "step": 4652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027582573541066276,
+      "loss": 1.4579,
+      "step": 4653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002758154011481697,
+      "loss": 0.791,
+      "step": 4654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002758050648709318,
+      "loss": 0.4922,
+      "step": 4655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002757947265791147,
+      "loss": 0.8339,
+      "step": 4656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027578438627288374,
+      "loss": 0.294,
+      "step": 4657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027577404395240466,
+      "loss": 0.9584,
+      "step": 4658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000275763699617843,
+      "loss": 0.4774,
+      "step": 4659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027575335326936445,
+      "loss": 0.5848,
+      "step": 4660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027574300490713465,
+      "loss": 0.5713,
+      "step": 4661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027573265453131933,
+      "loss": 0.4393,
+      "step": 4662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002757223021420843,
+      "loss": 0.4246,
+      "step": 4663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002757119477395953,
+      "loss": 0.2423,
+      "step": 4664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027570159132401807,
+      "loss": 0.069,
+      "step": 4665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027569123289551846,
+      "loss": 0.8783,
+      "step": 4666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027568087245426247,
+      "loss": 0.0685,
+      "step": 4667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002756705100004159,
+      "loss": 0.0695,
+      "step": 4668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027566014553414474,
+      "loss": 0.6239,
+      "step": 4669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027564977905561493,
+      "loss": 0.5238,
+      "step": 4670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002756394105649925,
+      "loss": 0.1801,
+      "step": 4671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027562904006244343,
+      "loss": 1.2727,
+      "step": 4672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027561866754813384,
+      "loss": 0.4046,
+      "step": 4673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002756082930222298,
+      "loss": 0.0415,
+      "step": 4674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002755979164848975,
+      "loss": 0.3237,
+      "step": 4675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000275587537936303,
+      "loss": 0.6173,
+      "step": 4676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027557715737661266,
+      "loss": 1.125,
+      "step": 4677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002755667748059925,
+      "loss": 0.8653,
+      "step": 4678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027555639022460896,
+      "loss": 0.6549,
+      "step": 4679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027554600363262826,
+      "loss": 0.4592,
+      "step": 4680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027553561503021673,
+      "loss": 0.012,
+      "step": 4681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027552522441754075,
+      "loss": 0.131,
+      "step": 4682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002755148317947666,
+      "loss": 0.9353,
+      "step": 4683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002755044371620609,
+      "loss": 0.6177,
+      "step": 4684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002754940405195899,
+      "loss": 0.5473,
+      "step": 4685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002754836418675202,
+      "loss": 0.7061,
+      "step": 4686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002754732412060183,
+      "loss": 0.4406,
+      "step": 4687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002754628385352508,
+      "loss": 0.7043,
+      "step": 4688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027545243385538416,
+      "loss": 0.6108,
+      "step": 4689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027544202716658515,
+      "loss": 0.5092,
+      "step": 4690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027543161846902027,
+      "loss": 0.6531,
+      "step": 4691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002754212077628563,
+      "loss": 0.2792,
+      "step": 4692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027541079504825994,
+      "loss": 0.0057,
+      "step": 4693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002754003803253979,
+      "loss": 0.9297,
+      "step": 4694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027538996359443695,
+      "loss": 3.6166,
+      "step": 4695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027537954485554395,
+      "loss": 4.293,
+      "step": 4696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027536912410888563,
+      "loss": 0.5778,
+      "step": 4697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000275358701354629,
+      "loss": 0.5039,
+      "step": 4698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002753482765929409,
+      "loss": 0.012,
+      "step": 4699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027533784982398826,
+      "loss": 0.618,
+      "step": 4700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027532742104793807,
+      "loss": 0.7735,
+      "step": 4701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027531699026495733,
+      "loss": 0.1862,
+      "step": 4702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002753065574752131,
+      "loss": 1.042,
+      "step": 4703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027529612267887234,
+      "loss": 1.1202,
+      "step": 4704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027528568587610223,
+      "loss": 0.2502,
+      "step": 4705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027527524706706994,
+      "loss": 0.0444,
+      "step": 4706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027526480625194254,
+      "loss": 0.967,
+      "step": 4707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002752543634308872,
+      "loss": 0.2039,
+      "step": 4708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027524391860407133,
+      "loss": 0.4194,
+      "step": 4709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000275233471771662,
+      "loss": 0.9743,
+      "step": 4710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002752230229338266,
+      "loss": 0.4346,
+      "step": 4711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002752125720907324,
+      "loss": 0.8012,
+      "step": 4712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027520211924254677,
+      "loss": 1.4912,
+      "step": 4713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002751916643894371,
+      "loss": 0.2543,
+      "step": 4714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027518120753157086,
+      "loss": 0.4594,
+      "step": 4715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002751707486691154,
+      "loss": 0.9184,
+      "step": 4716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002751602878022383,
+      "loss": 0.2917,
+      "step": 4717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000275149824931107,
+      "loss": 0.6251,
+      "step": 4718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027513936005588906,
+      "loss": 0.3292,
+      "step": 4719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027512889317675215,
+      "loss": 0.4515,
+      "step": 4720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002751184242938637,
+      "loss": 2.0607,
+      "step": 4721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002751079534073915,
+      "loss": 0.6278,
+      "step": 4722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002750974805175032,
+      "loss": 0.1351,
+      "step": 4723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027508700562436656,
+      "loss": 0.3836,
+      "step": 4724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002750765287281492,
+      "loss": 0.2381,
+      "step": 4725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002750660498290189,
+      "loss": 0.4457,
+      "step": 4726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027505556892714353,
+      "loss": 0.6027,
+      "step": 4727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027504508602269094,
+      "loss": 0.4864,
+      "step": 4728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027503460111582894,
+      "loss": 0.0041,
+      "step": 4729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027502411420672547,
+      "loss": 0.14,
+      "step": 4730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002750136252955484,
+      "loss": 2.817,
+      "step": 4731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027500313438246583,
+      "loss": 0.0493,
+      "step": 4732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002749926414676456,
+      "loss": 0.1506,
+      "step": 4733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027498214655125584,
+      "loss": 0.0587,
+      "step": 4734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027497164963346453,
+      "loss": 0.1424,
+      "step": 4735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027496115071443984,
+      "loss": 0.3757,
+      "step": 4736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002749506497943499,
+      "loss": 0.5527,
+      "step": 4737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027494014687336275,
+      "loss": 0.1445,
+      "step": 4738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027492964195164666,
+      "loss": 0.1132,
+      "step": 4739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002749191350293699,
+      "loss": 0.7832,
+      "step": 4740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002749086261067006,
+      "loss": 0.7999,
+      "step": 4741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027489811518380714,
+      "loss": 0.0719,
+      "step": 4742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027488760226085786,
+      "loss": 0.7838,
+      "step": 4743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000274877087338021,
+      "loss": 0.2902,
+      "step": 4744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027486657041546505,
+      "loss": 2.3262,
+      "step": 4745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002748560514933584,
+      "loss": 0.0038,
+      "step": 4746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002748455305718694,
+      "loss": 1.9088,
+      "step": 4747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002748350076511666,
+      "loss": 1.0031,
+      "step": 4748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027482448273141854,
+      "loss": 0.7066,
+      "step": 4749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027481395581279366,
+      "loss": 0.0082,
+      "step": 4750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027480342689546066,
+      "loss": 0.2861,
+      "step": 4751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002747928959795881,
+      "loss": 1.064,
+      "step": 4752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027478236306534455,
+      "loss": 0.3487,
+      "step": 4753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027477182815289873,
+      "loss": 0.1907,
+      "step": 4754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027476129124241934,
+      "loss": 0.5524,
+      "step": 4755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002747507523340751,
+      "loss": 0.2912,
+      "step": 4756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027474021142803484,
+      "loss": 0.2136,
+      "step": 4757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027472966852446726,
+      "loss": 0.7964,
+      "step": 4758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027471912362354124,
+      "loss": 0.7936,
+      "step": 4759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027470857672542564,
+      "loss": 1.2169,
+      "step": 4760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002746980278302893,
+      "loss": 0.2389,
+      "step": 4761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027468747693830127,
+      "loss": 0.7077,
+      "step": 4762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027467692404963034,
+      "loss": 1.2584,
+      "step": 4763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027466636916444567,
+      "loss": 0.9652,
+      "step": 4764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002746558122829161,
+      "loss": 2.6954,
+      "step": 4765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002746452534052109,
+      "loss": 0.3592,
+      "step": 4766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027463469253149894,
+      "loss": 0.1239,
+      "step": 4767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002746241296619494,
+      "loss": 0.2166,
+      "step": 4768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027461356479673157,
+      "loss": 0.8415,
+      "step": 4769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027460299793601446,
+      "loss": 0.4783,
+      "step": 4770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027459242907996735,
+      "loss": 0.8305,
+      "step": 4771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027458185822875947,
+      "loss": 1.3848,
+      "step": 4772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002745712853825601,
+      "loss": 0.0085,
+      "step": 4773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027456071054153856,
+      "loss": 1.0671,
+      "step": 4774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002745501337058642,
+      "loss": 2.9146,
+      "step": 4775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002745395548757063,
+      "loss": 1.2473,
+      "step": 4776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002745289740512344,
+      "loss": 0.7451,
+      "step": 4777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002745183912326178,
+      "loss": 0.4809,
+      "step": 4778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027450780642002614,
+      "loss": 1.3155,
+      "step": 4779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002744972196136287,
+      "loss": 3.3868,
+      "step": 4780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027448663081359526,
+      "loss": 0.5588,
+      "step": 4781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002744760400200952,
+      "loss": 0.0057,
+      "step": 4782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027446544723329813,
+      "loss": 0.2865,
+      "step": 4783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002744548524533738,
+      "loss": 0.5808,
+      "step": 4784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027444425568049174,
+      "loss": 0.5499,
+      "step": 4785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027443365691482167,
+      "loss": 0.4741,
+      "step": 4786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002744230561565334,
+      "loss": 0.5561,
+      "step": 4787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002744124534057966,
+      "loss": 0.4342,
+      "step": 4788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027440184866278107,
+      "loss": 0.7738,
+      "step": 4789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002743912419276566,
+      "loss": 0.0334,
+      "step": 4790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002743806332005931,
+      "loss": 0.0221,
+      "step": 4791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002743700224817604,
+      "loss": 0.4243,
+      "step": 4792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002743594097713285,
+      "loss": 0.1018,
+      "step": 4793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027434879506946724,
+      "loss": 0.0681,
+      "step": 4794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002743381783763467,
+      "loss": 0.3346,
+      "step": 4795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027432755969213676,
+      "loss": 0.8675,
+      "step": 4796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002743169390170076,
+      "loss": 6.9629,
+      "step": 4797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002743063163511292,
+      "loss": 0.0362,
+      "step": 4798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027429569169467173,
+      "loss": 0.8943,
+      "step": 4799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027428506504780534,
+      "loss": 0.622,
+      "step": 4800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002742744364107001,
+      "loss": 0.0377,
+      "step": 4801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002742638057835263,
+      "loss": 0.0243,
+      "step": 4802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027425317316645407,
+      "loss": 0.2744,
+      "step": 4803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027424253855965386,
+      "loss": 0.5951,
+      "step": 4804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027423190196329577,
+      "loss": 0.4969,
+      "step": 4805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027422126337755024,
+      "loss": 1.6406,
+      "step": 4806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002742106228025876,
+      "loss": 0.5822,
+      "step": 4807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741999802385783,
+      "loss": 0.0178,
+      "step": 4808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027418933568569266,
+      "loss": 0.2156,
+      "step": 4809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741786891441012,
+      "loss": 0.4591,
+      "step": 4810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741680406139744,
+      "loss": 0.3907,
+      "step": 4811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741573900954828,
+      "loss": 0.8634,
+      "step": 4812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741467375887969,
+      "loss": 0.0036,
+      "step": 4813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741360830940874,
+      "loss": 0.0086,
+      "step": 4814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027412542661152476,
+      "loss": 0.266,
+      "step": 4815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741147681412797,
+      "loss": 1.0233,
+      "step": 4816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002741041076835229,
+      "loss": 0.7314,
+      "step": 4817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027409344523842506,
+      "loss": 0.4349,
+      "step": 4818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027408278080615696,
+      "loss": 0.7939,
+      "step": 4819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002740721143868893,
+      "loss": 0.6234,
+      "step": 4820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000274061445980793,
+      "loss": 0.1688,
+      "step": 4821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027405077558803885,
+      "loss": 0.5925,
+      "step": 4822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002740401032087976,
+      "loss": 0.0503,
+      "step": 4823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027402942884324034,
+      "loss": 0.0141,
+      "step": 4824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002740187524915379,
+      "loss": 1.665,
+      "step": 4825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027400807415386123,
+      "loss": 0.8385,
+      "step": 4826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002739973938303814,
+      "loss": 2.6535,
+      "step": 4827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002739867115212694,
+      "loss": 0.7462,
+      "step": 4828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027397602722669626,
+      "loss": 0.2765,
+      "step": 4829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002739653409468332,
+      "loss": 0.5416,
+      "step": 4830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027395465268185114,
+      "loss": 0.8032,
+      "step": 4831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027394396243192134,
+      "loss": 0.4546,
+      "step": 4832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027393327019721505,
+      "loss": 0.642,
+      "step": 4833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027392257597790346,
+      "loss": 0.2761,
+      "step": 4834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027391187977415777,
+      "loss": 0.3131,
+      "step": 4835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002739011815861493,
+      "loss": 4.2344,
+      "step": 4836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027389048141404935,
+      "loss": 0.6007,
+      "step": 4837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027387977925802924,
+      "loss": 0.7195,
+      "step": 4838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002738690751182604,
+      "loss": 1.0042,
+      "step": 4839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027385836899491425,
+      "loss": 0.7874,
+      "step": 4840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027384766088816214,
+      "loss": 0.3711,
+      "step": 4841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002738369507981757,
+      "loss": 0.2568,
+      "step": 4842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002738262387251263,
+      "loss": 0.5631,
+      "step": 4843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027381552466918555,
+      "loss": 0.0066,
+      "step": 4844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027380480863052495,
+      "loss": 0.2871,
+      "step": 4845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027379409060931617,
+      "loss": 0.8681,
+      "step": 4846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027378337060573083,
+      "loss": 0.8178,
+      "step": 4847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027377264861994056,
+      "loss": 1.1026,
+      "step": 4848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002737619246521171,
+      "loss": 0.341,
+      "step": 4849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027375119870243214,
+      "loss": 0.0097,
+      "step": 4850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027374047077105746,
+      "loss": 0.9279,
+      "step": 4851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027372974085816486,
+      "loss": 0.2899,
+      "step": 4852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027371900896392613,
+      "loss": 1.5195,
+      "step": 4853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027370827508851317,
+      "loss": 0.6507,
+      "step": 4854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002736975392320978,
+      "loss": 0.0165,
+      "step": 4855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027368680139485204,
+      "loss": 1.4561,
+      "step": 4856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002736760615769478,
+      "loss": 0.328,
+      "step": 4857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027366531977855696,
+      "loss": 0.5408,
+      "step": 4858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002736545759998517,
+      "loss": 0.5016,
+      "step": 4859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027364383024100394,
+      "loss": 1.502,
+      "step": 4860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027363308250218576,
+      "loss": 1.2394,
+      "step": 4861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002736223327835694,
+      "loss": 0.914,
+      "step": 4862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002736115810853269,
+      "loss": 0.9011,
+      "step": 4863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002736008274076304,
+      "loss": 0.5735,
+      "step": 4864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002735900717506522,
+      "loss": 0.6193,
+      "step": 4865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002735793141145644,
+      "loss": 0.4128,
+      "step": 4866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002735685544995394,
+      "loss": 1.4158,
+      "step": 4867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002735577929057495,
+      "loss": 0.3083,
+      "step": 4868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002735470293333669,
+      "loss": 0.0344,
+      "step": 4869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002735362637825641,
+      "loss": 0.6275,
+      "step": 4870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027352549625351343,
+      "loss": 0.449,
+      "step": 4871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027351472674638734,
+      "loss": 0.0046,
+      "step": 4872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002735039552613582,
+      "loss": 0.0423,
+      "step": 4873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002734931817985986,
+      "loss": 0.5048,
+      "step": 4874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002734824063582811,
+      "loss": 0.1596,
+      "step": 4875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027347162894057807,
+      "loss": 0.7776,
+      "step": 4876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002734608495456622,
+      "loss": 0.646,
+      "step": 4877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027345006817370627,
+      "loss": 0.0373,
+      "step": 4878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027343928482488264,
+      "loss": 0.4276,
+      "step": 4879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027342849949936416,
+      "loss": 0.6492,
+      "step": 4880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002734177121973235,
+      "loss": 2.0039,
+      "step": 4881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002734069229189334,
+      "loss": 1.2055,
+      "step": 4882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002733961316643666,
+      "loss": 0.9631,
+      "step": 4883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000273385338433796,
+      "loss": 0.005,
+      "step": 4884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002733745432273944,
+      "loss": 0.4519,
+      "step": 4885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027336374604533455,
+      "loss": 0.2222,
+      "step": 4886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002733529468877895,
+      "loss": 1.114,
+      "step": 4887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002733421457549322,
+      "loss": 0.2547,
+      "step": 4888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002733313426469355,
+      "loss": 0.491,
+      "step": 4889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027332053756397236,
+      "loss": 0.0197,
+      "step": 4890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000273309730506216,
+      "loss": 0.3727,
+      "step": 4891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027329892147383934,
+      "loss": 0.1748,
+      "step": 4892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002732881104670155,
+      "loss": 0.0126,
+      "step": 4893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027327729748591757,
+      "loss": 0.3502,
+      "step": 4894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027326648253071876,
+      "loss": 0.8976,
+      "step": 4895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002732556656015922,
+      "loss": 0.127,
+      "step": 4896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027324484669871116,
+      "loss": 0.2634,
+      "step": 4897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002732340258222489,
+      "loss": 0.0018,
+      "step": 4898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002732232029723786,
+      "loss": 0.5519,
+      "step": 4899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002732123781492736,
+      "loss": 0.9941,
+      "step": 4900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027320155135310736,
+      "loss": 1.0035,
+      "step": 4901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027319072258405317,
+      "loss": 0.8511,
+      "step": 4902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027317989184228445,
+      "loss": 0.2233,
+      "step": 4903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027316905912797457,
+      "loss": 0.5012,
+      "step": 4904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027315822444129716,
+      "loss": 0.2864,
+      "step": 4905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027314738778242553,
+      "loss": 2.457,
+      "step": 4906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002731365491515333,
+      "loss": 0.6246,
+      "step": 4907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027312570854879406,
+      "loss": 0.4109,
+      "step": 4908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002731148659743814,
+      "loss": 0.0027,
+      "step": 4909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027310402142846895,
+      "loss": 2.127,
+      "step": 4910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730931749112303,
+      "loss": 0.9322,
+      "step": 4911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027308232642283924,
+      "loss": 0.6542,
+      "step": 4912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730714759634694,
+      "loss": 0.7795,
+      "step": 4913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730606235332946,
+      "loss": 0.8784,
+      "step": 4914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730497691324886,
+      "loss": 2.2207,
+      "step": 4915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730389127612252,
+      "loss": 0.426,
+      "step": 4916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730280544196783,
+      "loss": 0.0403,
+      "step": 4917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730171941080217,
+      "loss": 1.2227,
+      "step": 4918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002730063318264294,
+      "loss": 0.2917,
+      "step": 4919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027299546757507534,
+      "loss": 0.0301,
+      "step": 4920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027298460135413336,
+      "loss": 0.4721,
+      "step": 4921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002729737331637776,
+      "loss": 1.0159,
+      "step": 4922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002729628630041821,
+      "loss": 0.0467,
+      "step": 4923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027295199087552086,
+      "loss": 0.8412,
+      "step": 4924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000272941116777968,
+      "loss": 1.5362,
+      "step": 4925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027293024071169765,
+      "loss": 0.3249,
+      "step": 4926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000272919362676884,
+      "loss": 0.2499,
+      "step": 4927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002729084826737013,
+      "loss": 0.296,
+      "step": 4928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027289760070232363,
+      "loss": 1.2305,
+      "step": 4929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027288671676292534,
+      "loss": 0.3735,
+      "step": 4930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027287583085568076,
+      "loss": 0.6428,
+      "step": 4931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027286494298076416,
+      "loss": 0.4776,
+      "step": 4932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027285405313834984,
+      "loss": 0.2786,
+      "step": 4933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027284316132861224,
+      "loss": 0.4104,
+      "step": 4934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027283226755172576,
+      "loss": 0.5358,
+      "step": 4935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002728213718078649,
+      "loss": 0.6052,
+      "step": 4936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027281047409720405,
+      "loss": 0.0271,
+      "step": 4937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002727995744199179,
+      "loss": 0.4802,
+      "step": 4938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027278867277618073,
+      "loss": 0.5239,
+      "step": 4939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027277776916616726,
+      "loss": 0.4303,
+      "step": 4940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027276686359005214,
+      "loss": 0.0595,
+      "step": 4941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027275595604800987,
+      "loss": 0.3832,
+      "step": 4942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002727450465402152,
+      "loss": 0.8376,
+      "step": 4943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002727341350668429,
+      "loss": 0.4991,
+      "step": 4944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027272322162806757,
+      "loss": 0.3574,
+      "step": 4945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027271230622406405,
+      "loss": 1.6417,
+      "step": 4946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027270138885500714,
+      "loss": 0.4026,
+      "step": 4947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027269046952107156,
+      "loss": 0.8391,
+      "step": 4948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002726795482224323,
+      "loss": 0.3403,
+      "step": 4949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002726686249592642,
+      "loss": 0.0131,
+      "step": 4950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002726576997317421,
+      "loss": 0.1545,
+      "step": 4951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002726467725400411,
+      "loss": 0.1672,
+      "step": 4952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002726358433843361,
+      "loss": 1.9727,
+      "step": 4953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027262491226480207,
+      "loss": 0.4103,
+      "step": 4954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002726139791816141,
+      "loss": 0.2517,
+      "step": 4955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027260304413494733,
+      "loss": 0.266,
+      "step": 4956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027259210712497674,
+      "loss": 0.2906,
+      "step": 4957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002725811681518776,
+      "loss": 0.4815,
+      "step": 4958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000272570227215825,
+      "loss": 0.0149,
+      "step": 4959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002725592843169941,
+      "loss": 0.0135,
+      "step": 4960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002725483394555603,
+      "loss": 0.3552,
+      "step": 4961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027253739263169874,
+      "loss": 0.2171,
+      "step": 4962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027252644384558473,
+      "loss": 0.2128,
+      "step": 4963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027251549309739357,
+      "loss": 0.4463,
+      "step": 4964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002725045403873007,
+      "loss": 0.5123,
+      "step": 4965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027249358571548144,
+      "loss": 0.4418,
+      "step": 4966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027248262908211127,
+      "loss": 0.7798,
+      "step": 4967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002724716704873656,
+      "loss": 0.9526,
+      "step": 4968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027246070993141994,
+      "loss": 1.4038,
+      "step": 4969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002724497474144498,
+      "loss": 0.3216,
+      "step": 4970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027243878293663074,
+      "loss": 0.4654,
+      "step": 4971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002724278164981383,
+      "loss": 0.0053,
+      "step": 4972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002724168480991481,
+      "loss": 0.0422,
+      "step": 4973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002724058777398358,
+      "loss": 0.4291,
+      "step": 4974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002723949054203771,
+      "loss": 0.2068,
+      "step": 4975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002723839311409477,
+      "loss": 0.3646,
+      "step": 4976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002723729549017233,
+      "loss": 0.3246,
+      "step": 4977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002723619767028797,
+      "loss": 0.4006,
+      "step": 4978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002723509965445926,
+      "loss": 0.0094,
+      "step": 4979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000272340014427038,
+      "loss": 0.5591,
+      "step": 4980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002723290303503916,
+      "loss": 0.8188,
+      "step": 4981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027231804431482944,
+      "loss": 0.8479,
+      "step": 4982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027230705632052735,
+      "loss": 0.0123,
+      "step": 4983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002722960663676613,
+      "loss": 0.7712,
+      "step": 4984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002722850744564073,
+      "loss": 1.0293,
+      "step": 4985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002722740805869414,
+      "loss": 0.7461,
+      "step": 4986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002722630847594395,
+      "loss": 0.4057,
+      "step": 4987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027225208697407785,
+      "loss": 0.1177,
+      "step": 4988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002722410872310325,
+      "loss": 0.4867,
+      "step": 4989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027223008553047966,
+      "loss": 1.1861,
+      "step": 4990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002722190818725953,
+      "loss": 0.1751,
+      "step": 4991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002722080762575559,
+      "loss": 0.6969,
+      "step": 4992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002721970686855375,
+      "loss": 1.5541,
+      "step": 4993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027218605915671647,
+      "loss": 0.0031,
+      "step": 4994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002721750476712691,
+      "loss": 1.969,
+      "step": 4995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002721640342293716,
+      "loss": 1.442,
+      "step": 4996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002721530188312005,
+      "loss": 0.2917,
+      "step": 4997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027214200147693214,
+      "loss": 0.3055,
+      "step": 4998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027213098216674294,
+      "loss": 0.2686,
+      "step": 4999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002721199609008093,
+      "loss": 0.215,
+      "step": 5000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002721089376793078,
+      "loss": 0.414,
+      "step": 5001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027209791250241495,
+      "loss": 0.0051,
+      "step": 5002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027208688537030723,
+      "loss": 0.7594,
+      "step": 5003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002720758562831613,
+      "loss": 0.6186,
+      "step": 5004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002720648252411537,
+      "loss": 0.0332,
+      "step": 5005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027205379224446113,
+      "loss": 1.767,
+      "step": 5006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002720427572932603,
+      "loss": 0.6813,
+      "step": 5007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002720317203877278,
+      "loss": 0.3675,
+      "step": 5008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027202068152804053,
+      "loss": 0.6247,
+      "step": 5009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002720096407143751,
+      "loss": 0.7072,
+      "step": 5010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002719985979469084,
+      "loss": 0.5464,
+      "step": 5011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027198755322581727,
+      "loss": 0.0139,
+      "step": 5012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002719765065512785,
+      "loss": 0.3164,
+      "step": 5013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002719654579234691,
+      "loss": 1.0372,
+      "step": 5014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002719544073425659,
+      "loss": 0.4983,
+      "step": 5015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027194335480874594,
+      "loss": 0.734,
+      "step": 5016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027193230032218613,
+      "loss": 0.7844,
+      "step": 5017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027192124388306354,
+      "loss": 0.9148,
+      "step": 5018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002719101854915552,
+      "loss": 0.4116,
+      "step": 5019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027189912514783814,
+      "loss": 0.8859,
+      "step": 5020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002718880628520896,
+      "loss": 0.2659,
