diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,28158 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 4688,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.2697,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1727,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 1.1512,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4116,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.5147,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.5487,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.3461,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.3519,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 1.4732,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.2603,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.4051,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 1.5126,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.4592,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.3586,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 1.568,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.2941,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.4179,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 1.4248,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 1.4829,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.6275,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 1.3551,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.391,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.5285,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 1.412,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 1.2196,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.358,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.3668,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.3352,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8e-05,
+      "loss": 1.3057,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 1.4492,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2e-05,
+      "loss": 1.2966,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.2441,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6e-05,
+      "loss": 1.6387,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.4246,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-05,
+      "loss": 1.5533,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-05,
+      "loss": 1.6343,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-05,
+      "loss": 1.3978,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6e-05,
+      "loss": 1.2502,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.4378,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 1.6146,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.2e-05,
+      "loss": 1.5379,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-05,
+      "loss": 1.5665,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.6e-05,
+      "loss": 1.3731,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.3499,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9e-05,
+      "loss": 1.4851,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.2945,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.4e-05,
+      "loss": 1.3099,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.6e-05,
+      "loss": 1.4015,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.8e-05,
+      "loss": 1.3672,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 1.2962,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 1.2735,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.3815,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010600000000000002,
+      "loss": 1.3348,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.3422,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.3699,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.5097,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 1.4185,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000116,
+      "loss": 1.4951,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000118,
+      "loss": 1.2166,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012,
+      "loss": 1.2781,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000122,
+      "loss": 1.3036,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000124,
+      "loss": 1.2765,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000126,
+      "loss": 1.6547,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.4887,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.334,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000132,
+      "loss": 1.206,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000134,
+      "loss": 1.628,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.5388,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000138,
+      "loss": 1.2602,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014,
+      "loss": 1.2329,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000142,
+      "loss": 1.1837,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000144,
+      "loss": 1.4647,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000146,
+      "loss": 1.3805,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000148,
+      "loss": 1.4361,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.5632,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000152,
+      "loss": 1.3057,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000154,
+      "loss": 1.3901,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.5926,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015800000000000002,
+      "loss": 1.4285,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016,
+      "loss": 1.1802,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000162,
+      "loss": 1.3302,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000164,
+      "loss": 1.5157,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000166,
+      "loss": 1.4044,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000168,
+      "loss": 1.3849,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017,
+      "loss": 1.4373,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000172,
+      "loss": 1.4188,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000174,
+      "loss": 1.4036,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.2989,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017800000000000002,
+      "loss": 1.2206,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018,
+      "loss": 1.32,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000182,
+      "loss": 1.6151,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.3669,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018600000000000002,
+      "loss": 1.3845,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000188,
+      "loss": 1.467,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019,
+      "loss": 1.4345,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000192,
+      "loss": 1.3821,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000194,
+      "loss": 1.218,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000196,
+      "loss": 1.3816,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 1.4841,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 1.1609,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019999997655647495,
+      "loss": 1.2078,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999999062259108,
+      "loss": 1.2759,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999997890083405,
+      "loss": 1.5263,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019999962490381906,
+      "loss": 1.4527,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019999941391242335,
+      "loss": 1.239,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999915603425235,
+      "loss": 1.3728,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999885126942694,
+      "loss": 1.2909,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999849961809007,
+      "loss": 1.5023,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999810108040658,
+      "loss": 1.2558,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999765565656332,
+      "loss": 1.2637,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999716334676918,
+      "loss": 1.3563,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999662415125493,
+      "loss": 1.1959,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999603807027342,
+      "loss": 1.3656,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999540510409945,
+      "loss": 1.3907,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999472525302977,
+      "loss": 1.3825,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999399851738318,
+      "loss": 1.1964,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999932248975004,
+      "loss": 1.3115,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999240439374418,
+      "loss": 1.357,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999153700649917,
+      "loss": 1.4453,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019999062273617216,
+      "loss": 1.4053,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019998966158319174,
+      "loss": 1.1877,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999886535480086,
+      "loss": 1.5058,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999875986310954,
+      "loss": 1.3083,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019998649683294672,
+      "loss": 1.3857,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019998534815407918,
+      "loss": 1.4058,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019998415259503137,
+      "loss": 1.2968,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019998291015636382,
+      "loss": 1.3864,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019998162083865908,
+      "loss": 1.5208,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999802846425217,
+      "loss": 1.2557,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997890156857814,
+      "loss": 1.2605,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997747161747695,
+      "loss": 1.465,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997599478988853,
+      "loss": 1.5193,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997447108650537,
+      "loss": 1.3152,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997290050804183,
+      "loss": 1.6536,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019997128305523435,
+      "loss": 1.306,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999696187288413,
+      "loss": 1.3149,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019996790752964305,
+      "loss": 1.5695,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999661494584419,
+      "loss": 1.1744,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019996434451606217,
+      "loss": 1.438,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019996249270335016,
+      "loss": 1.3835,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019996059402117408,
+      "loss": 1.3183,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999586484704242,
+      "loss": 1.2612,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019995665605201276,
+      "loss": 1.4624,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999546167668739,
+      "loss": 1.3976,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999525306159638,
+      "loss": 1.2524,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019995039760026059,
+      "loss": 1.7054,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019994821772076438,
+      "loss": 1.1012,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019994599097849727,
+      "loss": 1.201,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019994371737450326,
+      "loss": 1.1381,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019994139690984844,
+      "loss": 1.5642,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019993902958562076,
+      "loss": 1.4132,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019993661540293025,
+      "loss": 1.352,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019993415436290877,
+      "loss": 1.3132,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019993164646671025,
+      "loss": 1.4066,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992909171551062,
+      "loss": 1.2489,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999264901105077,
+      "loss": 1.3159,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992384165292126,
+      "loss": 1.8481,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019992114634399312,
+      "loss": 1.5619,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019991840418498704,
+      "loss": 1.6583,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019991561517718876,
+      "loss": 1.2153,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999127793219059,
+      "loss": 1.4563,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019990989662046818,
+      "loss": 1.0691,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019990696707422712,
+      "loss": 1.2684,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999039906845564,
+      "loss": 1.4371,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999009674528515,
+      "loss": 1.4676,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019989789738052993,
+      "loss": 1.3093,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019989478046903117,
+      "loss": 1.7313,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019989161671981665,
+      "loss": 1.3445,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988840613436971,
+      "loss": 1.3778,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988514871419579,
+      "loss": 1.4462,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988184446082213,
+      "loss": 1.1789,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199878493375798,
+      "loss": 1.278,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987509546069468,
+      "loss": 1.2531,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987165071710527,
+      "loss": 1.4409,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986815914664502,
+      "loss": 1.2921,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019986462075095087,
+      "loss": 1.1371,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199861035531682,
+      "loss": 1.2737,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019985740349051935,
+      "loss": 1.4239,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019985372462916594,
+      "loss": 1.5384,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998499989493466,
+      "loss": 1.4647,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019984622645280822,
+      "loss": 1.5626,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019984240714131964,
+      "loss": 1.229,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998385410166716,
+      "loss": 1.4062,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998346280806768,
+      "loss": 1.3019,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998306683351699,
+      "loss": 1.3104,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982666178200754,
+      "loss": 1.4765,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982260842306825,
+      "loss": 1.4878,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981850826025253,
+      "loss": 1.3439,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981436129548285,
+      "loss": 1.4246,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981016753070358,
+      "loss": 1.7005,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980592696788106,
+      "loss": 1.3029,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980163960900353,
+      "loss": 1.3054,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979730545608126,
+      "loss": 1.406,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979292451114636,
+      "loss": 1.5447,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019978849677625298,
+      "loss": 1.2113,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019978402225347708,
+      "loss": 1.4416,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997795009449167,
+      "loss": 1.6918,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019977493285269171,
+      "loss": 1.618,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019977031797894398,
+      "loss": 1.4757,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976565632583726,
+      "loss": 1.1542,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997609478955573,
+      "loss": 1.1996,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975619269031167,
+      "loss": 1.429,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975139071233002,
+      "loss": 1.3136,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997465419638638,
+      "loss": 1.1705,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974164644718655,
+      "loss": 1.4794,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997367041645935,
+      "loss": 1.7285,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019973171511840202,
+      "loss": 0.9202,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997266793109513,
+      "loss": 1.2903,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997215967446025,
+      "loss": 1.4761,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019971646742173864,
+      "loss": 1.5668,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019971129134476473,
+      "loss": 1.3141,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019970606851610776,
+      "loss": 1.3681,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019970079893821642,
+      "loss": 1.1602,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019969548261356156,
+      "loss": 1.262,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019969011954463584,
+      "loss": 1.5745,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001996847097339538,
+      "loss": 1.2598,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019967925318405195,
+      "loss": 1.3814,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019967374989748876,
+      "loss": 1.4596,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019966819987684447,
+      "loss": 1.3089,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019966260312472142,
+      "loss": 1.212,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019965695964374366,
+      "loss": 1.5613,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001996512694365573,
+      "loss": 1.3927,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019964553250583037,
+      "loss": 1.288,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019963974885425266,
+      "loss": 1.125,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199633918484536,
+      "loss": 1.4194,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019962804139941401,
+      "loss": 1.3439,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996221176016424,
+      "loss": 1.415,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019961614709399856,
+      "loss": 1.3788,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019961012987928194,
+      "loss": 1.5081,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019960406596031377,
+      "loss": 1.2891,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019959795533993735,
+      "loss": 1.2042,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995917980210177,
+      "loss": 1.337,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019958559400644178,
+      "loss": 1.1964,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019957934329911854,
+      "loss": 1.3295,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995730459019787,
+      "loss": 1.6325,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019956670181797494,
+      "loss": 1.2354,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019956031105008184,
+      "loss": 1.3751,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995538736012958,
+      "loss": 1.5724,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019954738947463514,
+      "loss": 1.469,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995408586731401,
+      "loss": 1.5339,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019953428119987284,
+      "loss": 1.3512,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019952765705791726,
+      "loss": 1.3061,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019952098625037926,
+      "loss": 1.3643,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019951426878038654,
+      "loss": 1.1866,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995075046510888,
+      "loss": 1.2705,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019950069386565747,
+      "loss": 1.2855,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019949383642728596,
+      "loss": 1.4329,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019948693233918952,
+      "loss": 1.3507,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994799816046053,
+      "loss": 1.266,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994729842267922,
+      "loss": 1.26,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994659402090312,
+      "loss": 1.0397,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019945884955462496,
+      "loss": 1.361,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019945171226689812,
+      "loss": 1.4752,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994445283491971,
+      "loss": 1.3548,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019943729780489027,
+      "loss": 1.2586,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019943002063736777,
+      "loss": 1.2311,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019942269685004172,
+      "loss": 1.2416,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019941532644634596,
+      "loss": 1.1483,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940790942973627,
+      "loss": 1.3454,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994004458036903,
+      "loss": 1.7435,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939293557170749,
+      "loss": 1.3464,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938537873730918,
+      "loss": 1.3681,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937777530403857,
+      "loss": 1.4342,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937012527546064,
+      "loss": 1.5978,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993624286551623,
+      "loss": 1.1366,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019935468544675223,
+      "loss": 1.4353,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019934689565386107,
+      "loss": 1.3891,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019933905928014114,
+      "loss": 1.4346,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019933117632926672,
+      "loss": 1.586,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001993232468049339,
+      "loss": 1.4322,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019931527071086053,
+      "loss": 1.2098,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019930724805078648,
+      "loss": 1.248,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019929917882847327,
+      "loss": 1.4682,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019929106304770433,
+      "loss": 1.3645,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992829007122849,
+      "loss": 1.212,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992746918260421,
+      "loss": 1.0879,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992664363928248,
+      "loss": 1.4744,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992581344165037,
+      "loss": 1.4301,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019924978590097142,
+      "loss": 1.2859,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992413908501423,
+      "loss": 1.6858,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992329492679525,
+      "loss": 1.673,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992244611583601,
+      "loss": 1.367,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019921592652534487,
+      "loss": 1.3955,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019920734537290846,
+      "loss": 1.3693,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991987177050743,
+      "loss": 1.171,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019919004352588767,
+      "loss": 1.434,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019918132283941567,
+      "loss": 1.2412,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019917255564974714,
+      "loss": 1.46,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019916374196099273,
+      "loss": 1.3371,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019915488177728498,
+      "loss": 1.3846,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019914597510277808,
+      "loss": 1.1572,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019913702194164818,
+      "loss": 1.4684,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019912802229809316,
+      "loss": 1.4511,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019911897617633264,
+      "loss": 1.3199,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001991098835806081,
+      "loss": 1.5227,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019910074451518278,
+      "loss": 1.577,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019909155898434176,
+      "loss": 1.4729,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019908232699239182,
+      "loss": 1.3416,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019907304854366155,
+      "loss": 1.3886,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019906372364250142,
+      "loss": 1.2705,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019905435229328352,
+      "loss": 1.4403,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019904493450040184,
+      "loss": 1.4774,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001990354702682721,
+      "loss": 1.0685,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019902595960133183,
+      "loss": 1.3515,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019901640250404024,
+      "loss": 1.4533,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019900679898087836,
+      "loss": 1.4645,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001989971490363491,
+      "loss": 1.3927,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019898745267497693,
+      "loss": 1.4126,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019897770990130826,
+      "loss": 1.5104,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019896792071991115,
+      "loss": 1.2231,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019895808513537548,
+      "loss": 1.1963,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019894820315231285,
+      "loss": 1.4894,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019893827477535661,
+      "loss": 1.4547,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019892830000916193,
+      "loss": 1.4203,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019891827885840563,
+      "loss": 1.5367,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019890821132778638,
+      "loss": 1.4016,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019889809742202455,
+      "loss": 1.3598,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988879371458622,
+      "loss": 1.4626,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019887773050406325,
+      "loss": 1.3549,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019886747750141323,
+      "loss": 1.4731,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019885717814271953,
+      "loss": 1.3145,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019884683243281116,
+      "loss": 1.4063,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019883644037653895,
+      "loss": 1.2046,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019882600197877542,
+      "loss": 1.3465,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019881551724441483,
+      "loss": 1.2616,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019880498617837316,
+      "loss": 1.2946,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019879440878558814,
+      "loss": 1.4456,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019878378507101916,
+      "loss": 1.3954,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019877311503964737,
+      "loss": 1.3976,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019876239869647563,
+      "loss": 1.2658,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019875163604652856,
+      "loss": 1.1574,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019874082709485244,
+      "loss": 1.3922,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019872997184651524,
+      "loss": 1.3728,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019871907030660663,
+      "loss": 1.4011,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001987081224802381,
+      "loss": 1.3951,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019869712837254275,
+      "loss": 1.3519,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019868608798867535,
+      "loss": 1.3,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019867500133381242,
+      "loss": 1.2851,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001986638684131522,
+      "loss": 1.5353,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019865268923191452,
+      "loss": 1.4713,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019864146379534108,
+      "loss": 1.3012,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019863019210869504,
+      "loss": 1.1734,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019861887417726144,
+      "loss": 1.3834,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019860751000634686,
+      "loss": 1.2181,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019859609960127972,
+      "loss": 1.2273,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985846429674099,
+      "loss": 1.4122,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019857314011010918,
+      "loss": 1.1426,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019856159103477086,
+      "loss": 1.5258,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019854999574680994,
+      "loss": 1.5196,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985383542516632,
+      "loss": 1.5498,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019852666655478888,
+      "loss": 1.2968,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019851493266166708,
+      "loss": 1.207,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019850315257779943,
+      "loss": 1.3014,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019849132630870927,
+      "loss": 1.626,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019847945385994162,
+      "loss": 1.3975,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984675352370631,
+      "loss": 1.2587,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019845557044566198,
+      "loss": 1.2148,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019844355949134824,
+      "loss": 1.3351,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019843150237975344,
+      "loss": 1.3228,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019841939911653079,
+      "loss": 1.1848,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019840724970735516,
+      "loss": 1.5046,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019839505415792305,
+      "loss": 1.2445,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019838281247395263,
+      "loss": 1.2699,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019837052466118363,
+      "loss": 1.3367,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019835819072537743,
+      "loss": 1.6542,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019834581067231706,
+      "loss": 1.4403,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001983333845078072,
+      "loss": 1.5518,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019832091223767403,
+      "loss": 1.2868,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019830839386776554,
+      "loss": 1.2765,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019829582940395112,
+      "loss": 1.2057,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019828321885212194,
+      "loss": 1.3395,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019827056221819066,
+      "loss": 1.367,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019825785950809168,
+      "loss": 1.2541,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019824511072778086,
+      "loss": 1.167,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019823231588323578,
+      "loss": 1.1125,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019821947498045551,
+      "loss": 1.3131,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019820658802546082,
+      "loss": 1.3104,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000198193655024294,
+      "loss": 1.1958,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001981806759830189,
+      "loss": 1.4517,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019816765090772113,
+      "loss": 1.4419,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019815457980450766,
+      "loss": 1.3935,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001981414626795072,
+      "loss": 1.1211,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019812829953886994,
+      "loss": 1.3516,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019811509038876772,
+      "loss": 1.5352,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019810183523539393,
+      "loss": 1.33,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019808853408496347,
+      "loss": 1.4587,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001980751869437129,
+      "loss": 1.32,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001980617938179003,
+      "loss": 1.4307,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001980483547138053,
+      "loss": 1.449,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019803486963772908,
+      "loss": 1.1975,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019802133859599444,
+      "loss": 1.4869,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019800776159494568,
+      "loss": 1.2179,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001979941386409486,
+      "loss": 1.5145,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019798046974039068,
+      "loss": 1.3346,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001979667548996808,
+      "loss": 1.5098,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019795299412524945,
+      "loss": 1.3657,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001979391874235487,
+      "loss": 1.3052,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019792533480105203,
+      "loss": 1.1437,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019791143626425464,
+      "loss": 1.4016,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019789749181967303,
+      "loss": 1.4929,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019788350147384538,
+      "loss": 1.3315,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019786946523333136,
+      "loss": 1.2659,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019785538310471213,
+      "loss": 1.1567,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019784125509459042,
+      "loss": 1.4537,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001978270812095904,
+      "loss": 1.314,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019781286145635784,
+      "loss": 1.1826,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019779859584155985,
+      "loss": 1.1266,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001977842843718853,
+      "loss": 1.4691,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001977699270540443,
+      "loss": 1.4814,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019775552389476864,
+      "loss": 1.1741,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001977410749008115,
+      "loss": 1.5014,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019772658007894761,
+      "loss": 1.3707,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019771203943597313,
+      "loss": 1.2719,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001976974529787058,
+      "loss": 1.6717,
+      "step": 414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019768282071398475,
+      "loss": 1.5176,
+      "step": 415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019766814264867056,
+      "loss": 1.2179,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019765341878964546,
+      "loss": 1.3192,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001976386491438129,
+      "loss": 1.3347,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019762383371809805,
+      "loss": 1.4218,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019760897251944737,
+      "loss": 1.1055,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001975940655548288,
+      "loss": 1.2017,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019757911283123188,
+      "loss": 1.4414,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019756411435566739,
+      "loss": 1.347,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019754907013516774,
+      "loss": 1.1534,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001975339801767867,
+      "loss": 1.3568,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001975188444875995,
+      "loss": 1.3925,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019750366307470283,
+      "loss": 1.117,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019748843594521483,
+      "loss": 1.3123,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000197473163106275,
+      "loss": 1.3439,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019745784456504434,
+      "loss": 1.4131,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001974424803287053,
+      "loss": 1.5072,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019742707040446166,
+      "loss": 1.3618,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001974116147995387,
+      "loss": 1.2824,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019739611352118311,
+      "loss": 1.4661,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000197380566576663,
+      "loss": 1.4706,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019736497397326787,
+      "loss": 1.4687,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019734933571830858,
+      "loss": 1.3718,
+      "step": 437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019733365181911748,
+      "loss": 1.4003,
+      "step": 438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001973179222830483,
+      "loss": 1.3384,
+      "step": 439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001973021471174761,
+      "loss": 1.4246,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019728632632979746,
+      "loss": 1.5286,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019727045992743025,
+      "loss": 1.4301,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019725454791781377,
+      "loss": 1.5251,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019723859030840868,
+      "loss": 1.4184,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019722258710669704,
+      "loss": 1.4078,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019720653832018225,
+      "loss": 1.318,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019719044395638915,
+      "loss": 1.4852,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001971743040228639,
+      "loss": 1.4642,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019715811852717404,
+      "loss": 1.288,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019714188747690847,
+      "loss": 1.2902,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019712561087967742,
+      "loss": 1.3431,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019710928874311256,
+      "loss": 1.4854,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019709292107486683,
+      "loss": 1.34,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019707650788261457,
+      "loss": 1.1896,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019706004917405137,
+      "loss": 1.4435,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019704354495689438,
+      "loss": 1.2793,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019702699523888178,
+      "loss": 1.2755,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019701040002777332,
+      "loss": 1.3251,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019699375933134996,
+      "loss": 1.3081,
+      "step": 459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019697707315741408,
+      "loss": 1.3727,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019696034151378937,
+      "loss": 1.4429,
+      "step": 461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019694356440832074,
+      "loss": 1.4455,
+      "step": 462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019692674184887448,
+      "loss": 1.2044,
+      "step": 463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001969098738433382,
+      "loss": 1.4188,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019689296039962083,
+      "loss": 1.3443,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001968760015256526,
+      "loss": 1.3237,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019685899722938496,
+      "loss": 1.2883,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019684194751879076,
+      "loss": 1.1434,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019682485240186414,
+      "loss": 1.3259,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019680771188662044,
+      "loss": 1.259,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019679052598109637,
+      "loss": 1.4396,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001967732946933499,
+      "loss": 1.1806,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019675601803146025,
+      "loss": 1.3035,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019673869600352796,
+      "loss": 1.4981,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019672132861767478,
+      "loss": 1.2473,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019670391588204382,
+      "loss": 1.2557,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019668645780479937,
+      "loss": 1.2101,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000196668954394127,
+      "loss": 1.4006,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019665140565823354,
+      "loss": 1.4939,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019663381160534712,
+      "loss": 1.5405,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019661617224371703,
+      "loss": 1.2365,
+      "step": 481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019659848758161381,
+      "loss": 1.1786,
+      "step": 482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019658075762732935,
+      "loss": 1.1711,
+      "step": 483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019656298238917669,
+      "loss": 1.5076,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019654516187549007,
+      "loss": 1.3375,
+      "step": 485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019652729609462505,
+      "loss": 1.3006,
+      "step": 486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019650938505495836,
+      "loss": 1.5258,
+      "step": 487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019649142876488793,
+      "loss": 1.5839,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019647342723283292,
+      "loss": 1.291,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019645538046723378,
+      "loss": 1.4815,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001964372884765521,
+      "loss": 1.4554,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001964191512692706,
+      "loss": 1.0798,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019640096885389338,
+      "loss": 1.3211,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001963827412389456,
+      "loss": 1.249,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019636446843297362,
+      "loss": 1.3174,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019634615044454505,
+      "loss": 1.4271,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019632778728224866,
+      "loss": 1.6242,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001963093789546944,
+      "loss": 1.3387,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019629092547051333,
+      "loss": 1.2556,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019627242683835784,
+      "loss": 1.2534,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001962538830669013,
+      "loss": 1.4623,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019623529416483844,
+      "loss": 1.3385,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019621666014088494,
+      "loss": 1.1915,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001961979810037778,
+      "loss": 1.254,
+      "step": 504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019617925676227514,
+      "loss": 1.1632,
+      "step": 505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019616048742515616,
+      "loss": 1.2422,
+      "step": 506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019614167300122126,
+      "loss": 1.3515,
+      "step": 507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019612281349929198,
+      "loss": 1.6473,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019610390892821095,
+      "loss": 1.5952,
+      "step": 509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019608495929684204,
+      "loss": 1.3409,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019606596461407009,
+      "loss": 1.4643,
+      "step": 511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019604692488880117,
+      "loss": 1.2316,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019602784012996246,
+      "loss": 1.2831,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019600871034650225,
+      "loss": 1.157,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019598953554738992,
+      "loss": 1.4214,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019597031574161597,
+      "loss": 1.1743,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019595105093819199,
+      "loss": 1.3691,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019593174114615068,
+      "loss": 1.4748,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019591238637454582,
+      "loss": 1.3636,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019589298663245233,
+      "loss": 1.27,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019587354192896617,
+      "loss": 1.4354,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019585405227320435,
+      "loss": 1.5259,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019583451767430502,
+      "loss": 1.4016,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019581493814142738,
+      "loss": 1.4925,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019579531368375168,
+      "loss": 1.3406,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019577564431047927,
+      "loss": 1.0784,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001957559300308325,
+      "loss": 1.3967,
+      "step": 527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019573617085405487,
+      "loss": 1.506,
+      "step": 528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019571636678941082,
+      "loss": 1.3839,
+      "step": 529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019569651784618594,
+      "loss": 1.6833,
+      "step": 530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019567662403368677,
+      "loss": 1.399,
+      "step": 531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019565668536124098,
+      "loss": 1.1977,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019563670183819721,
+      "loss": 1.2918,
+      "step": 533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019561667347392508,
+      "loss": 1.5529,
+      "step": 534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001955966002778154,
+      "loss": 1.4461,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019557648225927986,
+      "loss": 1.1558,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019555631942775117,
+      "loss": 1.3226,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001955361117926831,
+      "loss": 1.3726,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019551585936355048,
+      "loss": 1.3469,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019549556214984898,
+      "loss": 1.3815,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019547522016109543,
+      "loss": 1.4868,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954548334068276,
+      "loss": 1.0253,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019543440189660416,
+      "loss": 1.2434,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019541392564000488,
+      "loss": 1.2296,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019539340464663052,
+      "loss": 1.6079,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019537283892610268,
+      "loss": 1.4194,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019535222848806414,
+      "loss": 1.1505,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019533157334217843,
+      "loss": 1.4645,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019531087349813016,
+      "loss": 1.3686,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019529012896562486,
+      "loss": 1.2075,
+      "step": 550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019526933975438907,
+      "loss": 1.344,
+      "step": 551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019524850587417022,
+      "loss": 1.2596,
+      "step": 552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001952276273347367,
+      "loss": 1.1239,
+      "step": 553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019520670414587782,
+      "loss": 1.3228,
+      "step": 554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951857363174039,
+      "loss": 1.3776,
+      "step": 555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019516472385914608,
+      "loss": 1.4115,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951436667809565,
+      "loss": 1.4728,
+      "step": 557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019512256509270822,
+      "loss": 1.3402,
+      "step": 558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019510141880429519,
+      "loss": 1.5459,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001950802279256323,
+      "loss": 1.4809,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019505899246665526,
+      "loss": 1.219,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019503771243732082,
+      "loss": 1.3116,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019501638784760652,
+      "loss": 1.164,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019499501870751088,
+      "loss": 1.4821,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001949736050270532,
+      "loss": 1.3288,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019495214681627374,
+      "loss": 1.3373,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019493064408523364,
+      "loss": 1.1355,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001949090968440149,
+      "loss": 1.4333,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019488750510272037,
+      "loss": 1.3623,
+      "step": 569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019486586887147378,
+      "loss": 1.23,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001948441881604197,
+      "loss": 1.4201,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001948224629797236,
+      "loss": 1.6193,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001948006933395718,
+      "loss": 1.4618,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019477887925017142,
+      "loss": 1.3388,
+      "step": 574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019475702072175043,
+      "loss": 1.47,
+      "step": 575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019473511776455765,
+      "loss": 1.3976,
+      "step": 576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019471317038886277,
+      "loss": 1.2228,
+      "step": 577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001946911786049562,
+      "loss": 1.6908,
+      "step": 578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001946691424231493,
+      "loss": 1.4054,
+      "step": 579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019464706185377416,
+      "loss": 1.2664,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001946249369071837,
+      "loss": 1.2545,
+      "step": 581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019460276759375165,
+      "loss": 1.5681,
+      "step": 582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001945805539238726,
+      "loss": 1.639,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019455829590796182,
+      "loss": 1.3944,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019453599355645547,
+      "loss": 1.4034,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019451364687981046,
+      "loss": 1.4483,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019449125588850446,
+      "loss": 1.3809,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019446882059303598,
+      "loss": 1.199,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001944463410039243,
+      "loss": 1.5345,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019442381713170935,
+      "loss": 1.2709,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019440124898695196,
+      "loss": 1.5108,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019437863658023366,
+      "loss": 1.5733,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019435597992215677,
+      "loss": 1.3626,
+      "step": 593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001943332790233443,
+      "loss": 1.2718,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019431053389443997,
+      "loss": 1.5351,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019428774454610843,
+      "loss": 1.2982,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019426491098903484,
+      "loss": 1.1998,
+      "step": 597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001942420332339252,
+      "loss": 1.2371,
+      "step": 598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019421911129150626,
+      "loss": 1.4122,
+      "step": 599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019419614517252534,
+      "loss": 1.3614,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001941731348877507,
+      "loss": 1.3438,
+      "step": 601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019415008044797107,
+      "loss": 1.2874,
+      "step": 602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001941269818639961,
+      "loss": 1.2912,
+      "step": 603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019410383914665597,
+      "loss": 1.4172,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001940806523068016,
+      "loss": 1.204,
+      "step": 605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019405742135530468,
+      "loss": 1.416,
+      "step": 606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019403414630305744,
+      "loss": 1.0793,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019401082716097296,
+      "loss": 1.0979,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019398746393998483,
+      "loss": 1.5641,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001939640566510474,
+      "loss": 1.3772,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019394060530513563,
+      "loss": 1.4949,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019391710991324522,
+      "loss": 1.4204,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019389357048639241,
+      "loss": 1.366,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019386998703561415,
+      "loss": 1.3435,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019384635957196808,
+      "loss": 1.5886,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019382268810653238,
+      "loss": 1.2406,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019379897265040587,
+      "loss": 1.3207,
+      "step": 617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019377521321470805,
+      "loss": 1.3138,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019375140981057904,
+      "loss": 1.2768,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019372756244917955,
+      "loss": 1.259,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001937036711416909,
+      "loss": 1.4693,
+      "step": 621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000193679735899315,
+      "loss": 1.3161,
+      "step": 622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019365575673327442,
+      "loss": 1.2069,
+      "step": 623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001936317336548122,
+      "loss": 1.1388,
+      "step": 624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019360766667519217,
+      "loss": 1.3786,
+      "step": 625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019358355580569855,
+      "loss": 1.4336,
+      "step": 626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001935594010576362,
+      "loss": 1.4619,
+      "step": 627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019353520244233066,
+      "loss": 1.4225,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019351095997112788,
+      "loss": 1.4785,
+      "step": 629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019348667365539442,
+      "loss": 1.1769,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019346234350651743,
+      "loss": 1.3008,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019343796953590466,
+      "loss": 1.3286,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019341355175498426,
+      "loss": 1.3772,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019338909017520509,
+      "loss": 1.5824,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019336458480803638,
+      "loss": 1.3823,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019334003566496804,
+      "loss": 1.4937,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001933154427575104,
+      "loss": 1.2706,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019329080609719435,
+      "loss": 1.4844,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019326612569557132,
+      "loss": 1.437,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019324140156421323,
+      "loss": 1.2783,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019321663371471245,
+      "loss": 1.3026,
+      "step": 641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019319182215868193,
+      "loss": 1.3505,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019316696690775502,
+      "loss": 1.4792,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001931420679735857,
+      "loss": 1.4992,
+      "step": 644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019311712536784832,
+      "loss": 1.2387,
+      "step": 645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001930921391022377,
+      "loss": 1.075,
+      "step": 646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019306710918846914,
+      "loss": 1.6459,
+      "step": 647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001930420356382785,
+      "loss": 1.2903,
+      "step": 648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019301691846342197,
+      "loss": 1.5642,
+      "step": 649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019299175767567635,
+      "loss": 1.1244,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019296655328683864,
+      "loss": 1.3721,
+      "step": 651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019294130530872656,
+      "loss": 1.4528,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019291601375317807,
+      "loss": 1.3534,
+      "step": 653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019289067863205165,
+      "loss": 1.3614,
+      "step": 654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019286529995722623,
+      "loss": 1.2653,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019283987774060108,
+      "loss": 1.2853,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019281441199409593,
+      "loss": 1.4053,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019278890272965096,
+      "loss": 1.3962,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019276334995922665,
+      "loss": 1.4482,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019273775369480395,
+      "loss": 1.1937,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019271211394838425,
+      "loss": 1.5829,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001926864307319892,
+      "loss": 1.4489,
+      "step": 662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019266070405766096,
+      "loss": 1.1634,
+      "step": 663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000192634933937462,
+      "loss": 1.2996,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019260912038347513,
+      "loss": 1.347,
+      "step": 665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019258326340780357,
+      "loss": 1.2672,
+      "step": 666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019255736302257095,
+      "loss": 1.4947,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019253141923992114,
+      "loss": 1.2732,
+      "step": 668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019250543207201842,
+      "loss": 1.4866,
+      "step": 669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001924794015310474,
+      "loss": 1.2846,
+      "step": 670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001924533276292131,
+      "loss": 1.4845,
+      "step": 671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019242721037874075,
+      "loss": 1.3548,
+      "step": 672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019240104979187595,
+      "loss": 1.2205,
+      "step": 673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019237484588088464,
+      "loss": 1.1225,
+      "step": 674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019234859865805307,
+      "loss": 1.353,
+      "step": 675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019232230813568773,
+      "loss": 1.3966,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019229597432611557,
+      "loss": 1.2891,
+      "step": 677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019226959724168367,
+      "loss": 1.4354,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019224317689475948,
+      "loss": 1.5464,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019221671329773074,
+      "loss": 1.2485,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001921902064630054,
+      "loss": 1.2931,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001921636564030118,
+      "loss": 1.1249,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019213706313019846,
+      "loss": 1.2673,
+      "step": 683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019211042665703414,
+      "loss": 1.4006,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019208374699600792,
+      "loss": 1.5024,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001920570241596291,
+      "loss": 1.4841,
+      "step": 686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019203025816042727,
+      "loss": 1.4847,
+      "step": 687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019200344901095222,
+      "loss": 1.391,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001919765967237739,
+      "loss": 1.4098,
+      "step": 689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001919497013114826,
+      "loss": 1.2918,
+      "step": 690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019192276278668874,
+      "loss": 1.5963,
+      "step": 691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019189578116202307,
+      "loss": 1.3048,
+      "step": 692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019186875645013646,
+      "loss": 1.1464,
+      "step": 693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019184168866369995,
+      "loss": 1.3087,
+      "step": 694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019181457781540487,
+      "loss": 1.4277,
+      "step": 695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019178742391796273,
+      "loss": 1.3276,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001917602269841051,
+      "loss": 1.6823,
+      "step": 697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001917329870265839,
+      "loss": 1.3506,
+      "step": 698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019170570405817113,
+      "loss": 1.4204,
+      "step": 699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001916783780916589,
+      "loss": 1.5691,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019165100913985964,
+      "loss": 1.1591,
+      "step": 701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019162359721560578,
+      "loss": 1.5427,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019159614233175,
+      "loss": 1.5062,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019156864450116512,
+      "loss": 1.2789,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019154110373674395,
+      "loss": 1.2373,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019151352005139963,
+      "loss": 1.2251,
+      "step": 706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019148589345806532,
+      "loss": 1.3109,
+      "step": 707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001914582239696943,
+      "loss": 1.1562,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019143051159926003,
+      "loss": 1.5951,
+      "step": 709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019140275635975594,
+      "loss": 1.2056,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019137495826419566,
+      "loss": 1.3293,
+      "step": 711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019134711732561295,
+      "loss": 1.2847,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019131923355706156,
+      "loss": 1.5616,
+      "step": 713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001912913069716154,
+      "loss": 1.3434,
+      "step": 714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001912633375823684,
+      "loss": 1.585,
+      "step": 715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019123532540243454,
+      "loss": 1.3745,
+      "step": 716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000191207270444948,
+      "loss": 1.1986,
+      "step": 717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019117917272306283,
+      "loss": 1.3848,
+      "step": 718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019115103224995328,
+      "loss": 1.3152,
+      "step": 719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001911228490388136,
+      "loss": 1.3506,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000191094623102858,
+      "loss": 1.2777,
+      "step": 721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019106635445532085,
+      "loss": 1.1251,
+      "step": 722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019103804310945644,
+      "loss": 1.3919,
+      "step": 723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019100968907853914,
+      "loss": 1.2643,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019098129237586335,
+      "loss": 1.3758,
+      "step": 725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001909528530147434,
+      "loss": 1.3364,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001909243710085137,
+      "loss": 1.4626,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019089584637052863,
+      "loss": 1.3768,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001908672791141625,
+      "loss": 1.3963,
+      "step": 729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001908386692528097,
+      "loss": 1.6673,
+      "step": 730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019081001679988453,
+      "loss": 1.5695,
+      "step": 731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019078132176882128,
+      "loss": 1.2874,
+      "step": 732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019075258417307423,
+      "loss": 1.6256,
+      "step": 733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019072380402611759,
+      "loss": 1.552,
+      "step": 734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019069498134144546,
+      "loss": 1.3225,
+      "step": 735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019066611613257202,
+      "loss": 1.2265,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019063720841303126,
+      "loss": 1.4763,
+      "step": 737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019060825819637716,
+      "loss": 1.2046,
+      "step": 738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019057926549618368,
+      "loss": 1.182,
+      "step": 739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001905502303260446,
+      "loss": 1.2386,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019052115269957366,
+      "loss": 1.4668,
+      "step": 741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001904920326304045,
+      "loss": 1.3732,
+      "step": 742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019046287013219066,
+      "loss": 1.3399,
+      "step": 743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019043366521860557,
+      "loss": 1.3884,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019040441790334258,
+      "loss": 1.4312,
+      "step": 745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019037512820011486,
+      "loss": 1.1485,
+      "step": 746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019034579612265553,
+      "loss": 1.452,
+      "step": 747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019031642168471747,
+      "loss": 1.3649,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019028700490007357,
+      "loss": 1.3386,
+      "step": 749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019025754578251637,
+      "loss": 1.4414,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019022804434585852,
+      "loss": 1.3633,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019019850060393231,
+      "loss": 1.1304,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019016891457058991,
+      "loss": 1.2424,
+      "step": 753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001901392862597034,
+      "loss": 1.2757,
+      "step": 754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019010961568516452,
+      "loss": 1.1852,
+      "step": 755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019007990286088502,
+      "loss": 1.3361,
+      "step": 756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019005014780079633,
+      "loss": 1.271,
+      "step": 757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019002035051884973,
+      "loss": 1.2148,
+      "step": 758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001899905110290163,
+      "loss": 1.3183,
+      "step": 759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018996062934528683,
+      "loss": 1.3641,
+      "step": 760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018993070548167204,
+      "loss": 1.3204,
+      "step": 761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001899007394522023,
+      "loss": 1.5319,
+      "step": 762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018987073127092781,
+      "loss": 1.3603,
+      "step": 763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018984068095191856,
+      "loss": 1.3878,
+      "step": 764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018981058850926416,
+      "loss": 1.2604,
+      "step": 765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018978045395707418,
+      "loss": 1.3741,
+      "step": 766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018975027730947774,
+      "loss": 1.4638,
+      "step": 767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018972005858062385,
+      "loss": 1.2516,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001896897977846811,
+      "loss": 1.3471,
+      "step": 769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018965949493583795,
+      "loss": 1.0956,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018962915004830247,
+      "loss": 1.2528,
+      "step": 771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018959876313630255,
+      "loss": 1.362,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018956833421408562,
+      "loss": 1.1317,
+      "step": 773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018953786329591895,
+      "loss": 1.3617,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018950735039608948,
+      "loss": 1.2047,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018947679552890377,
+      "loss": 1.6061,
+      "step": 776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018944619870868814,
+      "loss": 1.359,
+      "step": 777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018941555994978848,
+      "loss": 1.5402,
+      "step": 778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018938487926657043,
+      "loss": 1.4686,
+      "step": 779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018935415667341925,
+      "loss": 1.4052,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001893233921847399,
+      "loss": 1.4225,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018929258581495685,
+      "loss": 1.4834,
+      "step": 782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001892617375785144,
+      "loss": 1.1202,
+      "step": 783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001892308474898763,
+      "loss": 1.019,
+      "step": 784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018919991556352606,
+      "loss": 1.2792,
+      "step": 785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001891689418139667,
+      "loss": 1.1487,
+      "step": 786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018913792625572092,
+      "loss": 1.6332,
+      "step": 787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000189106868903331,
+      "loss": 1.1438,
+      "step": 788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001890757697713588,
+      "loss": 1.267,
+      "step": 789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001890446288743858,
+      "loss": 1.3624,
+      "step": 790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018901344622701305,
+      "loss": 1.1852,
+      "step": 791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018898222184386118,
+      "loss": 1.4427,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018895095573957037,
+      "loss": 1.2725,
+      "step": 793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018891964792880036,
+      "loss": 1.1068,
+      "step": 794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018888829842623052,
+      "loss": 1.4201,
+      "step": 795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001888569072465596,
+      "loss": 1.4044,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001888254744045061,
+      "loss": 1.5088,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018879399991480792,
+      "loss": 1.4335,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018876248379222252,
+      "loss": 1.477,
+      "step": 799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018873092605152685,
+      "loss": 1.537,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018869932670751742,
+      "loss": 1.4241,
+      "step": 801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018866768577501027,
+      "loss": 1.288,
+      "step": 802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018863600326884082,
+      "loss": 1.5275,
+      "step": 803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018860427920386417,
+      "loss": 1.142,
+      "step": 804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018857251359495467,
+      "loss": 1.2301,
+      "step": 805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001885407064570064,
+      "loss": 1.2561,
+      "step": 806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001885088578049327,
+      "loss": 1.4862,
+      "step": 807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001884769676536665,
+      "loss": 1.4809,
+      "step": 808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018844503601816016,
+      "loss": 1.352,
+      "step": 809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018841306291338548,
+      "loss": 1.4796,
+      "step": 810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001883810483543337,
+      "loss": 1.269,
+      "step": 811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018834899235601547,
+      "loss": 1.4198,
+      "step": 812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018831689493346095,
+      "loss": 1.1933,
+      "step": 813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018828475610171963,
+      "loss": 1.2836,
+      "step": 814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001882525758758605,
+      "loss": 1.346,
+      "step": 815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018822035427097192,
+      "loss": 1.3758,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001881880913021616,
+      "loss": 1.7343,
+      "step": 817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018815578698455675,
+      "loss": 1.296,
+      "step": 818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018812344133330391,
+      "loss": 1.2553,
+      "step": 819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018809105436356894,
+      "loss": 1.2906,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001880586260905372,
+      "loss": 1.6559,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001880261565294133,
+      "loss": 1.2458,