+      "step": 5021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002718769986044867,
+      "loss": 2.3031,
+      "step": 5022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027186593240520647,
+      "loss": 0.6975,
+      "step": 5023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027185486425442636,
+      "loss": 0.3944,
+      "step": 5024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027184379415232335,
+      "loss": 0.727,
+      "step": 5025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002718327220990749,
+      "loss": 0.2972,
+      "step": 5026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027182164809485827,
+      "loss": 0.3164,
+      "step": 5027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027181057213985073,
+      "loss": 0.4184,
+      "step": 5028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027179949423422975,
+      "loss": 1.7531,
+      "step": 5029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002717884143781726,
+      "loss": 0.6555,
+      "step": 5030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002717773325718568,
+      "loss": 0.3719,
+      "step": 5031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027176624881545977,
+      "loss": 0.4932,
+      "step": 5032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027175516310915904,
+      "loss": 1.0633,
+      "step": 5033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027174407545313206,
+      "loss": 0.0032,
+      "step": 5034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002717329858475565,
+      "loss": 4.0898,
+      "step": 5035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027172189429260974,
+      "loss": 0.0055,
+      "step": 5036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027171080078846965,
+      "loss": 0.5547,
+      "step": 5037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002716997053353136,
+      "loss": 0.0311,
+      "step": 5038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002716886079333196,
+      "loss": 0.3363,
+      "step": 5039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002716775085826651,
+      "loss": 0.4997,
+      "step": 5040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027166640728352786,
+      "loss": 0.2984,
+      "step": 5041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002716553040360857,
+      "loss": 0.0046,
+      "step": 5042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027164419884051646,
+      "loss": 0.0167,
+      "step": 5043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027163309169699794,
+      "loss": 0.0781,
+      "step": 5044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000271621982605708,
+      "loss": 0.3799,
+      "step": 5045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027161087156682455,
+      "loss": 0.528,
+      "step": 5046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002715997585805255,
+      "loss": 0.0022,
+      "step": 5047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002715886436469888,
+      "loss": 0.1259,
+      "step": 5048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027157752676639245,
+      "loss": 0.7868,
+      "step": 5049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027156640793891447,
+      "loss": 0.3035,
+      "step": 5050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027155528716473295,
+      "loss": 0.3669,
+      "step": 5051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027154416444402586,
+      "loss": 0.7415,
+      "step": 5052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027153303977697145,
+      "loss": 0.0027,
+      "step": 5053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002715219131637478,
+      "loss": 0.1052,
+      "step": 5054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027151078460453307,
+      "loss": 0.2668,
+      "step": 5055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002714996540995055,
+      "loss": 0.5208,
+      "step": 5056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002714885216488433,
+      "loss": 0.4799,
+      "step": 5057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002714773872527248,
+      "loss": 1.1736,
+      "step": 5058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027146625091132817,
+      "loss": 2.9192,
+      "step": 5059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027145511262483186,
+      "loss": 0.0083,
+      "step": 5060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002714439723934143,
+      "loss": 0.6565,
+      "step": 5061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002714328302172536,
+      "loss": 0.4844,
+      "step": 5062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027142168609652854,
+      "loss": 0.7202,
+      "step": 5063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002714105400314173,
+      "loss": 1.548,
+      "step": 5064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002713993920220985,
+      "loss": 0.5505,
+      "step": 5065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027138824206875063,
+      "loss": 0.1686,
+      "step": 5066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027137709017155225,
+      "loss": 1.1727,
+      "step": 5067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027136593633068196,
+      "loss": 0.4322,
+      "step": 5068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002713547805463183,
+      "loss": 0.3421,
+      "step": 5069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027134362281863993,
+      "loss": 0.0198,
+      "step": 5070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027133246314782564,
+      "loss": 0.0023,
+      "step": 5071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000271321301534054,
+      "loss": 0.3839,
+      "step": 5072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002713101379775038,
+      "loss": 0.4803,
+      "step": 5073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002712989724783538,
+      "loss": 0.3562,
+      "step": 5074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002712878050367828,
+      "loss": 0.2093,
+      "step": 5075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027127663565296964,
+      "loss": 0.4603,
+      "step": 5076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027126546432709313,
+      "loss": 0.249,
+      "step": 5077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027125429105933225,
+      "loss": 0.5766,
+      "step": 5078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002712431158498659,
+      "loss": 0.7111,
+      "step": 5079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002712319386988729,
+      "loss": 0.3979,
+      "step": 5080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027122075960653243,
+      "loss": 0.0052,
+      "step": 5081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002712095785730234,
+      "loss": 0.3125,
+      "step": 5082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027119839559852486,
+      "loss": 0.594,
+      "step": 5083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000271187210683216,
+      "loss": 0.3727,
+      "step": 5084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027117602382727574,
+      "loss": 0.0031,
+      "step": 5085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027116483503088337,
+      "loss": 0.888,
+      "step": 5086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027115364429421796,
+      "loss": 0.6689,
+      "step": 5087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002711424516174588,
+      "loss": 0.2603,
+      "step": 5088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027113125700078505,
+      "loss": 0.0134,
+      "step": 5089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002711200604443761,
+      "loss": 0.5516,
+      "step": 5090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002711088619484111,
+      "loss": 0.3815,
+      "step": 5091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002710976615130694,
+      "loss": 0.3958,
+      "step": 5092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002710864591385304,
+      "loss": 0.5315,
+      "step": 5093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027107525482497353,
+      "loss": 0.3884,
+      "step": 5094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002710640485725781,
+      "loss": 0.4723,
+      "step": 5095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002710528403815237,
+      "loss": 0.3273,
+      "step": 5096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027104163025198967,
+      "loss": 0.2883,
+      "step": 5097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002710304181841556,
+      "loss": 0.0564,
+      "step": 5098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027101920417820104,
+      "loss": 0.3169,
+      "step": 5099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027100798823430556,
+      "loss": 1.3761,
+      "step": 5100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027099677035264877,
+      "loss": 0.8277,
+      "step": 5101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027098555053341026,
+      "loss": 0.7294,
+      "step": 5102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027097432877676975,
+      "loss": 0.6623,
+      "step": 5103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002709631050829069,
+      "loss": 0.5063,
+      "step": 5104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027095187945200147,
+      "loss": 0.916,
+      "step": 5105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027094065188423323,
+      "loss": 3.0999,
+      "step": 5106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027092942237978197,
+      "loss": 0.1649,
+      "step": 5107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00027091819093882745,
+      "loss": 0.7795,
+      "step": 5108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002709069575615496,
+      "loss": 0.3781,
+      "step": 5109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027089572224812833,
+      "loss": 0.702,
+      "step": 5110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027088448499874347,
+      "loss": 0.3973,
+      "step": 5111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000270873245813575,
+      "loss": 1.8497,
+      "step": 5112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027086200469280286,
+      "loss": 0.0541,
+      "step": 5113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027085076163660713,
+      "loss": 1.1935,
+      "step": 5114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002708395166451678,
+      "loss": 1.2148,
+      "step": 5115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000270828269718665,
+      "loss": 2.4962,
+      "step": 5116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027081702085727887,
+      "loss": 0.4765,
+      "step": 5117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002708057700611894,
+      "loss": 0.2961,
+      "step": 5118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002707945173305768,
+      "loss": 0.9672,
+      "step": 5119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002707832626656213,
+      "loss": 0.9065,
+      "step": 5120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027077200606650316,
+      "loss": 0.0646,
+      "step": 5121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027076074753340257,
+      "loss": 0.8794,
+      "step": 5122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027074948706649985,
+      "loss": 0.4423,
+      "step": 5123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027073822466597535,
+      "loss": 0.8058,
+      "step": 5124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027072696033200933,
+      "loss": 0.7195,
+      "step": 5125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027071569406478225,
+      "loss": 0.6618,
+      "step": 5126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027070442586447446,
+      "loss": 0.3325,
+      "step": 5127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002706931557312665,
+      "loss": 0.2873,
+      "step": 5128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027068188366533876,
+      "loss": 0.017,
+      "step": 5129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002706706096668718,
+      "loss": 0.2939,
+      "step": 5130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002706593337360461,
+      "loss": 0.5671,
+      "step": 5131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027064805587304226,
+      "loss": 0.3266,
+      "step": 5132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027063677607804085,
+      "loss": 1.0926,
+      "step": 5133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027062549435122253,
+      "loss": 0.7719,
+      "step": 5134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000270614210692768,
+      "loss": 0.552,
+      "step": 5135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002706029251028578,
+      "loss": 1.3692,
+      "step": 5136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002705916375816728,
+      "loss": 0.2909,
+      "step": 5137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002705803481293938,
+      "loss": 0.462,
+      "step": 5138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002705690567462014,
+      "loss": 0.4587,
+      "step": 5139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002705577634322765,
+      "loss": 0.4725,
+      "step": 5140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002705464681878,
+      "loss": 0.595,
+      "step": 5141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027053517101295263,
+      "loss": 0.7256,
+      "step": 5142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002705238719079155,
+      "loss": 0.3883,
+      "step": 5143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002705125708728694,
+      "loss": 0.4414,
+      "step": 5144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027050126790799535,
+      "loss": 0.0295,
+      "step": 5145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002704899630134743,
+      "loss": 0.3996,
+      "step": 5146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002704786561894874,
+      "loss": 0.2559,
+      "step": 5147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027046734743621563,
+      "loss": 1.1359,
+      "step": 5148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027045603675384,
+      "loss": 0.6631,
+      "step": 5149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002704447241425418,
+      "loss": 0.3537,
+      "step": 5150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000270433409602502,
+      "loss": 0.2166,
+      "step": 5151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027042209313390205,
+      "loss": 0.2566,
+      "step": 5152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002704107747369229,
+      "loss": 0.9312,
+      "step": 5153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002703994544117459,
+      "loss": 0.4554,
+      "step": 5154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027038813215855237,
+      "loss": 0.3764,
+      "step": 5155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027037680797752355,
+      "loss": 0.4714,
+      "step": 5156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002703654818688409,
+      "loss": 0.2031,
+      "step": 5157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027035415383268557,
+      "loss": 1.1316,
+      "step": 5158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002703428238692392,
+      "loss": 0.966,
+      "step": 5159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027033149197868306,
+      "loss": 0.4395,
+      "step": 5160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027032015816119866,
+      "loss": 0.0311,
+      "step": 5161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002703088224169675,
+      "loss": 0.4664,
+      "step": 5162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027029748474617117,
+      "loss": 0.5671,
+      "step": 5163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027028614514899114,
+      "loss": 3.0045,
+      "step": 5164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027027480362560896,
+      "loss": 0.8462,
+      "step": 5165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027026346017620636,
+      "loss": 0.3459,
+      "step": 5166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002702521148009649,
+      "loss": 0.154,
+      "step": 5167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002702407675000663,
+      "loss": 0.0469,
+      "step": 5168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027022941827369235,
+      "loss": 0.6406,
+      "step": 5169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002702180671220246,
+      "loss": 1.0015,
+      "step": 5170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027020671404524496,
+      "loss": 2.3184,
+      "step": 5171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002701953590435352,
+      "loss": 0.137,
+      "step": 5172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027018400211707716,
+      "loss": 0.1569,
+      "step": 5173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002701726432660527,
+      "loss": 0.149,
+      "step": 5174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027016128249064366,
+      "loss": 0.0115,
+      "step": 5175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002701499197910321,
+      "loss": 0.3667,
+      "step": 5176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027013855516739985,
+      "loss": 0.5227,
+      "step": 5177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027012718861992897,
+      "loss": 0.4261,
+      "step": 5178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002701158201488014,
+      "loss": 0.3587,
+      "step": 5179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027010444975419934,
+      "loss": 1.5499,
+      "step": 5180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027009307743630466,
+      "loss": 0.2272,
+      "step": 5181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002700817031952996,
+      "loss": 0.4869,
+      "step": 5182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002700703270313663,
+      "loss": 0.2982,
+      "step": 5183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002700589489446869,
+      "loss": 0.7136,
+      "step": 5184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002700475689354436,
+      "loss": 0.4295,
+      "step": 5185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002700361870038187,
+      "loss": 1.0464,
+      "step": 5186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027002480314999443,
+      "loss": 0.6562,
+      "step": 5187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000270013417374153,
+      "loss": 0.6703,
+      "step": 5188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027000202967647686,
+      "loss": 0.7066,
+      "step": 5189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026999064005714824,
+      "loss": 0.6726,
+      "step": 5190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002699792485163497,
+      "loss": 1.5038,
+      "step": 5191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002699678550542635,
+      "loss": 0.6065,
+      "step": 5192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002699564596710721,
+      "loss": 2.1863,
+      "step": 5193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002699450623669581,
+      "loss": 1.4364,
+      "step": 5194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002699336631421039,
+      "loss": 0.5962,
+      "step": 5195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002699222619966921,
+      "loss": 0.6432,
+      "step": 5196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002699108589309053,
+      "loss": 0.2823,
+      "step": 5197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000269899453944926,
+      "loss": 1.2481,
+      "step": 5198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002698880470389369,
+      "loss": 0.4015,
+      "step": 5199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026987663821312065,
+      "loss": 0.3033,
+      "step": 5200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026986522746765993,
+      "loss": 4.9241,
+      "step": 5201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002698538148027375,
+      "loss": 0.5229,
+      "step": 5202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002698424002185361,
+      "loss": 0.4157,
+      "step": 5203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026983098371523854,
+      "loss": 0.3099,
+      "step": 5204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026981956529302764,
+      "loss": 0.008,
+      "step": 5205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026980814495208616,
+      "loss": 0.0047,
+      "step": 5206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026979672269259705,
+      "loss": 0.1613,
+      "step": 5207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002697852985147433,
+      "loss": 0.3621,
+      "step": 5208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026977387241870765,
+      "loss": 1.6095,
+      "step": 5209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026976244440467327,
+      "loss": 0.4214,
+      "step": 5210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026975101447282305,
+      "loss": 0.3521,
+      "step": 5211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026973958262334,
+      "loss": 0.4248,
+      "step": 5212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026972814885640735,
+      "loss": 0.0126,
+      "step": 5213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026971671317220806,
+      "loss": 1.2097,
+      "step": 5214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002697052755709252,
+      "loss": 0.6315,
+      "step": 5215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026969383605274206,
+      "loss": 1.2428,
+      "step": 5216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026968239461784175,
+      "loss": 1.4551,
+      "step": 5217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002696709512664075,
+      "loss": 0.3401,
+      "step": 5218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002696595059986226,
+      "loss": 0.4584,
+      "step": 5219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002696480588146703,
+      "loss": 0.3943,
+      "step": 5220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026963660971473385,
+      "loss": 0.1584,
+      "step": 5221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002696251586989967,
+      "loss": 0.7672,
+      "step": 5222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002696137057676421,
+      "loss": 0.3504,
+      "step": 5223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002696022509208536,
+      "loss": 0.1306,
+      "step": 5224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026959079415881456,
+      "loss": 0.0056,
+      "step": 5225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002695793354817084,
+      "loss": 0.1923,
+      "step": 5226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026956787488971863,
+      "loss": 0.0771,
+      "step": 5227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026955641238302877,
+      "loss": 0.7919,
+      "step": 5228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002695449479618224,
+      "loss": 0.6207,
+      "step": 5229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026953348162628315,
+      "loss": 0.3058,
+      "step": 5230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002695220133765945,
+      "loss": 0.1591,
+      "step": 5231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002695105432129403,
+      "loss": 0.8658,
+      "step": 5232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026949907113550404,
+      "loss": 0.8408,
+      "step": 5233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026948759714446945,
+      "loss": 0.3608,
+      "step": 5234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002694761212400204,
+      "loss": 0.4757,
+      "step": 5235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002694646434223406,
+      "loss": 0.5935,
+      "step": 5236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002694531636916137,
+      "loss": 0.004,
+      "step": 5237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026944168204802373,
+      "loss": 0.1961,
+      "step": 5238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026943019849175453,
+      "loss": 0.3013,
+      "step": 5239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026941871302298994,
+      "loss": 1.2863,
+      "step": 5240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002694072256419138,
+      "loss": 0.4762,
+      "step": 5241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002693957363487102,
+      "loss": 0.8075,
+      "step": 5242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000269384245143563,
+      "loss": 1.0787,
+      "step": 5243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026937275202665637,
+      "loss": 0.2623,
+      "step": 5244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026936125699817423,
+      "loss": 0.9414,
+      "step": 5245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002693497600583007,
+      "loss": 0.6723,
+      "step": 5246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002693382612072199,
+      "loss": 0.0034,
+      "step": 5247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026932676044511596,
+      "loss": 0.3038,
+      "step": 5248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000269315257772173,
+      "loss": 0.4563,
+      "step": 5249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026930375318857524,
+      "loss": 0.7673,
+      "step": 5250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002692922466945069,
+      "loss": 0.4377,
+      "step": 5251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026928073829015237,
+      "loss": 0.6639,
+      "step": 5252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026926922797569576,
+      "loss": 0.2943,
+      "step": 5253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002692577157513215,
+      "loss": 0.6372,
+      "step": 5254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026924620161721387,
+      "loss": 0.513,
+      "step": 5255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026923468557355727,
+      "loss": 0.2722,
+      "step": 5256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002692231676205361,
+      "loss": 0.6129,
+      "step": 5257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002692116477583349,
+      "loss": 0.9037,
+      "step": 5258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002692001259871381,
+      "loss": 0.689,
+      "step": 5259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026918860230713007,
+      "loss": 2.5001,
+      "step": 5260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002691770767184955,
+      "loss": 0.6996,
+      "step": 5261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026916554922141897,
+      "loss": 0.5809,
+      "step": 5262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026915401981608494,
+      "loss": 0.1609,
+      "step": 5263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002691424885026781,
+      "loss": 0.618,
+      "step": 5264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026913095528138315,
+      "loss": 0.4191,
+      "step": 5265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026911942015238474,
+      "loss": 1.2627,
+      "step": 5266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002691078831158676,
+      "loss": 1.1788,
+      "step": 5267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026909634417201643,
+      "loss": 3.6446,
+      "step": 5268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026908480332101604,
+      "loss": 0.2226,
+      "step": 5269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026907326056305127,
+      "loss": 0.4881,
+      "step": 5270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026906171589830696,
+      "loss": 0.5733,
+      "step": 5271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026905016932696793,
+      "loss": 0.2333,
+      "step": 5272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002690386208492191,
+      "loss": 0.63,
+      "step": 5273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002690270704652454,
+      "loss": 0.3544,
+      "step": 5274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026901551817523183,
+      "loss": 0.9786,
+      "step": 5275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002690039639793633,
+      "loss": 0.4536,
+      "step": 5276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026899240787782496,
+      "loss": 0.2141,
+      "step": 5277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026898084987080176,
+      "loss": 0.797,
+      "step": 5278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002689692899584788,
+      "loss": 0.5984,
+      "step": 5279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026895772814104123,
+      "loss": 0.2884,
+      "step": 5280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002689461644186742,
+      "loss": 0.2932,
+      "step": 5281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026893459879156285,
+      "loss": 0.3264,
+      "step": 5282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002689230312598923,
+      "loss": 0.3919,
+      "step": 5283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000268911461823848,
+      "loss": 3.2369,
+      "step": 5284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026889989048361503,
+      "loss": 0.1846,
+      "step": 5285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002688883172393788,
+      "loss": 0.0762,
+      "step": 5286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026887674209132457,
+      "loss": 0.5065,
+      "step": 5287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026886516503963774,
+      "loss": 0.9943,
+      "step": 5288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002688535860845037,
+      "loss": 0.857,
+      "step": 5289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002688420052261078,
+      "loss": 0.0364,
+      "step": 5290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002688304224646356,
+      "loss": 0.4896,
+      "step": 5291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002688188378002725,
+      "loss": 0.33,
+      "step": 5292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026880725123320403,
+      "loss": 1.3077,
+      "step": 5293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002687956627636157,
+      "loss": 1.1622,
+      "step": 5294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026878407239169317,
+      "loss": 0.2962,
+      "step": 5295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000268772480117622,
+      "loss": 0.3257,
+      "step": 5296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002687608859415878,
+      "loss": 0.3188,
+      "step": 5297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026874928986377625,
+      "loss": 0.5823,
+      "step": 5298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000268737691884373,
+      "loss": 1.1319,
+      "step": 5299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026872609200356384,
+      "loss": 0.3356,
+      "step": 5300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002687144902215345,
+      "loss": 0.0028,
+      "step": 5301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002687028865384708,
+      "loss": 0.2925,
+      "step": 5302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026869128095455844,
+      "loss": 0.9093,
+      "step": 5303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002686796734699834,
+      "loss": 0.3799,
+      "step": 5304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002686680640849315,
+      "loss": 0.1776,
+      "step": 5305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026865645279958866,
+      "loss": 0.3699,
+      "step": 5306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002686448396141408,
+      "loss": 0.4009,
+      "step": 5307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002686332245287739,
+      "loss": 0.0179,
+      "step": 5308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026862160754367393,
+      "loss": 0.2594,
+      "step": 5309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026860998865902694,
+      "loss": 0.576,
+      "step": 5310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000268598367875019,
+      "loss": 0.294,
+      "step": 5311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002685867451918362,
+      "loss": 0.6305,
+      "step": 5312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002685751206096646,
+      "loss": 2.3575,
+      "step": 5313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002685634941286905,
+      "loss": 0.4648,
+      "step": 5314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002685518657490999,
+      "loss": 1.3284,
+      "step": 5315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002685402354710791,
+      "loss": 0.8474,
+      "step": 5316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026852860329481436,
+      "loss": 2.5704,
+      "step": 5317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002685169692204919,
+      "loss": 0.4805,
+      "step": 5318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026850533324829803,
+      "loss": 0.5606,
+      "step": 5319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002684936953784192,
+      "loss": 0.6657,
+      "step": 5320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026848205561104155,
+      "loss": 0.0801,
+      "step": 5321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002684704139463516,
+      "loss": 1.7422,
+      "step": 5322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026845877038453585,
+      "loss": 0.5296,
+      "step": 5323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026844712492578066,
+      "loss": 0.4269,
+      "step": 5324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026843547757027247,
+      "loss": 0.1891,
+      "step": 5325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002684238283181979,
+      "loss": 0.3454,
+      "step": 5326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002684121771697435,
+      "loss": 0.0649,
+      "step": 5327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002684005241250957,
+      "loss": 0.601,
+      "step": 5328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002683888691844412,
+      "loss": 0.7187,
+      "step": 5329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002683772123479667,
+      "loss": 0.2996,
+      "step": 5330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002683655536158588,
+      "loss": 0.3693,
+      "step": 5331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026835389298830416,
+      "loss": 0.3369,
+      "step": 5332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026834223046548955,
+      "loss": 0.2014,
+      "step": 5333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026833056604760175,
+      "loss": 0.4297,
+      "step": 5334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002683188997348275,
+      "loss": 0.5795,
+      "step": 5335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002683072315273537,
+      "loss": 0.852,
+      "step": 5336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002682955614253671,
+      "loss": 1.1702,
+      "step": 5337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002682838894290546,
+      "loss": 0.3264,
+      "step": 5338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026827221553860314,
+      "loss": 0.2603,
+      "step": 5339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026826053975419965,
+      "loss": 0.4701,
+      "step": 5340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026824886207603107,
+      "loss": 0.2345,
+      "step": 5341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002682371825042845,
+      "loss": 0.669,
+      "step": 5342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002682255010391469,
+      "loss": 0.6379,
+      "step": 5343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002682138176808053,
+      "loss": 0.7395,
+      "step": 5344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026820213242944685,
+      "loss": 0.8577,
+      "step": 5345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002681904452852586,
+      "loss": 0.0133,
+      "step": 5346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026817875624842775,
+      "loss": 0.8394,
+      "step": 5347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002681670653191415,
+      "loss": 2.4528,
+      "step": 5348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026815537249758704,
+      "loss": 2.1485,
+      "step": 5349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026814367778395167,
+      "loss": 0.019,
+      "step": 5350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026813198117842254,
+      "loss": 1.5479,
+      "step": 5351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000268120282681187,
+      "loss": 0.2561,
+      "step": 5352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002681085822924325,
+      "loss": 0.84,
+      "step": 5353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026809688001234625,
+      "loss": 0.4201,
+      "step": 5354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002680851758411157,
+      "loss": 0.6751,
+      "step": 5355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002680734697789283,
+      "loss": 0.3545,
+      "step": 5356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002680617618259715,
+      "loss": 0.2078,
+      "step": 5357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002680500519824327,
+      "loss": 0.8314,
+      "step": 5358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002680383402484996,
+      "loss": 0.599,
+      "step": 5359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026802662662435954,
+      "loss": 0.4058,
+      "step": 5360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002680149111102002,
+      "loss": 1.7032,
+      "step": 5361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026800319370620917,
+      "loss": 0.9379,
+      "step": 5362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026799147441257414,
+      "loss": 0.5892,
+      "step": 5363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002679797532294827,
+      "loss": 0.4544,
+      "step": 5364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026796803015712256,
+      "loss": 0.3688,
+      "step": 5365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026795630519568147,
+      "loss": 0.3301,
+      "step": 5366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002679445783453472,
+      "loss": 0.4323,
+      "step": 5367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002679328496063075,
+      "loss": 0.7034,
+      "step": 5368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026792111897875015,
+      "loss": 0.4994,
+      "step": 5369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026790938646286316,
+      "loss": 0.0943,
+      "step": 5370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026789765205883424,
+      "loss": 0.4635,
+      "step": 5371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002678859157668514,
+      "loss": 0.7838,
+      "step": 5372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002678741775871025,
+      "loss": 0.4908,
+      "step": 5373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026786243751977557,
+      "loss": 0.4053,
+      "step": 5374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002678506955650586,
+      "loss": 1.1606,
+      "step": 5375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002678389517231396,
+      "loss": 0.5101,
+      "step": 5376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002678272059942066,
+      "loss": 0.1985,
+      "step": 5377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002678154583784478,
+      "loss": 0.9099,
+      "step": 5378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002678037088760512,
+      "loss": 0.4567,
+      "step": 5379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000267791957487205,
+      "loss": 0.4331,
+      "step": 5380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026778020421209744,
+      "loss": 0.5019,
+      "step": 5381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002677684490509166,
+      "loss": 0.0421,
+      "step": 5382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002677566920038509,
+      "loss": 0.6336,
+      "step": 5383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002677449330710884,
+      "loss": 0.3754,
+      "step": 5384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002677331722528175,
+      "loss": 0.7366,
+      "step": 5385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002677214095492266,
+      "loss": 1.5245,
+      "step": 5386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000267709644960504,
+      "loss": 0.2858,
+      "step": 5387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002676978784868381,
+      "loss": 2.0918,
+      "step": 5388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026768611012841723,
+      "loss": 0.5204,
+      "step": 5389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026767433988543,