+      "step": 822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001879936456954213,
+      "loss": 1.1819,
+      "step": 823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018796109360380458,
+      "loss": 1.1734,
+      "step": 824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001879285002698258,
+      "loss": 1.609,
+      "step": 825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018789586570876706,
+      "loss": 1.4253,
+      "step": 826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018786318993592972,
+      "loss": 1.2763,
+      "step": 827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018783047296663452,
+      "loss": 1.362,
+      "step": 828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018779771481622145,
+      "loss": 1.214,
+      "step": 829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018776491550004983,
+      "loss": 1.1917,
+      "step": 830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018773207503349832,
+      "loss": 1.2842,
+      "step": 831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018769919343196483,
+      "loss": 1.1538,
+      "step": 832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018766627071086655,
+      "loss": 1.1884,
+      "step": 833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018763330688564004,
+      "loss": 1.566,
+      "step": 834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018760030197174102,
+      "loss": 1.3903,
+      "step": 835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001875672559846445,
+      "loss": 1.5377,
+      "step": 836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018753416893984478,
+      "loss": 1.2489,
+      "step": 837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018750104085285546,
+      "loss": 1.3959,
+      "step": 838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018746787173920927,
+      "loss": 1.3558,
+      "step": 839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018743466161445823,
+      "loss": 1.3097,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018740141049417355,
+      "loss": 1.4625,
+      "step": 841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001873681183939458,
+      "loss": 1.2441,
+      "step": 842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001873347853293846,
+      "loss": 1.3556,
+      "step": 843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018730141131611882,
+      "loss": 1.4471,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018726799636979659,
+      "loss": 1.4405,
+      "step": 845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018723454050608516,
+      "loss": 1.3663,
+      "step": 846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018720104374067104,
+      "loss": 1.4557,
+      "step": 847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018716750608925987,
+      "loss": 1.2564,
+      "step": 848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018713392756757643,
+      "loss": 1.3068,
+      "step": 849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001871003081913647,
+      "loss": 1.3891,
+      "step": 850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018706664797638785,
+      "loss": 1.3352,
+      "step": 851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018703294693842812,
+      "loss": 1.3482,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000186999205093287,
+      "loss": 1.2049,
+      "step": 853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018696542245678495,
+      "loss": 1.3008,
+      "step": 854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018693159904476173,
+      "loss": 1.3547,
+      "step": 855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001868977348730761,
+      "loss": 1.3183,
+      "step": 856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018686382995760598,
+      "loss": 1.1686,
+      "step": 857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001868298843142484,
+      "loss": 1.1715,
+      "step": 858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018679589795891945,
+      "loss": 1.4196,
+      "step": 859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018676187090755435,
+      "loss": 1.4057,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018672780317610736,
+      "loss": 1.2263,
+      "step": 861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018669369478055184,
+      "loss": 1.3298,
+      "step": 862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018665954573688024,
+      "loss": 1.3387,
+      "step": 863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000186625356061104,
+      "loss": 1.4492,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018659112576925363,
+      "loss": 1.3176,
+      "step": 865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018655685487737876,
+      "loss": 1.3837,
+      "step": 866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000186522543401548,
+      "loss": 1.5514,
+      "step": 867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018648819135784893,
+      "loss": 1.421,
+      "step": 868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018645379876238826,
+      "loss": 1.576,
+      "step": 869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018641936563129163,
+      "loss": 1.5007,
+      "step": 870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018638489198070376,
+      "loss": 1.4116,
+      "step": 871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018635037782678833,
+      "loss": 1.6024,
+      "step": 872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018631582318572796,
+      "loss": 1.3328,
+      "step": 873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018628122807372432,
+      "loss": 1.3297,
+      "step": 874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018624659250699805,
+      "loss": 1.4165,
+      "step": 875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018621191650178875,
+      "loss": 1.4214,
+      "step": 876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018617720007435497,
+      "loss": 1.4169,
+      "step": 877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001861424432409742,
+      "loss": 1.1858,
+      "step": 878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018610764601794292,
+      "loss": 1.3243,
+      "step": 879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001860728084215765,
+      "loss": 1.3077,
+      "step": 880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001860379304682093,
+      "loss": 1.5214,
+      "step": 881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001860030121741945,
+      "loss": 1.3361,
+      "step": 882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018596805355590432,
+      "loss": 1.3517,
+      "step": 883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018593305462972977,
+      "loss": 1.306,
+      "step": 884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001858980154120809,
+      "loss": 1.266,
+      "step": 885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018586293591938647,
+      "loss": 1.425,
+      "step": 886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018582781616809424,
+      "loss": 1.3522,
+      "step": 887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018579265617467086,
+      "loss": 1.3204,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018575745595560177,
+      "loss": 1.1705,
+      "step": 889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018572221552739138,
+      "loss": 1.5063,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018568693490656282,
+      "loss": 1.3798,
+      "step": 891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018565161410965817,
+      "loss": 1.2199,
+      "step": 892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018561625315323833,
+      "loss": 1.3058,
+      "step": 893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001855808520538829,
+      "loss": 1.027,
+      "step": 894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018554541082819057,
+      "loss": 1.2991,
+      "step": 895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018550992949277858,
+      "loss": 1.5249,
+      "step": 896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018547440806428308,
+      "loss": 1.4381,
+      "step": 897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001854388465593591,
+      "loss": 1.4692,
+      "step": 898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001854032449946803,
+      "loss": 1.2554,
+      "step": 899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018536760338693924,
+      "loss": 1.1782,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018533192175284718,
+      "loss": 1.3458,
+      "step": 901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018529620010913426,
+      "loss": 1.6592,
+      "step": 902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001852604384725492,
+      "loss": 1.4548,
+      "step": 903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001852246368598597,
+      "loss": 1.534,
+      "step": 904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000185188795287852,
+      "loss": 1.2578,
+      "step": 905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018515291377333112,
+      "loss": 1.2049,
+      "step": 906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018511699233312096,
+      "loss": 1.2617,
+      "step": 907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018508103098406395,
+      "loss": 1.4421,
+      "step": 908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018504502974302128,
+      "loss": 1.2185,
+      "step": 909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018500898862687293,
+      "loss": 1.2258,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001849729076525175,
+      "loss": 1.478,
+      "step": 911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001849367868368723,
+      "loss": 1.4848,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018490062619687327,
+      "loss": 1.204,
+      "step": 913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018486442574947511,
+      "loss": 1.4108,
+      "step": 914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001848281855116511,
+      "loss": 1.3489,
+      "step": 915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018479190550039332,
+      "loss": 1.3842,
+      "step": 916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018475558573271227,
+      "loss": 1.2441,
+      "step": 917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018471922622563728,
+      "loss": 1.299,
+      "step": 918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018468282699621625,
+      "loss": 1.2675,
+      "step": 919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018464638806151574,
+      "loss": 1.5702,
+      "step": 920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018460990943862079,
+      "loss": 1.4698,
+      "step": 921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018457339114463523,
+      "loss": 1.4438,
+      "step": 922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001845368331966814,
+      "loss": 1.2813,
+      "step": 923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018450023561190023,
+      "loss": 1.184,
+      "step": 924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018446359840745125,
+      "loss": 1.4716,
+      "step": 925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018442692160051258,
+      "loss": 1.4517,
+      "step": 926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018439020520828088,
+      "loss": 1.454,
+      "step": 927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018435344924797134,
+      "loss": 1.4315,
+      "step": 928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018431665373681787,
+      "loss": 1.198,
+      "step": 929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018427981869207266,
+      "loss": 1.4456,
+      "step": 930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018424294413100665,
+      "loss": 1.307,
+      "step": 931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001842060300709092,
+      "loss": 1.4983,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018416907652908825,
+      "loss": 1.1479,
+      "step": 933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018413208352287022,
+      "loss": 1.3698,
+      "step": 934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018409505106960005,
+      "loss": 1.3,
+      "step": 935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018405797918664117,
+      "loss": 1.2168,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018402086789137546,
+      "loss": 1.3893,
+      "step": 937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001839837172012033,
+      "loss": 1.2419,
+      "step": 938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018394652713354362,
+      "loss": 1.3838,
+      "step": 939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018390929770583365,
+      "loss": 1.3255,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018387202893552926,
+      "loss": 1.2985,
+      "step": 941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018383472084010464,
+      "loss": 1.0394,
+      "step": 942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018379737343705247,
+      "loss": 1.1666,
+      "step": 943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001837599867438838,
+      "loss": 1.2492,
+      "step": 944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018372256077812818,
+      "loss": 1.3293,
+      "step": 945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018368509555733359,
+      "loss": 1.4657,
+      "step": 946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001836475910990663,
+      "loss": 1.3864,
+      "step": 947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018361004742091104,
+      "loss": 1.184,
+      "step": 948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018357246454047096,
+      "loss": 1.2917,
+      "step": 949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018353484247536758,
+      "loss": 1.4229,
+      "step": 950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018349718124324076,
+      "loss": 1.192,
+      "step": 951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018345948086174871,
+      "loss": 1.309,
+      "step": 952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018342174134856806,
+      "loss": 1.1719,
+      "step": 953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018338396272139375,
+      "loss": 1.1143,
+      "step": 954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001833461449979391,
+      "loss": 1.2656,
+      "step": 955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018330828819593565,
+      "loss": 1.2463,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018327039233313337,
+      "loss": 1.3258,
+      "step": 957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018323245742730054,
+      "loss": 1.34,
+      "step": 958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018319448349622368,
+      "loss": 1.1604,
+      "step": 959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018315647055770767,
+      "loss": 1.2151,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018311841862957563,
+      "loss": 1.3412,
+      "step": 961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018308032772966902,
+      "loss": 1.3108,
+      "step": 962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001830421978758475,
+      "loss": 1.5548,
+      "step": 963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018300402908598908,
+      "loss": 1.4594,
+      "step": 964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018296582137798998,
+      "loss": 1.4365,
+      "step": 965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001829275747697646,
+      "loss": 1.0814,
+      "step": 966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001828892892792457,
+      "loss": 1.5619,
+      "step": 967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018285096492438424,
+      "loss": 1.254,
+      "step": 968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018281260172314928,
+      "loss": 1.4068,
+      "step": 969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018277419969352834,
+      "loss": 1.634,
+      "step": 970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018273575885352687,
+      "loss": 1.415,
+      "step": 971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001826972792211687,
+      "loss": 1.3569,
+      "step": 972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001826587608144958,
+      "loss": 1.4315,
+      "step": 973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001826202036515683,
+      "loss": 1.2855,
+      "step": 974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001825816077504645,
+      "loss": 1.4127,
+      "step": 975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018254297312928093,
+      "loss": 1.2949,
+      "step": 976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018250429980613218,
+      "loss": 1.4985,
+      "step": 977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018246558779915105,
+      "loss": 1.0886,
+      "step": 978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018242683712648843,
+      "loss": 1.2773,
+      "step": 979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001823880478063134,
+      "loss": 1.237,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018234921985681312,
+      "loss": 1.5661,
+      "step": 981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018231035329619287,
+      "loss": 1.4495,
+      "step": 982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018227144814267603,
+      "loss": 1.2687,
+      "step": 983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001822325044145041,
+      "loss": 1.2093,
+      "step": 984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001821935221299366,
+      "loss": 1.4216,
+      "step": 985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018215450130725124,
+      "loss": 1.4153,
+      "step": 986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018211544196474364,
+      "loss": 1.2756,
+      "step": 987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018207634412072764,
+      "loss": 1.3412,
+      "step": 988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018203720779353506,
+      "loss": 1.4588,
+      "step": 989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018199803300151577,
+      "loss": 1.2848,
+      "step": 990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018195881976303764,
+      "loss": 1.4384,
+      "step": 991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018191956809648663,
+      "loss": 1.339,
+      "step": 992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018188027802026668,
+      "loss": 1.4205,
+      "step": 993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018184094955279979,
+      "loss": 1.3135,
+      "step": 994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018180158271252582,
+      "loss": 1.3907,
+      "step": 995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001817621775179028,
+      "loss": 1.2911,
+      "step": 996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018172273398740668,
+      "loss": 1.074,
+      "step": 997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001816832521395313,
+      "loss": 1.0686,
+      "step": 998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018164373199278856,
+      "loss": 1.5471,
+      "step": 999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001816041735657083,
+      "loss": 1.2625,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018156457687683832,
+      "loss": 1.2167,
+      "step": 1001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001815249419447443,
+      "loss": 1.3692,
+      "step": 1002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018148526878800992,
+      "loss": 1.1092,
+      "step": 1003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018144555742523671,
+      "loss": 1.4075,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001814058078750442,
+      "loss": 1.4667,
+      "step": 1005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018136602015606976,
+      "loss": 1.349,
+      "step": 1006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001813261942869687,
+      "loss": 1.1715,
+      "step": 1007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018128633028641414,
+      "loss": 1.343,
+      "step": 1008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018124642817309722,
+      "loss": 1.2903,
+      "step": 1009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001812064879657268,
+      "loss": 1.4198,
+      "step": 1010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018116650968302966,
+      "loss": 1.3107,
+      "step": 1011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001811264933437505,
+      "loss": 1.573,
+      "step": 1012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018108643896665172,
+      "loss": 1.3994,
+      "step": 1013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001810463465705137,
+      "loss": 1.2693,
+      "step": 1014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018100621617413455,
+      "loss": 1.3185,
+      "step": 1015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018096604779633025,
+      "loss": 1.5288,
+      "step": 1016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018092584145593455,
+      "loss": 1.3129,
+      "step": 1017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000180885597171799,
+      "loss": 1.4255,
+      "step": 1018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018084531496279297,
+      "loss": 1.4408,
+      "step": 1019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018080499484780368,
+      "loss": 1.4807,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001807646368457359,
+      "loss": 1.4738,
+      "step": 1021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001807242409755124,
+      "loss": 1.3143,
+      "step": 1022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018068380725607362,
+      "loss": 1.2862,
+      "step": 1023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018064333570637765,
+      "loss": 1.3817,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018060282634540053,
+      "loss": 0.9002,
+      "step": 1025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001805622791921358,
+      "loss": 1.4231,
+      "step": 1026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018052169426559486,
+      "loss": 1.4664,
+      "step": 1027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001804810715848068,
+      "loss": 1.4685,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018044041116881837,
+      "loss": 1.1508,
+      "step": 1029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018039971303669407,
+      "loss": 1.157,
+      "step": 1030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018035897720751603,
+      "loss": 1.3468,
+      "step": 1031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001803182037003841,
+      "loss": 1.5237,
+      "step": 1032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018027739253441573,
+      "loss": 1.3218,
+      "step": 1033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001802365437287461,
+      "loss": 1.4045,
+      "step": 1034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018019565730252804,
+      "loss": 1.3296,
+      "step": 1035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018015473327493195,
+      "loss": 1.2342,
+      "step": 1036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018011377166514593,
+      "loss": 1.4577,
+      "step": 1037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018007277249237563,
+      "loss": 1.492,
+      "step": 1038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018003173577584437,
+      "loss": 1.2304,
+      "step": 1039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017999066153479308,
+      "loss": 1.2945,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017994954978848023,
+      "loss": 1.4534,
+      "step": 1041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017990840055618192,
+      "loss": 1.2989,
+      "step": 1042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017986721385719178,
+      "loss": 1.2317,
+      "step": 1043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001798259897108211,
+      "loss": 1.4349,
+      "step": 1044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017978472813639859,
+      "loss": 1.3465,
+      "step": 1045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017974342915327068,
+      "loss": 1.6644,
+      "step": 1046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017970209278080114,
+      "loss": 1.1605,
+      "step": 1047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017966071903837146,
+      "loss": 1.1496,
+      "step": 1048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017961930794538052,
+      "loss": 1.261,
+      "step": 1049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017957785952124478,
+      "loss": 1.36,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017953637378539817,
+      "loss": 1.3873,
+      "step": 1051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017949485075729213,
+      "loss": 1.3573,
+      "step": 1052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017945329045639558,
+      "loss": 1.3407,
+      "step": 1053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017941169290219494,
+      "loss": 1.32,
+      "step": 1054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017937005811419408,
+      "loss": 1.304,
+      "step": 1055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017932838611191428,
+      "loss": 1.2948,
+      "step": 1056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017928667691489433,
+      "loss": 1.3517,
+      "step": 1057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017924493054269046,
+      "loss": 1.4746,
+      "step": 1058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017920314701487633,
+      "loss": 1.3573,
+      "step": 1059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017916132635104298,
+      "loss": 1.6497,
+      "step": 1060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017911946857079888,
+      "loss": 1.2851,
+      "step": 1061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017907757369376985,
+      "loss": 1.4312,
+      "step": 1062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001790356417395993,
+      "loss": 1.5041,
+      "step": 1063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017899367272794777,
+      "loss": 1.3559,
+      "step": 1064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017895166667849334,
+      "loss": 1.5288,
+      "step": 1065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017890962361093139,
+      "loss": 1.2793,
+      "step": 1066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017886754354497472,
+      "loss": 1.3869,
+      "step": 1067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017882542650035334,
+      "loss": 1.3138,
+      "step": 1068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017878327249681476,
+      "loss": 1.3121,
+      "step": 1069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017874108155412375,
+      "loss": 1.2116,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017869885369206236,
+      "loss": 1.3774,
+      "step": 1071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017865658893043005,
+      "loss": 1.481,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017861428728904345,
+      "loss": 1.5106,
+      "step": 1073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017857194878773656,
+      "loss": 1.5332,
+      "step": 1074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017852957344636073,
+      "loss": 1.3154,
+      "step": 1075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001784871612847844,
+      "loss": 1.4778,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017844471232289347,
+      "loss": 1.3508,
+      "step": 1077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017840222658059096,
+      "loss": 1.2339,
+      "step": 1078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001783597040777972,
+      "loss": 1.307,
+      "step": 1079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017831714483444974,
+      "loss": 1.3228,
+      "step": 1080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017827454887050332,
+      "loss": 1.4146,
+      "step": 1081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017823191620593,
+      "loss": 1.4845,
+      "step": 1082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017818924686071886,
+      "loss": 1.4756,
+      "step": 1083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001781465408548764,
+      "loss": 1.5245,
+      "step": 1084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001781037982084262,
+      "loss": 1.6796,
+      "step": 1085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000178061018941409,
+      "loss": 1.3382,
+      "step": 1086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017801820307388272,
+      "loss": 1.4163,
+      "step": 1087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017797535062592248,
+      "loss": 1.4664,
+      "step": 1088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017793246161762052,
+      "loss": 1.3108,
+      "step": 1089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017788953606908626,
+      "loss": 1.5612,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001778465740004462,
+      "loss": 1.459,
+      "step": 1091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017780357543184397,
+      "loss": 1.2087,
+      "step": 1092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017776054038344034,
+      "loss": 1.3913,
+      "step": 1093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001777174688754132,
+      "loss": 1.2834,
+      "step": 1094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017767436092795745,
+      "loss": 1.4763,
+      "step": 1095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017763121656128524,
+      "loss": 1.2623,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017758803579562559,
+      "loss": 1.1324,
+      "step": 1097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017754481865122474,
+      "loss": 1.2854,
+      "step": 1098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001775015651483459,
+      "loss": 1.4576,
+      "step": 1099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017745827530726938,
+      "loss": 1.5175,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017741494914829252,
+      "loss": 1.438,
+      "step": 1101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017737158669172963,
+      "loss": 1.1,
+      "step": 1102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001773281879579122,
+      "loss": 1.154,
+      "step": 1103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017728475296718848,
+      "loss": 1.2672,
+      "step": 1104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017724128173992392,
+      "loss": 1.4924,
+      "step": 1105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017719777429650092,
+      "loss": 1.3804,
+      "step": 1106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017715423065731877,
+      "loss": 1.2627,
+      "step": 1107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017711065084279385,
+      "loss": 1.5176,
+      "step": 1108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017706703487335945,
+      "loss": 1.2709,
+      "step": 1109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017702338276946578,
+      "loss": 1.3873,
+      "step": 1110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017697969455158008,
+      "loss": 1.2115,
+      "step": 1111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001769359702401864,
+      "loss": 1.5822,
+      "step": 1112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001768922098557858,
+      "loss": 1.4614,
+      "step": 1113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017684841341889627,
+      "loss": 1.4419,
+      "step": 1114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001768045809500526,
+      "loss": 1.381,
+      "step": 1115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001767607124698066,
+      "loss": 1.398,
+      "step": 1116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017671680799872687,
+      "loss": 1.3646,
+      "step": 1117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017667286755739895,
+      "loss": 1.3362,
+      "step": 1118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001766288911664252,
+      "loss": 1.2492,
+      "step": 1119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017658487884642486,
+      "loss": 1.412,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000176540830618034,
+      "loss": 1.3433,
+      "step": 1121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017649674650190556,
+      "loss": 1.3165,
+      "step": 1122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017645262651870926,
+      "loss": 1.4109,
+      "step": 1123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017640847068913166,
+      "loss": 1.326,
+      "step": 1124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017636427903387612,
+      "loss": 1.434,
+      "step": 1125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001763200515736628,
+      "loss": 1.4028,
+      "step": 1126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017627578832922865,
+      "loss": 1.1727,
+      "step": 1127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017623148932132744,
+      "loss": 1.3004,
+      "step": 1128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017618715457072963,
+      "loss": 1.3148,
+      "step": 1129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001761427840982225,
+      "loss": 1.3297,
+      "step": 1130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017609837792461,
+      "loss": 1.2863,
+      "step": 1131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017605393607071296,
+      "loss": 1.107,
+      "step": 1132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017600945855736878,
+      "loss": 1.2169,
+      "step": 1133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001759649454054317,
+      "loss": 1.4083,
+      "step": 1134
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001759203966357726,
+      "loss": 1.2543,
+      "step": 1135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001758758122692791,
+      "loss": 1.215,
+      "step": 1136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001758311923268555,
+      "loss": 1.1605,
+      "step": 1137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017578653682942275,
+      "loss": 1.107,
+      "step": 1138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017574184579791848,
+      "loss": 1.1875,
+      "step": 1139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017569711925329707,
+      "loss": 1.1998,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001756523572165294,
+      "loss": 1.544,
+      "step": 1141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001756075597086031,
+      "loss": 1.2275,
+      "step": 1142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017556272675052238,
+      "loss": 1.332,
+      "step": 1143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017551785836330815,
+      "loss": 1.318,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001754729545679978,
+      "loss": 1.2209,
+      "step": 1145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017542801538564544,
+      "loss": 1.2301,
+      "step": 1146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017538304083732174,
+      "loss": 1.3096,
+      "step": 1147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001753380309441139,
+      "loss": 1.3432,
+      "step": 1148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017529298572712575,
+      "loss": 1.228,
+      "step": 1149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017524790520747764,
+      "loss": 1.7282,
+      "step": 1150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017520278940630655,
+      "loss": 1.3388,
+      "step": 1151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017515763834476586,
+      "loss": 1.5004,
+      "step": 1152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017511245204402567,
+      "loss": 1.3449,
+      "step": 1153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017506723052527242,
+      "loss": 1.4859,
+      "step": 1154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017502197380970925,
+      "loss": 1.2827,
+      "step": 1155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017497668191855556,
+      "loss": 1.3932,
+      "step": 1156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001749313548730475,
+      "loss": 1.406,
+      "step": 1157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017488599269443753,
+      "loss": 1.2545,
+      "step": 1158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017484059540399463,
+      "loss": 1.3898,
+      "step": 1159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001747951630230043,
+      "loss": 1.2447,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017474969557276836,
+      "loss": 1.472,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017470419307460524,
+      "loss": 1.3224,
+      "step": 1162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017465865554984968,
+      "loss": 1.1745,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001746130830198529,
+      "loss": 1.5307,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017456747550598247,
+      "loss": 1.2556,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017452183302962246,
+      "loss": 1.5292,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017447615561217328,
+      "loss": 1.4945,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001744304432750517,
+      "loss": 1.2399,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017438469603969087,
+      "loss": 1.4573,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017433891392754035,
+      "loss": 1.5827,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017429309696006599,
+      "loss": 1.5317,
+      "step": 1171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017424724515875006,
+      "loss": 1.2734,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001742013585450911,
+      "loss": 1.3448,
+      "step": 1173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017415543714060396,
+      "loss": 1.1183,
+      "step": 1174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017410948096681987,
+      "loss": 1.3791,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001740634900452863,
+      "loss": 1.4799,
+      "step": 1176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017401746439756706,
+      "loss": 1.3687,
+      "step": 1177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001739714040452422,
+      "loss": 1.3326,
+      "step": 1178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017392530900990806,
+      "loss": 1.2617,
+      "step": 1179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017387917931317723,
+      "loss": 1.3822,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017383301497667862,
+      "loss": 1.2806,
+      "step": 1181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017378681602205726,
+      "loss": 1.2379,
+      "step": 1182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001737405824709745,
+      "loss": 1.3678,
+      "step": 1183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017369431434510792,
+      "loss": 1.2725,
+      "step": 1184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017364801166615124,
+      "loss": 1.1707,
+      "step": 1185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017360167445581442,
+      "loss": 1.4593,
+      "step": 1186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017355530273582362,
+      "loss": 1.3689,
+      "step": 1187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001735088965279212,
+      "loss": 1.4481,
+      "step": 1188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001734624558538656,
+      "loss": 1.3016,
+      "step": 1189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017341598073543154,
+      "loss": 1.1466,
+      "step": 1190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001733694711944098,
+      "loss": 1.5006,
+      "step": 1191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017332292725260736,
+      "loss": 1.3168,
+      "step": 1192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017327634893184727,
+      "loss": 1.4329,
+      "step": 1193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001732297362539687,
+      "loss": 1.4672,
+      "step": 1194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001731830892408271,
+      "loss": 1.4762,
+      "step": 1195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017313640791429374,
+      "loss": 1.5273,
+      "step": 1196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017308969229625612,
+      "loss": 1.2902,
+      "step": 1197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001730429424086179,
+      "loss": 1.4199,
+      "step": 1198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017299615827329868,
+      "loss": 1.5259,
+      "step": 1199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017294933991223412,
+      "loss": 1.5053,
+      "step": 1200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017290248734737603,
+      "loss": 1.4796,
+      "step": 1201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017285560060069214,
+      "loss": 1.054,
+      "step": 1202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017280867969416634,
+      "loss": 1.2627,
+      "step": 1203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017276172464979837,
+      "loss": 1.0529,
+      "step": 1204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017271473548960417,
+      "loss": 1.3296,
+      "step": 1205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017266771223561548,
+      "loss": 1.0949,
+      "step": 1206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017262065490988016,
+      "loss": 1.4863,
+      "step": 1207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017257356353446194,
+      "loss": 1.3666,
+      "step": 1208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017252643813144068,
+      "loss": 1.2563,
+      "step": 1209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000172479278722912,
+      "loss": 1.5398,
+      "step": 1210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017243208533098762,
+      "loss": 1.3793,
+      "step": 1211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001723848579777951,
+      "loss": 1.2145,
+      "step": 1212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017233759668547794,
+      "loss": 1.3612,
+      "step": 1213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001722903014761956,
+      "loss": 1.3451,
+      "step": 1214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017224297237212333,
+      "loss": 1.1862,
+      "step": 1215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017219560939545246,
+      "loss": 1.2306,
+      "step": 1216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017214821256839003,
+      "loss": 1.2795,
+      "step": 1217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017210078191315904,
+      "loss": 1.5159,
+      "step": 1218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017205331745199828,
+      "loss": 1.0025,
+      "step": 1219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001720058192071625,
+      "loss": 1.4711,
+      "step": 1220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017195828720092217,
+      "loss": 1.2732,
+      "step": 1221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017191072145556366,
+      "loss": 1.4406,
+      "step": 1222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017186312199338918,
+      "loss": 1.4379,
+      "step": 1223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017181548883671668,
+      "loss": 1.2145,
+      "step": 1224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017176782200787995,
+      "loss": 1.3599,
+      "step": 1225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017172012152922856,
+      "loss": 1.3246,
+      "step": 1226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017167238742312783,
+      "loss": 1.2451,
+      "step": 1227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017162461971195894,
+      "loss": 1.3643,
+      "step": 1228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001715768184181187,
+      "loss": 1.1753,
+      "step": 1229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017152898356401973,
+      "loss": 1.5163,
+      "step": 1230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017148111517209042,
+      "loss": 1.1718,
+      "step": 1231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017143321326477481,
+      "loss": 1.5463,
+      "step": 1232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017138527786453276,
+      "loss": 1.3914,
+      "step": 1233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017133730899383966,
+      "loss": 1.1009,
+      "step": 1234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017128930667518674,
+      "loss": 1.1711,
+      "step": 1235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017124127093108094,
+      "loss": 1.3224,
+      "step": 1236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001711932017840447,
+      "loss": 1.2709,
+      "step": 1237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001711450992566163,
+      "loss": 1.2252,
+      "step": 1238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017109696337134952,
+      "loss": 1.5048,
+      "step": 1239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017104879415081394,
+      "loss": 1.3966,
+      "step": 1240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017100059161759466,
+      "loss": 1.4742,
+      "step": 1241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001709523557942924,
+      "loss": 1.3079,
+      "step": 1242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017090408670352353,
+      "loss": 1.3998,
+      "step": 1243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017085578436792,
+      "loss": 1.5939,
+      "step": 1244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017080744881012937,
+      "loss": 1.2065,
+      "step": 1245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001707590800528147,
+      "loss": 1.266,
+      "step": 1246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 1.5008,
+      "step": 1247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017066224303034373,
+      "loss": 1.1996,
+      "step": 1248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001706137748105914,
+      "loss": 1.4393,
+      "step": 1249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017056527348212311,
+      "loss": 1.2781,
+      "step": 1250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017051673906767967,
+      "loss": 1.6442,
+      "step": 1251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017046817159001746,
+      "loss": 1.4474,
+      "step": 1252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017041957107190834,
+      "loss": 1.3999,
+      "step": 1253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017037093753613966,
+      "loss": 1.305,
+      "step": 1254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001703222710055142,
+      "loss": 1.3147,
+      "step": 1255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017027357150285036,
+      "loss": 1.5337,
+      "step": 1256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017022483905098179,
+      "loss": 1.2923,
+      "step": 1257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017017607367275777,
+      "loss": 1.5389,
+      "step": 1258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017012727539104293,
+      "loss": 1.3086,
+      "step": 1259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017007844422871734,
+      "loss": 1.4933,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017002958020867652,
+      "loss": 1.3073,
+      "step": 1261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001699806833538313,
+      "loss": 1.2746,
+      "step": 1262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016993175368710804,
+      "loss": 1.0203,
+      "step": 1263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016988279123144836,
+      "loss": 1.1616,
+      "step": 1264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016983379600980938,
+      "loss": 1.4114,
+      "step": 1265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016978476804516346,
+      "loss": 1.0895,
+      "step": 1266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001697357073604984,
+      "loss": 1.3311,
+      "step": 1267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016968661397881728,
+      "loss": 1.5283,
+      "step": 1268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016963748792313854,
+      "loss": 1.3653,
+      "step": 1269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016958832921649597,
+      "loss": 1.3084,
+      "step": 1270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016953913788193858,
+      "loss": 1.3986,
+      "step": 1271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016948991394253083,
+      "loss": 1.2899,
+      "step": 1272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016944065742135225,
+      "loss": 0.9964,
+      "step": 1273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016939136834149786,
+      "loss": 1.4448,
+      "step": 1274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016934204672607782,
+      "loss": 1.339,
+      "step": 1275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016929269259821763,
+      "loss": 1.2805,
+      "step": 1276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016924330598105792,
+      "loss": 1.4053,
+      "step": 1277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016919388689775464,
+      "loss": 1.3803,
+      "step": 1278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016914443537147892,
+      "loss": 1.3496,
+      "step": 1279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001690949514254172,
+      "loss": 1.3279,
+      "step": 1280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016904543508277094,
+      "loss": 1.2899,
+      "step": 1281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016899588636675695,
+      "loss": 1.619,
+      "step": 1282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016894630530060717,
+      "loss": 1.1934,
+      "step": 1283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016889669190756868,
+      "loss": 1.4453,
+      "step": 1284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001688470462109037,
+      "loss": 1.2257,
+      "step": 1285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001687973682338897,
+      "loss": 1.2945,
+      "step": 1286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001687476579998192,
+      "loss": 1.3868,
+      "step": 1287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016869791553199984,
+      "loss": 1.3905,
+      "step": 1288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001686481408537544,
+      "loss": 1.5397,
+      "step": 1289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016859833398842077,
+      "loss": 1.3711,
+      "step": 1290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016854849495935188,
+      "loss": 1.1891,
+      "step": 1291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016849862378991585,
+      "loss": 1.3227,
+      "step": 1292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016844872050349578,
+      "loss": 1.2551,
+      "step": 1293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001683987851234898,
+      "loss": 1.3033,
+      "step": 1294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001683488176733112,
+      "loss": 1.6458,
+      "step": 1295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016829881817638818,
+      "loss": 1.2854,
+      "step": 1296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016824878665616408,
+      "loss": 1.3206,
+      "step": 1297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016819872313609716,
+      "loss": 1.4086,
+      "step": 1298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016814862763966083,
+      "loss": 1.2919,
+      "step": 1299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016809850019034323,
+      "loss": 1.263,
+      "step": 1300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016804834081164777,
+      "loss": 1.2774,
+      "step": 1301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016799814952709268,
+      "loss": 1.3222,
+      "step": 1302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001679479263602111,
+      "loss": 1.5479,
+      "step": 1303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016789767133455129,
+      "loss": 1.1457,
+      "step": 1304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001678473844736763,
+      "loss": 1.1009,
+      "step": 1305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016779706580116415,
+      "loss": 1.3292,
+      "step": 1306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016774671534060777,
+      "loss": 1.4957,
+      "step": 1307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016769633311561504,
+      "loss": 1.3609,
+      "step": 1308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001676459191498087,
+      "loss": 1.5986,
+      "step": 1309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016759547346682636,
+      "loss": 1.5753,
+      "step": 1310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001675449960903205,
+      "loss": 1.2072,
+      "step": 1311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016749448704395846,
+      "loss": 1.2716,
+      "step": 1312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001674439463514225,
+      "loss": 1.205,
+      "step": 1313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016739337403640957,
+      "loss": 1.3144,
+      "step": 1314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016734277012263164,
+      "loss": 1.0703,
+      "step": 1315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016729213463381534,
+      "loss": 1.2255,
+      "step": 1316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016724146759370214,
+      "loss": 1.5176,
+      "step": 1317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016719076902604836,
+      "loss": 1.422,
+      "step": 1318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000167140038954625,
+      "loss": 1.473,
+      "step": 1319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000167089277403218,
+      "loss": 1.5645,
+      "step": 1320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016703848439562785,
+      "loss": 1.1216,
+      "step": 1321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016698765995566995,
+      "loss": 1.2049,
+      "step": 1322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001669368041071744,
+      "loss": 1.38,
+      "step": 1323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016688591687398589,
+      "loss": 1.3724,
+      "step": 1324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016683499827996407,
+      "loss": 1.2202,
+      "step": 1325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016678404834898314,
+      "loss": 1.2665,
+      "step": 1326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000166733067104932,
+      "loss": 1.2341,
+      "step": 1327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016668205457171423,
+      "loss": 1.3822,
+      "step": 1328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001666310107732481,
+      "loss": 1.3867,
+      "step": 1329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001665799357334666,
+      "loss": 1.3465,
+      "step": 1330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016652882947631726,
+      "loss": 1.2221,
+      "step": 1331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016647769202576232,
+      "loss": 1.1887,
+      "step": 1332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016642652340577863,
+      "loss": 1.585,
+      "step": 1333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016637532364035762,
+      "loss": 1.2257,
+      "step": 1334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016632409275350534,
+      "loss": 1.3754,
+      "step": 1335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016627283076924249,
+      "loss": 1.3292,
+      "step": 1336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016622153771160425,
+      "loss": 1.2716,
+      "step": 1337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016617021360464045,
+      "loss": 1.2857,
+      "step": 1338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016611885847241545,
+      "loss": 1.4954,
+      "step": 1339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016606747233900815,
+      "loss": 1.353,
+      "step": 1340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016601605522851198,
+      "loss": 1.4431,
+      "step": 1341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016596460716503496,
+      "loss": 1.1736,
+      "step": 1342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001659131281726995,
+      "loss": 1.1343,
+      "step": 1343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016586161827564263,
+      "loss": 1.2244,
+      "step": 1344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016581007749801575,
+      "loss": 1.4564,
+      "step": 1345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001657585058639849,
+      "loss": 1.3401,
+      "step": 1346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016570690339773049,
+      "loss": 1.4598,
+      "step": 1347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016565527012344733,
+      "loss": 1.4148,
+      "step": 1348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016560360606534478,
+      "loss": 1.2535,
+      "step": 1349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016555191124764657,
+      "loss": 1.2001,
+      "step": 1350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001655001856945909,
+      "loss": 1.47,
+      "step": 1351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016544842943043035,
+      "loss": 1.4335,
+      "step": 1352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001653966424794319,
+      "loss": 1.1288,
+      "step": 1353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016534482486587693,
+      "loss": 1.3912,
+      "step": 1354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016529297661406118,
+      "loss": 1.133,
+      "step": 1355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016524109774829478,
+      "loss": 1.4559,
+      "step": 1356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001651891882929022,
+      "loss": 1.7451,
+      "step": 1357
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016513724827222227,
+      "loss": 1.2179,
+      "step": 1358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016508527771060805,
+      "loss": 1.3161,
+      "step": 1359
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016503327663242704,
+      "loss": 1.4139,
+      "step": 1360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001649812450620611,
+      "loss": 1.5517,
+      "step": 1361
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001649291830239062,
+      "loss": 1.3365,
+      "step": 1362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016487709054237274,
+      "loss": 1.3173,
+      "step": 1363
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016482496764188534,
+      "loss": 1.4017,
+      "step": 1364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016477281434688286,
+      "loss": 1.1664,
+      "step": 1365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001647206306818185,
+      "loss": 1.5537,
+      "step": 1366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016466841667115955,
+      "loss": 1.2685,
+      "step": 1367
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001646161723393877,
+      "loss": 1.6056,
+      "step": 1368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016456389771099878,
+      "loss": 1.1462,
+      "step": 1369
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016451159281050278,
+      "loss": 1.2643,
+      "step": 1370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016445925766242391,
+      "loss": 1.1056,
+      "step": 1371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001644068922913006,
+      "loss": 1.1597,
+      "step": 1372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016435449672168548,
+      "loss": 1.1643,
+      "step": 1373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016430207097814522,
+      "loss": 1.5693,
+      "step": 1374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016424961508526074,
+      "loss": 1.501,
+      "step": 1375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016419712906762702,
+      "loss": 1.2478,
+      "step": 1376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016414461294985322,
+      "loss": 1.2316,
+      "step": 1377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016409206675656264,
+      "loss": 1.3668,
+      "step": 1378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001640394905123926,
+      "loss": 1.3146,
+      "step": 1379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016398688424199454,
+      "loss": 1.1769,
+      "step": 1380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000163934247970034,
+      "loss": 1.4559,
+      "step": 1381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016388158172119064,
+      "loss": 1.4745,
+      "step": 1382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000163828885520158,
+      "loss": 1.4059,
+      "step": 1383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016377615939164387,
+      "loss": 1.2211,
+      "step": 1384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016372340336036988,
+      "loss": 1.3545,
+      "step": 1385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016367061745107186,
+      "loss": 1.7128,
+      "step": 1386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016361780168849954,
+      "loss": 1.3966,
+      "step": 1387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016356495609741667,
+      "loss": 1.3383,
+      "step": 1388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016351208070260098,
+      "loss": 1.4856,
+      "step": 1389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001634591755288442,
+      "loss": 1.3384,
+      "step": 1390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016340624060095195,
+      "loss": 1.2121,
+      "step": 1391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016335327594374394,
+      "loss": 1.4871,
+      "step": 1392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001633002815820537,
+      "loss": 1.6195,
+      "step": 1393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016324725754072872,
+      "loss": 1.2072,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001631942038446304,
+      "loss": 1.3441,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016314112051863408,
+      "loss": 1.3421,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016308800758762894,
+      "loss": 1.2423,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016303486507651807,
+      "loss": 1.2584,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016298169301021842,
+      "loss": 1.1673,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016292849141366083,
+      "loss": 1.3822,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016287526031178993,
+      "loss": 1.2357,
+      "step": 1401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016282199972956425,
+      "loss": 1.1926,
+      "step": 1402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016276870969195607,
+      "loss": 1.4876,
+      "step": 1403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001627153902239515,
+      "loss": 1.4294,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016266204135055053,
+      "loss": 1.3908,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016260866309676683,
+      "loss": 1.5708,
+      "step": 1406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001625552554876279,
+      "loss": 1.4247,
+      "step": 1407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000162501818548175,
+      "loss": 1.2475,
+      "step": 1408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016244835230346308,
+      "loss": 1.2387,
+      "step": 1409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016239485677856097,
+      "loss": 1.313,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016234133199855106,
+      "loss": 1.1244,
+      "step": 1411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016228777798852964,
+      "loss": 1.4072,
+      "step": 1412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016223419477360648,
+      "loss": 1.6203,
+      "step": 1413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016218058237890527,
+      "loss": 1.3461,
+      "step": 1414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016212694082956325,
+      "loss": 1.0983,
+      "step": 1415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016207327015073134,
+      "loss": 1.4656,
+      "step": 1416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001620195703675742,
+      "loss": 1.5248,
+      "step": 1417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016196584150527,
+      "loss": 1.4353,
+      "step": 1418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001619120835890106,
+      "loss": 1.3581,
+      "step": 1419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016185829664400162,
+      "loss": 1.4448,
+      "step": 1420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016180448069546207,
+      "loss": 1.3862,
+      "step": 1421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001617506357686247,
+      "loss": 1.3019,
+      "step": 1422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016169676188873582,
+      "loss": 1.4175,
+      "step": 1423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016164285908105524,
+      "loss": 1.3996,
+      "step": 1424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016158892737085646,
+      "loss": 1.2409,
+      "step": 1425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016153496678342644,
+      "loss": 1.3485,
+      "step": 1426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001614809773440657,
+      "loss": 1.3336,
+      "step": 1427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016142695907808835,
+      "loss": 1.3489,
+      "step": 1428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001613729120108219,
+      "loss": 1.4934,
+      "step": 1429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016131883616760746,
+      "loss": 1.1298,
+      "step": 1430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016126473157379958,
+      "loss": 1.3556,
+      "step": 1431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001612105982547663,
+      "loss": 1.4542,
+      "step": 1432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016115643623588915,
+      "loss": 1.266,
+      "step": 1433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016110224554256313,
+      "loss": 1.2662,
+      "step": 1434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001610480262001966,
+      "loss": 1.4906,
+      "step": 1435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016099377823421143,
+      "loss": 1.442,
+      "step": 1436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001609395016700429,
+      "loss": 1.2087,
+      "step": 1437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001608851965331397,
+      "loss": 1.2296,
+      "step": 1438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016083086284896393,
+      "loss": 1.2564,
+      "step": 1439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016077650064299097,
+      "loss": 1.3597,
+      "step": 1440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016072210994070972,
+      "loss": 1.3193,
+      "step": 1441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016066769076762238,
+      "loss": 1.53,
+      "step": 1442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016061324314924446,
+      "loss": 1.5225,
+      "step": 1443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016055876711110488,
+      "loss": 1.5777,
+      "step": 1444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001605042626787458,
+      "loss": 1.4051,
+      "step": 1445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016044972987772272,
+      "loss": 1.4165,
+      "step": 1446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001603951687336046,
+      "loss": 1.3286,
+      "step": 1447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016034057927197338,
+      "loss": 1.381,
+      "step": 1448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016028596151842456,
+      "loss": 1.1985,
+      "step": 1449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016023131549856677,
+      "loss": 1.2433,
+      "step": 1450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016017664123802188,
+      "loss": 1.3927,
+      "step": 1451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016012193876242505,
+      "loss": 1.2892,
+      "step": 1452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001600672080974247,
+      "loss": 1.3034,
+      "step": 1453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001600124492686824,
+      "loss": 1.4235,
+      "step": 1454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015995766230187294,
+      "loss": 1.0743,
+      "step": 1455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001599028472226843,
+      "loss": 1.3567,
+      "step": 1456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015984800405681769,
+      "loss": 1.4866,
+      "step": 1457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001597931328299874,
+      "loss": 1.4722,
+      "step": 1458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015973823356792098,
+      "loss": 1.2234,
+      "step": 1459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001596833062963591,
+      "loss": 1.1937,
+      "step": 1460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015962835104105545,
+      "loss": 1.5077,
+      "step": 1461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015957336782777696,
+      "loss": 1.2589,
+      "step": 1462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015951835668230366,
+      "loss": 1.5252,
+      "step": 1463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015946331763042867,
+      "loss": 1.2508,
+      "step": 1464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015940825069795812,
+      "loss": 1.1963,
+      "step": 1465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015935315591071133,
+      "loss": 1.239,
+      "step": 1466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001592980332945206,
+      "loss": 1.2952,
+      "step": 1467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015924288287523126,
+      "loss": 1.6851,
+      "step": 1468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001591877046787017,
+      "loss": 1.5357,
+      "step": 1469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015913249873080346,
+      "loss": 1.1051,
+      "step": 1470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015907726505742093,
+      "loss": 1.6035,
+      "step": 1471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015902200368445152,
+      "loss": 1.4277,
+      "step": 1472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015896671463780563,
+      "loss": 1.3915,
+      "step": 1473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001589113979434067,
+      "loss": 1.33,
+      "step": 1474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015885605362719115,
+      "loss": 1.2595,
+      "step": 1475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015880068171510826,
+      "loss": 1.3928,
+      "step": 1476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015874528223312021,
+      "loss": 1.4225,
+      "step": 1477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001586898552072023,
+      "loss": 1.3915,
+      "step": 1478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015863440066334254,
+      "loss": 1.3942,
+      "step": 1479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015857891862754197,
+      "loss": 1.4878,
+      "step": 1480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015852340912581448,
+      "loss": 1.4771,
+      "step": 1481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015846787218418685,
+      "loss": 1.2385,
+      "step": 1482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001584123078286987,
+      "loss": 1.3068,
+      "step": 1483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015835671608540247,
+      "loss": 1.4742,
+      "step": 1484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015830109698036356,
+      "loss": 1.4043,
+      "step": 1485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015824545053966008,
+      "loss": 1.3989,
+      "step": 1486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015818977678938307,
+      "loss": 1.2144,
+      "step": 1487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015813407575563623,
+      "loss": 1.27,
+      "step": 1488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015807834746453615,
+      "loss": 1.5951,
+      "step": 1489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015802259194221222,
+      "loss": 1.4543,
+      "step": 1490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015796680921480654,
+      "loss": 1.315,
+      "step": 1491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015791099930847398,
+      "loss": 1.1748,
+      "step": 1492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015785516224938215,
+      "loss": 1.1892,
+      "step": 1493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001577992980637114,
+      "loss": 1.4371,
+      "step": 1494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001577434067776548,
+      "loss": 1.3037,
+      "step": 1495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001576874884174182,
+      "loss": 1.2099,
+      "step": 1496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015763154300921993,
+      "loss": 1.3271,
+      "step": 1497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015757557057929127,
+      "loss": 1.1339,
+      "step": 1498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015751957115387594,
+      "loss": 1.2669,
+      "step": 1499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015746354475923048,
+      "loss": 1.2237,
+      "step": 1500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015740749142162402,
+      "loss": 1.4127,
+      "step": 1501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015735141116733826,
+      "loss": 1.4449,
+      "step": 1502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015729530402266764,
+      "loss": 1.3536,
+      "step": 1503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001572391700139191,
+      "loss": 1.3507,
+      "step": 1504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015718300916741225,
+      "loss": 1.36,
+      "step": 1505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015712682150947923,
+      "loss": 1.4082,
+      "step": 1506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001570706070664648,
+      "loss": 1.3571,
+      "step": 1507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015701436586472624,
+      "loss": 1.3912,
+      "step": 1508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001569580979306334,
+      "loss": 1.3672,
+      "step": 1509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015690180329056862,
+      "loss": 1.2854,
+      "step": 1510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015684548197092687,
+      "loss": 1.1184,
+      "step": 1511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015678913399811543,
+      "loss": 1.2415,
+      "step": 1512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015673275939855435,
+      "loss": 1.3751,
+      "step": 1513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015667635819867592,
+      "loss": 1.4839,
+      "step": 1514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015661993042492505,
+      "loss": 1.3633,
+      "step": 1515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000156563476103759,
+      "loss": 1.3428,
+      "step": 1516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001565069952616476,
+      "loss": 1.2353,
+      "step": 1517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015645048792507297,
+      "loss": 1.3976,
+      "step": 1518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015639395412052985,
+      "loss": 1.25,
+      "step": 1519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015633739387452516,
+      "loss": 1.5122,
+      "step": 1520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001562808072135784,