+      "loss": 0.9762,
+      "step": 5390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002676625677580648,
+      "loss": 0.0425,
+      "step": 5391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026765079374651017,
+      "loss": 0.0273,
+      "step": 5392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026763901785095467,
+      "loss": 0.1676,
+      "step": 5393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002676272400715868,
+      "loss": 0.1877,
+      "step": 5394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002676154604085953,
+      "loss": 0.2754,
+      "step": 5395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002676036788621686,
+      "loss": 0.272,
+      "step": 5396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002675918954324956,
+      "loss": 0.2238,
+      "step": 5397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026758011011976477,
+      "loss": 0.5024,
+      "step": 5398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026756832292416504,
+      "loss": 0.0502,
+      "step": 5399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000267556533845885,
+      "loss": 0.9217,
+      "step": 5400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002675447428851135,
+      "loss": 0.0453,
+      "step": 5401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026753295004203935,
+      "loss": 0.6314,
+      "step": 5402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002675211553168514,
+      "loss": 0.0874,
+      "step": 5403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026750935870973854,
+      "loss": 0.3272,
+      "step": 5404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002674975602208897,
+      "loss": 0.4336,
+      "step": 5405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002674857598504937,
+      "loss": 0.939,
+      "step": 5406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026747395759873957,
+      "loss": 0.219,
+      "step": 5407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002674621534658164,
+      "loss": 1.2029,
+      "step": 5408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000267450347451913,
+      "loss": 0.2999,
+      "step": 5409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026743853955721867,
+      "loss": 0.1183,
+      "step": 5410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002674267297819223,
+      "loss": 0.357,
+      "step": 5411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026741491812621313,
+      "loss": 0.3573,
+      "step": 5412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026740310459028027,
+      "loss": 2.2657,
+      "step": 5413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002673912891743129,
+      "loss": 1.5705,
+      "step": 5414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026737947187850013,
+      "loss": 0.9312,
+      "step": 5415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026736765270303134,
+      "loss": 0.0199,
+      "step": 5416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026735583164809576,
+      "loss": 0.1629,
+      "step": 5417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002673440087138826,
+      "loss": 0.0223,
+      "step": 5418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002673321839005813,
+      "loss": 0.0444,
+      "step": 5419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026732035720838113,
+      "loss": 0.4048,
+      "step": 5420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026730852863747153,
+      "loss": 0.4271,
+      "step": 5421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002672966981880419,
+      "loss": 0.3899,
+      "step": 5422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002672848658602817,
+      "loss": 1.9162,
+      "step": 5423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026727303165438034,
+      "loss": 0.0237,
+      "step": 5424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002672611955705274,
+      "loss": 0.7174,
+      "step": 5425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002672493576089125,
+      "loss": 0.4343,
+      "step": 5426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026723751776972497,
+      "loss": 1.0501,
+      "step": 5427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002672256760531546,
+      "loss": 0.681,
+      "step": 5428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026721383245939095,
+      "loss": 0.0058,
+      "step": 5429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026720198698862365,
+      "loss": 0.2324,
+      "step": 5430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002671901396410425,
+      "loss": 0.2105,
+      "step": 5431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002671782904168371,
+      "loss": 1.1403,
+      "step": 5432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026716643931619724,
+      "loss": 0.1758,
+      "step": 5433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002671545863393127,
+      "loss": 0.0428,
+      "step": 5434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026714273148637326,
+      "loss": 0.7647,
+      "step": 5435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026713087475756876,
+      "loss": 1.5645,
+      "step": 5436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002671190161530892,
+      "loss": 0.0434,
+      "step": 5437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002671071556731243,
+      "loss": 0.1111,
+      "step": 5438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026709529331786405,
+      "loss": 1.7327,
+      "step": 5439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026708342908749843,
+      "loss": 0.7881,
+      "step": 5440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002670715629822174,
+      "loss": 0.011,
+      "step": 5441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026705969500221094,
+      "loss": 0.4664,
+      "step": 5442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026704782514766923,
+      "loss": 0.9047,
+      "step": 5443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026703595341878223,
+      "loss": 1.0668,
+      "step": 5444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026702407981574005,
+      "loss": 0.3375,
+      "step": 5445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002670122043387329,
+      "loss": 0.2352,
+      "step": 5446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002670003269879509,
+      "loss": 0.4915,
+      "step": 5447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026698844776358425,
+      "loss": 0.0075,
+      "step": 5448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002669765666658232,
+      "loss": 0.7483,
+      "step": 5449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026696468369485796,
+      "loss": 0.7692,
+      "step": 5450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002669527988508789,
+      "loss": 0.2391,
+      "step": 5451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026694091213407624,
+      "loss": 0.5048,
+      "step": 5452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026692902354464036,
+      "loss": 0.6993,
+      "step": 5453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026691713308276175,
+      "loss": 0.769,
+      "step": 5454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002669052407486306,
+      "loss": 0.4241,
+      "step": 5455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002668933465424376,
+      "loss": 0.3509,
+      "step": 5456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000266881450464373,
+      "loss": 0.0042,
+      "step": 5457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002668695525146274,
+      "loss": 0.6369,
+      "step": 5458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026685765269339137,
+      "loss": 0.2164,
+      "step": 5459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002668457510008553,
+      "loss": 0.0013,
+      "step": 5460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026683384743720994,
+      "loss": 0.5716,
+      "step": 5461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026682194200264585,
+      "loss": 0.1546,
+      "step": 5462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026681003469735374,
+      "loss": 0.4197,
+      "step": 5463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002667981255215242,
+      "loss": 0.523,
+      "step": 5464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026678621447534794,
+      "loss": 2.8133,
+      "step": 5465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002667743015590157,
+      "loss": 0.317,
+      "step": 5466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002667623867727184,
+      "loss": 0.1106,
+      "step": 5467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026675047011664657,
+      "loss": 0.5836,
+      "step": 5468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026673855159099123,
+      "loss": 0.189,
+      "step": 5469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002667266311959432,
+      "loss": 0.3522,
+      "step": 5470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026671470893169333,
+      "loss": 0.4839,
+      "step": 5471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026670278479843263,
+      "loss": 0.3838,
+      "step": 5472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026669085879635194,
+      "loss": 0.3783,
+      "step": 5473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002666789309256422,
+      "loss": 0.6567,
+      "step": 5474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002666670011864946,
+      "loss": 0.4799,
+      "step": 5475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026665506957910005,
+      "loss": 0.2919,
+      "step": 5476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026664313610364957,
+      "loss": 3.598,
+      "step": 5477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026663120076033433,
+      "loss": 1.4423,
+      "step": 5478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026661926354934554,
+      "loss": 1.3551,
+      "step": 5479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002666073244708742,
+      "loss": 0.0096,
+      "step": 5480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002665953835251115,
+      "loss": 0.4268,
+      "step": 5481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026658344071224885,
+      "loss": 1.144,
+      "step": 5482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002665714960324773,
+      "loss": 0.919,
+      "step": 5483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026655954948598816,
+      "loss": 0.2671,
+      "step": 5484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026654760107297276,
+      "loss": 0.7987,
+      "step": 5485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002665356507936225,
+      "loss": 2.961,
+      "step": 5486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026652369864812867,
+      "loss": 0.0098,
+      "step": 5487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002665117446366827,
+      "loss": 0.0057,
+      "step": 5488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000266499788759476,
+      "loss": 0.3567,
+      "step": 5489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002664878310167,
+      "loss": 0.6006,
+      "step": 5490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026647587140854614,
+      "loss": 1.2958,
+      "step": 5491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002664639099352061,
+      "loss": 0.185,
+      "step": 5492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002664519465968713,
+      "loss": 0.06,
+      "step": 5493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026643998139373336,
+      "loss": 0.3065,
+      "step": 5494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002664280143259839,
+      "loss": 0.3581,
+      "step": 5495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026641604539381447,
+      "loss": 0.2479,
+      "step": 5496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002664040745974168,
+      "loss": 0.8986,
+      "step": 5497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026639210193698257,
+      "loss": 0.0586,
+      "step": 5498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002663801274127035,
+      "loss": 0.2407,
+      "step": 5499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002663681510247714,
+      "loss": 1.9321,
+      "step": 5500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000266356172773378,
+      "loss": 0.3389,
+      "step": 5501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002663441926587151,
+      "loss": 2.3635,
+      "step": 5502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026633221068097455,
+      "loss": 0.036,
+      "step": 5503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002663202268403482,
+      "loss": 0.4173,
+      "step": 5504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002663082411370281,
+      "loss": 0.9021,
+      "step": 5505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000266296253571206,
+      "loss": 0.501,
+      "step": 5506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002662842641430739,
+      "loss": 0.3068,
+      "step": 5507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002662722728528239,
+      "loss": 2.2949,
+      "step": 5508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002662602797006479,
+      "loss": 0.6985,
+      "step": 5509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000266248284686738,
+      "loss": 0.8566,
+      "step": 5510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026623628781128635,
+      "loss": 2.9541,
+      "step": 5511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026622428907448496,
+      "loss": 2.0647,
+      "step": 5512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026621228847652597,
+      "loss": 1.2902,
+      "step": 5513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026620028601760166,
+      "loss": 0.2296,
+      "step": 5514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002661882816979041,
+      "loss": 0.5915,
+      "step": 5515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002661762755176256,
+      "loss": 0.5072,
+      "step": 5516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002661642674769584,
+      "loss": 0.0092,
+      "step": 5517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002661522575760948,
+      "loss": 0.769,
+      "step": 5518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026614024581522716,
+      "loss": 0.7514,
+      "step": 5519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002661282321945477,
+      "loss": 0.3787,
+      "step": 5520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000266116216714249,
+      "loss": 0.0256,
+      "step": 5521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002661041993745232,
+      "loss": 0.5377,
+      "step": 5522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000266092180175563,
+      "loss": 0.4487,
+      "step": 5523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002660801591175608,
+      "loss": 0.0069,
+      "step": 5524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026606813620070897,
+      "loss": 0.8953,
+      "step": 5525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026605611142520015,
+      "loss": 0.7779,
+      "step": 5526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026604408479122694,
+      "loss": 0.3774,
+      "step": 5527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026603205629898184,
+      "loss": 0.8944,
+      "step": 5528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026602002594865745,
+      "loss": 0.6213,
+      "step": 5529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026600799374044656,
+      "loss": 4.6896,
+      "step": 5530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026599595967454165,
+      "loss": 0.6279,
+      "step": 5531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002659839237511356,
+      "loss": 0.5865,
+      "step": 5532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002659718859704211,
+      "loss": 0.7039,
+      "step": 5533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026595984633259083,
+      "loss": 0.1343,
+      "step": 5534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002659478048378377,
+      "loss": 0.0165,
+      "step": 5535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026593576148635446,
+      "loss": 0.3843,
+      "step": 5536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000265923716278334,
+      "loss": 0.7082,
+      "step": 5537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002659116692139692,
+      "loss": 0.2677,
+      "step": 5538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026589962029345294,
+      "loss": 0.3956,
+      "step": 5539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002658875695169783,
+      "loss": 0.4227,
+      "step": 5540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002658755168847381,
+      "loss": 0.5371,
+      "step": 5541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026586346239692535,
+      "loss": 0.7336,
+      "step": 5542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002658514060537332,
+      "loss": 0.0063,
+      "step": 5543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002658393478553547,
+      "loss": 0.007,
+      "step": 5544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026582728780198277,
+      "loss": 0.6149,
+      "step": 5545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026581522589381074,
+      "loss": 0.3336,
+      "step": 5546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002658031621310317,
+      "loss": 0.4898,
+      "step": 5547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002657910965138387,
+      "loss": 0.0127,
+      "step": 5548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002657790290424252,
+      "loss": 0.7222,
+      "step": 5549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026576695971698423,
+      "loss": 0.0239,
+      "step": 5550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026575488853770917,
+      "loss": 0.6981,
+      "step": 5551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026574281550479327,
+      "loss": 0.3653,
+      "step": 5552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002657307406184299,
+      "loss": 0.3818,
+      "step": 5553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026571866387881246,
+      "loss": 0.3818,
+      "step": 5554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002657065852861342,
+      "loss": 0.1669,
+      "step": 5555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026569450484058873,
+      "loss": 0.5682,
+      "step": 5556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002656824225423693,
+      "loss": 0.6906,
+      "step": 5557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002656703383916695,
+      "loss": 0.2699,
+      "step": 5558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002656582523886829,
+      "loss": 0.6384,
+      "step": 5559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026564616453360286,
+      "loss": 0.503,
+      "step": 5560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026563407482662314,
+      "loss": 0.7014,
+      "step": 5561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026562198326793724,
+      "loss": 1.0608,
+      "step": 5562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002656098898577388,
+      "loss": 0.013,
+      "step": 5563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026559779459622146,
+      "loss": 0.2676,
+      "step": 5564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026558569748357896,
+      "loss": 2.3829,
+      "step": 5565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026557359852000495,
+      "loss": 0.1762,
+      "step": 5566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002655614977056932,
+      "loss": 0.6785,
+      "step": 5567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026554939504083757,
+      "loss": 1.6163,
+      "step": 5568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002655372905256317,
+      "loss": 0.1739,
+      "step": 5569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026552518416026957,
+      "loss": 0.2668,
+      "step": 5570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000265513075944945,
+      "loss": 0.1858,
+      "step": 5571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002655009658798519,
+      "loss": 0.5063,
+      "step": 5572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026548885396518413,
+      "loss": 0.3505,
+      "step": 5573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002654767402011357,
+      "loss": 0.7825,
+      "step": 5574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002654646245879006,
+      "loss": 0.2978,
+      "step": 5575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002654525071256728,
+      "loss": 0.3954,
+      "step": 5576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026544038781464637,
+      "loss": 0.7083,
+      "step": 5577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026542826665501535,
+      "loss": 0.3952,
+      "step": 5578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026541614364697396,
+      "loss": 0.0929,
+      "step": 5579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002654040187907162,
+      "loss": 1.9708,
+      "step": 5580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026539189208643626,
+      "loss": 0.3527,
+      "step": 5581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026537976353432837,
+      "loss": 0.2514,
+      "step": 5582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002653676331345867,
+      "loss": 0.3298,
+      "step": 5583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026535550088740555,
+      "loss": 0.7406,
+      "step": 5584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026534336679297915,
+      "loss": 1.0579,
+      "step": 5585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026533123085150187,
+      "loss": 1.1444,
+      "step": 5586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026531909306316804,
+      "loss": 0.0937,
+      "step": 5587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002653069534281719,
+      "loss": 0.3469,
+      "step": 5588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002652948119467081,
+      "loss": 0.8589,
+      "step": 5589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026528266861897083,
+      "loss": 0.6113,
+      "step": 5590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026527052344515464,
+      "loss": 0.8873,
+      "step": 5591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000265258376425454,
+      "loss": 0.762,
+      "step": 5592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002652462275600635,
+      "loss": 0.9296,
+      "step": 5593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026523407684917756,
+      "loss": 0.8063,
+      "step": 5594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002652219242929909,
+      "loss": 0.4092,
+      "step": 5595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026520976989169795,
+      "loss": 0.6565,
+      "step": 5596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026519761364549347,
+      "loss": 0.3957,
+      "step": 5597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002651854555545721,
+      "loss": 0.2035,
+      "step": 5598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002651732956191286,
+      "loss": 1.1846,
+      "step": 5599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026516113383935754,
+      "loss": 0.4088,
+      "step": 5600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026514897021545376,
+      "loss": 0.3095,
+      "step": 5601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002651368047476121,
+      "loss": 0.2869,
+      "step": 5602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026512463743602724,
+      "loss": 0.4348,
+      "step": 5603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002651124682808941,
+      "loss": 0.393,
+      "step": 5604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026510029728240756,
+      "loss": 0.3377,
+      "step": 5605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002650881244407625,
+      "loss": 0.3109,
+      "step": 5606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026507594975615386,
+      "loss": 0.5403,
+      "step": 5607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026506377322877653,
+      "loss": 0.42,
+      "step": 5608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026505159485882564,
+      "loss": 4.1254,
+      "step": 5609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002650394146464961,
+      "loss": 0.343,
+      "step": 5610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026502723259198293,
+      "loss": 0.848,
+      "step": 5611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002650150486954813,
+      "loss": 1.4893,
+      "step": 5612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002650028629571863,
+      "loss": 0.7835,
+      "step": 5613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026499067537729304,
+      "loss": 0.2249,
+      "step": 5614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026497848595599666,
+      "loss": 0.0271,
+      "step": 5615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002649662946934924,
+      "loss": 1.3565,
+      "step": 5616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026495410158997544,
+      "loss": 0.2453,
+      "step": 5617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026494190664564114,
+      "loss": 0.3931,
+      "step": 5618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002649297098606847,
+      "loss": 0.3271,
+      "step": 5619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026491751123530136,
+      "loss": 0.1528,
+      "step": 5620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002649053107696866,
+      "loss": 0.2288,
+      "step": 5621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026489310846403575,
+      "loss": 0.079,
+      "step": 5622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026488090431854415,
+      "loss": 0.0719,
+      "step": 5623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026486869833340735,
+      "loss": 0.0767,
+      "step": 5624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002648564905088206,
+      "loss": 0.6749,
+      "step": 5625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002648442808449797,
+      "loss": 0.0773,
+      "step": 5626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026483206934207993,
+      "loss": 0.8949,
+      "step": 5627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026481985600031687,
+      "loss": 0.0062,
+      "step": 5628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002648076408198862,
+      "loss": 1.2764,
+      "step": 5629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002647954238009834,
+      "loss": 0.5771,
+      "step": 5630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026478320494380427,
+      "loss": 0.2582,
+      "step": 5631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026477098424854427,
+      "loss": 0.2218,
+      "step": 5632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002647587617153993,
+      "loss": 0.0142,
+      "step": 5633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002647465373445649,
+      "loss": 0.2855,
+      "step": 5634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000264734311136237,
+      "loss": 0.0066,
+      "step": 5635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002647220830906112,
+      "loss": 0.4785,
+      "step": 5636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002647098532078835,
+      "loss": 1.239,
+      "step": 5637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026469762148824963,
+      "loss": 0.3771,
+      "step": 5638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002646853879319055,
+      "loss": 0.5727,
+      "step": 5639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000264673152539047,
+      "loss": 0.2253,
+      "step": 5640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026466091530987,
+      "loss": 0.007,
+      "step": 5641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002646486762445706,
+      "loss": 1.2354,
+      "step": 5642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026463643534334466,
+      "loss": 0.9444,
+      "step": 5643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002646241926063883,
+      "loss": 0.5984,
+      "step": 5644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026461194803389747,
+      "loss": 1.124,
+      "step": 5645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002645997016260683,
+      "loss": 0.9825,
+      "step": 5646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026458745338309695,
+      "loss": 0.7355,
+      "step": 5647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002645752033051795,
+      "loss": 0.5966,
+      "step": 5648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002645629513925121,
+      "loss": 0.6704,
+      "step": 5649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026455069764529094,
+      "loss": 0.866,
+      "step": 5650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026453844206371233,
+      "loss": 0.5221,
+      "step": 5651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002645261846479725,
+      "loss": 0.2055,
+      "step": 5652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002645139253982676,
+      "loss": 0.0035,
+      "step": 5653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002645016643147941,
+      "loss": 0.4753,
+      "step": 5654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026448940139774834,
+      "loss": 0.0442,
+      "step": 5655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026447713664732653,
+      "loss": 1.3765,
+      "step": 5656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026446487006372523,
+      "loss": 0.0024,
+      "step": 5657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026445260164714083,
+      "loss": 1.5645,
+      "step": 5658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002644403313977698,
+      "loss": 0.7166,
+      "step": 5659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002644280593158086,
+      "loss": 0.4859,
+      "step": 5660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002644157854014538,
+      "loss": 0.9595,
+      "step": 5661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026440350965490185,
+      "loss": 0.6802,
+      "step": 5662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002643912320763494,
+      "loss": 1.1673,
+      "step": 5663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026437895266599303,
+      "loss": 0.3396,
+      "step": 5664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002643666714240295,
+      "loss": 0.3873,
+      "step": 5665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026435438835065525,
+      "loss": 0.5128,
+      "step": 5666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002643421034460671,
+      "loss": 0.1248,
+      "step": 5667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026432981671046184,
+      "loss": 0.3231,
+      "step": 5668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026431752814403607,
+      "loss": 0.2673,
+      "step": 5669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002643052377469867,
+      "loss": 0.266,
+      "step": 5670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026429294551951046,
+      "loss": 2.1544,
+      "step": 5671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002642806514618043,
+      "loss": 0.2961,
+      "step": 5672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026426835557406494,
+      "loss": 0.4332,
+      "step": 5673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002642560578564894,
+      "loss": 1.0128,
+      "step": 5674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026424375830927453,
+      "loss": 0.6778,
+      "step": 5675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026423145693261737,
+      "loss": 0.7006,
+      "step": 5676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026421915372671484,
+      "loss": 0.437,
+      "step": 5677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000264206848691764,
+      "loss": 0.3902,
+      "step": 5678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002641945418279618,
+      "loss": 0.4595,
+      "step": 5679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002641822331355055,
+      "loss": 0.1923,
+      "step": 5680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002641699226145921,
+      "loss": 0.3024,
+      "step": 5681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026415761026541866,
+      "loss": 1.2288,
+      "step": 5682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002641452960881825,
+      "loss": 1.0265,
+      "step": 5683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026413298008308065,
+      "loss": 0.5068,
+      "step": 5684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026412066225031043,
+      "loss": 0.0067,
+      "step": 5685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002641083425900691,
+      "loss": 0.7042,
+      "step": 5686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026409602110255396,
+      "loss": 0.2132,
+      "step": 5687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002640836977879622,
+      "loss": 0.0147,
+      "step": 5688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002640713726464913,
+      "loss": 1.5071,
+      "step": 5689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002640590456783385,
+      "loss": 0.0843,
+      "step": 5690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026404671688370126,
+      "loss": 2.2267,
+      "step": 5691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026403438626277704,
+      "loss": 0.5867,
+      "step": 5692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002640220538157633,
+      "loss": 0.447,
+      "step": 5693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002640097195428574,
+      "loss": 0.7049,
+      "step": 5694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026399738344425706,
+      "loss": 1.465,
+      "step": 5695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026398504552015964,
+      "loss": 0.085,
+      "step": 5696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002639727057707628,
+      "loss": 0.6358,
+      "step": 5697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002639603641962641,
+      "loss": 0.6089,
+      "step": 5698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026394802079686126,
+      "loss": 0.6478,
+      "step": 5699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026393567557275185,
+      "loss": 0.3666,
+      "step": 5700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002639233285241335,
+      "loss": 1.6602,
+      "step": 5701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026391097965120414,
+      "loss": 0.5042,
+      "step": 5702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026389862895416136,
+      "loss": 0.5682,
+      "step": 5703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026388627643320295,
+      "loss": 0.8292,
+      "step": 5704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026387392208852676,
+      "loss": 0.3038,
+      "step": 5705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002638615659203306,
+      "loss": 0.4086,
+      "step": 5706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002638492079288124,
+      "loss": 0.6633,
+      "step": 5707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002638368481141699,
+      "loss": 0.4378,
+      "step": 5708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002638244864766012,
+      "loss": 0.0946,
+      "step": 5709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026381212301630416,
+      "loss": 0.4374,
+      "step": 5710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026379975773347677,
+      "loss": 0.2052,
+      "step": 5711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026378739062831705,
+      "loss": 0.896,
+      "step": 5712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026377502170102303,
+      "loss": 1.4325,
+      "step": 5713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002637626509517928,
+      "loss": 0.5507,
+      "step": 5714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026375027838082446,
+      "loss": 0.4014,
+      "step": 5715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002637379039883161,
+      "loss": 0.5438,
+      "step": 5716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002637255277744659,
+      "loss": 1.0293,
+      "step": 5717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026371314973947207,
+      "loss": 0.6124,
+      "step": 5718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026370076988353286,
+      "loss": 0.7583,
+      "step": 5719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002636883882068464,
+      "loss": 0.1772,
+      "step": 5720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026367600470961097,
+      "loss": 0.8267,
+      "step": 5721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000263663619392025,
+      "loss": 1.0812,
+      "step": 5722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026365123225428677,
+      "loss": 0.3607,
+      "step": 5723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026363884329659457,
+      "loss": 0.7836,
+      "step": 5724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026362645251914687,
+      "loss": 0.5798,
+      "step": 5725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002636140599221421,
+      "loss": 0.2857,
+      "step": 5726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002636016655057786,
+      "loss": 0.8928,
+      "step": 5727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026358926927025496,
+      "loss": 0.3722,
+      "step": 5728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002635768712157696,
+      "loss": 0.2363,
+      "step": 5729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026356447134252115,
+      "loss": 0.5108,
+      "step": 5730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002635520696507081,
+      "loss": 0.6161,
+      "step": 5731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002635396661405291,
+      "loss": 0.2065,
+      "step": 5732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002635272608121828,
+      "loss": 0.3296,
+      "step": 5733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026351485366586766,
+      "loss": 0.3828,
+      "step": 5734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002635024447017826,
+      "loss": 0.4292,
+      "step": 5735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002634900339201262,
+      "loss": 0.69,
+      "step": 5736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026347762132109724,
+      "loss": 0.5092,
+      "step": 5737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002634652069048945,
+      "loss": 1.084,
+      "step": 5738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002634527906717167,
+      "loss": 0.4775,
+      "step": 5739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002634403726217628,
+      "loss": 1.2544,
+      "step": 5740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002634279527552315,
+      "loss": 0.4922,
+      "step": 5741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026341553107232183,
+      "loss": 0.9375,
+      "step": 5742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002634031075732326,
+      "loss": 0.8388,
+      "step": 5743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026339068225816284,
+      "loss": 0.4014,
+      "step": 5744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026337825512731146,
+      "loss": 0.4025,
+      "step": 5745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002633658261808775,
+      "loss": 0.8751,
+      "step": 5746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026335339541905993,
+      "loss": 0.6274,
+      "step": 5747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026334096284205786,
+      "loss": 0.6754,
+      "step": 5748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002633285284500704,
+      "loss": 0.6675,
+      "step": 5749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026331609224329664,
+      "loss": 0.3241,
+      "step": 5750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002633036542219357,
+      "loss": 0.0115,
+      "step": 5751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002632912143861868,
+      "loss": 0.6226,
+      "step": 5752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026327877273624916,
+      "loss": 0.3789,
+      "step": 5753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026326632927232193,
+      "loss": 0.1635,
+      "step": 5754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026325388399460443,
+      "loss": 0.6678,
+      "step": 5755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002632414369032959,
+      "loss": 0.4334,
+      "step": 5756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002632289879985958,
+      "loss": 0.048,
+      "step": 5757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026321653728070336,
+      "loss": 2.1211,
+      "step": 5758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000263204084749818,
+      "loss": 0.3057,
+      "step": 5759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002631916304061391,
+      "loss": 0.5707,
+      "step": 5760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002631791742498661,
+      "loss": 0.9201,
+      "step": 5761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026316671628119853,
+      "loss": 0.0102,
+      "step": 5762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002631542565003358,
+      "loss": 0.4461,
+      "step": 5763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002631417949074775,
+      "loss": 0.2523,
+      "step": 5764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026312933150282315,
+      "loss": 0.2958,
+      "step": 5765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026311686628657235,
+      "loss": 0.2544,
+      "step": 5766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026310439925892465,
+      "loss": 0.3043,
+      "step": 5767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026309193042007976,
+      "loss": 0.3443,
+      "step": 5768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002630794597702374,
+      "loss": 0.0735,
+      "step": 5769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002630669873095971,
+      "loss": 0.0688,
+      "step": 5770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026305451303835874,
+      "loss": 0.4408,
+      "step": 5771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026304203695672203,
+      "loss": 0.0702,
+      "step": 5772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026302955906488674,
+      "loss": 0.5171,
+      "step": 5773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002630170793630527,
+      "loss": 0.2857,
+      "step": 5774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026300459785141974,
+      "loss": 0.3306,
+      "step": 5775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002629921145301877,
+      "loss": 0.1914,
+      "step": 5776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002629796293995566,
+      "loss": 0.2673,
+      "step": 5777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026296714245972627,
+      "loss": 0.1562,
+      "step": 5778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002629546537108967,
+      "loss": 1.4614,
+      "step": 5779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026294216315326786,
+      "loss": 0.4539,
+      "step": 5780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026292967078703975,
+      "loss": 0.8104,
+      "step": 5781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026291717661241246,
+      "loss": 0.4194,
+      "step": 5782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026290468062958606,
+      "loss": 0.5295,
+      "step": 5783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026289218283876065,
+      "loss": 0.8313,
+      "step": 5784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002628796832401364,
+      "loss": 0.237,
+      "step": 5785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002628671818339134,
+      "loss": 0.0618,
+      "step": 5786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002628546786202919,
+      "loss": 0.0021,
+      "step": 5787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026284217359947204,
+      "loss": 0.4667,
+      "step": 5788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026282966677165407,
+      "loss": 1.2786,
+      "step": 5789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026281715813703845,
+      "loss": 0.584,
+      "step": 5790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026280464769582525,
+      "loss": 0.3922,
+      "step": 5791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000262792135448215,
+      "loss": 1.2522,
+      "step": 5792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002627796213944079,
+      "loss": 2.0645,
+      "step": 5793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002627671055346045,
+      "loss": 0.3853,
+      "step": 5794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002627545878690051,
+      "loss": 0.7081,
+      "step": 5795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002627420683978102,
+      "loss": 0.6748,
+      "step": 5796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026272954712122025,
+      "loss": 0.007,
+      "step": 5797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026271702403943585,
+      "loss": 0.06,
+      "step": 5798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002627044991526574,
+      "loss": 0.1737,
+      "step": 5799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002626919724610856,
+      "loss": 0.5375,
+      "step": 5800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026267944396492096,
+      "loss": 0.2663,
+      "step": 5801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026266691366436416,
+      "loss": 0.3092,
+      "step": 5802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002626543815596158,
+      "loss": 0.4834,
+      "step": 5803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002626418476508766,
+      "loss": 0.4743,
+      "step": 5804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026262931193834723,
+      "loss": 0.7136,
+      "step": 5805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026261677442222857,
+      "loss": 1.1129,
+      "step": 5806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026260423510272114,
+      "loss": 0.5759,
+      "step": 5807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026259169398002596,
+      "loss": 0.0316,
+      "step": 5808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026257915105434376,
+      "loss": 0.3307,
+      "step": 5809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002625666063258754,
+      "loss": 0.0298,
+      "step": 5810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026255405979482184,
+      "loss": 0.0685,
+      "step": 5811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026254151146138386,
+      "loss": 1.9578,
+      "step": 5812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002625289613257625,
+      "loss": 0.3447,
+      "step": 5813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002625164093881587,
+      "loss": 1.0424,
+      "step": 5814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002625038556487735,
+      "loss": 0.3069,
+      "step": 5815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026249130010780786,
+      "loss": 0.5055,
+      "step": 5816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002624787427654629,
+      "loss": 0.007,
+      "step": 5817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026246618362193964,
+      "loss": 0.0128,
+      "step": 5818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002624536226774393,
+      "loss": 0.4852,
+      "step": 5819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002624410599321629,
+      "loss": 4.7412,
+      "step": 5820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026242849538631166,
+      "loss": 0.014,
+      "step": 5821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002624159290400869,
+      "loss": 0.0067,
+      "step": 5822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002624033608936897,
+      "loss": 0.0634,
+      "step": 5823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026239079094732135,
+      "loss": 0.5168,
+      "step": 5824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026237821920118315,
+      "loss": 0.3707,
+      "step": 5825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026236564565547643,
+      "loss": 0.3484,
+      "step": 5826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002623530703104026,
+      "loss": 0.0093,
+      "step": 5827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026234049316616287,
+      "loss": 1.4932,
+      "step": 5828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026232791422295875,
+      "loss": 1.1062,
+      "step": 5829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026231533348099173,
+      "loss": 0.0039,
+      "step": 5830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026230275094046314,
+      "loss": 0.0256,
+      "step": 5831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002622901666015746,
+      "loss": 0.5801,
+      "step": 5832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026227758046452757,
+      "loss": 0.3892,
+      "step": 5833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026226499252952353,
+      "loss": 0.2341,
+      "step": 5834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002622524027967642,
+      "loss": 0.5934,
+      "step": 5835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026223981126645104,
+      "loss": 0.3539,
+      "step": 5836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026222721793878575,
+      "loss": 0.1678,
+      "step": 5837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026221462281397,
+      "loss": 0.843,
+      "step": 5838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002622020258922055,
+      "loss": 0.2319,
+      "step": 5839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002621894271736939,
+      "loss": 0.4036,
+      "step": 5840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002621768266586371,
+      "loss": 0.4581,
+      "step": 5841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026216422434723665,
+      "loss": 1.2901,
+      "step": 5842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002621516202396946,
+      "loss": 0.1361,
+      "step": 5843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026213901433621255,
+      "loss": 0.1104,
+      "step": 5844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026212640663699253,
+      "loss": 0.4992,
+      "step": 5845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002621137971422364,
+      "loss": 1.0512,
+      "step": 5846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026210118585214605,
+      "loss": 0.2401,
+      "step": 5847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002620885727669235,
+      "loss": 0.0182,
+      "step": 5848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002620759578867706,
+      "loss": 0.5635,
+      "step": 5849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002620633412118895,
+      "loss": 0.2643,
+      "step": 5850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002620507227424822,
+      "loss": 0.1339,
+      "step": 5851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002620381024787506,
+      "loss": 0.3118,
+      "step": 5852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026202548042089706,
+      "loss": 0.9502,
+      "step": 5853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002620128565691235,
+      "loss": 0.2101,
+      "step": 5854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00026200023092363224,
+      "loss": 0.3894,
+      "step": 5855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002619876034846254,
+      "loss": 1.1029,
+      "step": 5856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000261974974252305,
+      "loss": 0.0313,
+      "step": 5857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026196234322687354,
+      "loss": 0.9858,
+      "step": 5858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026194971040853314,
+      "loss": 0.3536,
+      "step": 5859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002619370757974862,
+      "loss": 1.8643,
+      "step": 5860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026192443939393494,
+      "loss": 0.4938,
+      "step": 5861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002619118011980818,
+      "loss": 1.0845,
+      "step": 5862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002618991612101291,
+      "loss": 0.1458,
+      "step": 5863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002618865194302792,
+      "loss": 0.1591,
+      "step": 5864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002618738758587347,
+      "loss": 0.56,
+      "step": 5865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000261861230495698,
+      "loss": 0.2768,
+      "step": 5866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002618485833413715,
+      "loss": 0.3722,
+      "step": 5867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002618359343959578,
+      "loss": 0.7039,
+      "step": 5868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026182328365965946,
+      "loss": 0.3355,
+      "step": 5869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026181063113267907,
+      "loss": 0.5071,
+      "step": 5870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026179797681521923,
+      "loss": 0.3851,
+      "step": 5871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002617853207074826,
+      "loss": 1.2149,
+      "step": 5872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026177266280967174,
+      "loss": 0.1414,
+      "step": 5873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002617600031219895,
+      "loss": 0.5786,
+      "step": 5874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002617473416446386,
+      "loss": 0.5348,
+      "step": 5875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002617346783778216,
+      "loss": 0.4309,
+      "step": 5876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002617220133217415,
+      "loss": 1.0506,
+      "step": 5877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000261709346476601,
+      "loss": 0.0282,
+      "step": 5878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026169667784260297,
+      "loss": 0.4379,
+      "step": 5879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026168400741995027,
+      "loss": 0.3945,
+      "step": 5880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002616713352088458,
+      "loss": 0.0022,
+      "step": 5881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026165866120949255,
+      "loss": 2.2618,
+      "step": 5882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026164598542209334,
+      "loss": 0.5163,
+      "step": 5883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002616333078468513,
+      "loss": 1.6818,
+      "step": 5884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026162062848396935,
+      "loss": 0.5133,
+      "step": 5885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002616079473336505,
+      "loss": 0.8463,
+      "step": 5886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026159526439609794,
+      "loss": 0.5826,
+      "step": 5887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002615825796715147,
+      "loss": 0.2557,
+      "step": 5888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002615698931601039,
+      "loss": 0.0486,
+      "step": 5889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026155720486206865,
+      "loss": 0.5561,
+      "step": 5890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026154451477761225,
+      "loss": 0.9544,
+      "step": 5891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002615318229069378,
+      "loss": 0.7517,
+      "step": 5892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002615191292502487,
+      "loss": 0.7169,
+      "step": 5893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026150643380774797,
+      "loss": 0.7872,
+      "step": 5894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026149373657963915,
+      "loss": 0.2281,
+      "step": 5895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002614810375661254,
+      "loss": 0.2968,
+      "step": 5896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002614683367674102,
+      "loss": 0.0133,
+      "step": 5897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026145563418369683,
+      "loss": 0.1495,
+      "step": 5898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026144292981518873,
+      "loss": 0.6649,
+      "step": 5899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026143022366208944,
+      "loss": 0.0163,
+      "step": 5900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002614175157246023,
+      "loss": 0.2932,
+      "step": 5901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026140480600293087,
+      "loss": 0.2181,
+      "step": 5902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002613920944972787,
+      "loss": 2.8184,
+      "step": 5903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026137938120784926,
+      "loss": 1.7178,
+      "step": 5904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026136666613484625,
+      "loss": 0.1884,
+      "step": 5905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026135394927847313,
+      "loss": 0.8113,
+      "step": 5906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026134123063893374,
+      "loss": 0.9957,
+      "step": 5907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002613285102164315,
+      "loss": 0.4598,
+      "step": 5908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002613157880111704,
+      "loss": 0.2123,
+      "step": 5909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026130306402335395,
+      "loss": 0.8738,
+      "step": 5910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026129033825318595,
+      "loss": 0.0209,
+      "step": 5911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002612776107008702,
+      "loss": 0.6367,
+      "step": 5912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002612648813666106,
+      "loss": 0.7332,
+      "step": 5913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002612521502506108,
+      "loss": 2.435,
+      "step": 5914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026123941735307487,
+      "loss": 0.2492,
+      "step": 5915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002612266826742066,
+      "loss": 0.7809,
+      "step": 5916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002612139462142099,
+      "loss": 0.3045,
+      "step": 5917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002612012079732888,
+      "loss": 0.603,
+      "step": 5918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002611884679516472,
+      "loss": 0.0312,
+      "step": 5919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002611757261494892,
+      "loss": 0.0511,
+      "step": 5920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002611629825670188,
+      "loss": 0.6671,
+      "step": 5921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026115023720444004,
+      "loss": 0.2292,
+      "step": 5922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026113749006195704,
+      "loss": 0.2476,
+      "step": 5923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002611247411397739,
+      "loss": 0.4548,
+      "step": 5924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002611119904380949,
+      "loss": 0.6519,
+      "step": 5925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026109923795712405,
+      "loss": 0.3302,
+      "step": 5926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026108648369706564,
+      "loss": 0.3467,
+      "step": 5927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002610737276581239,
+      "loss": 0.6338,
+      "step": 5928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002610609698405032,
+      "loss": 0.4657,
+      "step": 5929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026104821024440766,
+      "loss": 0.3594,
+      "step": 5930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026103544887004173,
+      "loss": 0.0122,
+      "step": 5931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002610226857176097,
+      "loss": 0.6286,
+      "step": 5932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026100992078731606,
+      "loss": 0.3576,
+      "step": 5933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026099715407936505,
+      "loss": 0.298,
+      "step": 5934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026098438559396124,
+      "loss": 0.0552,
+      "step": 5935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026097161533130905,
+      "loss": 0.4508,
+      "step": 5936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000260958843291613,
+      "loss": 0.196,
+      "step": 5937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002609460694750776,
+      "loss": 0.3478,
+      "step": 5938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002609332938819074,
+      "loss": 0.5441,
+      "step": 5939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000260920516512307,
+      "loss": 0.2328,
+      "step": 5940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002609077373664809,
+      "loss": 0.7882,
+      "step": 5941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026089495644463395,
+      "loss": 0.3749,
+      "step": 5942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026088217374697063,
+      "loss": 1.2696,
+      "step": 5943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026086938927369574,
+      "loss": 0.0173,
+      "step": 5944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000260856603025014,
+      "loss": 0.6553,
+      "step": 5945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002608438150011301,
+      "loss": 0.4971,
+      "step": 5946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026083102520224887,
+      "loss": 0.3366,
+      "step": 5947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002608182336285751,
+      "loss": 0.668,
+      "step": 5948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002608054402803137,
+      "loss": 1.0264,
+      "step": 5949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026079264515766935,
+      "loss": 0.1513,
+      "step": 5950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026077984826084713,
+      "loss": 0.8847,
+      "step": 5951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026076704959005186,
+      "loss": 1.0264,
+      "step": 5952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026075424914548856,
+      "loss": 0.8011,
+      "step": 5953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002607414469273622,
+      "loss": 1.1661,
+      "step": 5954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002607286429358777,
+      "loss": 0.848,
+      "step": 5955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002607158371712402,
+      "loss": 0.4924,
+      "step": 5956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002607030296336548,
+      "loss": 0.526,
+      "step": 5957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002606902203233264,
+      "loss": 0.8598,
+      "step": 5958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002606774092404603,
+      "loss": 0.265,
+      "step": 5959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002606645963852616,
+      "loss": 0.2982,
+      "step": 5960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002606517817579355,
+      "loss": 0.7018,
+      "step": 5961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026063896535868713,
+      "loss": 0.209,
+      "step": 5962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002606261471877218,
+      "loss": 0.2355,
+      "step": 5963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002606133272452447,
+      "loss": 0.4606,
+      "step": 5964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002606005055314612,
+      "loss": 0.0968,
+      "step": 5965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026058768204657666,
+      "loss": 0.4701,
+      "step": 5966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002605748567907963,
+      "loss": 1.2551,
+      "step": 5967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026056202976432554,
+      "loss": 0.4131,
+      "step": 5968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026054920096736986,
+      "loss": 0.4732,
+      "step": 5969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002605363704001346,
+      "loss": 0.2899,
+      "step": 5970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002605235380628252,
+      "loss": 0.3216,
+      "step": 5971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026051070395564734,
+      "loss": 0.3896,
+      "step": 5972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026049786807880637,
+      "loss": 0.5823,
+      "step": 5973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026048503043250783,
+      "loss": 0.721,
+      "step": 5974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002604721910169574,
+      "loss": 0.4804,
+      "step": 5975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026045934983236056,
+      "loss": 0.7918,
+      "step": 5976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000260446506878923,
+      "loss": 0.3646,
+      "step": 5977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002604336621568504,
+      "loss": 0.3436,
+      "step": 5978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002604208156663485,
+      "loss": 0.6238,
+      "step": 5979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002604079674076229,
+      "loss": 1.0117,
+      "step": 5980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026039511738087945,
+      "loss": 0.0833,
+      "step": 5981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002603822655863239,
+      "loss": 0.3487,
+      "step": 5982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026036941202416195,
+      "loss": 0.2645,
+      "step": 5983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026035655669459954,
+      "loss": 1.2022,
+      "step": 5984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002603436995978425,
+      "loss": 0.9463,
+      "step": 5985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002603308407340967,
+      "loss": 0.3864,
+      "step": 5986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026031798010356806,
+      "loss": 0.5237,
+      "step": 5987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002603051177064626,
+      "loss": 1.0138,
+      "step": 5988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026029225354298616,
+      "loss": 0.559,
+      "step": 5989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002602793876133448,
+      "loss": 0.4502,
+      "step": 5990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026026651991774454,
+      "loss": 0.5618,
+      "step": 5991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002602536504563915,
+      "loss": 1.0394,
+      "step": 5992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002602407792294917,
+      "loss": 0.9112,
+      "step": 5993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026022790623725123,
+      "loss": 1.0748,
+      "step": 5994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002602150314798763,
+      "loss": 0.0055,
+      "step": 5995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002602021549575731,
+      "loss": 1.6202,
+      "step": 5996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002601892766705477,
+      "loss": 0.58,
+      "step": 5997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026017639661900645,
+      "loss": 0.1929,
+      "step": 5998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026016351480315556,
+      "loss": 0.6722,
+      "step": 5999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002601506312232013,
+      "loss": 0.6363,
+      "step": 6000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026013774587935,
+      "loss": 0.8608,
+      "step": 6001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000260124858771808,
+      "loss": 0.3725,
+      "step": 6002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026011196990078164,
+      "loss": 0.543,
+      "step": 6003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026009907926647735,
+      "loss": 2.2141,
+      "step": 6004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026008618686910155,
+      "loss": 0.4775,
+      "step": 6005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026007329270886066,
+      "loss": 1.0373,
+      "step": 6006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002600603967859612,
+      "loss": 0.2995,
+      "step": 6007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026004749910060965,
+      "loss": 0.5742,
+      "step": 6008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002600345996530126,
+      "loss": 0.0494,
+      "step": 6009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002600216984433765,
+      "loss": 1.0243,
+      "step": 6010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00026000879547190807,
+      "loss": 0.5889,
+      "step": 6011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002599958907388139,
+      "loss": 0.3191,
+      "step": 6012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002599829842443006,
+      "loss": 0.4867,
+      "step": 6013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002599700759885748,
+      "loss": 0.497,
+      "step": 6014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025995716597184337,
+      "loss": 1.0391,
+      "step": 6015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002599442541943129,
+      "loss": 0.0326,
+      "step": 6016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025993134065619024,
+      "loss": 0.6313,
+      "step": 6017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002599184253576821,
+      "loss": 0.3638,
+      "step": 6018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002599055082989954,
+      "loss": 0.2729,
+      "step": 6019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002598925894803369,
+      "loss": 0.5051,
+      "step": 6020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025987966890191354,
+      "loss": 0.6741,
+      "step": 6021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025986674656393213,
+      "loss": 0.3952,
+      "step": 6022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025985382246659965,
+      "loss": 0.8054,
+      "step": 6023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025984089661012307,
+      "loss": 0.3258,
+      "step": 6024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025982796899470947,
+      "loss": 0.5311,
+      "step": 6025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025981503962056566,
+      "loss": 0.7238,
+      "step": 6026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002598021084878989,
+      "loss": 0.1642,
+      "step": 6027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002597891755969161,
+      "loss": 0.555,
+      "step": 6028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002597762409478244,
+      "loss": 0.253,
+      "step": 6029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025976330454083107,
+      "loss": 0.3327,
+      "step": 6030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000259750366376143,
+      "loss": 2.1267,
+      "step": 6031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025973742645396764,
+      "loss": 0.2716,
+      "step": 6032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025972448477451206,
+      "loss": 0.066,
+      "step": 6033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002597115413379835,
+      "loss": 0.2109,
+      "step": 6034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025969859614458925,
+      "loss": 0.7317,
+      "step": 6035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002596856491945367,
+      "loss": 0.0349,
+      "step": 6036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000259672700488033,
+      "loss": 1.0347,
+      "step": 6037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002596597500252856,
+      "loss": 0.7271,
+      "step": 6038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025964679780650195,
+      "loss": 0.5219,
+      "step": 6039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025963384383188935,
+      "loss": 0.0166,
+      "step": 6040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025962088810165527,
+      "loss": 1.2323,
+      "step": 6041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002596079306160072,
+      "loss": 0.0334,
+      "step": 6042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002595949713751526,
+      "loss": 0.8106,
+      "step": 6043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025958201037929905,
+      "loss": 0.0422,
+      "step": 6044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000259569047628654,
+      "loss": 0.254,
+      "step": 6045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025955608312342517,
+      "loss": 0.4777,
+      "step": 6046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025954311686382004,
+      "loss": 2.2679,
+      "step": 6047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002595301488500463,
+      "loss": 0.7599,
+      "step": 6048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002595171790823116,
+      "loss": 0.0174,
+      "step": 6049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025950420756082365,
+      "loss": 0.7514,
+      "step": 6050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002594912342857902,
+      "loss": 0.5205,
+      "step": 6051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002594782592574189,
+      "loss": 0.3242,
+      "step": 6052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002594652824759176,
+      "loss": 0.25,
+      "step": 6053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002594523039414941,
+      "loss": 0.2332,
+      "step": 6054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025943932365435617,
+      "loss": 0.1445,
+      "step": 6055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025942634161471177,
+      "loss": 0.2531,
+      "step": 6056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025941335782276876,
+      "loss": 0.2994,
+      "step": 6057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000259400372278735,
+      "loss": 0.0072,
+      "step": 6058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002593873849828184,
+      "loss": 0.6095,
+      "step": 6059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025937439593522703,
+      "loss": 0.2567,
+      "step": 6060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025936140513616886,
+      "loss": 0.7412,
+      "step": 6061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025934841258585194,
+      "loss": 0.3664,
+      "step": 6062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025933541828448433,
+      "loss": 0.0954,
+      "step": 6063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025932242223227403,
+      "loss": 0.0472,
+      "step": 6064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002593094244294292,
+      "loss": 0.143,
+      "step": 6065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000259296424876158,
+      "loss": 0.0092,
+      "step": 6066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002592834235726686,
+      "loss": 0.496,
+      "step": 6067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002592704205191692,
+      "loss": 1.0961,
+      "step": 6068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000259257415715868,
+      "loss": 0.9671,
+      "step": 6069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025924440916297325,
+      "loss": 1.0786,
+      "step": 6070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025923140086069323,
+      "loss": 0.3191,
+      "step": 6071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002592183908092363,
+      "loss": 0.0078,
+      "step": 6072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002592053790088107,
+      "loss": 0.4218,
+      "step": 6073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002591923654596249,
+      "loss": 0.2495,
+      "step": 6074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025917935016188723,
+      "loss": 1.491,
+      "step": 6075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002591663331158061,
+      "loss": 0.7908,
+      "step": 6076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025915331432159,
+      "loss": 1.1125,
+      "step": 6077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025914029377944737,
+      "loss": 1.7452,
+      "step": 6078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002591272714895867,
+      "loss": 0.5378,
+      "step": 6079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002591142474522166,
+      "loss": 0.0052,
+      "step": 6080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002591012216675456,
+      "loss": 1.1652,
+      "step": 6081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025908819413578225,
+      "loss": 0.3855,
+      "step": 6082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002590751648571352,
+      "loss": 0.0687,
+      "step": 6083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025906213383181304,
+      "loss": 0.0331,
+      "step": 6084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002590491010600245,
+      "loss": 0.3106,
+      "step": 6085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025903606654197834,
+      "loss": 0.9116,
+      "step": 6086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025902303027788316,
+      "loss": 0.3991,
+      "step": 6087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025900999226794775,
+      "loss": 0.0255,
+      "step": 6088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002589969525123809,
+      "loss": 0.0038,
+      "step": 6089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002589839110113915,
+      "loss": 0.4511,
+      "step": 6090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002589708677651883,
+      "loss": 0.4101,
+      "step": 6091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002589578227739802,
+      "loss": 0.6295,
+      "step": 6092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025894477603797606,
+      "loss": 0.3421,
+      "step": 6093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002589317275573849,
+      "loss": 0.2559,
+      "step": 6094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025891867733241554,
+      "loss": 0.2594,
+      "step": 6095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002589056253632771,
+      "loss": 1.0034,
+      "step": 6096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025889257165017846,
+      "loss": 0.339,
+      "step": 6097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002588795161933287,
+      "loss": 0.4406,
+      "step": 6098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025886645899293693,
+      "loss": 0.4891,
+      "step": 6099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025885340004921215,
+      "loss": 0.3565,
+      "step": 6100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002588403393623636,
+      "loss": 0.3218,
+      "step": 6101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002588272769326003,
+      "loss": 0.129,
+      "step": 6102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002588142127601315,
+      "loss": 2.668,
+      "step": 6103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025880114684516647,
+      "loss": 0.3237,
+      "step": 6104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002587880791879143,
+      "loss": 1.5649,
+      "step": 6105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002587750097885843,
+      "loss": 0.079,
+      "step": 6106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025876193864738577,
+      "loss": 0.0947,
+      "step": 6107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025874886576452804,
+      "loss": 0.2987,
+      "step": 6108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002587357911402204,
+      "loss": 0.6529,
+      "step": 6109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025872271477467225,
+      "loss": 0.0043,
+      "step": 6110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000258709636668093,
+      "loss": 0.845,
+      "step": 6111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025869655682069214,
+      "loss": 0.7469,
+      "step": 6112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000258683475232679,
+      "loss": 0.4095,
+      "step": 6113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002586703919042631,
+      "loss": 1.292,
+      "step": 6114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000258657306835654,
+      "loss": 0.8651,
+      "step": 6115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002586442200270612,
+      "loss": 0.2005,
+      "step": 6116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025863113147869425,
+      "loss": 0.3522,
+      "step": 6117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002586180411907628,
+      "loss": 0.5691,
+      "step": 6118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025860494916347634,
+      "loss": 0.5802,
+      "step": 6119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025859185539704466,
+      "loss": 0.4053,
+      "step": 6120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025857875989167744,
+      "loss": 0.9073,
+      "step": 6121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002585656626475843,
+      "loss": 0.4838,
+      "step": 6122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000258552563664975,
+      "loss": 0.4305,
+      "step": 6123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002585394629440593,
+      "loss": 1.4721,
+      "step": 6124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000258526360485047,
+      "loss": 0.5658,
+      "step": 6125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002585132562881479,
+      "loss": 0.4341,
+      "step": 6126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025850015035357187,
+      "loss": 0.0053,