+      "loss": 1.5914,
+      "step": 1521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015622419416422135,
+      "loss": 1.1751,
+      "step": 1522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015616755475299823,
+      "loss": 1.3446,
+      "step": 1523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015611088900646558,
+      "loss": 1.4582,
+      "step": 1524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015605419695119226,
+      "loss": 1.3812,
+      "step": 1525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015599747861375955,
+      "loss": 1.39,
+      "step": 1526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000155940734020761,
+      "loss": 1.2084,
+      "step": 1527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015588396319880243,
+      "loss": 1.2421,
+      "step": 1528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015582716617450207,
+      "loss": 1.4228,
+      "step": 1529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015577034297449031,
+      "loss": 1.3832,
+      "step": 1530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001557134936254099,
+      "loss": 1.3851,
+      "step": 1531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001556566181539158,
+      "loss": 1.6114,
+      "step": 1532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001555997165866753,
+      "loss": 1.0902,
+      "step": 1533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015554278895036781,
+      "loss": 1.3391,
+      "step": 1534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015548583527168506,
+      "loss": 1.4806,
+      "step": 1535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015542885557733092,
+      "loss": 1.4368,
+      "step": 1536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015537184989402147,
+      "loss": 1.4617,
+      "step": 1537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015531481824848504,
+      "loss": 1.1512,
+      "step": 1538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015525776066746205,
+      "loss": 1.3349,
+      "step": 1539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015520067717770514,
+      "loss": 1.2503,
+      "step": 1540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015514356780597908,
+      "loss": 1.5371,
+      "step": 1541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015508643257906073,
+      "loss": 1.4209,
+      "step": 1542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015502927152373914,
+      "loss": 1.0174,
+      "step": 1543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015497208466681546,
+      "loss": 1.5133,
+      "step": 1544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015491487203510287,
+      "loss": 1.2716,
+      "step": 1545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015485763365542677,
+      "loss": 1.3801,
+      "step": 1546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015480036955462442,
+      "loss": 1.5941,
+      "step": 1547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001547430797595454,
+      "loss": 1.416,
+      "step": 1548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015468576429705113,
+      "loss": 1.5057,
+      "step": 1549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015462842319401515,
+      "loss": 1.4073,
+      "step": 1550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015457105647732303,
+      "loss": 1.384,
+      "step": 1551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001545136641738723,
+      "loss": 1.2286,
+      "step": 1552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015445624631057255,
+      "loss": 1.3697,
+      "step": 1553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001543988029143453,
+      "loss": 1.2903,
+      "step": 1554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015434133401212406,
+      "loss": 1.151,
+      "step": 1555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015428383963085433,
+      "loss": 1.1355,
+      "step": 1556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015422631979749354,
+      "loss": 1.6978,
+      "step": 1557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015416877453901096,
+      "loss": 1.3527,
+      "step": 1558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015411120388238798,
+      "loss": 1.2174,
+      "step": 1559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001540536078546177,
+      "loss": 1.2966,
+      "step": 1560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001539959864827052,
+      "loss": 1.5117,
+      "step": 1561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015393833979366752,
+      "loss": 1.3012,
+      "step": 1562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015388066781453342,
+      "loss": 1.4302,
+      "step": 1563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015382297057234357,
+      "loss": 1.16,
+      "step": 1564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015376524809415057,
+      "loss": 1.2921,
+      "step": 1565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015370750040701877,
+      "loss": 1.4025,
+      "step": 1566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015364972753802432,
+      "loss": 1.4693,
+      "step": 1567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015359192951425527,
+      "loss": 1.4962,
+      "step": 1568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001535341063628114,
+      "loss": 1.2495,
+      "step": 1569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001534762581108042,
+      "loss": 1.2435,
+      "step": 1570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015341838478535706,
+      "loss": 1.2972,
+      "step": 1571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015336048641360512,
+      "loss": 1.0488,
+      "step": 1572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001533025630226952,
+      "loss": 1.1645,
+      "step": 1573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015324461463978582,
+      "loss": 1.2745,
+      "step": 1574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001531866412920473,
+      "loss": 1.4374,
+      "step": 1575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001531286430066616,
+      "loss": 1.4613,
+      "step": 1576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015307061981082245,
+      "loss": 1.4443,
+      "step": 1577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015301257173173523,
+      "loss": 1.4113,
+      "step": 1578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015295449879661695,
+      "loss": 1.2162,
+      "step": 1579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015289640103269625,
+      "loss": 1.3983,
+      "step": 1580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001528382784672135,
+      "loss": 1.4609,
+      "step": 1581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015278013112742067,
+      "loss": 1.2221,
+      "step": 1582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015272195904058133,
+      "loss": 1.3753,
+      "step": 1583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015266376223397061,
+      "loss": 1.23,
+      "step": 1584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015260554073487533,
+      "loss": 1.252,
+      "step": 1585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015254729457059378,
+      "loss": 1.3014,
+      "step": 1586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001524890237684359,
+      "loss": 1.4382,
+      "step": 1587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015243072835572318,
+      "loss": 1.5672,
+      "step": 1588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001523724083597886,
+      "loss": 1.3609,
+      "step": 1589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001523140638079766,
+      "loss": 1.4196,
+      "step": 1590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015225569472764337,
+      "loss": 1.3708,
+      "step": 1591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015219730114615632,
+      "loss": 1.4364,
+      "step": 1592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015213888309089456,
+      "loss": 1.416,
+      "step": 1593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001520804405892485,
+      "loss": 1.3125,
+      "step": 1594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015202197366862024,
+      "loss": 1.2388,
+      "step": 1595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001519634823564231,
+      "loss": 1.3599,
+      "step": 1596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015190496668008194,
+      "loss": 1.2795,
+      "step": 1597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015184642666703308,
+      "loss": 1.4429,
+      "step": 1598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015178786234472415,
+      "loss": 1.3342,
+      "step": 1599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015172927374061427,
+      "loss": 1.2256,
+      "step": 1600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001516706608821739,
+      "loss": 1.4688,
+      "step": 1601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015161202379688484,
+      "loss": 1.5329,
+      "step": 1602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015155336251224032,
+      "loss": 1.2503,
+      "step": 1603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015149467705574492,
+      "loss": 1.4216,
+      "step": 1604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001514359674549145,
+      "loss": 1.3662,
+      "step": 1605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001513772337372762,
+      "loss": 1.3543,
+      "step": 1606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015131847593036863,
+      "loss": 1.3279,
+      "step": 1607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015125969406174153,
+      "loss": 1.4119,
+      "step": 1608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015120088815895598,
+      "loss": 1.5733,
+      "step": 1609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015114205824958435,
+      "loss": 1.4651,
+      "step": 1610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015108320436121026,
+      "loss": 1.6005,
+      "step": 1611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015102432652142853,
+      "loss": 1.281,
+      "step": 1612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015096542475784522,
+      "loss": 1.4618,
+      "step": 1613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015090649909807772,
+      "loss": 1.4299,
+      "step": 1614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001508475495697545,
+      "loss": 1.3793,
+      "step": 1615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015078857620051518,
+      "loss": 1.499,
+      "step": 1616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015072957901801076,
+      "loss": 1.3108,
+      "step": 1617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015067055804990318,
+      "loss": 1.3067,
+      "step": 1618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015061151332386566,
+      "loss": 1.3155,
+      "step": 1619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015055244486758247,
+      "loss": 1.3675,
+      "step": 1620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001504933527087492,
+      "loss": 1.4295,
+      "step": 1621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015043423687507236,
+      "loss": 1.3802,
+      "step": 1622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015037509739426955,
+      "loss": 1.3002,
+      "step": 1623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015031593429406961,
+      "loss": 1.1799,
+      "step": 1624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015025674760221235,
+      "loss": 1.4199,
+      "step": 1625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015019753734644867,
+      "loss": 1.412,
+      "step": 1626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015013830355454052,
+      "loss": 1.4751,
+      "step": 1627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015007904625426085,
+      "loss": 1.2261,
+      "step": 1628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015001976547339364,
+      "loss": 1.4152,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014996046123973397,
+      "loss": 1.3821,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001499011335810878,
+      "loss": 1.668,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014984178252527212,
+      "loss": 1.3118,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001497824081001149,
+      "loss": 1.5022,
+      "step": 1633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014972301033345503,
+      "loss": 1.151,
+      "step": 1634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014966358925314242,
+      "loss": 1.6384,
+      "step": 1635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014960414488703782,
+      "loss": 1.4106,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014954467726301297,
+      "loss": 1.6064,
+      "step": 1637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001494851864089505,
+      "loss": 1.0679,
+      "step": 1638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001494256723527438,
+      "loss": 1.4999,
+      "step": 1639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001493661351222974,
+      "loss": 1.4782,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001493065747455265,
+      "loss": 1.3514,
+      "step": 1641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014924699125035717,
+      "loss": 1.3124,
+      "step": 1642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001491873846647264,
+      "loss": 1.5797,
+      "step": 1643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014912775501658193,
+      "loss": 1.6183,
+      "step": 1644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001490681023338823,
+      "loss": 1.5718,
+      "step": 1645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000149008426644597,
+      "loss": 1.2576,
+      "step": 1646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014894872797670614,
+      "loss": 1.3271,
+      "step": 1647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014888900635820067,
+      "loss": 1.267,
+      "step": 1648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001488292618170823,
+      "loss": 1.6038,
+      "step": 1649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014876949438136347,
+      "loss": 1.4779,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014870970407906736,
+      "loss": 1.3973,
+      "step": 1651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001486498909382279,
+      "loss": 1.4589,
+      "step": 1652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014859005498688973,
+      "loss": 1.3481,
+      "step": 1653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014853019625310813,
+      "loss": 1.299,
+      "step": 1654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014847031476494907,
+      "loss": 1.3419,
+      "step": 1655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014841041055048925,
+      "loss": 1.3947,
+      "step": 1656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000148350483637816,
+      "loss": 1.2313,
+      "step": 1657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014829053405502723,
+      "loss": 1.3977,
+      "step": 1658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001482305618302316,
+      "loss": 1.5334,
+      "step": 1659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014817056699154826,
+      "loss": 1.3967,
+      "step": 1660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014811054956710704,
+      "loss": 1.1272,
+      "step": 1661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014805050958504833,
+      "loss": 1.4361,
+      "step": 1662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001479904470735231,
+      "loss": 1.3932,
+      "step": 1663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014793036206069298,
+      "loss": 1.2915,
+      "step": 1664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014787025457472993,
+      "loss": 1.3523,
+      "step": 1665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014781012464381665,
+      "loss": 1.1828,
+      "step": 1666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001477499722961463,
+      "loss": 1.4536,
+      "step": 1667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014768979755992246,
+      "loss": 1.3766,
+      "step": 1668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014762960046335935,
+      "loss": 1.6301,
+      "step": 1669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014756938103468168,
+      "loss": 1.6053,
+      "step": 1670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014750913930212443,
+      "loss": 1.5425,
+      "step": 1671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014744887529393323,
+      "loss": 1.3415,
+      "step": 1672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014738858903836412,
+      "loss": 1.4159,
+      "step": 1673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014732828056368354,
+      "loss": 1.4464,
+      "step": 1674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001472679498981683,
+      "loss": 1.4792,
+      "step": 1675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014720759707010573,
+      "loss": 1.2264,
+      "step": 1676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014714722210779348,
+      "loss": 1.3151,
+      "step": 1677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014708682503953956,
+      "loss": 1.3257,
+      "step": 1678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001470264058936624,
+      "loss": 1.1246,
+      "step": 1679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014696596469849075,
+      "loss": 1.3138,
+      "step": 1680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001469055014823637,
+      "loss": 1.5017,
+      "step": 1681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014684501627363068,
+      "loss": 1.6901,
+      "step": 1682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001467845091006514,
+      "loss": 1.6006,
+      "step": 1683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001467239799917959,
+      "loss": 1.8561,
+      "step": 1684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014666342897544447,
+      "loss": 1.2185,
+      "step": 1685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014660285607998776,
+      "loss": 1.5256,
+      "step": 1686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014654226133382656,
+      "loss": 1.1782,
+      "step": 1687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014648164476537193,
+      "loss": 1.5459,
+      "step": 1688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014642100640304526,
+      "loss": 1.5223,
+      "step": 1689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001463603462752781,
+      "loss": 1.0324,
+      "step": 1690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014629966441051208,
+      "loss": 1.4615,
+      "step": 1691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014623896083719925,
+      "loss": 1.589,
+      "step": 1692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001461782355838017,
+      "loss": 1.4456,
+      "step": 1693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014611748867879166,
+      "loss": 1.3502,
+      "step": 1694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001460567201506516,
+      "loss": 1.3339,
+      "step": 1695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001459959300278741,
+      "loss": 1.1755,
+      "step": 1696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014593511833896187,
+      "loss": 1.2918,
+      "step": 1697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014587428511242766,
+      "loss": 1.2057,
+      "step": 1698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014581343037679443,
+      "loss": 1.364,
+      "step": 1699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014575255416059512,
+      "loss": 1.3253,
+      "step": 1700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014569165649237282,
+      "loss": 1.3406,
+      "step": 1701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014563073740068064,
+      "loss": 1.3375,
+      "step": 1702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001455697969140818,
+      "loss": 1.3619,
+      "step": 1703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001455088350611494,
+      "loss": 1.196,
+      "step": 1704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001454478518704667,
+      "loss": 1.3467,
+      "step": 1705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000145386847370627,
+      "loss": 1.1601,
+      "step": 1706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014532582159023337,
+      "loss": 1.2508,
+      "step": 1707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014526477455789909,
+      "loss": 1.2411,
+      "step": 1708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014520370630224724,
+      "loss": 1.399,
+      "step": 1709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000145142616851911,
+      "loss": 1.1868,
+      "step": 1710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001450815062355334,
+      "loss": 1.5596,
+      "step": 1711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014502037448176734,
+      "loss": 1.2139,
+      "step": 1712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014495922161927575,
+      "loss": 1.3267,
+      "step": 1713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014489804767673143,
+      "loss": 1.5392,
+      "step": 1714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014483685268281698,
+      "loss": 1.1553,
+      "step": 1715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014477563666622493,
+      "loss": 1.1798,
+      "step": 1716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014471439965565769,
+      "loss": 1.4638,
+      "step": 1717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001446531416798275,
+      "loss": 1.2241,
+      "step": 1718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001445918627674564,
+      "loss": 1.3507,
+      "step": 1719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014453056294727622,
+      "loss": 1.4532,
+      "step": 1720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014446924224802873,
+      "loss": 1.1905,
+      "step": 1721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014440790069846532,
+      "loss": 1.1908,
+      "step": 1722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014434653832734728,
+      "loss": 1.1669,
+      "step": 1723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014428515516344561,
+      "loss": 1.3855,
+      "step": 1724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014422375123554102,
+      "loss": 1.2488,
+      "step": 1725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014416232657242403,
+      "loss": 1.1785,
+      "step": 1726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014410088120289487,
+      "loss": 1.4306,
+      "step": 1727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014403941515576344,
+      "loss": 1.229,
+      "step": 1728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001439779284598494,
+      "loss": 1.2515,
+      "step": 1729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014391642114398195,
+      "loss": 1.287,
+      "step": 1730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001438548932370001,
+      "loss": 1.3581,
+      "step": 1731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014379334476775254,
+      "loss": 1.1554,
+      "step": 1732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014373177576509745,
+      "loss": 1.4246,
+      "step": 1733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014367018625790275,
+      "loss": 1.24,
+      "step": 1734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014360857627504598,
+      "loss": 1.109,
+      "step": 1735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014354694584541415,
+      "loss": 1.5166,
+      "step": 1736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000143485294997904,
+      "loss": 1.4247,
+      "step": 1737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014342362376142185,
+      "loss": 1.3589,
+      "step": 1738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014336193216488345,
+      "loss": 1.5386,
+      "step": 1739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001433002202372142,
+      "loss": 1.2867,
+      "step": 1740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014323848800734895,
+      "loss": 1.391,
+      "step": 1741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001431767355042322,
+      "loss": 1.4031,
+      "step": 1742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014311496275681783,
+      "loss": 1.2572,
+      "step": 1743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001430531697940693,
+      "loss": 1.2381,
+      "step": 1744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014299135664495944,
+      "loss": 1.1326,
+      "step": 1745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014292952333847065,
+      "loss": 1.3607,
+      "step": 1746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014286766990359478,
+      "loss": 1.0812,
+      "step": 1747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014280579636933303,
+      "loss": 1.3289,
+      "step": 1748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014274390276469609,
+      "loss": 1.3614,
+      "step": 1749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014268198911870401,
+      "loss": 1.4436,
+      "step": 1750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014262005546038635,
+      "loss": 1.3662,
+      "step": 1751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014255810181878193,
+      "loss": 1.3015,
+      "step": 1752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014249612822293895,
+      "loss": 1.4794,
+      "step": 1753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014243413470191508,
+      "loss": 1.3931,
+      "step": 1754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001423721212847772,
+      "loss": 1.3246,
+      "step": 1755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014231008800060158,
+      "loss": 1.2125,
+      "step": 1756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014224803487847382,
+      "loss": 1.4413,
+      "step": 1757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014218596194748875,
+      "loss": 1.5046,
+      "step": 1758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014212386923675058,
+      "loss": 1.1167,
+      "step": 1759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014206175677537275,
+      "loss": 1.4961,
+      "step": 1760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014199962459247794,
+      "loss": 1.3104,
+      "step": 1761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001419374727171981,
+      "loss": 1.6551,
+      "step": 1762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001418753011786744,
+      "loss": 1.3931,
+      "step": 1763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014181311000605732,
+      "loss": 1.5124,
+      "step": 1764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014175089922850633,
+      "loss": 1.3758,
+      "step": 1765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001416886688751903,
+      "loss": 1.3185,
+      "step": 1766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014162641897528725,
+      "loss": 1.3958,
+      "step": 1767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014156414955798427,
+      "loss": 1.2747,
+      "step": 1768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014150186065247766,
+      "loss": 1.0998,
+      "step": 1769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014143955228797283,
+      "loss": 1.1384,
+      "step": 1770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014137722449368437,
+      "loss": 1.2722,
+      "step": 1771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014131487729883598,
+      "loss": 1.2717,
+      "step": 1772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014125251073266031,
+      "loss": 1.2117,
+      "step": 1773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001411901248243993,
+      "loss": 1.4773,
+      "step": 1774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001411277196033038,
+      "loss": 1.4718,
+      "step": 1775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001410652950986338,
+      "loss": 1.0623,
+      "step": 1776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014100285133965832,
+      "loss": 1.3979,
+      "step": 1777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001409403883556554,
+      "loss": 1.4625,
+      "step": 1778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014087790617591209,
+      "loss": 1.451,
+      "step": 1779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014081540482972443,
+      "loss": 1.3051,
+      "step": 1780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014075288434639745,
+      "loss": 1.454,
+      "step": 1781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014069034475524514,
+      "loss": 1.064,
+      "step": 1782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014062778608559052,
+      "loss": 1.3828,
+      "step": 1783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001405652083667655,
+      "loss": 1.1291,
+      "step": 1784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001405026116281109,
+      "loss": 1.4426,
+      "step": 1785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014043999589897647,
+      "loss": 1.2443,
+      "step": 1786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014037736120872093,
+      "loss": 1.3573,
+      "step": 1787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014031470758671182,
+      "loss": 1.4231,
+      "step": 1788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014025203506232555,
+      "loss": 1.2851,
+      "step": 1789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014018934366494742,
+      "loss": 1.3761,
+      "step": 1790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001401266334239716,
+      "loss": 1.3317,
+      "step": 1791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014006390436880107,
+      "loss": 1.4383,
+      "step": 1792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014000115652884759,
+      "loss": 1.4786,
+      "step": 1793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013993838993353188,
+      "loss": 1.2481,
+      "step": 1794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001398756046122832,
+      "loss": 1.3427,
+      "step": 1795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013981280059453986,
+      "loss": 1.2347,
+      "step": 1796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013974997790974872,
+      "loss": 1.2588,
+      "step": 1797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013968713658736552,
+      "loss": 1.3701,
+      "step": 1798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013962427665685476,
+      "loss": 1.404,
+      "step": 1799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001395613981476895,
+      "loss": 1.4017,
+      "step": 1800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013949850108935168,
+      "loss": 1.2372,
+      "step": 1801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013943558551133186,
+      "loss": 1.2845,
+      "step": 1802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013937265144312933,
+      "loss": 1.3098,
+      "step": 1803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013930969891425197,
+      "loss": 1.2932,
+      "step": 1804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013924672795421637,
+      "loss": 1.224,
+      "step": 1805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001391837385925478,
+      "loss": 1.4211,
+      "step": 1806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013912073085878008,
+      "loss": 1.3874,
+      "step": 1807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013905770478245567,
+      "loss": 1.2935,
+      "step": 1808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013899466039312566,
+      "loss": 1.3093,
+      "step": 1809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001389315977203497,
+      "loss": 1.317,
+      "step": 1810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013886851679369603,
+      "loss": 1.3893,
+      "step": 1811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013880541764274136,
+      "loss": 1.3172,
+      "step": 1812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001387423002970711,
+      "loss": 1.5613,
+      "step": 1813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001386791647862791,
+      "loss": 1.2968,
+      "step": 1814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013861601113996773,
+      "loss": 1.0993,
+      "step": 1815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013855283938774788,
+      "loss": 1.4054,
+      "step": 1816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001384896495592389,
+      "loss": 1.3522,
+      "step": 1817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013842644168406864,
+      "loss": 1.2725,
+      "step": 1818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001383632157918734,
+      "loss": 1.4336,
+      "step": 1819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013829997191229797,
+      "loss": 1.6374,
+      "step": 1820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013823671007499554,
+      "loss": 1.2858,
+      "step": 1821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013817343030962767,
+      "loss": 1.2721,
+      "step": 1822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013811013264586444,
+      "loss": 1.2327,
+      "step": 1823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013804681711338418,
+      "loss": 1.2543,
+      "step": 1824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013798348374187373,
+      "loss": 1.3047,
+      "step": 1825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013792013256102827,
+      "loss": 1.599,
+      "step": 1826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001378567636005512,
+      "loss": 1.1474,
+      "step": 1827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013779337689015443,
+      "loss": 1.4336,
+      "step": 1828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013772997245955812,
+      "loss": 1.3392,
+      "step": 1829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013766655033849073,
+      "loss": 1.4535,
+      "step": 1830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013760311055668897,
+      "loss": 1.2585,
+      "step": 1831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013753965314389796,
+      "loss": 1.2923,
+      "step": 1832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013747617812987097,
+      "loss": 1.3826,
+      "step": 1833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013741268554436951,
+      "loss": 1.5358,
+      "step": 1834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001373491754171635,
+      "loss": 1.4111,
+      "step": 1835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013728564777803088,
+      "loss": 1.2752,
+      "step": 1836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001372221026567579,
+      "loss": 1.4708,
+      "step": 1837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000137158540083139,
+      "loss": 1.4269,
+      "step": 1838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001370949600869768,
+      "loss": 1.1871,
+      "step": 1839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013703136269808207,
+      "loss": 1.1854,
+      "step": 1840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013696774794627373,
+      "loss": 1.5748,
+      "step": 1841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013690411586137895,
+      "loss": 1.3077,
+      "step": 1842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013684046647323284,
+      "loss": 1.3068,
+      "step": 1843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013677679981167877,
+      "loss": 1.3625,
+      "step": 1844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013671311590656815,
+      "loss": 1.4329,
+      "step": 1845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013664941478776048,
+      "loss": 1.1805,
+      "step": 1846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013658569648512329,
+      "loss": 1.4601,
+      "step": 1847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013652196102853235,
+      "loss": 1.4309,
+      "step": 1848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001364582084478712,
+      "loss": 1.5392,
+      "step": 1849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013639443877303155,
+      "loss": 1.4426,
+      "step": 1850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001363306520339132,
+      "loss": 1.183,
+      "step": 1851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013626684826042383,
+      "loss": 1.1242,
+      "step": 1852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013620302748247913,
+      "loss": 1.4604,
+      "step": 1853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001361391897300028,
+      "loss": 1.4575,
+      "step": 1854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013607533503292646,
+      "loss": 1.2553,
+      "step": 1855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013601146342118972,
+      "loss": 1.42,
+      "step": 1856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001359475749247401,
+      "loss": 1.328,
+      "step": 1857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000135883669573533,
+      "loss": 1.2052,
+      "step": 1858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013581974739753177,
+      "loss": 1.2629,
+      "step": 1859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013575580842670762,
+      "loss": 1.4886,
+      "step": 1860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001356918526910397,
+      "loss": 1.2556,
+      "step": 1861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001356278802205149,
+      "loss": 1.2297,
+      "step": 1862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013556389104512807,
+      "loss": 1.4505,
+      "step": 1863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001354998851948818,
+      "loss": 1.2603,
+      "step": 1864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013543586269978657,
+      "loss": 1.3243,
+      "step": 1865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013537182358986066,
+      "loss": 1.3055,
+      "step": 1866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001353077678951301,
+      "loss": 1.3824,
+      "step": 1867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001352436956456287,
+      "loss": 1.4016,
+      "step": 1868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013517960687139808,
+      "loss": 1.0678,
+      "step": 1869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013511550160248757,
+      "loss": 1.199,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001350513798689542,
+      "loss": 1.6122,
+      "step": 1871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013498724170086282,
+      "loss": 1.5709,
+      "step": 1872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013492308712828588,
+      "loss": 1.4257,
+      "step": 1873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013485891618130358,
+      "loss": 1.2289,
+      "step": 1874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001347947288900038,
+      "loss": 1.4194,
+      "step": 1875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013473052528448201,
+      "loss": 1.2479,
+      "step": 1876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001346663053948415,
+      "loss": 1.1951,
+      "step": 1877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013460206925119297,
+      "loss": 1.4269,
+      "step": 1878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001345378168836549,
+      "loss": 1.2004,
+      "step": 1879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001344735483223533,
+      "loss": 1.0719,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013440926359742188,
+      "loss": 1.4484,
+      "step": 1881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013434496273900177,
+      "loss": 1.3178,
+      "step": 1882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013428064577724173,
+      "loss": 1.0344,
+      "step": 1883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013421631274229817,
+      "loss": 1.2843,
+      "step": 1884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001341519636643349,
+      "loss": 1.3083,
+      "step": 1885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013408759857352337,
+      "loss": 1.3331,
+      "step": 1886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013402321750004235,
+      "loss": 1.4907,
+      "step": 1887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013395882047407832,
+      "loss": 0.9841,
+      "step": 1888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001338944075258251,
+      "loss": 1.2904,
+      "step": 1889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013382997868548408,
+      "loss": 0.9096,
+      "step": 1890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000133765533983264,
+      "loss": 1.1595,
+      "step": 1891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013370107344938106,
+      "loss": 1.4391,
+      "step": 1892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013363659711405893,
+      "loss": 1.3272,
+      "step": 1893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013357210500752867,
+      "loss": 1.2604,
+      "step": 1894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013350759716002872,
+      "loss": 1.2945,
+      "step": 1895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001334430736018049,
+      "loss": 1.5985,
+      "step": 1896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001333785343631104,
+      "loss": 1.3373,
+      "step": 1897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013331397947420576,
+      "loss": 1.4012,
+      "step": 1898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013324940896535886,
+      "loss": 1.3674,
+      "step": 1899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013318482286684498,
+      "loss": 1.3794,
+      "step": 1900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013312022120894653,
+      "loss": 1.1398,
+      "step": 1901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001330556040219534,
+      "loss": 1.3465,
+      "step": 1902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001329909713361626,
+      "loss": 1.5024,
+      "step": 1903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013292632318187856,
+      "loss": 1.5367,
+      "step": 1904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013286165958941291,
+      "loss": 1.4598,
+      "step": 1905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001327969805890844,
+      "loss": 1.3663,
+      "step": 1906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013273228621121924,
+      "loss": 1.5168,
+      "step": 1907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013266757648615061,
+      "loss": 1.3968,
+      "step": 1908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000132602851444219,
+      "loss": 1.2258,
+      "step": 1909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013253811111577215,
+      "loss": 1.4393,
+      "step": 1910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001324733555311648,
+      "loss": 1.3416,
+      "step": 1911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013240858472075896,
+      "loss": 1.2709,
+      "step": 1912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001323437987149238,
+      "loss": 1.3082,
+      "step": 1913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013227899754403554,
+      "loss": 1.4709,
+      "step": 1914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001322141812384775,
+      "loss": 1.3833,
+      "step": 1915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013214934982864018,
+      "loss": 1.305,
+      "step": 1916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013208450334492111,
+      "loss": 1.3434,
+      "step": 1917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013201964181772485,
+      "loss": 1.4275,
+      "step": 1918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013195476527746313,
+      "loss": 1.4461,
+      "step": 1919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013188987375455457,
+      "loss": 1.4812,
+      "step": 1920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013182496727942494,
+      "loss": 1.5598,
+      "step": 1921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001317600458825069,
+      "loss": 1.489,
+      "step": 1922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013169510959424035,
+      "loss": 1.4822,
+      "step": 1923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013163015844507183,
+      "loss": 1.5279,
+      "step": 1924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013156519246545505,
+      "loss": 1.2358,
+      "step": 1925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013150021168585068,
+      "loss": 1.1396,
+      "step": 1926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013143521613672628,
+      "loss": 1.1043,
+      "step": 1927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013137020584855634,
+      "loss": 1.5107,
+      "step": 1928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013130518085182225,
+      "loss": 1.4245,
+      "step": 1929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013124014117701232,
+      "loss": 1.4072,
+      "step": 1930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013117508685462175,
+      "loss": 1.5103,
+      "step": 1931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013111001791515263,
+      "loss": 1.106,
+      "step": 1932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001310449343891138,
+      "loss": 1.2547,
+      "step": 1933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000130979836307021,
+      "loss": 1.2402,
+      "step": 1934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001309147236993969,
+      "loss": 1.3063,
+      "step": 1935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013084959659677073,
+      "loss": 1.1793,
+      "step": 1936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013078445502967879,
+      "loss": 1.3153,
+      "step": 1937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000130719299028664,
+      "loss": 1.748,
+      "step": 1938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001306541286242761,
+      "loss": 1.6434,
+      "step": 1939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013058894384707152,
+      "loss": 1.4486,
+      "step": 1940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001305237447276135,
+      "loss": 1.4921,
+      "step": 1941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013045853129647203,
+      "loss": 1.404,
+      "step": 1942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001303933035842237,
+      "loss": 1.5371,
+      "step": 1943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013032806162145192,
+      "loss": 1.2532,
+      "step": 1944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001302628054387467,
+      "loss": 1.3486,
+      "step": 1945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001301975350667047,
+      "loss": 1.3088,
+      "step": 1946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013013225053592933,
+      "loss": 1.4957,
+      "step": 1947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001300669518770305,
+      "loss": 1.2723,
+      "step": 1948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013000163912062494,
+      "loss": 1.4636,
+      "step": 1949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012993631229733582,
+      "loss": 1.378,
+      "step": 1950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012987097143779292,
+      "loss": 1.6024,
+      "step": 1951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001298056165726327,
+      "loss": 1.2694,
+      "step": 1952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012974024773249808,
+      "loss": 1.4591,
+      "step": 1953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001296748649480386,
+      "loss": 1.0479,
+      "step": 1954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012960946824991035,
+      "loss": 1.4652,
+      "step": 1955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012954405766877591,
+      "loss": 1.4972,
+      "step": 1956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001294786332353043,
+      "loss": 1.2085,
+      "step": 1957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012941319498017116,
+      "loss": 1.3307,
+      "step": 1958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012934774293405857,
+      "loss": 1.2457,
+      "step": 1959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012928227712765504,
+      "loss": 1.4809,
+      "step": 1960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012921679759165557,
+      "loss": 1.3895,
+      "step": 1961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001291513043567616,
+      "loss": 1.4709,
+      "step": 1962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001290857974536809,
+      "loss": 1.3426,
+      "step": 1963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001290202769131278,
+      "loss": 1.3246,
+      "step": 1964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001289547427658229,
+      "loss": 1.5247,
+      "step": 1965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001288891950424933,
+      "loss": 1.362,
+      "step": 1966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012882363377387233,
+      "loss": 1.5301,
+      "step": 1967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012875805899069972,
+      "loss": 1.365,
+      "step": 1968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001286924707237216,
+      "loss": 1.4078,
+      "step": 1969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001286268690036904,
+      "loss": 1.1851,
+      "step": 1970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001285612538613647,
+      "loss": 1.4692,
+      "step": 1971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012849562532750967,
+      "loss": 1.2487,
+      "step": 1972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001284299834328965,
+      "loss": 1.2843,
+      "step": 1973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012836432820830272,
+      "loss": 1.3732,
+      "step": 1974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001282986596845122,
+      "loss": 1.3139,
+      "step": 1975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001282329778923149,
+      "loss": 1.2381,
+      "step": 1976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012816728286250717,
+      "loss": 1.2709,
+      "step": 1977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012810157462589137,
+      "loss": 1.3285,
+      "step": 1978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001280358532132762,
+      "loss": 1.2979,
+      "step": 1979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012797011865547646,
+      "loss": 1.1663,
+      "step": 1980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012790437098331322,
+      "loss": 1.1935,
+      "step": 1981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012783861022761355,
+      "loss": 1.0344,
+      "step": 1982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012777283641921076,
+      "loss": 1.1907,
+      "step": 1983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012770704958894424,
+      "loss": 1.2664,
+      "step": 1984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001276412497676595,
+      "loss": 1.5894,
+      "step": 1985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012757543698620815,
+      "loss": 1.4887,
+      "step": 1986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001275096112754478,
+      "loss": 1.219,
+      "step": 1987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001274437726662423,
+      "loss": 1.5533,
+      "step": 1988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001273779211894613,
+      "loss": 1.4091,
+      "step": 1989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012731205687598068,
+      "loss": 1.4472,
+      "step": 1990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001272461797566823,
+      "loss": 1.278,
+      "step": 1991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012718028986245393,
+      "loss": 1.6159,
+      "step": 1992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012711438722418946,
+      "loss": 1.3865,
+      "step": 1993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012704847187278867,
+      "loss": 1.1151,
+      "step": 1994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012698254383915734,
+      "loss": 1.293,
+      "step": 1995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012691660315420714,
+      "loss": 1.3011,
+      "step": 1996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012685064984885568,
+      "loss": 1.2151,
+      "step": 1997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012678468395402662,
+      "loss": 1.4304,
+      "step": 1998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012671870550064938,
+      "loss": 1.4034,
+      "step": 1999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012665271451965933,
+      "loss": 1.1947,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012658671104199764,
+      "loss": 1.3959,
+      "step": 2001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012652069509861143,
+      "loss": 1.3547,
+      "step": 2002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012645466672045358,
+      "loss": 1.4086,
+      "step": 2003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012638862593848294,
+      "loss": 1.4233,
+      "step": 2004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012632257278366403,
+      "loss": 1.2809,
+      "step": 2005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001262565072869672,
+      "loss": 1.453,
+      "step": 2006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012619042947936866,
+      "loss": 1.4384,
+      "step": 2007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012612433939185034,
+      "loss": 1.4357,
+      "step": 2008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012605823705539993,
+      "loss": 1.2694,
+      "step": 2009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012599212250101086,
+      "loss": 1.3467,
+      "step": 2010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001259259957596823,
+      "loss": 1.4257,
+      "step": 2011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012585985686241907,
+      "loss": 1.3038,
+      "step": 2012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001257937058402318,
+      "loss": 1.4061,
+      "step": 2013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012572754272413677,
+      "loss": 1.4886,
+      "step": 2014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001256613675451559,
+      "loss": 1.1685,
+      "step": 2015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012559518033431675,
+      "loss": 1.3364,
+      "step": 2016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001255289811226526,
+      "loss": 1.4285,
+      "step": 2017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012546276994120228,
+      "loss": 1.4334,
+      "step": 2018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012539654682101024,
+      "loss": 1.4504,
+      "step": 2019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001253303117931266,
+      "loss": 1.4753,
+      "step": 2020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012526406488860692,
+      "loss": 1.3923,
+      "step": 2021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012519780613851254,
+      "loss": 1.3695,
+      "step": 2022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012513153557391012,
+      "loss": 1.3464,
+      "step": 2023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012506525322587207,
+      "loss": 1.2054,
+      "step": 2024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012499895912547614,
+      "loss": 1.3958,
+      "step": 2025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012493265330380576,
+      "loss": 1.3195,
+      "step": 2026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012486633579194973,
+      "loss": 1.1871,
+      "step": 2027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012480000662100236,
+      "loss": 1.4471,
+      "step": 2028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012473366582206342,
+      "loss": 1.2356,
+      "step": 2029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012466731342623824,
+      "loss": 1.38,
+      "step": 2030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001246009494646375,
+      "loss": 1.4265,
+      "step": 2031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001245345739683772,
+      "loss": 1.1981,
+      "step": 2032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001244681869685789,
+      "loss": 1.3753,
+      "step": 2033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012440178849636957,
+      "loss": 1.096,
+      "step": 2034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001243353785828814,
+      "loss": 1.3547,
+      "step": 2035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012426895725925211,
+      "loss": 1.2351,
+      "step": 2036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012420252455662466,
+      "loss": 1.2054,
+      "step": 2037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001241360805061474,
+      "loss": 1.3721,
+      "step": 2038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012406962513897396,
+      "loss": 1.2666,
+      "step": 2039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001240031584862633,
+      "loss": 1.2646,
+      "step": 2040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012393668057917976,
+      "loss": 1.4575,
+      "step": 2041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012387019144889275,
+      "loss": 1.327,
+      "step": 2042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012380369112657716,
+      "loss": 1.3049,
+      "step": 2043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012373717964341295,
+      "loss": 1.374,
+      "step": 2044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012367065703058544,
+      "loss": 1.5505,
+      "step": 2045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012360412331928513,
+      "loss": 1.4629,
+      "step": 2046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001235375785407077,
+      "loss": 1.1411,
+      "step": 2047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012347102272605397,
+      "loss": 1.4244,
+      "step": 2048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012340445590653008,
+      "loss": 1.2676,
+      "step": 2049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012333787811334726,
+      "loss": 1.3754,
+      "step": 2050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012327128937772178,
+      "loss": 1.3649,
+      "step": 2051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001232046897308752,
+      "loss": 1.222,
+      "step": 2052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012313807920403419,
+      "loss": 1.5696,
+      "step": 2053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012307145782843032,
+      "loss": 1.3763,
+      "step": 2054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012300482563530046,
+      "loss": 1.5109,
+      "step": 2055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001229381826558865,
+      "loss": 1.1925,
+      "step": 2056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012287152892143533,
+      "loss": 1.4124,
+      "step": 2057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012280486446319895,
+      "loss": 1.5667,
+      "step": 2058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012273818931243436,
+      "loss": 1.3842,
+      "step": 2059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012267150350040354,
+      "loss": 1.1484,
+      "step": 2060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001226048070583735,
+      "loss": 1.4182,
+      "step": 2061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012253810001761626,
+      "loss": 1.1868,
+      "step": 2062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012247138240940877,
+      "loss": 1.3166,
+      "step": 2063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012240465426503297,
+      "loss": 1.2051,
+      "step": 2064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012233791561577567,
+      "loss": 1.4824,
+      "step": 2065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012227116649292868,
+      "loss": 1.1672,
+      "step": 2066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012220440692778874,
+      "loss": 1.1882,
+      "step": 2067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012213763695165735,
+      "loss": 1.3921,
+      "step": 2068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012207085659584108,
+      "loss": 1.3578,
+      "step": 2069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001220040658916512,
+      "loss": 1.296,
+      "step": 2070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012193726487040388,
+      "loss": 1.4216,
+      "step": 2071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012187045356342024,
+      "loss": 1.3902,
+      "step": 2072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012180363200202607,
+      "loss": 1.3583,
+      "step": 2073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012173680021755205,
+      "loss": 1.3115,
+      "step": 2074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001216699582413336,
+      "loss": 1.2567,
+      "step": 2075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012160310610471099,
+      "loss": 1.4038,
+      "step": 2076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012153624383902919,
+      "loss": 1.4342,
+      "step": 2077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012146937147563793,
+      "loss": 1.3358,
+      "step": 2078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012140248904589176,
+      "loss": 1.3511,
+      "step": 2079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012133559658114979,
+      "loss": 1.417,
+      "step": 2080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012126869411277596,
+      "loss": 1.4993,
+      "step": 2081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012120178167213888,
+      "loss": 1.1418,
+      "step": 2082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012113485929061179,
+      "loss": 1.1139,
+      "step": 2083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012106792699957263,
+      "loss": 1.4568,
+      "step": 2084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012100098483040401,
+      "loss": 1.0189,
+      "step": 2085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012093403281449311,
+      "loss": 1.0824,
+      "step": 2086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012086707098323171,
+      "loss": 1.3892,
+      "step": 2087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012080009936801633,
+      "loss": 1.3298,
+      "step": 2088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012073311800024793,
+      "loss": 1.2492,
+      "step": 2089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001206661269113321,
+      "loss": 1.3931,
+      "step": 2090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012059912613267901,
+      "loss": 1.1507,
+      "step": 2091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001205321156957033,
+      "loss": 1.358,
+      "step": 2092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012046509563182424,
+      "loss": 1.5573,
+      "step": 2093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001203980659724655,
+      "loss": 1.2294,
+      "step": 2094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012033102674905539,
+      "loss": 1.478,
+      "step": 2095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012026397799302658,
+      "loss": 1.123,
+      "step": 2096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012019691973581623,
+      "loss": 1.3688,
+      "step": 2097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012012985200886602,
+      "loss": 1.2821,
+      "step": 2098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012006277484362201,
+      "loss": 1.668,
+      "step": 2099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011999568827153471,
+      "loss": 1.46,
+      "step": 2100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011992859232405904,
+      "loss": 1.4293,
+      "step": 2101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001198614870326543,
+      "loss": 1.4472,
+      "step": 2102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011979437242878417,
+      "loss": 1.3636,
+      "step": 2103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001197272485439167,
+      "loss": 1.5649,
+      "step": 2104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011966011540952435,
+      "loss": 1.4515,
+      "step": 2105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011959297305708387,
+      "loss": 1.3001,
+      "step": 2106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011952582151807625,
+      "loss": 1.2542,
+      "step": 2107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011945866082398693,
+      "loss": 1.4966,
+      "step": 2108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011939149100630556,
+      "loss": 1.2383,
+      "step": 2109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011932431209652606,
+      "loss": 1.3004,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011925712412614668,
+      "loss": 1.2318,
+      "step": 2111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011918992712666992,
+      "loss": 1.2711,
+      "step": 2112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011912272112960234,
+      "loss": 1.275,
+      "step": 2113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011905550616645496,
+      "loss": 1.3892,
+      "step": 2114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011898828226874284,
+      "loss": 1.4064,
+      "step": 2115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001189210494679853,
+      "loss": 1.2975,
+      "step": 2116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011885380779570582,
+      "loss": 1.4062,
+      "step": 2117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011878655728343204,
+      "loss": 1.3243,
+      "step": 2118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011871929796269574,
+      "loss": 1.4995,
+      "step": 2119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011865202986503278,
+      "loss": 1.3912,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011858475302198328,
+      "loss": 1.3416,
+      "step": 2121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011851746746509131,
+      "loss": 1.4505,
+      "step": 2122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001184501732259051,
+      "loss": 1.3735,
+      "step": 2123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001183828703359769,
+      "loss": 1.2519,
+      "step": 2124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011831555882686309,
+      "loss": 1.4892,
+      "step": 2125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011824823873012404,
+      "loss": 1.2143,
+      "step": 2126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011818091007732415,
+      "loss": 1.3346,
+      "step": 2127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011811357290003187,
+      "loss": 1.0944,
+      "step": 2128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011804622722981956,
+      "loss": 1.6422,
+      "step": 2129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011797887309826364,
+      "loss": 1.3821,
+      "step": 2130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001179115105369445,
+      "loss": 1.5263,
+      "step": 2131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011784413957744642,
+      "loss": 1.2832,
+      "step": 2132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011777676025135768,
+      "loss": 1.3519,
+      "step": 2133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011770937259027048,
+      "loss": 1.4074,
+      "step": 2134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011764197662578086,
+      "loss": 1.3173,
+      "step": 2135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011757457238948882,
+      "loss": 1.174,
+      "step": 2136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001175071599129982,
+      "loss": 1.2073,
+      "step": 2137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011743973922791678,
+      "loss": 1.4606,
+      "step": 2138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011737231036585608,
+      "loss": 1.4116,
+      "step": 2139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011730487335843153,
+      "loss": 1.3681,
+      "step": 2140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011723742823726232,
+      "loss": 1.4845,
+      "step": 2141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011716997503397149,
+      "loss": 1.2476,
+      "step": 2142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011710251378018584,
+      "loss": 1.258,
+      "step": 2143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011703504450753604,
+      "loss": 1.0389,
+      "step": 2144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011696756724765636,
+      "loss": 1.2656,
+      "step": 2145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011690008203218493,
+      "loss": 1.5244,
+      "step": 2146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011683258889276354,
+      "loss": 1.2431,
+      "step": 2147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011676508786103779,
+      "loss": 1.3951,
+      "step": 2148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001166975789686569,
+      "loss": 1.2255,
+      "step": 2149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001166300622472738,
+      "loss": 1.3563,
+      "step": 2150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011656253772854507,
+      "loss": 1.6033,
+      "step": 2151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011649500544413095,
+      "loss": 1.2295,
+      "step": 2152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011642746542569538,
+      "loss": 1.5078,
+      "step": 2153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011635991770490588,
+      "loss": 1.3519,
+      "step": 2154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011629236231343357,
+      "loss": 1.1602,
+      "step": 2155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011622479928295318,
+      "loss": 1.1298,
+      "step": 2156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011615722864514303,
+      "loss": 1.4329,
+      "step": 2157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011608965043168496,
+      "loss": 1.4205,
+      "step": 2158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011602206467426442,
+      "loss": 1.5458,
+      "step": 2159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011595447140457046,
+      "loss": 1.6281,
+      "step": 2160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011588687065429544,
+      "loss": 1.3845,
+      "step": 2161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011581926245513541,
+      "loss": 1.1132,
+      "step": 2162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001157516468387899,
+      "loss": 1.2026,
+      "step": 2163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011568402383696182,
+      "loss": 1.2377,
+      "step": 2164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011561639348135764,
+      "loss": 1.3219,
+      "step": 2165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011554875580368723,
+      "loss": 1.3443,
+      "step": 2166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011548111083566389,
+      "loss": 1.3552,
+      "step": 2167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011541345860900437,
+      "loss": 1.2028,
+      "step": 2168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011534579915542874,
+      "loss": 1.4027,
+      "step": 2169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011527813250666062,
+      "loss": 1.1082,
+      "step": 2170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011521045869442686,
+      "loss": 1.4169,
+      "step": 2171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011514277775045768,
+      "loss": 1.4026,
+      "step": 2172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011507508970648672,
+      "loss": 1.3863,
+      "step": 2173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011500739459425089,
+      "loss": 1.1691,
+      "step": 2174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011493969244549045,
+      "loss": 1.4308,
+      "step": 2175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011487198329194895,
+      "loss": 1.3963,
+      "step": 2176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011480426716537315,
+      "loss": 1.4178,
+      "step": 2177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001147365440975132,
+      "loss": 1.3092,
+      "step": 2178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011466881412012245,
+      "loss": 1.4949,
+      "step": 2179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011460107726495746,
+      "loss": 1.1055,
+      "step": 2180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011453333356377806,
+      "loss": 1.7136,
+      "step": 2181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001144655830483473,
+      "loss": 1.1373,
+      "step": 2182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011439782575043133,
+      "loss": 1.5332,
+      "step": 2183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001143300617017996,
+      "loss": 1.6143,
+      "step": 2184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011426229093422465,
+      "loss": 1.4235,
+      "step": 2185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001141945134794822,
+      "loss": 1.3659,
+      "step": 2186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011412672936935115,
+      "loss": 1.5438,
+      "step": 2187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011405893863561336,
+      "loss": 1.1647,
+      "step": 2188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011399114131005402,
+      "loss": 1.2185,
+      "step": 2189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011392333742446122,
+      "loss": 1.3331,
+      "step": 2190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011385552701062623,
+      "loss": 1.2117,
+      "step": 2191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011378771010034333,
+      "loss": 1.3833,
+      "step": 2192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011371988672540992,
+      "loss": 1.2279,
+      "step": 2193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011365205691762635,
+      "loss": 1.3217,
+      "step": 2194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011358422070879597,
+      "loss": 1.3459,
+      "step": 2195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011351637813072525,
+      "loss": 1.2739,
+      "step": 2196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011344852921522353,
+      "loss": 1.2689,
+      "step": 2197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011338067399410319,
+      "loss": 1.5252,
+      "step": 2198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011331281249917954,
+      "loss": 1.2535,
+      "step": 2199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001132449447622708,
+      "loss": 1.4809,
+      "step": 2200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011317707081519819,
+      "loss": 1.2621,
+      "step": 2201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011310919068978578,
+      "loss": 1.4831,
+      "step": 2202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011304130441786058,
+      "loss": 1.4558,
+      "step": 2203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001129734120312524,
+      "loss": 1.4244,
+      "step": 2204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011290551356179406,
+      "loss": 1.1233,
+      "step": 2205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001128376090413211,
+      "loss": 1.236,
+      "step": 2206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011276969850167197,
+      "loss": 1.2994,
+      "step": 2207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011270178197468789,
+      "loss": 1.5724,
+      "step": 2208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011263385949221295,
+      "loss": 1.3744,
+      "step": 2209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011256593108609394,
+      "loss": 1.387,