+      "step": 6127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002584870426815288,
+      "loss": 0.115,
+      "step": 6128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025847393327222854,
+      "loss": 0.004,
+      "step": 6129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000258460822125881,
+      "loss": 0.558,
+      "step": 6130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002584477092426962,
+      "loss": 0.2299,
+      "step": 6131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025843459462288407,
+      "loss": 0.3775,
+      "step": 6132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025842147826665463,
+      "loss": 0.0441,
+      "step": 6133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000258408360174218,
+      "loss": 0.9039,
+      "step": 6134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025839524034578415,
+      "loss": 1.3209,
+      "step": 6135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025838211878156323,
+      "loss": 1.206,
+      "step": 6136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025836899548176525,
+      "loss": 0.3233,
+      "step": 6137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025835587044660053,
+      "loss": 0.0518,
+      "step": 6138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025834274367627914,
+      "loss": 0.2436,
+      "step": 6139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002583296151710113,
+      "loss": 0.7051,
+      "step": 6140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002583164849310073,
+      "loss": 1.3087,
+      "step": 6141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002583033529564773,
+      "loss": 0.8149,
+      "step": 6142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002582902192476316,
+      "loss": 0.4194,
+      "step": 6143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025827708380468065,
+      "loss": 0.0342,
+      "step": 6144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025826394662783466,
+      "loss": 0.4432,
+      "step": 6145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002582508077173041,
+      "loss": 0.6076,
+      "step": 6146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025823766707329924,
+      "loss": 0.8785,
+      "step": 6147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025822452469603064,
+      "loss": 0.8251,
+      "step": 6148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025821138058570867,
+      "loss": 0.0037,
+      "step": 6149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025819823474254383,
+      "loss": 0.9015,
+      "step": 6150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002581850871667467,
+      "loss": 0.2285,
+      "step": 6151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002581719378585277,
+      "loss": 0.6392,
+      "step": 6152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025815878681809747,
+      "loss": 0.4785,
+      "step": 6153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025814563404566665,
+      "loss": 0.0973,
+      "step": 6154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002581324795414457,
+      "loss": 0.0032,
+      "step": 6155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025811932330564545,
+      "loss": 0.3941,
+      "step": 6156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002581061653384765,
+      "loss": 0.5211,
+      "step": 6157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002580930056401495,
+      "loss": 0.9853,
+      "step": 6158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025807984421087534,
+      "loss": 0.2766,
+      "step": 6159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002580666810508646,
+      "loss": 0.2893,
+      "step": 6160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002580535161603282,
+      "loss": 0.622,
+      "step": 6161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002580403495394768,
+      "loss": 0.4279,
+      "step": 6162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025802718118852146,
+      "loss": 0.7636,
+      "step": 6163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025801401110767293,
+      "loss": 0.3198,
+      "step": 6164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002580008392971421,
+      "loss": 0.1575,
+      "step": 6165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002579876657571399,
+      "loss": 0.4311,
+      "step": 6166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025797449048787726,
+      "loss": 0.3243,
+      "step": 6167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025796131348956527,
+      "loss": 0.037,
+      "step": 6168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025794813476241485,
+      "loss": 0.314,
+      "step": 6169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002579349543066371,
+      "loss": 0.8732,
+      "step": 6170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025792177212244295,
+      "loss": 1.0067,
+      "step": 6171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025790858821004366,
+      "loss": 1.2227,
+      "step": 6172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025789540256965026,
+      "loss": 0.3125,
+      "step": 6173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002578822152014739,
+      "loss": 0.9597,
+      "step": 6174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002578690261057258,
+      "loss": 0.555,
+      "step": 6175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025785583528261714,
+      "loss": 0.2005,
+      "step": 6176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025784264273235913,
+      "loss": 0.235,
+      "step": 6177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025782944845516303,
+      "loss": 0.5624,
+      "step": 6178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025781625245124016,
+      "loss": 0.0936,
+      "step": 6179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002578030547208018,
+      "loss": 0.3133,
+      "step": 6180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025778985526405936,
+      "loss": 0.8515,
+      "step": 6181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002577766540812241,
+      "loss": 0.0019,
+      "step": 6182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002577634511725075,
+      "loss": 0.6898,
+      "step": 6183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000257750246538121,
+      "loss": 0.4124,
+      "step": 6184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002577370401782759,
+      "loss": 0.0044,
+      "step": 6185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025772383209318386,
+      "loss": 0.2049,
+      "step": 6186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002577106222830563,
+      "loss": 0.9918,
+      "step": 6187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025769741074810484,
+      "loss": 0.2607,
+      "step": 6188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002576841974885409,
+      "loss": 0.118,
+      "step": 6189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002576709825045762,
+      "loss": 0.6309,
+      "step": 6190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025765776579642226,
+      "loss": 0.8028,
+      "step": 6191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025764454736429084,
+      "loss": 0.007,
+      "step": 6192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002576313272083935,
+      "loss": 0.6699,
+      "step": 6193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025761810532894195,
+      "loss": 0.3273,
+      "step": 6194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000257604881726148,
+      "loss": 0.3734,
+      "step": 6195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025759165640022334,
+      "loss": 0.274,
+      "step": 6196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002575784293513798,
+      "loss": 0.3012,
+      "step": 6197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002575652005798292,
+      "loss": 0.3136,
+      "step": 6198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002575519700857833,
+      "loss": 0.7568,
+      "step": 6199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025753873786945404,
+      "loss": 0.9055,
+      "step": 6200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025752550393105323,
+      "loss": 0.1207,
+      "step": 6201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025751226827079295,
+      "loss": 0.5897,
+      "step": 6202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025749903088888494,
+      "loss": 0.258,
+      "step": 6203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025748579178554127,
+      "loss": 0.4646,
+      "step": 6204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002574725509609741,
+      "loss": 0.4634,
+      "step": 6205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002574593084153952,
+      "loss": 0.6171,
+      "step": 6206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025744606414901676,
+      "loss": 1.35,
+      "step": 6207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025743281816205084,
+      "loss": 0.4246,
+      "step": 6208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025741957045470967,
+      "loss": 0.5953,
+      "step": 6209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002574063210272052,
+      "loss": 3.2383,
+      "step": 6210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025739306987974974,
+      "loss": 0.0068,
+      "step": 6211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002573798170125554,
+      "loss": 1.0816,
+      "step": 6212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002573665624258345,
+      "loss": 0.596,
+      "step": 6213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002573533061197992,
+      "loss": 1.0931,
+      "step": 6214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002573400480946618,
+      "loss": 1.0832,
+      "step": 6215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002573267883506347,
+      "loss": 0.2659,
+      "step": 6216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002573135268879301,
+      "loss": 0.3659,
+      "step": 6217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025730026370676044,
+      "loss": 0.6608,
+      "step": 6218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025728699880733807,
+      "loss": 0.3426,
+      "step": 6219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025727373218987544,
+      "loss": 0.4554,
+      "step": 6220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025726046385458494,
+      "loss": 0.9682,
+      "step": 6221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002572471938016792,
+      "loss": 0.3003,
+      "step": 6222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002572339220313705,
+      "loss": 1.0809,
+      "step": 6223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025722064854387155,
+      "loss": 0.442,
+      "step": 6224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002572073733393948,
+      "loss": 0.2039,
+      "step": 6225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025719409641815284,
+      "loss": 0.3258,
+      "step": 6226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002571808177803583,
+      "loss": 0.3941,
+      "step": 6227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025716753742622384,
+      "loss": 0.5106,
+      "step": 6228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002571542553559621,
+      "loss": 0.6509,
+      "step": 6229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025714097156978577,
+      "loss": 0.0097,
+      "step": 6230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002571276860679076,
+      "loss": 0.3035,
+      "step": 6231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025711439885054034,
+      "loss": 0.4278,
+      "step": 6232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025710110991789665,
+      "loss": 0.1662,
+      "step": 6233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002570878192701895,
+      "loss": 0.1278,
+      "step": 6234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002570745269076316,
+      "loss": 0.5239,
+      "step": 6235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025706123283043586,
+      "loss": 0.0274,
+      "step": 6236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002570479370388151,
+      "loss": 1.2978,
+      "step": 6237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025703463953298235,
+      "loss": 0.859,
+      "step": 6238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002570213403131505,
+      "loss": 0.3858,
+      "step": 6239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002570080393795324,
+      "loss": 0.7022,
+      "step": 6240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002569947367323413,
+      "loss": 0.2556,
+      "step": 6241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025698143237178994,
+      "loss": 0.7745,
+      "step": 6242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002569681262980915,
+      "loss": 0.317,
+      "step": 6243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002569548185114591,
+      "loss": 0.5999,
+      "step": 6244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025694150901210585,
+      "loss": 1.1994,
+      "step": 6245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025692819780024473,
+      "loss": 2.546,
+      "step": 6246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025691488487608903,
+      "loss": 0.4703,
+      "step": 6247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025690157023985195,
+      "loss": 1.0255,
+      "step": 6248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002568882538917466,
+      "loss": 0.5075,
+      "step": 6249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002568749358319863,
+      "loss": 0.243,
+      "step": 6250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002568616160607843,
+      "loss": 0.6495,
+      "step": 6251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025684829457835396,
+      "loss": 0.056,
+      "step": 6252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002568349713849085,
+      "loss": 1.0684,
+      "step": 6253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002568216464806612,
+      "loss": 0.1948,
+      "step": 6254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002568083198658257,
+      "loss": 0.0232,
+      "step": 6255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002567949915406152,
+      "loss": 1.2492,
+      "step": 6256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025678166150524314,
+      "loss": 0.9098,
+      "step": 6257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002567683297599231,
+      "loss": 0.3508,
+      "step": 6258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025675499630486846,
+      "loss": 0.3929,
+      "step": 6259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025674166114029274,
+      "loss": 0.1836,
+      "step": 6260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002567283242664096,
+      "loss": 0.6949,
+      "step": 6261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025671498568343245,
+      "loss": 0.3975,
+      "step": 6262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025670164539157495,
+      "loss": 0.8349,
+      "step": 6263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002566883033910508,
+      "loss": 0.0289,
+      "step": 6264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025667495968207353,
+      "loss": 0.2878,
+      "step": 6265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025666161426485696,
+      "loss": 1.1268,
+      "step": 6266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025664826713961463,
+      "loss": 0.4279,
+      "step": 6267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002566349183065604,
+      "loss": 0.0084,
+      "step": 6268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000256621567765908,
+      "loss": 0.3434,
+      "step": 6269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025660821551787115,
+      "loss": 0.565,
+      "step": 6270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025659486156266373,
+      "loss": 0.1728,
+      "step": 6271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025658150590049964,
+      "loss": 2.7914,
+      "step": 6272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025656814853159267,
+      "loss": 0.5306,
+      "step": 6273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002565547894561567,
+      "loss": 1.6872,
+      "step": 6274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025654142867440576,
+      "loss": 0.3054,
+      "step": 6275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025652806618655364,
+      "loss": 0.9912,
+      "step": 6276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025651470199281447,
+      "loss": 0.3081,
+      "step": 6277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002565013360934022,
+      "loss": 0.0074,
+      "step": 6278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002564879684885309,
+      "loss": 0.4994,
+      "step": 6279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002564745991784146,
+      "loss": 0.9053,
+      "step": 6280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025646122816326734,
+      "loss": 0.2052,
+      "step": 6281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002564478554433033,
+      "loss": 1.1348,
+      "step": 6282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025643448101873666,
+      "loss": 0.2577,
+      "step": 6283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002564211048897815,
+      "loss": 1.1391,
+      "step": 6284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000256407727056652,
+      "loss": 0.4442,
+      "step": 6285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025639434751956256,
+      "loss": 0.7583,
+      "step": 6286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002563809662787273,
+      "loss": 1.038,
+      "step": 6287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025636758333436044,
+      "loss": 0.8222,
+      "step": 6288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025635419868667646,
+      "loss": 0.0226,
+      "step": 6289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025634081233588956,
+      "loss": 1.3429,
+      "step": 6290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002563274242822141,
+      "loss": 0.6926,
+      "step": 6291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025631403452586464,
+      "loss": 0.7001,
+      "step": 6292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002563006430670554,
+      "loss": 0.0034,
+      "step": 6293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025628724990600085,
+      "loss": 0.3529,
+      "step": 6294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025627385504291554,
+      "loss": 0.1341,
+      "step": 6295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000256260458478014,
+      "loss": 0.8677,
+      "step": 6296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002562470602115106,
+      "loss": 0.0234,
+      "step": 6297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025623366024362006,
+      "loss": 1.0967,
+      "step": 6298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002562202585745568,
+      "loss": 0.7377,
+      "step": 6299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025620685520453564,
+      "loss": 0.5886,
+      "step": 6300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025619345013377096,
+      "loss": 0.0111,
+      "step": 6301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002561800433624776,
+      "loss": 0.3133,
+      "step": 6302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025616663489087024,
+      "loss": 2.5051,
+      "step": 6303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002561532247191635,
+      "loss": 1.259,
+      "step": 6304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002561398128475722,
+      "loss": 0.5868,
+      "step": 6305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002561263992763111,
+      "loss": 0.0093,
+      "step": 6306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025611298400559496,
+      "loss": 0.5184,
+      "step": 6307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025609956703563863,
+      "loss": 0.4438,
+      "step": 6308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000256086148366657,
+      "loss": 1.2036,
+      "step": 6309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025607272799886495,
+      "loss": 0.004,
+      "step": 6310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002560593059324773,
+      "loss": 2.5607,
+      "step": 6311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000256045882167709,
+      "loss": 0.3034,
+      "step": 6312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025603245670477513,
+      "loss": 0.5475,
+      "step": 6313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025601902954389057,
+      "loss": 0.5466,
+      "step": 6314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025600560068527036,
+      "loss": 0.6142,
+      "step": 6315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025599217012912956,
+      "loss": 0.0108,
+      "step": 6316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025597873787568316,
+      "loss": 0.6418,
+      "step": 6317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002559653039251464,
+      "loss": 0.9479,
+      "step": 6318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002559518682777343,
+      "loss": 0.0044,
+      "step": 6319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025593843093366206,
+      "loss": 0.6389,
+      "step": 6320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025592499189314486,
+      "loss": 0.2462,
+      "step": 6321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002559115511563978,
+      "loss": 1.2182,
+      "step": 6322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025589810872363633,
+      "loss": 1.9983,
+      "step": 6323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002558846645950755,
+      "loss": 0.9038,
+      "step": 6324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025587121877093064,
+      "loss": 0.6327,
+      "step": 6325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025585777125141713,
+      "loss": 0.476,
+      "step": 6326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002558443220367503,
+      "loss": 0.4483,
+      "step": 6327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025583087112714554,
+      "loss": 0.7198,
+      "step": 6328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025581741852281816,
+      "loss": 0.5479,
+      "step": 6329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002558039642239836,
+      "loss": 0.2878,
+      "step": 6330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002557905082308574,
+      "loss": 1.258,
+      "step": 6331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000255777050543655,
+      "loss": 0.5118,
+      "step": 6332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002557635911625918,
+      "loss": 0.1693,
+      "step": 6333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002557501300878835,
+      "loss": 0.7386,
+      "step": 6334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025573666731974553,
+      "loss": 2.3454,
+      "step": 6335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025572320285839357,
+      "loss": 0.932,
+      "step": 6336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002557097367040431,
+      "loss": 0.442,
+      "step": 6337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002556962688569099,
+      "loss": 0.5312,
+      "step": 6338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025568279931720956,
+      "loss": 0.1945,
+      "step": 6339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002556693280851578,
+      "loss": 0.3702,
+      "step": 6340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025565585516097033,
+      "loss": 0.286,
+      "step": 6341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025564238054486297,
+      "loss": 0.7572,
+      "step": 6342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002556289042370514,
+      "loss": 0.3497,
+      "step": 6343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002556154262377514,
+      "loss": 0.9215,
+      "step": 6344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002556019465471789,
+      "loss": 0.3425,
+      "step": 6345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025558846516554975,
+      "loss": 0.4574,
+      "step": 6346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025557498209307977,
+      "loss": 0.2532,
+      "step": 6347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002555614973299849,
+      "loss": 0.7067,
+      "step": 6348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000255548010876481,
+      "loss": 0.8329,
+      "step": 6349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025553452273278417,
+      "loss": 0.4474,
+      "step": 6350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002555210328991104,
+      "loss": 0.0043,
+      "step": 6351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025550754137567555,
+      "loss": 0.4944,
+      "step": 6352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002554940481626958,
+      "loss": 0.6936,
+      "step": 6353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025548055326038725,
+      "loss": 0.1145,
+      "step": 6354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002554670566689659,
+      "loss": 0.5763,
+      "step": 6355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025545355838864787,
+      "loss": 0.2431,
+      "step": 6356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002554400584196495,
+      "loss": 3.2052,
+      "step": 6357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025542655676218676,
+      "loss": 0.1077,
+      "step": 6358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002554130534164759,
+      "loss": 1.0158,
+      "step": 6359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002553995483827332,
+      "loss": 2.261,
+      "step": 6360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025538604166117497,
+      "loss": 0.3241,
+      "step": 6361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025537253325201743,
+      "loss": 0.6693,
+      "step": 6362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002553590231554769,
+      "loss": 0.5348,
+      "step": 6363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002553455113717697,
+      "loss": 2.463,
+      "step": 6364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002553319979011123,
+      "loss": 0.2014,
+      "step": 6365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000255318482743721,
+      "loss": 0.5552,
+      "step": 6366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002553049658998123,
+      "loss": 1.155,
+      "step": 6367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025529144736960256,
+      "loss": 0.8787,
+      "step": 6368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002552779271533083,
+      "loss": 0.0208,
+      "step": 6369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002552644052511461,
+      "loss": 0.5809,
+      "step": 6370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002552508816633324,
+      "loss": 0.3491,
+      "step": 6371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002552373563900838,
+      "loss": 0.2922,
+      "step": 6372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002552238294316169,
+      "loss": 0.7052,
+      "step": 6373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002552103007881482,
+      "loss": 0.4723,
+      "step": 6374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025519677045989447,
+      "loss": 0.5668,
+      "step": 6375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025518323844707234,
+      "loss": 0.3159,
+      "step": 6376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025516970474989854,
+      "loss": 0.363,
+      "step": 6377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025515616936858977,
+      "loss": 0.7672,
+      "step": 6378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002551426323033627,
+      "loss": 0.3744,
+      "step": 6379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002551290935544342,
+      "loss": 0.7155,
+      "step": 6380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025511555312202104,
+      "loss": 0.88,
+      "step": 6381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025510201100634004,
+      "loss": 0.4947,
+      "step": 6382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002550884672076081,
+      "loss": 0.6813,
+      "step": 6383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025507492172604205,
+      "loss": 0.1428,
+      "step": 6384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025506137456185885,
+      "loss": 0.0211,
+      "step": 6385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002550478257152754,
+      "loss": 0.8307,
+      "step": 6386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002550342751865087,
+      "loss": 0.6231,
+      "step": 6387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025502072297577566,
+      "loss": 0.4428,
+      "step": 6388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002550071690832934,
+      "loss": 0.1908,
+      "step": 6389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002549936135092789,
+      "loss": 0.8768,
+      "step": 6390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025498005625394923,
+      "loss": 0.4323,
+      "step": 6391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025496649731752155,
+      "loss": 1.176,
+      "step": 6392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025495293670021297,
+      "loss": 0.6567,
+      "step": 6393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002549393744022405,
+      "loss": 0.3195,
+      "step": 6394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025492581042382156,
+      "loss": 1.1861,
+      "step": 6395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025491224476517323,
+      "loss": 0.3392,
+      "step": 6396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025489867742651276,
+      "loss": 0.2941,
+      "step": 6397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002548851084080573,
+      "loss": 1.7386,
+      "step": 6398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025487153771002434,
+      "loss": 0.4222,
+      "step": 6399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002548579653326311,
+      "loss": 0.5007,
+      "step": 6400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025484439127609486,
+      "loss": 0.3249,
+      "step": 6401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025483081554063307,
+      "loss": 0.2759,
+      "step": 6402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002548172381264631,
+      "loss": 0.6943,
+      "step": 6403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025480365903380236,
+      "loss": 0.0086,
+      "step": 6404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025479007826286826,
+      "loss": 0.7897,
+      "step": 6405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002547764958138784,
+      "loss": 0.6401,
+      "step": 6406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025476291168705014,
+      "loss": 1.1216,
+      "step": 6407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025474932588260115,
+      "loss": 2.1812,
+      "step": 6408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002547357384007489,
+      "loss": 0.7792,
+      "step": 6409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002547221492417109,
+      "loss": 0.3649,
+      "step": 6410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025470855840570495,
+      "loss": 2.014,
+      "step": 6411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025469496589294855,
+      "loss": 0.9571,
+      "step": 6412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025468137170365937,
+      "loss": 0.9263,
+      "step": 6413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025466777583805513,
+      "loss": 0.1944,
+      "step": 6414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025465417829635364,
+      "loss": 0.356,
+      "step": 6415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025464057907877244,
+      "loss": 0.044,
+      "step": 6416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002546269781855294,
+      "loss": 0.3138,
+      "step": 6417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002546133756168424,
+      "loss": 0.0958,
+      "step": 6418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002545997713729292,
+      "loss": 0.5653,
+      "step": 6419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025458616545400754,
+      "loss": 0.3563,
+      "step": 6420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002545725578602955,
+      "loss": 1.0659,
+      "step": 6421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002545589485920109,
+      "loss": 0.2388,
+      "step": 6422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002545453376493715,
+      "loss": 0.0126,
+      "step": 6423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025453172503259555,
+      "loss": 0.7707,
+      "step": 6424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002545181107419009,
+      "loss": 0.0039,
+      "step": 6425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002545044947775056,
+      "loss": 0.2855,
+      "step": 6426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025449087713962754,
+      "loss": 0.7046,
+      "step": 6427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000254477257828485,
+      "loss": 0.1637,
+      "step": 6428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002544636368442959,
+      "loss": 0.0057,
+      "step": 6429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002544500141872785,
+      "loss": 0.4642,
+      "step": 6430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025443638985765086,
+      "loss": 0.6154,
+      "step": 6431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025442276385563117,
+      "loss": 0.0235,
+      "step": 6432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025440913618143764,
+      "loss": 0.5061,
+      "step": 6433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025439550683528845,
+      "loss": 0.041,
+      "step": 6434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002543818758174019,
+      "loss": 0.2987,
+      "step": 6435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025436824312799635,
+      "loss": 0.3673,
+      "step": 6436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002543546087672899,
+      "loss": 0.3132,
+      "step": 6437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002543409727355011,
+      "loss": 1.7221,
+      "step": 6438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002543273350328482,
+      "loss": 0.2696,
+      "step": 6439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002543136956595496,
+      "loss": 0.2934,
+      "step": 6440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002543000546158237,
+      "loss": 1.4502,
+      "step": 6441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025428641190188895,
+      "loss": 0.0109,
+      "step": 6442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025427276751796384,
+      "loss": 0.4706,
+      "step": 6443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002542591214642668,
+      "loss": 0.5243,
+      "step": 6444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025424547374101647,
+      "loss": 0.627,
+      "step": 6445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025423182434843137,
+      "loss": 0.283,
+      "step": 6446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002542181732867299,
+      "loss": 0.8161,
+      "step": 6447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025420452055613087,
+      "loss": 2.8732,
+      "step": 6448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025419086615685284,
+      "loss": 0.0037,
+      "step": 6449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002541772100891144,
+      "loss": 0.695,
+      "step": 6450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002541635523531343,
+      "loss": 1.1693,
+      "step": 6451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002541498929491313,
+      "loss": 0.375,
+      "step": 6452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025413623187732406,
+      "loss": 0.4115,
+      "step": 6453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002541225691379313,
+      "loss": 0.2386,
+      "step": 6454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025410890473117194,
+      "loss": 0.9096,
+      "step": 6455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025409523865726466,
+      "loss": 0.0281,
+      "step": 6456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025408157091642836,
+      "loss": 3.0434,
+      "step": 6457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002540679015088819,
+      "loss": 0.238,
+      "step": 6458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002540542304348442,
+      "loss": 0.0133,
+      "step": 6459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025404055769453414,
+      "loss": 0.3087,
+      "step": 6460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002540268832881707,
+      "loss": 1.0115,
+      "step": 6461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002540132072159728,
+      "loss": 0.3341,
+      "step": 6462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025399952947815965,
+      "loss": 0.6891,
+      "step": 6463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025398585007494997,
+      "loss": 0.3882,
+      "step": 6464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000253972169006563,
+      "loss": 0.3798,
+      "step": 6465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002539584862732178,
+      "loss": 0.2271,
+      "step": 6466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025394480187513344,
+      "loss": 3.7645,
+      "step": 6467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002539311158125291,
+      "loss": 0.3073,
+      "step": 6468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002539174280856239,
+      "loss": 3.9351,
+      "step": 6469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000253903738694637,
+      "loss": 0.2527,
+      "step": 6470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002538900476397878,
+      "loss": 1.4803,
+      "step": 6471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002538763549212953,
+      "loss": 0.9539,
+      "step": 6472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002538626605393789,
+      "loss": 0.2982,
+      "step": 6473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025384896449425783,
+      "loss": 0.6885,
+      "step": 6474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002538352667861515,
+      "loss": 0.0028,
+      "step": 6475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025382156741527914,
+      "loss": 0.0849,
+      "step": 6476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025380786638186023,
+      "loss": 0.5625,
+      "step": 6477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025379416368611415,
+      "loss": 0.0113,
+      "step": 6478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002537804593282603,
+      "loss": 0.4616,
+      "step": 6479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025376675330851815,
+      "loss": 0.0169,
+      "step": 6480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002537530456271072,
+      "loss": 1.2512,
+      "step": 6481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002537393362842469,
+      "loss": 0.8229,
+      "step": 6482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002537256252801568,
+      "loss": 0.4949,
+      "step": 6483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025371191261505655,
+      "loss": 0.232,
+      "step": 6484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025369819828916564,
+      "loss": 0.6766,
+      "step": 6485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002536844823027037,
+      "loss": 0.304,
+      "step": 6486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002536707646558904,
+      "loss": 0.3208,
+      "step": 6487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002536570453489454,
+      "loss": 0.4091,
+      "step": 6488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025364332438208837,
+      "loss": 0.5747,
+      "step": 6489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000253629601755539,
+      "loss": 0.6064,
+      "step": 6490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002536158774695172,
+      "loss": 2.7385,
+      "step": 6491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025360215152424255,
+      "loss": 3.314,
+      "step": 6492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002535884239199349,
+      "loss": 1.0819,
+      "step": 6493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002535746946568142,
+      "loss": 0.2884,
+      "step": 6494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002535609637351001,
+      "loss": 0.597,
+      "step": 6495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025354723115501263,
+      "loss": 2.306,