+      "step": 2210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011249799678818054,
+      "loss": 1.4552,
+      "step": 2211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011243005663032515,
+      "loss": 1.3181,
+      "step": 2212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011236211064438286,
+      "loss": 1.1834,
+      "step": 2213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011229415886221157,
+      "loss": 1.0759,
+      "step": 2214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011222620131567185,
+      "loss": 1.3505,
+      "step": 2215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000112158238036627,
+      "loss": 1.5284,
+      "step": 2216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011209026905694298,
+      "loss": 1.3092,
+      "step": 2217
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011202229440848844,
+      "loss": 1.2621,
+      "step": 2218
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011195431412313472,
+      "loss": 1.4068,
+      "step": 2219
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011188632823275573,
+      "loss": 1.3345,
+      "step": 2220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011181833676922806,
+      "loss": 1.5332,
+      "step": 2221
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011175033976443092,
+      "loss": 1.3077,
+      "step": 2222
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001116823372502461,
+      "loss": 1.3573,
+      "step": 2223
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011161432925855792,
+      "loss": 1.1828,
+      "step": 2224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011154631582125343,
+      "loss": 1.3471,
+      "step": 2225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011147829697022203,
+      "loss": 1.2556,
+      "step": 2226
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011141027273735574,
+      "loss": 1.4178,
+      "step": 2227
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011134224315454918,
+      "loss": 1.316,
+      "step": 2228
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011127420825369938,
+      "loss": 1.1374,
+      "step": 2229
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011120616806670594,
+      "loss": 1.2774,
+      "step": 2230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011113812262547083,
+      "loss": 1.3818,
+      "step": 2231
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001110700719618986,
+      "loss": 1.5821,
+      "step": 2232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001110020161078962,
+      "loss": 1.3265,
+      "step": 2233
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011093395509537294,
+      "loss": 1.2095,
+      "step": 2234
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011086588895624072,
+      "loss": 1.2797,
+      "step": 2235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011079781772241371,
+      "loss": 1.3754,
+      "step": 2236
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011072974142580847,
+      "loss": 1.3415,
+      "step": 2237
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000110661660098344,
+      "loss": 1.2483,
+      "step": 2238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011059357377194161,
+      "loss": 1.5109,
+      "step": 2239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011052548247852496,
+      "loss": 1.1453,
+      "step": 2240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011045738625002013,
+      "loss": 1.318,
+      "step": 2241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011038928511835533,
+      "loss": 1.4995,
+      "step": 2242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011032117911546121,
+      "loss": 1.229,
+      "step": 2243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011025306827327067,
+      "loss": 1.3277,
+      "step": 2244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011018495262371887,
+      "loss": 1.3861,
+      "step": 2245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011011683219874323,
+      "loss": 1.5137,
+      "step": 2246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001100487070302834,
+      "loss": 1.0308,
+      "step": 2247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010998057715028129,
+      "loss": 1.2034,
+      "step": 2248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010991244259068097,
+      "loss": 1.244,
+      "step": 2249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001098443033834287,
+      "loss": 1.4917,
+      "step": 2250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010977615956047298,
+      "loss": 1.2594,
+      "step": 2251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010970801115376448,
+      "loss": 1.2059,
+      "step": 2252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010963985819525587,
+      "loss": 1.16,
+      "step": 2253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010957170071690213,
+      "loss": 1.2773,
+      "step": 2254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010950353875066028,
+      "loss": 1.4602,
+      "step": 2255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010943537232848946,
+      "loss": 1.2605,
+      "step": 2256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010936720148235089,
+      "loss": 1.3074,
+      "step": 2257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010929902624420787,
+      "loss": 1.3927,
+      "step": 2258
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010923084664602574,
+      "loss": 1.5318,
+      "step": 2259
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010916266271977192,
+      "loss": 1.4084,
+      "step": 2260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010909447449741583,
+      "loss": 1.2713,
+      "step": 2261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010902628201092896,
+      "loss": 1.5237,
+      "step": 2262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001089580852922847,
+      "loss": 1.143,
+      "step": 2263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001088898843734585,
+      "loss": 1.3654,
+      "step": 2264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010882167928642779,
+      "loss": 1.2601,
+      "step": 2265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010875347006317185,
+      "loss": 1.3853,
+      "step": 2266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010868525673567202,
+      "loss": 1.3187,
+      "step": 2267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010861703933591157,
+      "loss": 1.1044,
+      "step": 2268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001085488178958755,
+      "loss": 1.4403,
+      "step": 2269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010848059244755093,
+      "loss": 1.4756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010841236302292672,
+      "loss": 1.3464,
+      "step": 2271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010834412965399366,
+      "loss": 1.5048,
+      "step": 2272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010827589237274433,
+      "loss": 1.1052,
+      "step": 2273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010820765121117322,
+      "loss": 1.2539,
+      "step": 2274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010813940620127656,
+      "loss": 1.5024,
+      "step": 2275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010807115737505242,
+      "loss": 1.5605,
+      "step": 2276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001080029047645007,
+      "loss": 1.312,
+      "step": 2277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010793464840162302,
+      "loss": 1.3965,
+      "step": 2278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010786638831842277,
+      "loss": 1.5931,
+      "step": 2279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010779812454690508,
+      "loss": 1.3808,
+      "step": 2280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010772985711907682,
+      "loss": 1.3318,
+      "step": 2281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010766158606694658,
+      "loss": 1.3624,
+      "step": 2282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010759331142252462,
+      "loss": 1.3524,
+      "step": 2283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000107525033217823,
+      "loss": 1.3037,
+      "step": 2284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010745675148485522,
+      "loss": 1.2641,
+      "step": 2285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010738846625563663,
+      "loss": 1.2699,
+      "step": 2286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001073201775621842,
+      "loss": 1.397,
+      "step": 2287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010725188543651646,
+      "loss": 1.3429,
+      "step": 2288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010718358991065354,
+      "loss": 1.2301,
+      "step": 2289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010711529101661721,
+      "loss": 1.3566,
+      "step": 2290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010704698878643083,
+      "loss": 1.3693,
+      "step": 2291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010697868325211926,
+      "loss": 1.3414,
+      "step": 2292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010691037444570902,
+      "loss": 1.3321,
+      "step": 2293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010684206239922803,
+      "loss": 1.6089,
+      "step": 2294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010677374714470581,
+      "loss": 1.5004,
+      "step": 2295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010670542871417339,
+      "loss": 1.3691,
+      "step": 2296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010663710713966323,
+      "loss": 1.2045,
+      "step": 2297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010656878245320933,
+      "loss": 1.5804,
+      "step": 2298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001065004546868471,
+      "loss": 1.1269,
+      "step": 2299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010643212387261345,
+      "loss": 1.4764,
+      "step": 2300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010636379004254664,
+      "loss": 1.3632,
+      "step": 2301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010629545322868639,
+      "loss": 1.4777,
+      "step": 2302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010622711346307385,
+      "loss": 1.3764,
+      "step": 2303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010615877077775149,
+      "loss": 1.3493,
+      "step": 2304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001060904252047632,
+      "loss": 1.4438,
+      "step": 2305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010602207677615421,
+      "loss": 1.1731,
+      "step": 2306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010595372552397105,
+      "loss": 1.4628,
+      "step": 2307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010588537148026158,
+      "loss": 1.3681,
+      "step": 2308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010581701467707507,
+      "loss": 1.2082,
+      "step": 2309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010574865514646197,
+      "loss": 1.342,
+      "step": 2310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010568029292047408,
+      "loss": 1.3963,
+      "step": 2311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010561192803116437,
+      "loss": 1.2611,
+      "step": 2312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010554356051058715,
+      "loss": 1.1997,
+      "step": 2313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010547519039079796,
+      "loss": 1.3483,
+      "step": 2314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010540681770385348,
+      "loss": 1.085,
+      "step": 2315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001053384424818117,
+      "loss": 1.0556,
+      "step": 2316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010527006475673172,
+      "loss": 1.2715,
+      "step": 2317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010520168456067381,
+      "loss": 1.2902,
+      "step": 2318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010513330192569946,
+      "loss": 1.4248,
+      "step": 2319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010506491688387127,
+      "loss": 1.4829,
+      "step": 2320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010499652946725298,
+      "loss": 1.0807,
+      "step": 2321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001049281397079094,
+      "loss": 1.3335,
+      "step": 2322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010485974763790649,
+      "loss": 1.4234,
+      "step": 2323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010479135328931126,
+      "loss": 1.2546,
+      "step": 2324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010472295669419181,
+      "loss": 1.2061,
+      "step": 2325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001046545578846173,
+      "loss": 1.3023,
+      "step": 2326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010458615689265794,
+      "loss": 1.5643,
+      "step": 2327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010451775375038484,
+      "loss": 1.2302,
+      "step": 2328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010444934848987029,
+      "loss": 1.1701,
+      "step": 2329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001043809411431875,
+      "loss": 1.2954,
+      "step": 2330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001043125317424106,
+      "loss": 1.3663,
+      "step": 2331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010424412031961484,
+      "loss": 1.1569,
+      "step": 2332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010417570690687624,
+      "loss": 1.5717,
+      "step": 2333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010410729153627183,
+      "loss": 1.2938,
+      "step": 2334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001040388742398796,
+      "loss": 1.4464,
+      "step": 2335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010397045504977839,
+      "loss": 1.3973,
+      "step": 2336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001039020339980479,
+      "loss": 1.4622,
+      "step": 2337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001038336111167688,
+      "loss": 1.212,
+      "step": 2338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010376518643802254,
+      "loss": 1.2741,
+      "step": 2339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010369675999389142,
+      "loss": 1.373,
+      "step": 2340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010362833181645855,
+      "loss": 1.2946,
+      "step": 2341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010355990193780796,
+      "loss": 1.3959,
+      "step": 2342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010349147039002438,
+      "loss": 1.2811,
+      "step": 2343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001034230372051933,
+      "loss": 1.3235,
+      "step": 2344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010335460241540106,
+      "loss": 1.4344,
+      "step": 2345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010328616605273471,
+      "loss": 1.3471,
+      "step": 2346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010321772814928204,
+      "loss": 1.2383,
+      "step": 2347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010314928873713152,
+      "loss": 1.4786,
+      "step": 2348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010308084784837249,
+      "loss": 1.1781,
+      "step": 2349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010301240551509474,
+      "loss": 1.3312,
+      "step": 2350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010294396176938887,
+      "loss": 1.3711,
+      "step": 2351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010287551664334622,
+      "loss": 1.1275,
+      "step": 2352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010280707016905864,
+      "loss": 1.2165,
+      "step": 2353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010273862237861864,
+      "loss": 0.9758,
+      "step": 2354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010267017330411943,
+      "loss": 1.4252,
+      "step": 2355
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010260172297765469,
+      "loss": 1.3268,
+      "step": 2356
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010253327143131879,
+      "loss": 1.2323,
+      "step": 2357
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010246481869720665,
+      "loss": 1.4011,
+      "step": 2358
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001023963648074138,
+      "loss": 1.3467,
+      "step": 2359
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010232790979403609,
+      "loss": 1.4156,
+      "step": 2360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010225945368917019,
+      "loss": 1.2585,
+      "step": 2361
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010219099652491308,
+      "loss": 1.2505,
+      "step": 2362
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010212253833336237,
+      "loss": 1.3207,
+      "step": 2363
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000102054079146616,
+      "loss": 1.2983,
+      "step": 2364
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010198561899677257,
+      "loss": 1.1677,
+      "step": 2365
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010191715791593095,
+      "loss": 1.2528,
+      "step": 2366
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001018486959361905,
+      "loss": 1.357,
+      "step": 2367
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010178023308965107,
+      "loss": 1.2146,
+      "step": 2368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010171176940841289,
+      "loss": 1.4325,
+      "step": 2369
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010164330492457652,
+      "loss": 1.2936,
+      "step": 2370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010157483967024295,
+      "loss": 1.0563,
+      "step": 2371
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001015063736775135,
+      "loss": 1.1661,
+      "step": 2372
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010143790697848985,
+      "loss": 1.2879,
+      "step": 2373
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010136943960527407,
+      "loss": 1.3976,
+      "step": 2374
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010130097158996845,
+      "loss": 1.1913,
+      "step": 2375
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010123250296467564,
+      "loss": 1.3279,
+      "step": 2376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010116403376149852,
+      "loss": 1.4465,
+      "step": 2377
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010109556401254032,
+      "loss": 1.0903,
+      "step": 2378
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010102709374990447,
+      "loss": 1.1722,
+      "step": 2379
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010095862300569466,
+      "loss": 1.3093,
+      "step": 2380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010089015181201483,
+      "loss": 1.4897,
+      "step": 2381
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010082168020096903,
+      "loss": 1.2511,
+      "step": 2382
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001007532082046616,
+      "loss": 1.3868,
+      "step": 2383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010068473585519713,
+      "loss": 1.0906,
+      "step": 2384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010061626318468019,
+      "loss": 1.3598,
+      "step": 2385
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010054779022521561,
+      "loss": 1.2723,
+      "step": 2386
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010047931700890835,
+      "loss": 1.1959,
+      "step": 2387
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001004108435678635,
+      "loss": 1.1107,
+      "step": 2388
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010034236993418623,
+      "loss": 1.3116,
+      "step": 2389
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010027389613998177,
+      "loss": 1.3781,
+      "step": 2390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001002054222173555,
+      "loss": 1.27,
+      "step": 2391
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010013694819841284,
+      "loss": 1.2073,
+      "step": 2392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010006847411525914,
+      "loss": 1.4518,
+      "step": 2393
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001,
+      "loss": 1.4011,
+      "step": 2394
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.993152588474088e-05,
+      "loss": 1.5221,
+      "step": 2395
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.986305180158719e-05,
+      "loss": 1.1057,
+      "step": 2396
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.979457778264452e-05,
+      "loss": 1.1188,
+      "step": 2397
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.972610386001824e-05,
+      "loss": 1.2235,
+      "step": 2398
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.965763006581378e-05,
+      "loss": 1.0544,
+      "step": 2399
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.958915643213653e-05,
+      "loss": 1.2145,
+      "step": 2400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.952068299109166e-05,
+      "loss": 1.0554,
+      "step": 2401
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.945220977478441e-05,
+      "loss": 1.4011,
+      "step": 2402
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.938373681531981e-05,
+      "loss": 1.53,
+      "step": 2403
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.93152641448029e-05,
+      "loss": 1.3329,
+      "step": 2404
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.92467917953384e-05,
+      "loss": 1.3289,
+      "step": 2405
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.9178319799031e-05,
+      "loss": 1.3769,
+      "step": 2406
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.910984818798524e-05,
+      "loss": 1.4384,
+      "step": 2407
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.904137699430535e-05,
+      "loss": 1.3607,
+      "step": 2408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.897290625009554e-05,
+      "loss": 1.583,
+      "step": 2409
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.890443598745968e-05,
+      "loss": 1.3587,
+      "step": 2410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.88359662385015e-05,
+      "loss": 1.4904,
+      "step": 2411
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.87674970353244e-05,
+      "loss": 1.3891,
+      "step": 2412
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.869902841003154e-05,
+      "loss": 1.2836,
+      "step": 2413
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.863056039472594e-05,
+      "loss": 1.3277,
+      "step": 2414
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.856209302151014e-05,
+      "loss": 1.1022,
+      "step": 2415
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.849362632248652e-05,
+      "loss": 1.2777,
+      "step": 2416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.842516032975709e-05,
+      "loss": 1.1516,
+      "step": 2417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.83566950754235e-05,
+      "loss": 1.3876,
+      "step": 2418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.828823059158713e-05,
+      "loss": 1.2636,
+      "step": 2419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.821976691034891e-05,
+      "loss": 1.3035,
+      "step": 2420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.815130406380952e-05,
+      "loss": 1.1884,
+      "step": 2421
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.808284208406909e-05,
+      "loss": 1.2562,
+      "step": 2422
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.801438100322745e-05,
+      "loss": 1.2574,
+      "step": 2423
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794592085338401e-05,
+      "loss": 1.483,
+      "step": 2424
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.787746166663764e-05,
+      "loss": 0.9844,
+      "step": 2425
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.780900347508693e-05,
+      "loss": 1.5137,
+      "step": 2426
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.774054631082986e-05,
+      "loss": 1.4183,
+      "step": 2427
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.767209020596394e-05,
+      "loss": 1.4139,
+      "step": 2428
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.760363519258627e-05,
+      "loss": 1.3653,
+      "step": 2429
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.753518130279333e-05,
+      "loss": 1.1514,
+      "step": 2430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.746672856868123e-05,
+      "loss": 1.5165,
+      "step": 2431
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.739827702234535e-05,
+      "loss": 1.4025,
+      "step": 2432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.73298266958806e-05,
+      "loss": 1.4242,
+      "step": 2433
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.726137762138139e-05,
+      "loss": 1.0915,
+      "step": 2434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.719292983094138e-05,
+      "loss": 1.3396,
+      "step": 2435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.712448335665379e-05,
+      "loss": 1.3058,
+      "step": 2436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.705603823061111e-05,
+      "loss": 1.2976,
+      "step": 2437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.698759448490528e-05,
+      "loss": 1.408,
+      "step": 2438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.691915215162756e-05,
+      "loss": 1.1842,
+      "step": 2439
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.685071126286846e-05,
+      "loss": 1.1416,
+      "step": 2440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.678227185071799e-05,
+      "loss": 1.1788,
+      "step": 2441
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.671383394726528e-05,
+      "loss": 1.2128,
+      "step": 2442
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.664539758459896e-05,
+      "loss": 1.327,
+      "step": 2443
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.657696279480672e-05,
+      "loss": 1.2714,
+      "step": 2444
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.650852960997564e-05,
+      "loss": 1.3905,
+      "step": 2445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.644009806219207e-05,
+      "loss": 1.4073,
+      "step": 2446
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.637166818354146e-05,
+      "loss": 0.9863,
+      "step": 2447
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.630324000610862e-05,
+      "loss": 1.2915,
+      "step": 2448
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.623481356197753e-05,
+      "loss": 1.395,
+      "step": 2449
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.616638888323121e-05,
+      "loss": 1.312,
+      "step": 2450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.609796600195211e-05,
+      "loss": 1.3739,
+      "step": 2451
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.602954495022162e-05,
+      "loss": 1.2406,
+      "step": 2452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.596112576012041e-05,
+      "loss": 1.3288,
+      "step": 2453
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.589270846372819e-05,
+      "loss": 1.3316,
+      "step": 2454
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.582429309312377e-05,
+      "loss": 1.2947,
+      "step": 2455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.57558796803852e-05,
+      "loss": 1.4938,
+      "step": 2456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.568746825758938e-05,
+      "loss": 1.1724,
+      "step": 2457
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.561905885681252e-05,
+      "loss": 1.2279,
+      "step": 2458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.555065151012975e-05,
+      "loss": 1.4305,
+      "step": 2459
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.548224624961519e-05,
+      "loss": 1.3746,
+      "step": 2460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.541384310734211e-05,
+      "loss": 1.5133,
+      "step": 2461
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.534544211538269e-05,
+      "loss": 1.2928,
+      "step": 2462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.527704330580822e-05,
+      "loss": 1.5905,
+      "step": 2463
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.520864671068874e-05,
+      "loss": 1.523,
+      "step": 2464
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.514025236209353e-05,
+      "loss": 1.4138,
+      "step": 2465
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.507186029209064e-05,
+      "loss": 1.2153,
+      "step": 2466
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.500347053274704e-05,
+      "loss": 1.4496,
+      "step": 2467
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.493508311612874e-05,
+      "loss": 1.6821,
+      "step": 2468
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.486669807430054e-05,
+      "loss": 1.2478,
+      "step": 2469
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.479831543932621e-05,
+      "loss": 0.9271,
+      "step": 2470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.472993524326832e-05,
+      "loss": 1.2457,
+      "step": 2471
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.466155751818832e-05,
+      "loss": 1.4963,
+      "step": 2472
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.459318229614654e-05,
+      "loss": 1.0869,
+      "step": 2473
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.452480960920206e-05,
+      "loss": 1.5135,
+      "step": 2474
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.445643948941286e-05,
+      "loss": 1.2396,
+      "step": 2475
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.438807196883568e-05,
+      "loss": 1.2775,
+      "step": 2476
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.431970707952595e-05,
+      "loss": 1.3609,
+      "step": 2477
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.425134485353804e-05,
+      "loss": 1.3017,
+      "step": 2478
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.418298532292492e-05,
+      "loss": 1.2871,
+      "step": 2479
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.411462851973844e-05,
+      "loss": 1.2736,
+      "step": 2480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.404627447602901e-05,
+      "loss": 1.3469,
+      "step": 2481
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.397792322384583e-05,
+      "loss": 1.2192,
+      "step": 2482
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.390957479523684e-05,
+      "loss": 1.1235,
+      "step": 2483
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.38412292222485e-05,
+      "loss": 1.1493,
+      "step": 2484
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.377288653692617e-05,
+      "loss": 1.1034,
+      "step": 2485
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.370454677131363e-05,
+      "loss": 1.0551,
+      "step": 2486
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.363620995745337e-05,
+      "loss": 1.3318,
+      "step": 2487
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.356787612738659e-05,
+      "loss": 1.4806,
+      "step": 2488
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.34995453131529e-05,
+      "loss": 1.5647,
+      "step": 2489
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.343121754679069e-05,
+      "loss": 1.4368,
+      "step": 2490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.33628928603368e-05,
+      "loss": 1.303,
+      "step": 2491
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.329457128582663e-05,
+      "loss": 1.2806,
+      "step": 2492
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.32262528552942e-05,
+      "loss": 1.0971,
+      "step": 2493
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.315793760077197e-05,
+      "loss": 1.2662,
+      "step": 2494
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.308962555429101e-05,
+      "loss": 1.395,
+      "step": 2495
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.302131674788073e-05,
+      "loss": 1.1905,
+      "step": 2496
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.295301121356918e-05,
+      "loss": 1.2981,
+      "step": 2497
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.288470898338283e-05,
+      "loss": 1.3341,
+      "step": 2498
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.28164100893465e-05,
+      "loss": 1.5436,
+      "step": 2499
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.274811456348358e-05,
+      "loss": 1.4758,
+      "step": 2500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.267982243781579e-05,
+      "loss": 1.2972,
+      "step": 2501
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.261153374436338e-05,
+      "loss": 1.2694,
+      "step": 2502
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.254324851514482e-05,
+      "loss": 1.3429,
+      "step": 2503
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.247496678217703e-05,
+      "loss": 1.3591,
+      "step": 2504
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.24066885774754e-05,
+      "loss": 1.3674,
+      "step": 2505
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.233841393305344e-05,
+      "loss": 1.3291,
+      "step": 2506
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.227014288092321e-05,
+      "loss": 1.2671,
+      "step": 2507
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.220187545309498e-05,
+      "loss": 1.4542,
+      "step": 2508
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.213361168157726e-05,
+      "loss": 1.362,
+      "step": 2509
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.206535159837701e-05,
+      "loss": 1.2656,
+      "step": 2510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.199709523549929e-05,
+      "loss": 1.4024,
+      "step": 2511
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.192884262494759e-05,
+      "loss": 1.4188,
+      "step": 2512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.186059379872347e-05,
+      "loss": 1.3814,
+      "step": 2513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.17923487888268e-05,
+      "loss": 1.0734,
+      "step": 2514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.172410762725569e-05,
+      "loss": 1.5432,
+      "step": 2515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.165587034600635e-05,
+      "loss": 1.1896,
+      "step": 2516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.15876369770733e-05,
+      "loss": 1.1007,
+      "step": 2517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.151940755244912e-05,
+      "loss": 1.041,
+      "step": 2518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.145118210412452e-05,
+      "loss": 1.3814,
+      "step": 2519
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.138296066408848e-05,
+      "loss": 1.2229,
+      "step": 2520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.131474326432797e-05,
+      "loss": 1.4098,
+      "step": 2521
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.124652993682818e-05,
+      "loss": 1.3749,
+      "step": 2522
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.117832071357224e-05,
+      "loss": 1.2282,
+      "step": 2523
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.11101156265415e-05,
+      "loss": 1.2128,
+      "step": 2524
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.104191470771533e-05,
+      "loss": 1.4746,
+      "step": 2525
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.097371798907104e-05,
+      "loss": 1.4982,
+      "step": 2526
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.090552550258418e-05,
+      "loss": 1.3061,
+      "step": 2527
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.083733728022809e-05,
+      "loss": 1.5959,
+      "step": 2528
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.076915335397428e-05,
+      "loss": 1.5046,
+      "step": 2529
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.070097375579218e-05,
+      "loss": 1.2018,
+      "step": 2530
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.063279851764915e-05,
+      "loss": 1.3525,
+      "step": 2531
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.056462767151057e-05,
+      "loss": 1.154,
+      "step": 2532
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.049646124933973e-05,
+      "loss": 1.2433,
+      "step": 2533
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.04282992830979e-05,
+      "loss": 1.3988,
+      "step": 2534
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.036014180474417e-05,
+      "loss": 1.2147,
+      "step": 2535
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.029198884623556e-05,
+      "loss": 1.2923,
+      "step": 2536
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.022384043952703e-05,
+      "loss": 1.2147,
+      "step": 2537
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.01556966165713e-05,
+      "loss": 1.4101,
+      "step": 2538
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.008755740931905e-05,
+      "loss": 1.2574,
+      "step": 2539
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.001942284971875e-05,
+      "loss": 1.2443,
+      "step": 2540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.995129296971661e-05,
+      "loss": 1.3622,
+      "step": 2541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.98831678012568e-05,
+      "loss": 1.0291,
+      "step": 2542
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.981504737628114e-05,
+      "loss": 1.171,
+      "step": 2543
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.974693172672937e-05,
+      "loss": 1.117,
+      "step": 2544
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.967882088453883e-05,
+      "loss": 1.2522,
+      "step": 2545
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.961071488164469e-05,
+      "loss": 1.0349,
+      "step": 2546
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.954261374997992e-05,
+      "loss": 1.3344,
+      "step": 2547
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.947451752147503e-05,
+      "loss": 1.3537,
+      "step": 2548
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.94064262280584e-05,
+      "loss": 1.0394,
+      "step": 2549
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.9338339901656e-05,
+      "loss": 1.1777,