+      "step": 6496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002535334969167717,
+      "loss": 0.6371,
+      "step": 6497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002535197610205972,
+      "loss": 0.4124,
+      "step": 6498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025350602346670896,
+      "loss": 0.0215,
+      "step": 6499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025349228425532725,
+      "loss": 0.9634,
+      "step": 6500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025347854338667183,
+      "loss": 0.2838,
+      "step": 6501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002534648008609629,
+      "loss": 0.7892,
+      "step": 6502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002534510566784204,
+      "loss": 0.4297,
+      "step": 6503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025343731083926456,
+      "loss": 0.7924,
+      "step": 6504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002534235633437154,
+      "loss": 0.8108,
+      "step": 6505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002534098141919931,
+      "loss": 0.5259,
+      "step": 6506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025339606338431773,
+      "loss": 0.9629,
+      "step": 6507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002533823109209097,
+      "loss": 0.2242,
+      "step": 6508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025336855680198903,
+      "loss": 0.0175,
+      "step": 6509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025335480102777613,
+      "loss": 0.8807,
+      "step": 6510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025334104359849116,
+      "loss": 0.0265,
+      "step": 6511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025332728451435443,
+      "loss": 1.8916,
+      "step": 6512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002533135237755864,
+      "loss": 0.7596,
+      "step": 6513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025329976138240733,
+      "loss": 0.8454,
+      "step": 6514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025328599733503754,
+      "loss": 0.0204,
+      "step": 6515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025327223163369754,
+      "loss": 0.5592,
+      "step": 6516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002532584642786078,
+      "loss": 0.5859,
+      "step": 6517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002532446952699887,
+      "loss": 0.0283,
+      "step": 6518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002532309246080607,
+      "loss": 0.5509,
+      "step": 6519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002532171522930444,
+      "loss": 0.0968,
+      "step": 6520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002532033783251603,
+      "loss": 0.0236,
+      "step": 6521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000253189602704629,
+      "loss": 0.0405,
+      "step": 6522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002531758254316711,
+      "loss": 1.6274,
+      "step": 6523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025316204650650717,
+      "loss": 0.6326,
+      "step": 6524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002531482659293579,
+      "loss": 0.2168,
+      "step": 6525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002531344837004439,
+      "loss": 0.3704,
+      "step": 6526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025312069981998605,
+      "loss": 0.2643,
+      "step": 6527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002531069142882049,
+      "loss": 0.3465,
+      "step": 6528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002530931271053212,
+      "loss": 0.4584,
+      "step": 6529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025307933827155586,
+      "loss": 3.5586,
+      "step": 6530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002530655477871296,
+      "loss": 0.2096,
+      "step": 6531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025305175565226327,
+      "loss": 3.2761,
+      "step": 6532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025303796186717766,
+      "loss": 0.6692,
+      "step": 6533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002530241664320938,
+      "loss": 0.0081,
+      "step": 6534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025301036934723256,
+      "loss": 0.3289,
+      "step": 6535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025299657061281477,
+      "loss": 0.6849,
+      "step": 6536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002529827702290616,
+      "loss": 0.6338,
+      "step": 6537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025296896819619385,
+      "loss": 1.0297,
+      "step": 6538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002529551645144326,
+      "loss": 0.4744,
+      "step": 6539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002529413591839989,
+      "loss": 0.3844,
+      "step": 6540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002529275522051138,
+      "loss": 0.2672,
+      "step": 6541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002529137435779985,
+      "loss": 0.5912,
+      "step": 6542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000252899933302874,
+      "loss": 1.0303,
+      "step": 6543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002528861213799615,
+      "loss": 0.2164,
+      "step": 6544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002528723078094822,
+      "loss": 1.7707,
+      "step": 6545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025285849259165725,
+      "loss": 0.8184,
+      "step": 6546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025284467572670794,
+      "loss": 0.4772,
+      "step": 6547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025283085721485547,
+      "loss": 0.6848,
+      "step": 6548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002528170370563212,
+      "loss": 0.8682,
+      "step": 6549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002528032152513263,
+      "loss": 0.2428,
+      "step": 6550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025278939180009227,
+      "loss": 0.4286,
+      "step": 6551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002527755667028403,
+      "loss": 0.5398,
+      "step": 6552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000252761739959792,
+      "loss": 0.1621,
+      "step": 6553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002527479115711686,
+      "loss": 0.1643,
+      "step": 6554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002527340815371916,
+      "loss": 0.5768,
+      "step": 6555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025272024985808247,
+      "loss": 2.7004,
+      "step": 6556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002527064165340627,
+      "loss": 0.3112,
+      "step": 6557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002526925815653538,
+      "loss": 0.6827,
+      "step": 6558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025267874495217736,
+      "loss": 0.4907,
+      "step": 6559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025266490669475486,
+      "loss": 0.5592,
+      "step": 6560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025265106679330803,
+      "loss": 0.3761,
+      "step": 6561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002526372252480584,
+      "loss": 0.4632,
+      "step": 6562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002526233820592276,
+      "loss": 0.3664,
+      "step": 6563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025260953722703737,
+      "loss": 0.1897,
+      "step": 6564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002525956907517094,
+      "loss": 1.1718,
+      "step": 6565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025258184263346543,
+      "loss": 0.015,
+      "step": 6566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002525679928725272,
+      "loss": 0.2266,
+      "step": 6567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025255414146911647,
+      "loss": 1.3886,
+      "step": 6568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025254028842345514,
+      "loss": 0.5345,
+      "step": 6569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002525264337357649,
+      "loss": 0.4204,
+      "step": 6570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025251257740626775,
+      "loss": 0.0215,
+      "step": 6571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002524987194351855,
+      "loss": 0.9515,
+      "step": 6572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002524848598227401,
+      "loss": 2.5522,
+      "step": 6573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002524709985691535,
+      "loss": 0.8953,
+      "step": 6574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002524571356746476,
+      "loss": 0.2917,
+      "step": 6575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002524432711394445,
+      "loss": 0.0216,
+      "step": 6576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002524294049637661,
+      "loss": 0.9363,
+      "step": 6577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025241553714783454,
+      "loss": 0.0133,
+      "step": 6578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002524016676918718,
+      "loss": 0.1265,
+      "step": 6579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002523877965961001,
+      "loss": 0.8745,
+      "step": 6580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002523739238607415,
+      "loss": 0.5799,
+      "step": 6581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002523600494860181,
+      "loss": 1.0207,
+      "step": 6582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002523461734721522,
+      "loss": 1.5298,
+      "step": 6583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002523322958193658,
+      "loss": 0.714,
+      "step": 6584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025231841652788143,
+      "loss": 0.6866,
+      "step": 6585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025230453559792106,
+      "loss": 0.4121,
+      "step": 6586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002522906530297071,
+      "loss": 1.095,
+      "step": 6587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002522767688234619,
+      "loss": 0.3505,
+      "step": 6588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002522628829794077,
+      "loss": 2.4573,
+      "step": 6589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002522489954977669,
+      "loss": 0.2687,
+      "step": 6590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025223510637876193,
+      "loss": 0.0053,
+      "step": 6591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002522212156226152,
+      "loss": 1.0917,
+      "step": 6592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025220732322954904,
+      "loss": 0.7717,
+      "step": 6593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000252193429199786,
+      "loss": 4.0628,
+      "step": 6594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025217953353354856,
+      "loss": 1.2158,
+      "step": 6595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025216563623105926,
+      "loss": 0.4095,
+      "step": 6596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002521517372925406,
+      "loss": 1.1725,
+      "step": 6597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002521378367182152,
+      "loss": 0.6407,
+      "step": 6598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002521239345083057,
+      "loss": 0.6649,
+      "step": 6599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025211003066303455,
+      "loss": 1.5421,
+      "step": 6600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002520961251826245,
+      "loss": 0.3022,
+      "step": 6601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002520822180672983,
+      "loss": 1.0069,
+      "step": 6602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002520683093172786,
+      "loss": 0.6184,
+      "step": 6603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025205439893278804,
+      "loss": 0.3694,
+      "step": 6604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002520404869140494,
+      "loss": 0.4024,
+      "step": 6605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025202657326128564,
+      "loss": 0.0209,
+      "step": 6606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025201265797471933,
+      "loss": 0.9746,
+      "step": 6607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002519987410545734,
+      "loss": 0.7044,
+      "step": 6608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025198482250107074,
+      "loss": 1.5031,
+      "step": 6609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025197090231443423,
+      "loss": 0.3292,
+      "step": 6610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025195698049488674,
+      "loss": 0.364,
+      "step": 6611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002519430570426512,
+      "loss": 0.2532,
+      "step": 6612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002519291319579506,
+      "loss": 0.7526,
+      "step": 6613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025191520524100796,
+      "loss": 0.1781,
+      "step": 6614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025190127689204625,
+      "loss": 0.1674,
+      "step": 6615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002518873469112885,
+      "loss": 0.2938,
+      "step": 6616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025187341529895777,
+      "loss": 0.131,
+      "step": 6617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025185948205527726,
+      "loss": 0.8844,
+      "step": 6618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025184554718047,
+      "loss": 0.3935,
+      "step": 6619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025183161067475913,
+      "loss": 0.6024,
+      "step": 6620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002518176725383678,
+      "loss": 0.0392,
+      "step": 6621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002518037327715193,
+      "loss": 0.3518,
+      "step": 6622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025178979137443683,
+      "loss": 0.9208,
+      "step": 6623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002517758483473436,
+      "loss": 0.5875,
+      "step": 6624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002517619036904629,
+      "loss": 0.3099,
+      "step": 6625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000251747957404018,
+      "loss": 0.4198,
+      "step": 6626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025173400948823234,
+      "loss": 0.1125,
+      "step": 6627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002517200599433291,
+      "loss": 0.6485,
+      "step": 6628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025170610876953177,
+      "loss": 0.741,
+      "step": 6629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025169215596706384,
+      "loss": 0.1609,
+      "step": 6630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002516782015361486,
+      "loss": 2.4036,
+      "step": 6631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002516642454770095,
+      "loss": 1.8219,
+      "step": 6632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002516502877898702,
+      "loss": 0.8431,
+      "step": 6633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025163632847495407,
+      "loss": 0.8826,
+      "step": 6634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002516223675324846,
+      "loss": 0.8546,
+      "step": 6635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025160840496268546,
+      "loss": 0.0135,
+      "step": 6636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025159444076578027,
+      "loss": 0.7565,
+      "step": 6637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002515804749419925,
+      "loss": 1.1947,
+      "step": 6638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025156650749154585,
+      "loss": 0.0202,
+      "step": 6639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025155253841466414,
+      "loss": 0.4071,
+      "step": 6640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002515385677115709,
+      "loss": 0.1676,
+      "step": 6641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025152459538248977,
+      "loss": 0.177,
+      "step": 6642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025151062142764473,
+      "loss": 0.2526,
+      "step": 6643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002514966458472594,
+      "loss": 0.1236,
+      "step": 6644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025148266864155757,
+      "loss": 0.7906,
+      "step": 6645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025146868981076317,
+      "loss": 0.2465,
+      "step": 6646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025145470935509996,
+      "loss": 0.6085,
+      "step": 6647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025144072727479184,
+      "loss": 0.626,
+      "step": 6648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025142674357006274,
+      "loss": 0.1788,
+      "step": 6649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025141275824113656,
+      "loss": 0.3715,
+      "step": 6650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002513987712882372,
+      "loss": 1.6299,
+      "step": 6651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002513847827115888,
+      "loss": 0.8116,
+      "step": 6652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025137079251141516,
+      "loss": 0.557,
+      "step": 6653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002513568006879404,
+      "loss": 2.252,
+      "step": 6654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002513428072413887,
+      "loss": 0.4143,
+      "step": 6655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025132881217198395,
+      "loss": 0.3199,
+      "step": 6656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002513148154799504,
+      "loss": 0.1545,
+      "step": 6657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025130081716551205,
+      "loss": 0.7193,
+      "step": 6658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002512868172288932,
+      "loss": 0.5294,
+      "step": 6659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000251272815670318,
+      "loss": 0.1543,
+      "step": 6660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002512588124900107,
+      "loss": 0.941,
+      "step": 6661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002512448076881954,
+      "loss": 0.0323,
+      "step": 6662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025123080126509645,
+      "loss": 0.5004,
+      "step": 6663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002512167932209381,
+      "loss": 0.221,
+      "step": 6664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002512027835559448,
+      "loss": 0.3307,
+      "step": 6665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002511887722703408,
+      "loss": 0.2637,
+      "step": 6666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025117475936435043,
+      "loss": 1.7852,
+      "step": 6667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025116074483819817,
+      "loss": 0.2167,
+      "step": 6668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002511467286921084,
+      "loss": 0.5465,
+      "step": 6669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025113271092630556,
+      "loss": 2.2735,
+      "step": 6670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002511186915410141,
+      "loss": 0.395,
+      "step": 6671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002511046705364586,
+      "loss": 0.2875,
+      "step": 6672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025109064791286353,
+      "loss": 0.4564,
+      "step": 6673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002510766236704534,
+      "loss": 0.0173,
+      "step": 6674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002510625978094529,
+      "loss": 0.0942,
+      "step": 6675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025104857033008657,
+      "loss": 0.511,
+      "step": 6676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000251034541232579,
+      "loss": 0.2016,
+      "step": 6677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002510205105171549,
+      "loss": 0.2132,
+      "step": 6678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025100647818403893,
+      "loss": 1.1191,
+      "step": 6679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002509924442334558,
+      "loss": 0.4368,
+      "step": 6680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025097840866563026,
+      "loss": 0.6582,
+      "step": 6681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025096437148078705,
+      "loss": 0.7959,
+      "step": 6682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025095033267915094,
+      "loss": 0.1699,
+      "step": 6683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002509362922609467,
+      "loss": 0.6207,
+      "step": 6684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025092225022639933,
+      "loss": 0.244,
+      "step": 6685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025090820657573356,
+      "loss": 0.1402,
+      "step": 6686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002508941613091743,
+      "loss": 0.0556,
+      "step": 6687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002508801144269464,
+      "loss": 0.4042,
+      "step": 6688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000250866065929275,
+      "loss": 0.4801,
+      "step": 6689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002508520158163848,
+      "loss": 0.413,
+      "step": 6690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000250837964088501,
+      "loss": 0.2833,
+      "step": 6691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025082391074584854,
+      "loss": 1.0782,
+      "step": 6692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025080985578865246,
+      "loss": 0.624,
+      "step": 6693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002507957992171378,
+      "loss": 0.6182,
+      "step": 6694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002507817410315297,
+      "loss": 2.0314,
+      "step": 6695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002507676812320533,
+      "loss": 1.0968,
+      "step": 6696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002507536198189337,
+      "loss": 0.2978,
+      "step": 6697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002507395567923961,
+      "loss": 0.4151,
+      "step": 6698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025072549215266566,
+      "loss": 0.362,
+      "step": 6699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025071142589996767,
+      "loss": 0.5702,
+      "step": 6700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002506973580345273,
+      "loss": 0.6079,
+      "step": 6701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002506832885565699,
+      "loss": 0.5605,
+      "step": 6702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002506692174663207,
+      "loss": 0.4069,
+      "step": 6703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025065514476400506,
+      "loss": 0.547,
+      "step": 6704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002506410704498484,
+      "loss": 0.8013,
+      "step": 6705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000250626994524076,
+      "loss": 1.0477,
+      "step": 6706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025061291698691326,
+      "loss": 0.0564,
+      "step": 6707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002505988378385857,
+      "loss": 1.1389,
+      "step": 6708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025058475707931876,
+      "loss": 1.1692,
+      "step": 6709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002505706747093378,
+      "loss": 0.4691,
+      "step": 6710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002505565907288685,
+      "loss": 0.2545,
+      "step": 6711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002505425051381363,
+      "loss": 0.6327,
+      "step": 6712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002505284179373668,
+      "loss": 1.1813,
+      "step": 6713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025051432912678547,
+      "loss": 0.2034,
+      "step": 6714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025050023870661813,
+      "loss": 0.0803,
+      "step": 6715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002504861466770902,
+      "loss": 0.8307,
+      "step": 6716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002504720530384275,
+      "loss": 0.2514,
+      "step": 6717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025045795779085557,
+      "loss": 0.0027,
+      "step": 6718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002504438609346003,
+      "loss": 0.3593,
+      "step": 6719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025042976246988734,
+      "loss": 0.9719,
+      "step": 6720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025041566239694245,
+      "loss": 0.3573,
+      "step": 6721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025040156071599136,
+      "loss": 0.7524,
+      "step": 6722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025038745742726005,
+      "loss": 0.5005,
+      "step": 6723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002503733525309742,
+      "loss": 0.7097,
+      "step": 6724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002503592460273598,
+      "loss": 0.2721,
+      "step": 6725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025034513791664264,
+      "loss": 0.2703,
+      "step": 6726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025033102819904874,
+      "loss": 0.3648,
+      "step": 6727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000250316916874804,
+      "loss": 0.4894,
+      "step": 6728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025030280394413434,
+      "loss": 0.0615,
+      "step": 6729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025028868940726585,
+      "loss": 0.4345,
+      "step": 6730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025027457326442447,
+      "loss": 0.5727,
+      "step": 6731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025026045551583635,
+      "loss": 0.1357,
+      "step": 6732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025024633616172744,
+      "loss": 0.0555,
+      "step": 6733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000250232215202324,
+      "loss": 0.3952,
+      "step": 6734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025021809263785195,
+      "loss": 0.274,
+      "step": 6735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025020396846853763,
+      "loss": 0.8609,
+      "step": 6736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025018984269460713,
+      "loss": 0.5045,
+      "step": 6737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002501757153162866,
+      "loss": 0.3896,
+      "step": 6738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002501615863338024,
+      "loss": 1.1597,
+      "step": 6739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025014745574738074,
+      "loss": 1.589,
+      "step": 6740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002501333235572478,
+      "loss": 0.7823,
+      "step": 6741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025011918976363003,
+      "loss": 0.635,
+      "step": 6742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025010505436675364,
+      "loss": 0.0193,
+      "step": 6743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025009091736684504,
+      "loss": 0.0234,
+      "step": 6744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025007677876413067,
+      "loss": 0.8658,
+      "step": 6745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025006263855883684,
+      "loss": 0.7374,
+      "step": 6746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025004849675119004,
+      "loss": 0.3345,
+      "step": 6747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002500343533414167,
+      "loss": 0.0177,
+      "step": 6748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002500202083297433,
+      "loss": 0.8892,
+      "step": 6749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025000606171639647,
+      "loss": 0.8393,
+      "step": 6750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024999191350160257,
+      "loss": 0.6319,
+      "step": 6751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024997776368558826,
+      "loss": 0.7129,
+      "step": 6752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002499636122685801,
+      "loss": 5.2671,
+      "step": 6753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002499494592508047,
+      "loss": 0.3776,
+      "step": 6754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002499353046324888,
+      "loss": 0.5029,
+      "step": 6755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024992114841385885,
+      "loss": 0.152,
+      "step": 6756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002499069905951417,
+      "loss": 0.4112,
+      "step": 6757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002498928311765641,
+      "loss": 0.5773,
+      "step": 6758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002498786701583527,
+      "loss": 2.8068,
+      "step": 6759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024986450754073424,
+      "loss": 0.7678,
+      "step": 6760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024985034332393566,
+      "loss": 1.0945,
+      "step": 6761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024983617750818363,
+      "loss": 0.6977,
+      "step": 6762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002498220100937051,
+      "loss": 0.5358,
+      "step": 6763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002498078410807268,
+      "loss": 4.2771,
+      "step": 6764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002497936704694758,
+      "loss": 0.5908,
+      "step": 6765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002497794982601789,
+      "loss": 0.0714,
+      "step": 6766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024976532445306304,
+      "loss": 0.3699,
+      "step": 6767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002497511490483553,
+      "loss": 0.2326,
+      "step": 6768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002497369720462826,
+      "loss": 0.5193,
+      "step": 6769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024972279344707195,
+      "loss": 0.3121,
+      "step": 6770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024970861325095047,
+      "loss": 0.362,
+      "step": 6771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024969443145814516,
+      "loss": 0.2766,
+      "step": 6772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002496802480688832,
+      "loss": 0.6257,
+      "step": 6773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002496660630833916,
+      "loss": 1.8356,
+      "step": 6774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002496518765018976,
+      "loss": 1.2719,
+      "step": 6775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002496376883246284,
+      "loss": 0.0278,
+      "step": 6776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024962349855181114,
+      "loss": 0.3998,
+      "step": 6777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000249609307183673,
+      "loss": 0.2624,
+      "step": 6778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024959511422044134,
+      "loss": 0.7651,
+      "step": 6779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024958091966234336,
+      "loss": 0.5444,
+      "step": 6780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024956672350960645,
+      "loss": 0.8419,
+      "step": 6781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002495525257624579,
+      "loss": 0.4826,
+      "step": 6782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024953832642112504,
+      "loss": 0.7057,
+      "step": 6783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002495241254858352,
+      "loss": 0.216,
+      "step": 6784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000249509922956816,
+      "loss": 0.4803,
+      "step": 6785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002494957188342946,
+      "loss": 0.5811,
+      "step": 6786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024948151311849866,
+      "loss": 0.5971,
+      "step": 6787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002494673058096555,
+      "loss": 0.3309,
+      "step": 6788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002494530969079928,
+      "loss": 0.0935,
+      "step": 6789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024943888641373803,
+      "loss": 2.3265,
+      "step": 6790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024942467432711866,
+      "loss": 0.22,
+      "step": 6791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024941046064836243,
+      "loss": 0.4117,
+      "step": 6792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024939624537769677,
+      "loss": 0.2593,
+      "step": 6793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024938202851534946,
+      "loss": 0.3858,
+      "step": 6794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002493678100615481,
+      "loss": 1.0316,
+      "step": 6795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002493535900165204,
+      "loss": 0.3416,
+      "step": 6796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002493393683804941,
+      "loss": 0.3928,
+      "step": 6797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024932514515369684,
+      "loss": 0.4255,
+      "step": 6798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002493109203363565,
+      "loss": 0.3017,
+      "step": 6799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002492966939287008,
+      "loss": 0.2958,
+      "step": 6800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024928246593095753,
+      "loss": 0.4665,
+      "step": 6801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002492682363433546,
+      "loss": 0.5831,
+      "step": 6802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002492540051661199,
+      "loss": 0.2361,
+      "step": 6803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002492397723994812,
+      "loss": 0.2647,
+      "step": 6804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002492255380436665,
+      "loss": 1.2032,
+      "step": 6805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024921130209890367,
+      "loss": 1.4281,
+      "step": 6806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002491970645654208,
+      "loss": 0.4353,
+      "step": 6807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002491828254434458,
+      "loss": 1.2652,
+      "step": 6808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002491685847332067,
+      "loss": 0.1805,
+      "step": 6809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024915434243493156,
+      "loss": 0.4518,
+      "step": 6810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002491400985488484,
+      "loss": 1.7065,
+      "step": 6811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024912585307518536,
+      "loss": 0.0336,
+      "step": 6812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002491116060141706,
+      "loss": 0.4744,
+      "step": 6813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002490973573660322,
+      "loss": 0.8213,
+      "step": 6814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024908310713099834,
+      "loss": 0.4457,
+      "step": 6815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024906885530929715,
+      "loss": 0.8406,
+      "step": 6816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000249054601901157,
+      "loss": 0.255,
+      "step": 6817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002490403469068061,
+      "loss": 0.3266,
+      "step": 6818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002490260903264726,
+      "loss": 4.9414,
+      "step": 6819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024901183216038486,
+      "loss": 0.2265,
+      "step": 6820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002489975724087713,
+      "loss": 0.2601,
+      "step": 6821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024898331107186015,
+      "loss": 0.4454,
+      "step": 6822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002489690481498799,
+      "loss": 0.2998,
+      "step": 6823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002489547836430587,
+      "loss": 0.893,
+      "step": 6824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002489405175516253,
+      "loss": 3.5445,
+      "step": 6825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000248926249875808,
+      "loss": 0.2163,
+      "step": 6826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002489119806158352,
+      "loss": 0.1729,
+      "step": 6827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024889770977193555,
+      "loss": 0.9933,
+      "step": 6828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002488834373443375,
+      "loss": 0.7112,
+      "step": 6829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002488691633332696,
+      "loss": 0.1916,
+      "step": 6830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024885488773896037,
+      "loss": 0.1182,
+      "step": 6831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002488406105616385,
+      "loss": 0.0997,
+      "step": 6832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002488263318015326,
+      "loss": 0.0356,
+      "step": 6833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024881205145887135,
+      "loss": 0.4201,
+      "step": 6834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002487977695338833,
+      "loss": 0.4307,
+      "step": 6835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002487834860267973,
+      "loss": 1.5319,
+      "step": 6836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000248769200937842,
+      "loss": 0.4198,
+      "step": 6837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002487549142672463,
+      "loss": 0.4538,
+      "step": 6838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024874062601523873,
+      "loss": 0.7649,
+      "step": 6839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002487263361820483,
+      "loss": 0.4266,
+      "step": 6840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002487120447679037,
+      "loss": 0.7988,
+      "step": 6841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024869775177303383,
+      "loss": 0.4956,
+      "step": 6842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024868345719766767,
+      "loss": 0.9541,
+      "step": 6843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000248669161042034,
+      "loss": 0.4389,
+      "step": 6844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002486548633063618,
+      "loss": 0.7926,
+      "step": 6845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024864056399088,
+      "loss": 1.7551,
+      "step": 6846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024862626309581765,
+      "loss": 0.2904,
+      "step": 6847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002486119606214037,
+      "loss": 0.3466,
+      "step": 6848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002485976565678672,
+      "loss": 0.0122,
+      "step": 6849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024858335093543715,
+      "loss": 0.2016,
+      "step": 6850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002485690437243428,
+      "loss": 0.2711,
+      "step": 6851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00024855473493481305,
+      "loss": 0.7385,
+      "step": 6852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002485404245670772,
+      "loss": 0.4329,
+      "step": 6853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002485261126213643,
+      "loss": 1.1471,
+      "step": 6854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024851179909790355,
+      "loss": 0.4013,
+      "step": 6855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002484974839969242,
+      "loss": 0.0691,
+      "step": 6856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002484831673186555,
+      "loss": 0.3201,
+      "step": 6857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002484688490633266,
+      "loss": 0.3637,
+      "step": 6858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002484545292311669,
+      "loss": 0.7673,
+      "step": 6859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024844020782240577,
+      "loss": 0.4772,
+      "step": 6860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024842588483727233,
+      "loss": 0.5665,
+      "step": 6861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024841156027599615,
+      "loss": 0.4517,
+      "step": 6862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024839723413880646,
+      "loss": 0.8146,
+      "step": 6863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024838290642593275,
+      "loss": 0.0158,
+      "step": 6864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002483685771376044,
+      "loss": 0.3642,
+      "step": 6865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000248354246274051,
+      "loss": 0.4462,
+      "step": 6866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024833991383550193,
+      "loss": 0.7851,
+      "step": 6867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002483255798221867,
+      "loss": 0.1211,
+      "step": 6868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002483112442343349,
+      "loss": 0.6889,
+      "step": 6869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002482969070721761,
+      "loss": 0.0953,
+      "step": 6870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024828256833593984,
+      "loss": 0.1362,
+      "step": 6871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024826822802585574,
+      "loss": 0.6201,
+      "step": 6872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002482538861421534,
+      "loss": 0.8123,
+      "step": 6873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002482395426850626,
+      "loss": 0.9424,
+      "step": 6874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002482251976548129,
+      "loss": 0.0781,
+      "step": 6875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002482108510516341,
+      "loss": 0.0138,
+      "step": 6876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024819650287575596,
+      "loss": 0.1326,
+      "step": 6877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024818215312740816,
+      "loss": 0.5888,
+      "step": 6878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002481678018068205,
+      "loss": 0.0006,
+      "step": 6879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002481534489142229,
+      "loss": 0.1525,
+      "step": 6880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024813909444984506,
+      "loss": 0.1162,
+      "step": 6881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024812473841391694,
+      "loss": 0.6757,
+      "step": 6882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002481103808066684,
+      "loss": 0.4326,
+      "step": 6883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002480960216283293,
+      "loss": 0.5451,
+      "step": 6884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024808166087912967,
+      "loss": 1.6702,
+      "step": 6885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024806729855929944,
+      "loss": 1.1167,
+      "step": 6886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002480529346690686,
+      "loss": 1.0437,
+      "step": 6887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002480385692086672,
+      "loss": 0.188,
+      "step": 6888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002480242021783252,
+      "loss": 2.2782,
+      "step": 6889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024800983357827275,
+      "loss": 0.2826,
+      "step": 6890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002479954634087399,
+      "loss": 0.2869,
+      "step": 6891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024798109166995676,
+      "loss": 0.8113,
+      "step": 6892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002479667183621535,
+      "loss": 0.2399,
+      "step": 6893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024795234348556027,
+      "loss": 0.3654,
+      "step": 6894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002479379670404073,
+      "loss": 0.5109,
+      "step": 6895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002479235890269247,
+      "loss": 0.005,
+      "step": 6896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024790920944534284,
+      "loss": 0.3798,
+      "step": 6897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002478948282958919,
+      "loss": 1.0644,
+      "step": 6898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002478804455788022,
+      "loss": 0.8043,
+      "step": 6899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002478660612943041,
+      "loss": 0.7698,
+      "step": 6900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002478516754426279,
+      "loss": 1.2444,
+      "step": 6901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024783728802400394,
+      "loss": 0.8269,
+      "step": 6902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024782289903866266,
+      "loss": 0.9544,
+      "step": 6903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002478085084868345,
+      "loss": 0.2887,
+      "step": 6904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002477941163687498,
+      "loss": 0.0234,
+      "step": 6905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002477797226846392,
+      "loss": 0.4149,
+      "step": 6906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024776532743473297,
+      "loss": 0.4223,
+      "step": 6907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002477509306192618,
+      "loss": 0.7223,
+      "step": 6908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024773653223845614,
+      "loss": 0.4125,
+      "step": 6909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002477221322925466,
+      "loss": 0.3549,
+      "step": 6910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024770773078176383,
+      "loss": 0.749,
+      "step": 6911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024769332770633836,
+      "loss": 0.6268,
+      "step": 6912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002476789230665009,
+      "loss": 0.4868,
+      "step": 6913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024766451686248205,
+      "loss": 0.0906,
+      "step": 6914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002476501090945125,
+      "loss": 0.4793,
+      "step": 6915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024763569976282305,
+      "loss": 0.7382,
+      "step": 6916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002476212888676444,
+      "loss": 0.2756,
+      "step": 6917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002476068764092073,
+      "loss": 0.5813,
+      "step": 6918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002475924623877426,
+      "loss": 0.4752,
+      "step": 6919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024757804680348105,
+      "loss": 0.0037,
+      "step": 6920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024756362965665355,
+      "loss": 1.5612,
+      "step": 6921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002475492109474909,
+      "loss": 0.5254,
+      "step": 6922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024753479067622403,
+      "loss": 2.0874,
+      "step": 6923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002475203688430839,
+      "loss": 0.7689,
+      "step": 6924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002475059454483014,
+      "loss": 0.2021,
+      "step": 6925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002474915204921075,
+      "loss": 0.3816,
+      "step": 6926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002474770939747333,
+      "loss": 0.0006,
+      "step": 6927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002474626658964097,
+      "loss": 0.3795,
+      "step": 6928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024744823625736767,
+      "loss": 0.2723,
+      "step": 6929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002474338050578385,
+      "loss": 0.805,
+      "step": 6930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024741937229805314,
+      "loss": 0.4731,
+      "step": 6931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024740493797824277,
+      "loss": 0.3955,
+      "step": 6932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024739050209863844,
+      "loss": 1.1523,
+      "step": 6933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002473760646594714,
+      "loss": 0.7466,
+      "step": 6934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002473616256609728,
+      "loss": 0.6604,
+      "step": 6935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000247347185103374,
+      "loss": 0.3311,
+      "step": 6936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000247332742986906,
+      "loss": 0.5865,
+      "step": 6937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024731829931180023,
+      "loss": 0.0853,
+      "step": 6938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000247303854078288,
+      "loss": 0.6242,
+      "step": 6939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024728940728660053,
+      "loss": 0.4506,
+      "step": 6940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024727495893696924,
+      "loss": 2.3129,
+      "step": 6941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002472605090296255,
+      "loss": 0.4236,
+      "step": 6942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024724605756480065,
+      "loss": 1.2407,
+      "step": 6943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002472316045427261,
+      "loss": 1.0444,
+      "step": 6944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024721714996363335,
+      "loss": 0.2849,
+      "step": 6945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024720269382775385,
+      "loss": 0.4911,
+      "step": 6946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002471882361353191,
+      "loss": 0.0118,
+      "step": 6947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024717377688656056,
+      "loss": 0.3378,
+      "step": 6948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024715931608170985,
+      "loss": 0.4846,
+      "step": 6949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024714485372099855,
+      "loss": 0.0155,
+      "step": 6950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024713038980465814,
+      "loss": 0.0929,
+      "step": 6951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002471159243329203,
+      "loss": 2.6414,
+      "step": 6952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002471014573060168,
+      "loss": 0.0007,
+      "step": 6953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002470869887241791,
+      "loss": 0.31,
+      "step": 6954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000247072518587639,
+      "loss": 0.2711,
+      "step": 6955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024705804689662816,
+      "loss": 0.3718,
+      "step": 6956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002470435736513784,
+      "loss": 0.4503,
+      "step": 6957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002470290988521215,
+      "loss": 0.9581,
+      "step": 6958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002470146224990891,
+      "loss": 0.0014,
+      "step": 6959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024700014459251324,
+      "loss": 0.0195,
+      "step": 6960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024698566513262554,
+      "loss": 0.7444,
+      "step": 6961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000246971184119658,
+      "loss": 0.6958,
+      "step": 6962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024695670155384247,
+      "loss": 0.2458,
+      "step": 6963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002469422174354109,
+      "loss": 0.017,
+      "step": 6964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002469277317645952,
+      "loss": 2.4927,
+      "step": 6965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002469132445416273,
+      "loss": 0.2022,
+      "step": 6966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002468987557667393,
+      "loss": 3.0949,
+      "step": 6967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002468842654401631,
+      "loss": 0.5417,
+      "step": 6968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024686977356213084,
+      "loss": 0.7519,
+      "step": 6969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024685528013287456,
+      "loss": 0.7821,
+      "step": 6970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002468407851526263,
+      "loss": 0.6378,
+      "step": 6971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024682628862161815,
+      "loss": 0.2009,
+      "step": 6972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024681179054008234,
+      "loss": 0.0146,
+      "step": 6973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000246797290908251,
+      "loss": 0.5861,
+      "step": 6974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024678278972635633,
+      "loss": 0.2873,
+      "step": 6975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002467682869946305,
+      "loss": 0.531,
+      "step": 6976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024675378271330585,
+      "loss": 0.661,
+      "step": 6977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002467392768826145,
+      "loss": 0.5754,
+      "step": 6978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024672476950278883,
+      "loss": 0.3853,
+      "step": 6979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024671026057406116,
+      "loss": 0.8004,
+      "step": 6980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024669575009666383,
+      "loss": 0.5344,
+      "step": 6981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024668123807082917,
+      "loss": 0.4411,
+      "step": 6982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024666672449678953,
+      "loss": 0.8123,
+      "step": 6983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002466522093747774,
+      "loss": 0.3163,
+      "step": 6984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024663769270502525,
+      "loss": 0.0006,
+      "step": 6985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002466231744877654,
+      "loss": 0.8237,
+      "step": 6986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002466086547232305,
+      "loss": 0.6726,
+      "step": 6987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002465941334116529,
+      "loss": 2.2121,
+      "step": 6988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024657961055326525,
+      "loss": 0.238,
+      "step": 6989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002465650861483001,
+      "loss": 1.1923,
+      "step": 6990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002465505601969901,
+      "loss": 0.2538,
+      "step": 6991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002465360326995676,
+      "loss": 0.4141,
+      "step": 6992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024652150365626554,
+      "loss": 0.8267,
+      "step": 6993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024650697306731646,
+      "loss": 0.5889,
+      "step": 6994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000246492440932953,
+      "loss": 2.9564,
+      "step": 6995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002464779072534079,
+      "loss": 0.5922,
+      "step": 6996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024646337202891396,
+      "loss": 4.0059,
+      "step": 6997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024644883525970387,
+      "loss": 2.249,
+      "step": 6998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024643429694601044,
+      "loss": 0.3839,
+      "step": 6999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024641975708806645,
+      "loss": 0.4758,
+      "step": 7000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024640521568610477,
+      "loss": 0.0303,
+      "step": 7001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002463906727403582,
+      "loss": 0.4622,
+      "step": 7002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002463761282510597,
+      "loss": 0.8994,
+      "step": 7003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024636158221844214,
+      "loss": 2.0879,
+      "step": 7004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002463470346427385,
+      "loss": 1.2821,
+      "step": 7005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024633248552418166,
+      "loss": 0.7498,
+      "step": 7006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002463179348630047,
+      "loss": 0.0069,
+      "step": 7007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024630338265944045,
+      "loss": 0.7316,
+      "step": 7008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024628882891372216,
+      "loss": 0.2566,
+      "step": 7009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002462742736260827,
+      "loss": 0.6863,
+      "step": 7010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024625971679675533,
+      "loss": 0.4601,
+      "step": 7011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000246245158425973,
+      "loss": 0.4068,
+      "step": 7012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002462305985139689,
+      "loss": 0.0767,
+      "step": 7013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024621603706097625,
+      "loss": 0.9261,
+      "step": 7014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002462014740672281,
+      "loss": 0.6088,
+      "step": 7015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002461869095329577,
+      "loss": 0.3076,
+      "step": 7016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002461723434583984,
+      "loss": 2.3542,
+      "step": 7017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024615777584378326,
+      "loss": 0.8049,
+      "step": 7018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002461432066893457,
+      "loss": 0.9644,
+      "step": 7019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024612863599531894,
+      "loss": 0.265,
+      "step": 7020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024611406376193635,
+      "loss": 0.5692,
+      "step": 7021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002460994899894313,
+      "loss": 0.7085,
+      "step": 7022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002460849146780371,
+      "loss": 0.5139,
+      "step": 7023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002460703378279873,
+      "loss": 0.1767,
+      "step": 7024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002460557594395151,
+      "loss": 0.414,
+      "step": 7025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024604117951285414,
+      "loss": 0.0227,
+      "step": 7026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002460265980482378,
+      "loss": 0.901,
+      "step": 7027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024601201504589963,
+      "loss": 0.0146,
+      "step": 7028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002459974305060731,
+      "loss": 0.3494,
+      "step": 7029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024598284442899185,
+      "loss": 0.2329,
+      "step": 7030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024596825681488934,
+      "loss": 0.6865,
+      "step": 7031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024595366766399926,
+      "loss": 0.2984,
+      "step": 7032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002459390769765552,
+      "loss": 0.4552,
+      "step": 7033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024592448475279076,
+      "loss": 0.8611,
+      "step": 7034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002459098909929397,
+      "loss": 0.4576,
+      "step": 7035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002458952956972357,
+      "loss": 1.9207,
+      "step": 7036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002458806988659124,
+      "loss": 0.1716,
+      "step": 7037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002458661004992036,
+      "loss": 0.5852,
+      "step": 7038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002458515005973431,
+      "loss": 0.6501,
+      "step": 7039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024583689916056467,
+      "loss": 0.3121,
+      "step": 7040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024582229618910214,
+      "loss": 0.4411,
+      "step": 7041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002458076916831893,
+      "loss": 0.1753,
+      "step": 7042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024579308564306014,
+      "loss": 0.4314,
+      "step": 7043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024577847806894846,
+      "loss": 0.5531,
+      "step": 7044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024576386896108817,
+      "loss": 0.2841,
+      "step": 7045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024574925831971324,
+      "loss": 0.7177,
+      "step": 7046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024573464614505765,
+      "loss": 0.1352,
+      "step": 7047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024572003243735537,
+      "loss": 0.0388,
+      "step": 7048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002457054171968405,
+      "loss": 0.6362,
+      "step": 7049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024569080042374694,
+      "loss": 0.3863,
+      "step": 7050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002456761821183088,
+      "loss": 0.4866,
+      "step": 7051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002456615622807602,
+      "loss": 0.4171,
+      "step": 7052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024564694091133525,
+      "loss": 1.0318,
+      "step": 7053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002456323180102681,
+      "loss": 0.5723,
+      "step": 7054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002456176935777929,
+      "loss": 0.2184,
+      "step": 7055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002456030676141439,
+      "loss": 0.4307,
+      "step": 7056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024558844011955515,
+      "loss": 0.5292,
+      "step": 7057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000245573811094261,
+      "loss": 0.2929,
+      "step": 7058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002455591805384958,
+      "loss": 0.9302,
+      "step": 7059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024554454845249365,
+      "loss": 0.3704,
+      "step": 7060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000245529914836489,
+      "loss": 0.8311,
+      "step": 7061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002455152796907161,
+      "loss": 0.4689,
+      "step": 7062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002455006430154093,
+      "loss": 0.0034,
+      "step": 7063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002454860048108031,
+      "loss": 0.839,
+      "step": 7064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024547136507713186,
+      "loss": 0.3456,
+      "step": 7065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024545672381462994,
+      "loss": 1.0154,
+      "step": 7066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024544208102353186,
+      "loss": 0.4294,
+      "step": 7067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002454274367040721,
+      "loss": 0.0014,
+      "step": 7068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002454127908564851,
+      "loss": 0.1399,
+      "step": 7069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024539814348100553,
+      "loss": 2.9671,
+      "step": 7070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002453834945778678,
+      "loss": 0.4568,
+      "step": 7071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024536884414730657,
+      "loss": 0.7156,
+      "step": 7072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002453541921895565,
+      "loss": 0.746,
+      "step": 7073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024533953870485207,
+      "loss": 0.6926,
+      "step": 7074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024532488369342806,
+      "loss": 0.0017,
+      "step": 7075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024531022715551905,
+      "loss": 0.8131,
+      "step": 7076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002452955690913598,
+      "loss": 0.4339,
+      "step": 7077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002452809095011851,
+      "loss": 0.1032,
+      "step": 7078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024526624838522955,
+      "loss": 0.5049,
+      "step": 7079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024525158574372803,
+      "loss": 0.2118,
+      "step": 7080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002452369215769153,
+      "loss": 0.7615,
+      "step": 7081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024522225588502624,
+      "loss": 3.0391,
+      "step": 7082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024520758866829564,
+      "loss": 0.0004,
+      "step": 7083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024519291992695843,
+      "loss": 0.1521,
+      "step": 7084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024517824966124944,
+      "loss": 0.146,
+      "step": 7085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002451635778714036,
+      "loss": 0.5238,
+      "step": 7086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024514890455765593,
+      "loss": 0.7813,
+      "step": 7087
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002451342297202413,
+      "loss": 0.4035,
+      "step": 7088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002451195533593948,
+      "loss": 0.0123,
+      "step": 7089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002451048754753514,
+      "loss": 0.6002,
+      "step": 7090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024509019606834613,
+      "loss": 0.0027,
+      "step": 7091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002450755151386141,
+      "loss": 0.1632,
+      "step": 7092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002450608326863904,
+      "loss": 0.4236,
+      "step": 7093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002450461487119101,
+      "loss": 0.408,
+      "step": 7094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024503146321540835,
+      "loss": 0.4134,
+      "step": 7095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002450167761971204,
+      "loss": 2.1332,
+      "step": 7096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024500208765728136,
+      "loss": 0.0417,
+      "step": 7097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024498739759612644,
+      "loss": 0.394,
+      "step": 7098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002449727060138909,
+      "loss": 0.0156,
+      "step": 7099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024495801291081004,
+      "loss": 0.2415,
+      "step": 7100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002449433182871191,
+      "loss": 1.3225,
+      "step": 7101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024492862214305343,
+      "loss": 0.5346,
+      "step": 7102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00024491392447884827,
+      "loss": 0.6947,
+      "step": 7103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024489922529473916,
+      "loss": 1.3814,
+      "step": 7104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002448845245909613,
+      "loss": 0.2967,
+      "step": 7105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002448698223677502,
+      "loss": 0.1558,
+      "step": 7106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002448551186253413,
+      "loss": 0.0035,
+      "step": 7107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024484041336397,
+      "loss": 0.7107,
+      "step": 7108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002448257065838718,
+      "loss": 0.3494,
+      "step": 7109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002448109982852823,
+      "loss": 0.3274,
+      "step": 7110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002447962884684369,
+      "loss": 0.2904,
+      "step": 7111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002447815771335712,
+      "loss": 0.0231,
+      "step": 7112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024476686428092086,
+      "loss": 0.3028,
+      "step": 7113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002447521499107213,
+      "loss": 0.3031,
+      "step": 7114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024473743402320835,
+      "loss": 0.6175,
+      "step": 7115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024472271661861753,
+      "loss": 0.2198,
+      "step": 7116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024470799769718457,
+      "loss": 0.002,
+      "step": 7117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002446932772591452,
+      "loss": 0.9732,
+      "step": 7118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024467855530473505,
+      "loss": 3.2504,
+      "step": 7119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024466383183419,
+      "loss": 0.2965,
+      "step": 7120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002446491068477457,
+      "loss": 0.3536,
+      "step": 7121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024463438034563804,
+      "loss": 0.3732,
+      "step": 7122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002446196523281028,
+      "loss": 0.4633,
+      "step": 7123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002446049227953758,
+      "loss": 0.0058,
+      "step": 7124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024459019174769296,
+      "loss": 0.6168,
+      "step": 7125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024457545918529013,
+      "loss": 0.8526,
+      "step": 7126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024456072510840334,
+      "loss": 2.2129,
+      "step": 7127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002445459895172684,
+      "loss": 0.5467,
+      "step": 7128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002445312524121213,
+      "loss": 0.7988,
+      "step": 7129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002445165137931981,
+      "loss": 0.2209,
+      "step": 7130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024450177366073476,
+      "loss": 0.1546,
+      "step": 7131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002444870320149674,
+      "loss": 0.4724,
+      "step": 7132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024447228885613197,
+      "loss": 2.3887,
+      "step": 7133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024445754418446464,
+      "loss": 0.0134,
+      "step": 7134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002444427980002015,
+      "loss": 0.1889,
+      "step": 7135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024442805030357867,
+      "loss": 0.9952,
+      "step": 7136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002444133010948323,
+      "loss": 0.4584,
+      "step": 7137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002443985503741987,
+      "loss": 0.65,
+      "step": 7138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000244383798141914,
+      "loss": 0.6727,
+      "step": 7139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024436904439821425,
+      "loss": 1.1005,
+      "step": 7140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000244354289143336,
+      "loss": 0.0142,
+      "step": 7141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024433953237751544,
+      "loss": 0.5458,
+      "step": 7142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024432477410098885,
+      "loss": 0.4165,
+      "step": 7143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002443100143139925,
+      "loss": 0.0057,
+      "step": 7144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002442952530167629,
+      "loss": 0.6559,
+      "step": 7145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002442804902095363,
+      "loss": 0.058,
+      "step": 7146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002442657258925491,
+      "loss": 0.5093,
+      "step": 7147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002442509600660378,
+      "loss": 0.0584,
+      "step": 7148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024423619273023883,
+      "loss": 1.1182,
+      "step": 7149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002442214238853887,
+      "loss": 0.3081,
+      "step": 7150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002442066535317238,
+      "loss": 0.5095,
+      "step": 7151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024419188166948074,
+      "loss": 0.0356,
+      "step": 7152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002441771082988961,
+      "loss": 0.4173,
+      "step": 7153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002441623334202064,
+      "loss": 0.8032,
+      "step": 7154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002441475570336482,
+      "loss": 0.7482,
+      "step": 7155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024413277913945823,
+      "loss": 0.0209,
+      "step": 7156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024411799973787307,
+      "loss": 0.1422,
+      "step": 7157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024410321882912937,
+      "loss": 0.0018,
+      "step": 7158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024408843641346385,
+      "loss": 0.907,
+      "step": 7159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024407365249111326,
+      "loss": 0.3552,
+      "step": 7160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024405886706231427,
+      "loss": 0.0994,
+      "step": 7161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024404408012730376,
+      "loss": 3.1876,
+      "step": 7162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024402929168631837,
+      "loss": 0.3056,
+      "step": 7163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000244014501739595,
+      "loss": 0.4975,
+      "step": 7164
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024399971028737054,
+      "loss": 1.547,
+      "step": 7165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024398491732988175,
+      "loss": 0.3512,
+      "step": 7166
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002439701228673655,
+      "loss": 2.506,
+      "step": 7167
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024395532690005885,
+      "loss": 0.904,
+      "step": 7168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024394052942819863,
+      "loss": 0.524,
+      "step": 7169
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002439257304520218,
+      "loss": 0.4391,
+      "step": 7170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024391092997176535,
+      "loss": 0.4833,
+      "step": 7171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024389612798766635,
+      "loss": 0.6354,
+      "step": 7172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002438813244999617,
+      "loss": 1.0104,
+      "step": 7173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024386651950888857,
+      "loss": 0.6287,
+      "step": 7174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000243851713014684,
+      "loss": 0.4497,
+      "step": 7175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024383690501758514,
+      "loss": 1.8516,
+      "step": 7176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024382209551782904,
+      "loss": 0.8187,
+      "step": 7177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024380728451565286,
+      "loss": 3.0606,
+      "step": 7178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024379247201129385,
+      "loss": 0.0978,
+      "step": 7179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024377765800498913,
+      "loss": 0.0093,
+      "step": 7180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024376284249697598,
+      "loss": 0.2057,
+      "step": 7181
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002437480254874916,
+      "loss": 0.0164,
+      "step": 7182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024373320697677333,
+      "loss": 0.2939,
+      "step": 7183
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024371838696505839,
+      "loss": 0.6548,
+      "step": 7184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024370356545258415,
+      "loss": 0.9416,
+      "step": 7185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002436887424395879,
+      "loss": 0.8951,
+      "step": 7186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024367391792630712,
+      "loss": 0.651,
+      "step": 7187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024365909191297906,
+      "loss": 0.4152,
+      "step": 7188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024364426439984124,
+      "loss": 0.4494,
+      "step": 7189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024362943538713106,
+      "loss": 0.2932,
+      "step": 7190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024361460487508604,
+      "loss": 0.4334,
+      "step": 7191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024359977286394358,
+      "loss": 0.0066,
+      "step": 7192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024358493935394118,
+      "loss": 2.1218,
+      "step": 7193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024357010434531649,
+      "loss": 0.8543,
+      "step": 7194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024355526783830696,
+      "loss": 0.9116,
+      "step": 7195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024354042983315026,
+      "loss": 0.7211,
+      "step": 7196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024352559033008394,
+      "loss": 0.2289,
+      "step": 7197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002435107493293457,
+      "loss": 3.0118,
+      "step": 7198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002434959068311731,
+      "loss": 0.8221,
+      "step": 7199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002434810628358039,
+      "loss": 0.7337,
+      "step": 7200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024346621734347573,
+      "loss": 0.4428,
+      "step": 7201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002434513703544264,
+      "loss": 0.4898,
+      "step": 7202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024343652186889362,
+      "loss": 0.689,
+      "step": 7203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024342167188711516,
+      "loss": 0.7574,
+      "step": 7204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024340682040932887,
+      "loss": 0.9228,
+      "step": 7205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024339196743577244,
+      "loss": 0.4997,
+      "step": 7206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002433771129666839,
+      "loss": 0.3777,
+      "step": 7207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000243362257002301,
+      "loss": 0.6214,
+      "step": 7208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002433473995428617,
+      "loss": 1.7728,
+      "step": 7209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024333254058860385,
+      "loss": 0.6124,
+      "step": 7210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024331768013976547,
+      "loss": 0.0451,
+      "step": 7211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024330281819658447,
+      "loss": 0.3859,
+      "step": 7212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002432879547592989,
+      "loss": 0.4678,
+      "step": 7213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002432730898281467,
+      "loss": 0.4368,
+      "step": 7214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024325822340336595,
+      "loss": 2.2383,
+      "step": 7215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024324335548519472,
+      "loss": 0.017,
+      "step": 7216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002432284860738711,
+      "loss": 1.2217,
+      "step": 7217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024321361516963321,
+      "loss": 0.6574,
+      "step": 7218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024319874277271917,
+      "loss": 0.441,
+      "step": 7219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024318386888336712,
+      "loss": 0.0135,
+      "step": 7220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024316899350181525,
+      "loss": 0.3239,
+      "step": 7221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024315411662830175,
+      "loss": 0.4976,
+      "step": 7222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024313923826306492,
+      "loss": 0.4805,
+      "step": 7223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024312435840634293,
+      "loss": 0.3744,
+      "step": 7224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024310947705837417,
+      "loss": 1.0363,
+      "step": 7225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024309459421939682,
+      "loss": 0.9122,
+      "step": 7226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024307970988964926,
+      "loss": 1.7681,
+      "step": 7227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024306482406936985,
+      "loss": 0.4661,
+      "step": 7228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024304993675879695,
+      "loss": 0.3392,
+      "step": 7229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000243035047958169,
+      "loss": 0.7955,
+      "step": 7230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002430201576677243,
+      "loss": 0.7465,
+      "step": 7231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024300526588770142,
+      "loss": 0.0262,
+      "step": 7232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002429903726183388,
+      "loss": 0.3413,
+      "step": 7233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024297547785987494,
+      "loss": 0.2462,
+      "step": 7234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024296058161254832,
+      "loss": 0.561,
+      "step": 7235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024294568387659744,
+      "loss": 0.8225,
+      "step": 7236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024293078465226103,
+      "loss": 0.5538,
+      "step": 7237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024291588393977748,
+      "loss": 0.2242,
+      "step": 7238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024290098173938553,
+      "loss": 0.7223,
+      "step": 7239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002428860780513238,
+      "loss": 0.5002,
+      "step": 7240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024287117287583095,
+      "loss": 0.5283,
+      "step": 7241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024285626621314558,
+      "loss": 0.7762,
+      "step": 7242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024284135806350654,
+      "loss": 0.3126,
+      "step": 7243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024282644842715248,
+      "loss": 0.4739,
+      "step": 7244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024281153730432208,
+      "loss": 0.5997,
+      "step": 7245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024279662469525425,
+      "loss": 0.2238,
+      "step": 7246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024278171060018776,
+      "loss": 3.5521,
+      "step": 7247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002427667950193614,
+      "loss": 0.41,
+      "step": 7248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024275187795301406,
+      "loss": 0.2566,
+      "step": 7249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024273695940138458,
+      "loss": 0.0817,
+      "step": 7250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024272203936471186,
+      "loss": 0.5226,
+      "step": 7251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024270711784323484,
+      "loss": 0.2447,
+      "step": 7252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002426921948371925,
+      "loss": 1.1643,
+      "step": 7253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024267727034682374,
+      "loss": 0.1857,
+      "step": 7254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002426623443723676,
+      "loss": 0.8883,
+      "step": 7255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024264741691406306,
+      "loss": 0.7865,
+      "step": 7256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024263248797214918,
+      "loss": 0.3053,
+      "step": 7257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024261755754686503,
+      "loss": 0.2912,
+      "step": 7258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002426026256384497,
+      "loss": 0.4194,
+      "step": 7259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024258769224714234,
+      "loss": 0.8171,
+      "step": 7260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000242572757373182,
+      "loss": 0.2824,
+      "step": 7261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024255782101680789,
+      "loss": 2.6202,
+      "step": 7262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024254288317825917,
+      "loss": 0.7708,
+      "step": 7263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002425279438577751,
+      "loss": 0.3803,
+      "step": 7264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024251300305559484,
+      "loss": 0.6344,
+      "step": 7265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002424980607719577,
+      "loss": 0.5432,
+      "step": 7266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024248311700710292,
+      "loss": 0.7063,
+      "step": 7267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002424681717612698,
+      "loss": 0.5807,
+      "step": 7268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024245322503469775,
+      "loss": 0.3055,
+      "step": 7269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000242438276827626,
+      "loss": 0.1548,
+      "step": 7270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000242423327140294,
+      "loss": 0.1508,
+      "step": 7271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002424083759729411,
+      "loss": 0.9721,
+      "step": 7272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024239342332580676,
+      "loss": 0.2206,
+      "step": 7273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002423784691991304,
+      "loss": 0.2834,
+      "step": 7274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024236351359315154,
+      "loss": 1.0512,
+      "step": 7275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002423485565081096,
+      "loss": 0.5724,
+      "step": 7276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002423335979442441,
+      "loss": 0.3442,
+      "step": 7277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002423186379017946,
+      "loss": 0.369,
+      "step": 7278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002423036763810007,
+      "loss": 0.4919,
+      "step": 7279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002422887133821019,
+      "loss": 2.7995,
+      "step": 7280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002422737489053379,
+      "loss": 0.3897,
+      "step": 7281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024225878295094835,
+      "loss": 0.0031,
+      "step": 7282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024224381551917276,
+      "loss": 0.1774,
+      "step": 7283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024222884661025094,
+      "loss": 0.4639,
+      "step": 7284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024221387622442255,
+      "loss": 0.0908,
+      "step": 7285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024219890436192736,
+      "loss": 0.7037,
+      "step": 7286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024218393102300504,
+      "loss": 0.2875,
+      "step": 7287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002421689562078954,
+      "loss": 1.4165,
+      "step": 7288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024215397991683832,
+      "loss": 0.2432,
+      "step": 7289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024213900215007354,
+      "loss": 0.7146,
+      "step": 7290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024212402290784088,
+      "loss": 0.256,
+      "step": 7291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024210904219038027,
+      "loss": 0.6162,
+      "step": 7292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002420940599979316,
+      "loss": 1.5662,
+      "step": 7293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024207907633073475,
+      "loss": 0.1989,
+      "step": 7294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002420640911890297,
+      "loss": 0.9334,
+      "step": 7295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024204910457305636,
+      "loss": 0.2403,
+      "step": 7296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002420341164830548,
+      "loss": 0.376,
+      "step": 7297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000242019126919265,
+      "loss": 1.3544,
+      "step": 7298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024200413588192698,
+      "loss": 0.1733,
+      "step": 7299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024198914337128074,
+      "loss": 0.6026,
+      "step": 7300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024197414938756652,
+      "loss": 0.3226,
+      "step": 7301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024195915393102424,
+      "loss": 0.6408,
+      "step": 7302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024194415700189416,
+      "loss": 1.2999,
+      "step": 7303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024192915860041643,
+      "loss": 1.3382,
+      "step": 7304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024191415872683116,
+      "loss": 0.7824,
+      "step": 7305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024189915738137858,
+      "loss": 0.6152,
+      "step": 7306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002418841545642989,
+      "loss": 0.4267,
+      "step": 7307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002418691502758324,
+      "loss": 0.0052,
+      "step": 7308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024185414451621934,
+      "loss": 0.6037,
+      "step": 7309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002418391372857,
+      "loss": 0.267,
+      "step": 7310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024182412858451472,
+      "loss": 0.5916,
+      "step": 7311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002418091184129038,
+      "loss": 0.9159,
+      "step": 7312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024179410677110767,
+      "loss": 0.0143,
+      "step": 7313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024177909365936666,
+      "loss": 1.0027,
+      "step": 7314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024176407907792124,
+      "loss": 0.3156,
+      "step": 7315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024174906302701176,
+      "loss": 0.3892,
+      "step": 7316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024173404550687877,
+      "loss": 0.3377,
+      "step": 7317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024171902651776272,
+      "loss": 0.2368,
+      "step": 7318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024170400605990407,
+      "loss": 0.5329,
+      "step": 7319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024168898413354345,
+      "loss": 0.3496,
+      "step": 7320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002416739607389213,
+      "loss": 0.3505,
+      "step": 7321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024165893587627826,
+      "loss": 0.0074,
+      "step": 7322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024164390954585496,
+      "loss": 0.4414,
+      "step": 7323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024162888174789196,
+      "loss": 0.6819,
+      "step": 7324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024161385248262993,
+      "loss": 0.3276,
+      "step": 7325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024159882175030952,
+      "loss": 0.002,
+      "step": 7326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024158378955117153,
+      "loss": 0.154,
+      "step": 7327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024156875588545658,
+      "loss": 0.4939,
+      "step": 7328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002415537207534054,
+      "loss": 0.3116,
+      "step": 7329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024153868415525875,
+      "loss": 0.2659,
+      "step": 7330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002415236460912575,
+      "loss": 0.7935,
+      "step": 7331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024150860656164242,
+      "loss": 0.3315,
+      "step": 7332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002414935655666543,
+      "loss": 0.0065,
+      "step": 7333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002414785231065341,
+      "loss": 0.2772,
+      "step": 7334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024146347918152262,
+      "loss": 0.5359,
+      "step": 7335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024144843379186076,
+      "loss": 1.2198,
+      "step": 7336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002414333869377895,
+      "loss": 0.4726,
+      "step": 7337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024141833861954974,
+      "loss": 0.3655,
+      "step": 7338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024140328883738254,
+      "loss": 0.1636,
+      "step": 7339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024138823759152882,
+      "loss": 0.1121,
+      "step": 7340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002413731848822296,
+      "loss": 1.4652,
+      "step": 7341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000241358130709726,
+      "loss": 0.0062,
+      "step": 7342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024134307507425908,
+      "loss": 0.1679,
+      "step": 7343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002413280179760698,
+      "loss": 0.0263,
+      "step": 7344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024131295941539946,
+      "loss": 0.4279,
+      "step": 7345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024129789939248905,
+      "loss": 1.2714,
+      "step": 7346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024128283790757985,
+      "loss": 0.0046,
+      "step": 7347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024126777496091297,
+      "loss": 0.5966,
+      "step": 7348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024125271055272965,
+      "loss": 0.0041,
+      "step": 7349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024123764468327117,
+      "loss": 0.2548,
+      "step": 7350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024122257735277865,
+      "loss": 0.374,
+      "step": 7351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024120750856149355,
+      "loss": 0.4183,
+      "step": 7352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024119243830965703,
+      "loss": 0.2615,
+      "step": 7353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002411773665975105,
+      "loss": 0.5734,
+      "step": 7354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024116229342529527,
+      "loss": 0.0006,
+      "step": 7355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002411472187932527,
+      "loss": 0.6792,
+      "step": 7356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024113214270162427,
+      "loss": 0.001,
+      "step": 7357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002411170651506513,
+      "loss": 3.3399,
+      "step": 7358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024110198614057528,
+      "loss": 0.7808,
+      "step": 7359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024108690567163768,
+      "loss": 0.0095,
+      "step": 7360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024107182374408004,
+      "loss": 0.0085,
+      "step": 7361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024105674035814373,
+      "loss": 0.5612,
+      "step": 7362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024104165551407043,
+      "loss": 1.1444,
+      "step": 7363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024102656921210164,
+      "loss": 0.3731,
+      "step": 7364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024101148145247897,
+      "loss": 0.7139,
+      "step": 7365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024099639223544402,
+      "loss": 0.088,
+      "step": 7366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002409813015612384,
+      "loss": 0.2713,
+      "step": 7367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024096620943010378,
+      "loss": 0.3565,
+      "step": 7368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024095111584228186,
+      "loss": 0.0042,
+      "step": 7369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024093602079801427,
+      "loss": 0.6834,
+      "step": 7370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024092092429754282,
+      "loss": 1.6162,
+      "step": 7371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002409058263411092,
+      "loss": 0.5323,
+      "step": 7372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024089072692895527,
+      "loss": 0.5844,
+      "step": 7373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024087562606132264,
+      "loss": 0.3905,
+      "step": 7374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024086052373845332,
+      "loss": 0.5633,
+      "step": 7375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024084541996058907,
+      "loss": 0.0015,
+      "step": 7376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024083031472797175,
+      "loss": 0.0205,
+      "step": 7377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024081520804084322,
+      "loss": 0.9161,
+      "step": 7378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024080009989944549,
+      "loss": 0.2656,
+      "step": 7379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024078499030402042,
+      "loss": 2.2255,
+      "step": 7380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024076987925480996,
+      "loss": 0.3192,
+      "step": 7381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002407547667520561,
+      "loss": 0.3571,
+      "step": 7382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002407396527960009,
+      "loss": 0.672,
+      "step": 7383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024072453738688626,
+      "loss": 1.271,
+      "step": 7384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024070942052495438,
+      "loss": 0.4691,
+      "step": 7385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024069430221044724,
+      "loss": 0.0644,
+      "step": 7386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000240679182443607,
+      "loss": 0.4807,
+      "step": 7387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024066406122467566,
+      "loss": 0.0104,
+      "step": 7388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002406489385538955,
+      "loss": 1.1711,
+      "step": 7389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024063381443150867,
+      "loss": 0.2766,
+      "step": 7390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024061868885775727,
+      "loss": 0.2492,
+      "step": 7391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024060356183288362,
+      "loss": 0.2786,
+      "step": 7392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024058843335712985,
+      "loss": 0.1604,
+      "step": 7393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024057330343073827,
+      "loss": 0.3527,
+      "step": 7394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024055817205395123,
+      "loss": 0.3711,
+      "step": 7395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024054303922701095,
+      "loss": 0.6434,
+      "step": 7396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024052790495015977,
+      "loss": 1.6504,
+      "step": 7397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024051276922364004,
+      "loss": 0.4158,
+      "step": 7398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024049763204769416,
+      "loss": 0.3905,
+      "step": 7399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024048249342256453,
+      "loss": 3.2118,
+      "step": 7400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024046735334849354,
+      "loss": 1.8907,
+      "step": 7401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002404522118257237,
+      "loss": 0.4611,
+      "step": 7402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024043706885449744,
+      "loss": 0.0008,
+      "step": 7403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002404219244350572,
+      "loss": 0.0677,
+      "step": 7404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024040677856764558,
+      "loss": 0.3654,
+      "step": 7405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024039163125250508,
+      "loss": 0.0056,
+      "step": 7406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024037648248987825,
+      "loss": 0.7414,
+      "step": 7407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024036133228000768,
+      "loss": 0.0488,
+      "step": 7408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000240346180623136,
+      "loss": 0.0061,
+      "step": 7409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024033102751950584,
+      "loss": 0.1206,
+      "step": 7410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024031587296935986,
+      "loss": 0.3319,
+      "step": 7411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024030071697294064,
+      "loss": 0.3178,
+      "step": 7412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024028555953049108,
+      "loss": 0.7305,
+      "step": 7413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024027040064225367,
+      "loss": 0.4635,
+      "step": 7414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024025524030847132,
+      "loss": 0.1361,
+      "step": 7415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024024007852938676,
+      "loss": 0.4068,
+      "step": 7416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024022491530524271,
+      "loss": 0.3285,
+      "step": 7417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024020975063628215,
+      "loss": 0.3464,
+      "step": 7418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024019458452274775,
+      "loss": 0.4207,
+      "step": 7419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002401794169648825,
+      "loss": 0.6688,
+      "step": 7420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002401642479629292,
+      "loss": 0.7946,
+      "step": 7421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024014907751713076,
+      "loss": 0.4222,
+      "step": 7422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024013390562773013,
+      "loss": 0.369,
+      "step": 7423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002401187322949703,
+      "loss": 1.8839,
+      "step": 7424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024010355751909423,
+      "loss": 0.3378,
+      "step": 7425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002400883813003449,
+      "loss": 0.5632,
+      "step": 7426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024007320363896535,
+      "loss": 0.0069,
+      "step": 7427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024005802453519863,
+      "loss": 0.4494,
+      "step": 7428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002400428439892878,
+      "loss": 0.9101,
+      "step": 7429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024002766200147593,
+      "loss": 0.009,
+      "step": 7430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024001247857200622,
+      "loss": 0.2335,
+      "step": 7431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023999729370112173,
+      "loss": 0.7048,
+      "step": 7432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023998210738906565,
+      "loss": 0.0074,
+      "step": 7433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023996691963608117,
+      "loss": 2.326,
+      "step": 7434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002399517304424115,
+      "loss": 0.4084,
+      "step": 7435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023993653980829984,
+      "loss": 0.3635,
+      "step": 7436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023992134773398948,
+      "loss": 1.1045,
+      "step": 7437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002399061542197237,
+      "loss": 0.3905,
+      "step": 7438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023989095926574578,
+      "loss": 0.7959,
+      "step": 7439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023987576287229904,
+      "loss": 0.5296,
+      "step": 7440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002398605650396269,
+      "loss": 0.3348,
+      "step": 7441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023984536576797258,
+      "loss": 0.2659,
+      "step": 7442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023983016505757962,
+      "loss": 0.3323,
+      "step": 7443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023981496290869143,
+      "loss": 0.4353,
+      "step": 7444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023979975932155134,
+      "loss": 0.5294,
+      "step": 7445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023978455429640288,
+      "loss": 0.5703,
+      "step": 7446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023976934783348952,
+      "loss": 0.4562,
+      "step": 7447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023975413993305482,
+      "loss": 1.6831,
+      "step": 7448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023973893059534221,
+      "loss": 0.8889,
+      "step": 7449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023972371982059536,
+      "loss": 0.0686,
+      "step": 7450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002397085076090578,
+      "loss": 0.6512,
+      "step": 7451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023969329396097307,
+      "loss": 0.3744,
+      "step": 7452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023967807887658487,
+      "loss": 0.1919,
+      "step": 7453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023966286235613688,
+      "loss": 0.4845,
+      "step": 7454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002396476443998727,
+      "loss": 2.084,
+      "step": 7455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023963242500803597,
+      "loss": 0.195,
+      "step": 7456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002396172041808705,
+      "loss": 0.5895,
+      "step": 7457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023960198191862002,
+      "loss": 0.6425,
+      "step": 7458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023958675822152828,
+      "loss": 0.0163,
+      "step": 7459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023957153308983906,
+      "loss": 0.346,
+      "step": 7460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023955630652379613,
+      "loss": 0.495,
+      "step": 7461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023954107852364336,
+      "loss": 0.2631,
+      "step": 7462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023952584908962463,
+      "loss": 1.6757,
+      "step": 7463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023951061822198374,
+      "loss": 0.331,
+      "step": 7464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002394953859209647,
+      "loss": 0.3127,
+      "step": 7465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023948015218681133,
+      "loss": 1.7505,
+      "step": 7466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023946491701976763,
+      "loss": 3.1896,
+      "step": 7467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023944968042007752,
+      "loss": 0.2749,
+      "step": 7468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023943444238798506,
+      "loss": 0.7405,
+      "step": 7469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002394192029237342,
+      "loss": 1.0899,
+      "step": 7470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023940396202756903,
+      "loss": 0.5396,
+      "step": 7471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023938871969973356,
+      "loss": 0.0705,
+      "step": 7472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002393734759404719,
+      "loss": 0.3968,
+      "step": 7473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002393582307500282,
+      "loss": 0.2003,
+      "step": 7474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002393429841286465,
+      "loss": 0.3229,
+      "step": 7475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023932773607657099,
+      "loss": 0.2703,
+      "step": 7476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023931248659404586,
+      "loss": 1.8665,
+      "step": 7477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023929723568131528,
+      "loss": 0.3091,
+      "step": 7478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023928198333862352,
+      "loss": 0.4581,
+      "step": 7479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023926672956621476,
+      "loss": 0.4305,
+      "step": 7480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002392514743643333,
+      "loss": 0.387,
+      "step": 7481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023923621773322344,
+      "loss": 0.3487,
+      "step": 7482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023922095967312943,
+      "loss": 0.7781,
+      "step": 7483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023920570018429573,
+      "loss": 2.4124,
+      "step": 7484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023919043926696657,
+      "loss": 0.3287,
+      "step": 7485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023917517692138642,
+      "loss": 0.5436,
+      "step": 7486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023915991314779958,
+      "loss": 1.1887,
+      "step": 7487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023914464794645062,
+      "loss": 2.0439,
+      "step": 7488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023912938131758385,
+      "loss": 0.0163,
+      "step": 7489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023911411326144389,
+      "loss": 0.6463,
+      "step": 7490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023909884377827507,
+      "loss": 0.3822,
+      "step": 7491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023908357286832197,
+      "loss": 0.6765,
+      "step": 7492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023906830053182918,
+      "loss": 0.4323,
+      "step": 7493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023905302676904122,
+      "loss": 0.574,
+      "step": 7494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002390377515802027,
+      "loss": 0.6216,
+      "step": 7495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002390224749655582,
+      "loss": 1.556,
+      "step": 7496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023900719692535236,
+      "loss": 0.5541,
+      "step": 7497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023899191745982987,
+      "loss": 0.3794,
+      "step": 7498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023897663656923538,
+      "loss": 0.4356,
+      "step": 7499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023896135425381356,
+      "loss": 0.0332,
+      "step": 7500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002389460705138092,
+      "loss": 0.3133,
+      "step": 7501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023893078534946698,
+      "loss": 0.1624,
+      "step": 7502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002389154987610317,
+      "loss": 0.4666,
+      "step": 7503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023890021074874816,
+      "loss": 0.7078,
+      "step": 7504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023888492131286118,
+      "loss": 0.419,
+      "step": 7505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023886963045361556,
+      "loss": 1.1997,
+      "step": 7506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002388543381712562,
+      "loss": 1.5294,
+      "step": 7507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023883904446602797,
+      "loss": 0.432,
+      "step": 7508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023882374933817575,
+      "loss": 1.0415,
+      "step": 7509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023880845278794448,
+      "loss": 0.0646,
+      "step": 7510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023879315481557917,
+      "loss": 0.8472,
+      "step": 7511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023877785542132467,
+      "loss": 1.1573,
+      "step": 7512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023876255460542612,
+      "loss": 0.6338,
+      "step": 7513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023874725236812844,
+      "loss": 0.3815,
+      "step": 7514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023873194870967676,
+      "loss": 0.4384,
+      "step": 7515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023871664363031602,
+      "loss": 0.5096,
+      "step": 7516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002387013371302914,
+      "loss": 0.5487,
+      "step": 7517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023868602920984797,
+      "loss": 0.0247,
+      "step": 7518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023867071986923092,
+      "loss": 0.7243,
+      "step": 7519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023865540910868538,
+      "loss": 0.2763,
+      "step": 7520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023864009692845642,
+      "loss": 0.3061,
+      "step": 7521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023862478332878944,
+      "loss": 0.6115,
+      "step": 7522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002386094683099295,
+      "loss": 0.5056,
+      "step": 7523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000238594151872122,
+      "loss": 2.3577,
+      "step": 7524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023857883401561198,
+      "loss": 0.7804,
+      "step": 7525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023856351474064498,
+      "loss": 0.3971,
+      "step": 7526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023854819404746616,
+      "loss": 0.5058,
+      "step": 7527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002385328719363209,
+      "loss": 0.3223,
+      "step": 7528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002385175484074546,
+      "loss": 1.1656,
+      "step": 7529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002385022234611126,
+      "loss": 0.7911,
+      "step": 7530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023848689709754032,
+      "loss": 1.2368,
+      "step": 7531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023847156931698312,
+      "loss": 0.3622,
+      "step": 7532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002384562401196866,
+      "loss": 0.7032,
+      "step": 7533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002384409095058961,
+      "loss": 0.9037,
+      "step": 7534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002384255774758572,
+      "loss": 0.6926,
+      "step": 7535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023841024402981533,
+      "loss": 0.6494,
+      "step": 7536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002383949091680161,
+      "loss": 0.8902,
+      "step": 7537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002383795728907051,
+      "loss": 0.0886,
+      "step": 7538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023836423519812787,
+      "loss": 1.2459,
+      "step": 7539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023834889609053,
+      "loss": 0.0588,
+      "step": 7540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023833355556815718,
+      "loss": 0.219,
+      "step": 7541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023831821363125505,
+      "loss": 0.0211,
+      "step": 7542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023830287028006922,
+      "loss": 0.931,
+      "step": 7543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023828752551484546,
+      "loss": 0.3411,
+      "step": 7544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002382721793358295,
+      "loss": 0.7965,
+      "step": 7545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002382568317432671,
+      "loss": 1.6321,
+      "step": 7546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023824148273740394,
+      "loss": 2.7715,
+      "step": 7547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023822613231848589,
+      "loss": 0.4543,
+      "step": 7548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023821078048675876,
+      "loss": 0.3887,
+      "step": 7549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023819542724246834,
+      "loss": 0.3686,
+      "step": 7550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023818007258586053,
+      "loss": 0.3241,
+      "step": 7551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023816471651718114,
+      "loss": 0.1681,
+      "step": 7552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023814935903667618,
+      "loss": 0.4087,
+      "step": 7553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023813400014459157,
+      "loss": 0.0851,
+      "step": 7554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023811863984117315,
+      "loss": 1.0314,
+      "step": 7555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000238103278126667,
+      "loss": 0.3619,
+      "step": 7556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023808791500131904,
+      "loss": 0.6103,
+      "step": 7557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023807255046537536,
+      "loss": 0.7349,
+      "step": 7558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023805718451908194,
+      "loss": 0.4368,
+      "step": 7559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023804181716268486,
+      "loss": 2.4767,
+      "step": 7560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023802644839643024,
+      "loss": 2.5255,
+      "step": 7561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002380110782205641,
+      "loss": 0.7602,
+      "step": 7562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002379957066353327,
+      "loss": 0.0423,
+      "step": 7563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023798033364098208,
+      "loss": 0.9575,
+      "step": 7564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023796495923775847,
+      "loss": 0.4476,
+      "step": 7565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023794958342590806,
+      "loss": 0.7078,
+      "step": 7566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023793420620567705,
+      "loss": 0.3598,
+      "step": 7567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002379188275773117,
+      "loss": 0.0142,
+      "step": 7568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023790344754105828,
+      "loss": 0.3979,
+      "step": 7569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023788806609716304,
+      "loss": 0.1437,
+      "step": 7570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002378726832458724,
+      "loss": 0.0501,
+      "step": 7571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023785729898743256,
+      "loss": 0.5597,
+      "step": 7572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023784191332208992,
+      "loss": 0.7402,
+      "step": 7573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002378265262500909,
+      "loss": 0.2971,
+      "step": 7574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023781113777168186,
+      "loss": 0.6888,
+      "step": 7575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023779574788710924,
+      "loss": 1.9589,
+      "step": 7576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002377803565966195,
+      "loss": 0.468,
+      "step": 7577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023776496390045905,
+      "loss": 0.0136,
+      "step": 7578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023774956979887444,
+      "loss": 0.7419,
+      "step": 7579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023773417429211218,
+      "loss": 0.6816,
+      "step": 7580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023771877738041873,
+      "loss": 0.4938,
+      "step": 7581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023770337906404076,
+      "loss": 0.6057,
+      "step": 7582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002376879793432248,
+      "loss": 0.3293,
+      "step": 7583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023767257821821744,
+      "loss": 0.314,
+      "step": 7584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002376571756892653,
+      "loss": 0.5667,
+      "step": 7585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023764177175661507,
+      "loss": 0.6667,
+      "step": 7586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023762636642051343,
+      "loss": 0.9569,
+      "step": 7587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023761095968120695,
+      "loss": 0.331,
+      "step": 7588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023759555153894248,
+      "loss": 0.461,
+      "step": 7589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023758014199396675,
+      "loss": 0.4406,
+      "step": 7590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023756473104652646,
+      "loss": 0.2052,
+      "step": 7591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023754931869686841,
+      "loss": 0.328,
+      "step": 7592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023753390494523943,
+      "loss": 0.0217,
+      "step": 7593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023751848979188631,
+      "loss": 0.014,
+      "step": 7594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023750307323705595,
+      "loss": 0.3466,
+      "step": 7595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023748765528099514,
+      "loss": 0.0109,
+      "step": 7596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023747223592395089,
+      "loss": 0.0555,
+      "step": 7597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002374568151661701,
+      "loss": 1.6433,
+      "step": 7598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023744139300789957,
+      "loss": 0.0817,
+      "step": 7599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00023742596944938637,
+      "loss": 0.7099,
+      "step": 7600
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 24926,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 1.6759298804692255e+18,
+  "trial_name": null,
+  "trial_params": null
+}