+      "step": 2550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.927025857419156e-05,
+      "loss": 1.2338,
+      "step": 2551
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.920218227758633e-05,
+      "loss": 1.1146,
+      "step": 2552
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.913411104375929e-05,
+      "loss": 1.2729,
+      "step": 2553
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.906604490462707e-05,
+      "loss": 1.354,
+      "step": 2554
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.899798389210383e-05,
+      "loss": 1.1505,
+      "step": 2555
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.892992803810142e-05,
+      "loss": 1.3014,
+      "step": 2556
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.88618773745292e-05,
+      "loss": 1.4241,
+      "step": 2557
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.879383193329409e-05,
+      "loss": 1.4367,
+      "step": 2558
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.872579174630063e-05,
+      "loss": 1.2905,
+      "step": 2559
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.865775684545083e-05,
+      "loss": 1.3302,
+      "step": 2560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.858972726264429e-05,
+      "loss": 1.4992,
+      "step": 2561
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.852170302977803e-05,
+      "loss": 1.2359,
+      "step": 2562
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.845368417874659e-05,
+      "loss": 1.311,
+      "step": 2563
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.838567074144209e-05,
+      "loss": 1.2432,
+      "step": 2564
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.831766274975391e-05,
+      "loss": 1.4455,
+      "step": 2565
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.824966023556909e-05,
+      "loss": 1.095,
+      "step": 2566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.818166323077197e-05,
+      "loss": 1.3136,
+      "step": 2567
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.81136717672443e-05,
+      "loss": 1.1454,
+      "step": 2568
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.804568587686533e-05,
+      "loss": 1.3002,
+      "step": 2569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.797770559151157e-05,
+      "loss": 1.4022,
+      "step": 2570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.790973094305705e-05,
+      "loss": 0.9264,
+      "step": 2571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.784176196337305e-05,
+      "loss": 1.3017,
+      "step": 2572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.777379868432817e-05,
+      "loss": 1.3044,
+      "step": 2573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.770584113778846e-05,
+      "loss": 1.3338,
+      "step": 2574
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.763788935561715e-05,
+      "loss": 1.2939,
+      "step": 2575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.756994336967487e-05,
+      "loss": 1.4623,
+      "step": 2576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.750200321181945e-05,
+      "loss": 1.1173,
+      "step": 2577
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.743406891390607e-05,
+      "loss": 1.2136,
+      "step": 2578
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.73661405077871e-05,
+      "loss": 1.5109,
+      "step": 2579
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.729821802531212e-05,
+      "loss": 1.3039,
+      "step": 2580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.723030149832804e-05,
+      "loss": 1.3146,
+      "step": 2581
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.716239095867889e-05,
+      "loss": 1.3476,
+      "step": 2582
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.709448643820596e-05,
+      "loss": 1.2372,
+      "step": 2583
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.702658796874761e-05,
+      "loss": 1.3483,
+      "step": 2584
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.695869558213944e-05,
+      "loss": 1.2255,
+      "step": 2585
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.689080931021425e-05,
+      "loss": 1.5617,
+      "step": 2586
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.682292918480182e-05,
+      "loss": 1.2,
+      "step": 2587
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.67550552377292e-05,
+      "loss": 1.3436,
+      "step": 2588
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.66871875008205e-05,
+      "loss": 1.2136,
+      "step": 2589
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.661932600589682e-05,
+      "loss": 1.2996,
+      "step": 2590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.655147078477648e-05,
+      "loss": 1.1975,
+      "step": 2591
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.648362186927475e-05,
+      "loss": 1.4238,
+      "step": 2592
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.641577929120406e-05,
+      "loss": 1.3109,
+      "step": 2593
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.63479430823737e-05,
+      "loss": 1.3716,
+      "step": 2594
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.628011327459009e-05,
+      "loss": 1.4083,
+      "step": 2595
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.62122898996567e-05,
+      "loss": 1.0265,
+      "step": 2596
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.614447298937379e-05,
+      "loss": 1.623,
+      "step": 2597
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.607666257553881e-05,
+      "loss": 1.284,
+      "step": 2598
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.600885868994603e-05,
+      "loss": 1.4076,
+      "step": 2599
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.594106136438664e-05,
+      "loss": 1.372,
+      "step": 2600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.58732706306489e-05,
+      "loss": 1.1824,
+      "step": 2601
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.580548652051778e-05,
+      "loss": 1.1305,
+      "step": 2602
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.573770906577538e-05,
+      "loss": 1.3464,
+      "step": 2603
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.566993829820043e-05,
+      "loss": 1.4076,
+      "step": 2604
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.560217424956868e-05,
+      "loss": 1.4262,
+      "step": 2605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.553441695165277e-05,
+      "loss": 1.0481,
+      "step": 2606
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.546666643622195e-05,
+      "loss": 1.304,
+      "step": 2607
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.539892273504255e-05,
+      "loss": 1.1429,
+      "step": 2608
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.533118587987756e-05,
+      "loss": 1.1406,
+      "step": 2609
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.526345590248682e-05,
+      "loss": 1.144,
+      "step": 2610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.519573283462687e-05,
+      "loss": 1.5387,
+      "step": 2611
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.512801670805108e-05,
+      "loss": 1.3775,
+      "step": 2612
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.506030755450957e-05,
+      "loss": 1.4037,
+      "step": 2613
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.49926054057491e-05,
+      "loss": 1.2393,
+      "step": 2614
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.49249102935133e-05,
+      "loss": 1.3773,
+      "step": 2615
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.485722224954237e-05,
+      "loss": 1.5492,
+      "step": 2616
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.478954130557316e-05,
+      "loss": 1.3888,
+      "step": 2617
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.472186749333941e-05,
+      "loss": 1.3079,
+      "step": 2618
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.465420084457125e-05,
+      "loss": 1.1947,
+      "step": 2619
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.458654139099567e-05,
+      "loss": 1.2753,
+      "step": 2620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.451888916433615e-05,
+      "loss": 1.3631,
+      "step": 2621
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.445124419631278e-05,
+      "loss": 1.21,
+      "step": 2622
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.438360651864238e-05,
+      "loss": 1.0762,
+      "step": 2623
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.431597616303817e-05,
+      "loss": 1.5168,
+      "step": 2624
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.424835316121013e-05,
+      "loss": 1.4471,
+      "step": 2625
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.41807375448646e-05,
+      "loss": 1.1848,
+      "step": 2626
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.411312934570458e-05,
+      "loss": 1.2619,
+      "step": 2627
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.40455285954296e-05,
+      "loss": 1.206,
+      "step": 2628
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.397793532573556e-05,
+      "loss": 1.1488,
+      "step": 2629
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.391034956831506e-05,
+      "loss": 1.4595,
+      "step": 2630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.384277135485704e-05,
+      "loss": 1.2979,
+      "step": 2631
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.377520071704684e-05,
+      "loss": 1.2214,
+      "step": 2632
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.370763768656644e-05,
+      "loss": 1.3295,
+      "step": 2633
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.364008229509411e-05,
+      "loss": 1.4018,
+      "step": 2634
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.357253457430463e-05,
+      "loss": 1.4281,
+      "step": 2635
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.350499455586905e-05,
+      "loss": 1.1653,
+      "step": 2636
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.343746227145496e-05,
+      "loss": 1.28,
+      "step": 2637
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.336993775272624e-05,
+      "loss": 1.2469,
+      "step": 2638
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.330242103134311e-05,
+      "loss": 1.1612,
+      "step": 2639
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.323491213896222e-05,
+      "loss": 1.3723,
+      "step": 2640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.316741110723645e-05,
+      "loss": 1.4248,
+      "step": 2641
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.309991796781511e-05,
+      "loss": 1.2652,
+      "step": 2642
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.303243275234367e-05,
+      "loss": 1.4744,
+      "step": 2643
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.296495549246399e-05,
+      "loss": 1.3914,
+      "step": 2644
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.289748621981418e-05,
+      "loss": 1.3452,
+      "step": 2645
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.283002496602854e-05,
+      "loss": 1.3075,
+      "step": 2646
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.276257176273772e-05,
+      "loss": 1.2645,
+      "step": 2647
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.269512664156854e-05,
+      "loss": 1.3497,
+      "step": 2648
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.262768963414393e-05,
+      "loss": 1.247,
+      "step": 2649
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.256026077208323e-05,
+      "loss": 1.3302,
+      "step": 2650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.249284008700178e-05,
+      "loss": 1.365,
+      "step": 2651
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.24254276105112e-05,
+      "loss": 1.3271,
+      "step": 2652
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.235802337421919e-05,
+      "loss": 1.2529,
+      "step": 2653
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.229062740972954e-05,
+      "loss": 1.2883,
+      "step": 2654
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.222323974864235e-05,
+      "loss": 1.273,
+      "step": 2655
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.215586042255362e-05,
+      "loss": 1.5236,
+      "step": 2656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.208848946305554e-05,
+      "loss": 1.2507,
+      "step": 2657
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.202112690173643e-05,
+      "loss": 1.334,
+      "step": 2658
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.195377277018045e-05,
+      "loss": 1.2083,
+      "step": 2659
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.18864270999682e-05,
+      "loss": 1.1277,
+      "step": 2660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.181908992267586e-05,
+      "loss": 1.2136,
+      "step": 2661
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.175176126987596e-05,
+      "loss": 1.5287,
+      "step": 2662
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.16844411731369e-05,
+      "loss": 1.1566,
+      "step": 2663
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.16171296640231e-05,
+      "loss": 1.135,
+      "step": 2664
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.154982677409493e-05,
+      "loss": 1.3174,
+      "step": 2665
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.148253253490868e-05,
+      "loss": 1.2081,
+      "step": 2666
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.141524697801674e-05,
+      "loss": 1.2484,
+      "step": 2667
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.13479701349672e-05,
+      "loss": 1.4226,
+      "step": 2668
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.128070203730427e-05,
+      "loss": 1.4112,
+      "step": 2669
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.121344271656799e-05,
+      "loss": 1.5287,
+      "step": 2670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.114619220429419e-05,
+      "loss": 1.1998,
+      "step": 2671
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.107895053201471e-05,
+      "loss": 1.2517,
+      "step": 2672
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.101171773125716e-05,
+      "loss": 1.3228,
+      "step": 2673
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.094449383354508e-05,
+      "loss": 1.3275,
+      "step": 2674
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.087727887039769e-05,
+      "loss": 1.2485,
+      "step": 2675
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.081007287333011e-05,
+      "loss": 1.4442,
+      "step": 2676
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.074287587385332e-05,
+      "loss": 1.4918,
+      "step": 2677
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.067568790347394e-05,
+      "loss": 1.4124,
+      "step": 2678
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.060850899369447e-05,
+      "loss": 1.5446,
+      "step": 2679
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.054133917601312e-05,
+      "loss": 1.2405,
+      "step": 2680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.047417848192378e-05,
+      "loss": 1.1008,
+      "step": 2681
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.040702694291618e-05,
+      "loss": 1.3024,
+      "step": 2682
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.033988459047564e-05,
+      "loss": 1.1506,
+      "step": 2683
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.027275145608332e-05,
+      "loss": 1.4313,
+      "step": 2684
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.020562757121588e-05,
+      "loss": 1.321,
+      "step": 2685
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.013851296734574e-05,
+      "loss": 1.0884,
+      "step": 2686
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.0071407675941e-05,
+      "loss": 1.1677,
+      "step": 2687
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.000431172846531e-05,
+      "loss": 1.1607,
+      "step": 2688
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.993722515637802e-05,
+      "loss": 1.396,
+      "step": 2689
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.987014799113397e-05,
+      "loss": 1.2132,
+      "step": 2690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.980308026418379e-05,
+      "loss": 1.3151,
+      "step": 2691
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.973602200697346e-05,
+      "loss": 1.254,
+      "step": 2692
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.966897325094462e-05,
+      "loss": 1.333,
+      "step": 2693
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.960193402753451e-05,
+      "loss": 1.2779,
+      "step": 2694
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.95349043681758e-05,
+      "loss": 1.3003,
+      "step": 2695
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.946788430429672e-05,
+      "loss": 1.2087,
+      "step": 2696
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.940087386732104e-05,
+      "loss": 1.0884,
+      "step": 2697
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.933387308866793e-05,
+      "loss": 1.2636,
+      "step": 2698
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.926688199975209e-05,
+      "loss": 1.3964,
+      "step": 2699
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.919990063198368e-05,
+      "loss": 1.2726,
+      "step": 2700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.91329290167683e-05,
+      "loss": 1.1667,
+      "step": 2701
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.906596718550694e-05,
+      "loss": 1.197,
+      "step": 2702
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.8999015169596e-05,
+      "loss": 0.8579,
+      "step": 2703
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.89320730004274e-05,
+      "loss": 1.1718,
+      "step": 2704
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.88651407093882e-05,
+      "loss": 1.4263,
+      "step": 2705
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.879821832786115e-05,
+      "loss": 1.073,
+      "step": 2706
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.873130588722407e-05,
+      "loss": 1.1662,
+      "step": 2707
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.866440341885022e-05,
+      "loss": 1.2947,
+      "step": 2708
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.85975109541083e-05,
+      "loss": 1.5157,
+      "step": 2709
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.853062852436207e-05,
+      "loss": 1.3396,
+      "step": 2710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.846375616097083e-05,
+      "loss": 1.2537,
+      "step": 2711
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.839689389528906e-05,
+      "loss": 1.0238,
+      "step": 2712
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.833004175866642e-05,
+      "loss": 1.3365,
+      "step": 2713
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.826319978244798e-05,
+      "loss": 1.1814,
+      "step": 2714
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.819636799797391e-05,
+      "loss": 1.5051,
+      "step": 2715
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.812954643657977e-05,
+      "loss": 1.2936,
+      "step": 2716
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.80627351295961e-05,
+      "loss": 1.401,
+      "step": 2717
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.799593410834882e-05,
+      "loss": 1.1781,
+      "step": 2718
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.792914340415897e-05,
+      "loss": 1.1593,
+      "step": 2719
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.786236304834264e-05,
+      "loss": 1.1786,
+      "step": 2720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.779559307221129e-05,
+      "loss": 1.0686,
+      "step": 2721
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.77288335070713e-05,
+      "loss": 1.2163,
+      "step": 2722
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.766208438422436e-05,
+      "loss": 1.3011,
+      "step": 2723
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.759534573496707e-05,
+      "loss": 1.1897,
+      "step": 2724
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.752861759059125e-05,
+      "loss": 1.2822,
+      "step": 2725
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.746189998238378e-05,
+      "loss": 1.325,
+      "step": 2726
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.739519294162652e-05,
+      "loss": 1.1635,
+      "step": 2727
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.73284964995965e-05,
+      "loss": 1.5355,
+      "step": 2728
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.72618106875657e-05,
+      "loss": 1.2913,
+      "step": 2729
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.719513553680107e-05,
+      "loss": 1.3565,
+      "step": 2730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.712847107856468e-05,
+      "loss": 1.3125,
+      "step": 2731
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.70618173441135e-05,
+      "loss": 1.296,
+      "step": 2732
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.699517436469956e-05,
+      "loss": 1.0849,
+      "step": 2733
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.692854217156973e-05,
+      "loss": 1.0785,
+      "step": 2734
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.686192079596586e-05,
+      "loss": 1.2861,
+      "step": 2735
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.67953102691248e-05,
+      "loss": 1.2635,
+      "step": 2736
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.672871062227823e-05,
+      "loss": 1.4796,
+      "step": 2737
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.666212188665278e-05,
+      "loss": 1.3816,
+      "step": 2738
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.659554409346997e-05,
+      "loss": 1.3039,
+      "step": 2739
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.652897727394605e-05,
+      "loss": 1.4139,
+      "step": 2740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.646242145929235e-05,
+      "loss": 1.248,
+      "step": 2741
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.639587668071488e-05,
+      "loss": 1.0504,
+      "step": 2742
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.632934296941458e-05,
+      "loss": 1.2264,
+      "step": 2743
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.626282035658703e-05,
+      "loss": 1.2898,
+      "step": 2744
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.619630887342286e-05,
+      "loss": 1.4614,
+      "step": 2745
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.612980855110727e-05,
+      "loss": 1.2629,
+      "step": 2746
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.606331942082025e-05,
+      "loss": 1.1559,
+      "step": 2747
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.59968415137367e-05,
+      "loss": 1.3727,
+      "step": 2748
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.593037486102605e-05,
+      "loss": 1.6377,
+      "step": 2749
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.586391949385263e-05,
+      "loss": 1.3056,
+      "step": 2750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.579747544337539e-05,
+      "loss": 1.1528,
+      "step": 2751
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.573104274074791e-05,
+      "loss": 1.099,
+      "step": 2752
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.566462141711861e-05,
+      "loss": 1.4,
+      "step": 2753
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.559821150363044e-05,
+      "loss": 1.505,
+      "step": 2754
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.55318130314211e-05,
+      "loss": 1.4853,
+      "step": 2755
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.546542603162283e-05,
+      "loss": 1.4948,
+      "step": 2756
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.539905053536253e-05,
+      "loss": 1.5201,
+      "step": 2757
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.533268657376175e-05,
+      "loss": 1.3168,
+      "step": 2758
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.526633417793656e-05,
+      "loss": 1.3755,
+      "step": 2759
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.519999337899767e-05,
+      "loss": 1.3461,
+      "step": 2760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.513366420805032e-05,
+      "loss": 1.3064,
+      "step": 2761
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.506734669619426e-05,
+      "loss": 1.2522,
+      "step": 2762
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.500104087452387e-05,
+      "loss": 1.0341,
+      "step": 2763
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.493474677412794e-05,
+      "loss": 1.0938,
+      "step": 2764
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.486846442608989e-05,
+      "loss": 1.5744,
+      "step": 2765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.48021938614875e-05,
+      "loss": 1.192,
+      "step": 2766
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.473593511139307e-05,
+      "loss": 1.2488,
+      "step": 2767
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.466968820687346e-05,
+      "loss": 1.168,
+      "step": 2768
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.460345317898978e-05,
+      "loss": 1.3697,
+      "step": 2769
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.453723005879773e-05,
+      "loss": 1.2555,
+      "step": 2770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.447101887734744e-05,
+      "loss": 1.3346,
+      "step": 2771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.440481966568326e-05,
+      "loss": 1.4929,
+      "step": 2772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.433863245484412e-05,
+      "loss": 1.5321,
+      "step": 2773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.427245727586322e-05,
+      "loss": 1.3397,
+      "step": 2774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.420629415976821e-05,
+      "loss": 1.4851,
+      "step": 2775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.414014313758095e-05,
+      "loss": 1.2251,
+      "step": 2776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.407400424031774e-05,
+      "loss": 1.0805,
+      "step": 2777
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.400787749898919e-05,
+      "loss": 1.3705,
+      "step": 2778
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.394176294460006e-05,
+      "loss": 1.3512,
+      "step": 2779
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.387566060814965e-05,
+      "loss": 1.3701,
+      "step": 2780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.38095705206313e-05,
+      "loss": 1.2088,
+      "step": 2781
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.37434927130328e-05,
+      "loss": 1.3743,
+      "step": 2782
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.367742721633599e-05,
+      "loss": 1.2166,
+      "step": 2783
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.361137406151706e-05,
+      "loss": 1.3848,
+      "step": 2784
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.354533327954643e-05,
+      "loss": 1.5105,
+      "step": 2785
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.347930490138861e-05,
+      "loss": 1.2693,
+      "step": 2786
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.341328895800239e-05,
+      "loss": 1.2362,
+      "step": 2787
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.334728548034073e-05,
+      "loss": 1.4071,
+      "step": 2788
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.328129449935062e-05,
+      "loss": 1.4322,
+      "step": 2789
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.321531604597338e-05,
+      "loss": 1.355,
+      "step": 2790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.314935015114431e-05,
+      "loss": 1.3781,
+      "step": 2791
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.308339684579291e-05,
+      "loss": 1.3415,
+      "step": 2792
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.301745616084272e-05,
+      "loss": 1.304,
+      "step": 2793
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.295152812721134e-05,
+      "loss": 1.106,
+      "step": 2794
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.288561277581055e-05,
+      "loss": 1.423,
+      "step": 2795
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.281971013754605e-05,
+      "loss": 1.4137,
+      "step": 2796
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.275382024331772e-05,
+      "loss": 1.3649,
+      "step": 2797
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.268794312401934e-05,
+      "loss": 1.2753,
+      "step": 2798
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.262207881053871e-05,
+      "loss": 1.4393,
+      "step": 2799
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.255622733375775e-05,
+      "loss": 1.3952,
+      "step": 2800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.24903887245522e-05,
+      "loss": 1.3888,
+      "step": 2801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.242456301379187e-05,
+      "loss": 1.337,
+      "step": 2802
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.235875023234049e-05,
+      "loss": 1.2203,
+      "step": 2803
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.229295041105578e-05,
+      "loss": 1.124,
+      "step": 2804
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.222716358078927e-05,
+      "loss": 1.1679,
+      "step": 2805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.216138977238646e-05,
+      "loss": 1.4641,
+      "step": 2806
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.209562901668682e-05,
+      "loss": 1.188,
+      "step": 2807
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.202988134452354e-05,
+      "loss": 1.1793,
+      "step": 2808
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.196414678672383e-05,
+      "loss": 1.2862,
+      "step": 2809
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.18984253741087e-05,
+      "loss": 1.2332,
+      "step": 2810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.183271713749286e-05,
+      "loss": 1.3818,
+      "step": 2811
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.17670221076851e-05,
+      "loss": 1.4281,
+      "step": 2812
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.170134031548781e-05,
+      "loss": 1.3631,
+      "step": 2813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.16356717916973e-05,
+      "loss": 1.3155,
+      "step": 2814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.157001656710353e-05,
+      "loss": 1.4299,
+      "step": 2815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.150437467249034e-05,
+      "loss": 1.5238,
+      "step": 2816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.143874613863531e-05,
+      "loss": 1.5297,
+      "step": 2817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.137313099630965e-05,
+      "loss": 1.3449,
+      "step": 2818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.130752927627839e-05,
+      "loss": 1.4163,
+      "step": 2819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.124194100930031e-05,
+      "loss": 1.5476,
+      "step": 2820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.117636622612771e-05,
+      "loss": 1.2581,
+      "step": 2821
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.111080495750673e-05,
+      "loss": 1.3385,
+      "step": 2822
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.104525723417708e-05,
+      "loss": 1.3737,
+      "step": 2823
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.097972308687224e-05,
+      "loss": 1.4327,
+      "step": 2824
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.091420254631913e-05,
+      "loss": 1.4504,
+      "step": 2825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.084869564323844e-05,
+      "loss": 1.2056,
+      "step": 2826
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.078320240834446e-05,
+      "loss": 1.2147,
+      "step": 2827
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.071772287234497e-05,
+      "loss": 1.563,
+      "step": 2828
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.065225706594145e-05,
+      "loss": 1.5556,
+      "step": 2829
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.058680501982884e-05,
+      "loss": 1.3276,
+      "step": 2830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.052136676469575e-05,
+      "loss": 1.2711,
+      "step": 2831
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.045594233122415e-05,
+      "loss": 1.4288,
+      "step": 2832
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.039053175008965e-05,
+      "loss": 1.6248,
+      "step": 2833
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.032513505196142e-05,
+      "loss": 1.5592,
+      "step": 2834
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.025975226750192e-05,
+      "loss": 1.3062,
+      "step": 2835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.019438342736733e-05,
+      "loss": 1.0026,
+      "step": 2836
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.012902856220713e-05,
+      "loss": 1.1839,
+      "step": 2837
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.006368770266421e-05,
+      "loss": 1.1828,
+      "step": 2838
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.999836087937508e-05,
+      "loss": 1.3866,
+      "step": 2839
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.993304812296949e-05,
+      "loss": 1.1858,
+      "step": 2840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.986774946407069e-05,
+      "loss": 1.311,
+      "step": 2841
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.980246493329535e-05,
+      "loss": 1.2881,
+      "step": 2842
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.973719456125334e-05,
+      "loss": 1.2791,
+      "step": 2843
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.96719383785481e-05,
+      "loss": 1.1138,
+      "step": 2844
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.960669641577628e-05,
+      "loss": 1.2657,
+      "step": 2845
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.9541468703528e-05,
+      "loss": 1.412,
+      "step": 2846
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.947625527238653e-05,
+      "loss": 1.0117,
+      "step": 2847
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.941105615292852e-05,
+      "loss": 1.1866,
+      "step": 2848
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.934587137572396e-05,
+      "loss": 1.0958,
+      "step": 2849
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9280700971336e-05,
+      "loss": 1.2577,
+      "step": 2850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.921554497032121e-05,
+      "loss": 1.2963,
+      "step": 2851
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.915040340322929e-05,
+      "loss": 1.3373,
+      "step": 2852
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.908527630060314e-05,
+      "loss": 1.3646,
+      "step": 2853
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9020163692979e-05,
+      "loss": 1.1706,
+      "step": 2854
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.895506561088621e-05,
+      "loss": 1.3311,
+      "step": 2855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.88899820848474e-05,
+      "loss": 1.1815,
+      "step": 2856
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.882491314537823e-05,
+      "loss": 1.5131,
+      "step": 2857
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.875985882298767e-05,
+      "loss": 1.2887,
+      "step": 2858
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.869481914817779e-05,
+      "loss": 1.1659,
+      "step": 2859
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.86297941514437e-05,
+      "loss": 1.6373,
+      "step": 2860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.856478386327376e-05,
+      "loss": 1.326,
+      "step": 2861
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.849978831414933e-05,
+      "loss": 1.4355,
+      "step": 2862
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.843480753454499e-05,
+      "loss": 1.2579,
+      "step": 2863
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.836984155492822e-05,
+      "loss": 1.4733,
+      "step": 2864
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.830489040575966e-05,
+      "loss": 1.3657,
+      "step": 2865
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.823995411749308e-05,
+      "loss": 1.1602,
+      "step": 2866
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.817503272057509e-05,
+      "loss": 1.3711,
+      "step": 2867
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.811012624544545e-05,
+      "loss": 1.1518,
+      "step": 2868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.804523472253692e-05,
+      "loss": 1.4543,
+      "step": 2869
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.798035818227516e-05,
+      "loss": 1.428,
+      "step": 2870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.791549665507892e-05,
+      "loss": 1.4612,
+      "step": 2871
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.785065017135981e-05,
+      "loss": 1.2989,
+      "step": 2872
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.778581876152252e-05,
+      "loss": 1.22,
+      "step": 2873
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.772100245596449e-05,
+      "loss": 1.2898,
+      "step": 2874
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.765620128507619e-05,
+      "loss": 1.3606,
+      "step": 2875
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.759141527924105e-05,
+      "loss": 1.4191,
+      "step": 2876
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.752664446883522e-05,
+      "loss": 1.1985,
+      "step": 2877
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.746188888422788e-05,
+      "loss": 1.064,
+      "step": 2878
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.739714855578103e-05,
+      "loss": 1.1356,
+      "step": 2879
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.733242351384941e-05,
+      "loss": 1.1344,
+      "step": 2880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.726771378878081e-05,
+      "loss": 1.2335,
+      "step": 2881
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.72030194109156e-05,
+      "loss": 1.192,
+      "step": 2882
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.713834041058712e-05,
+      "loss": 1.2839,
+      "step": 2883
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.707367681812142e-05,
+      "loss": 1.2186,
+      "step": 2884
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.700902866383741e-05,
+      "loss": 1.4325,
+      "step": 2885
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.694439597804665e-05,
+      "loss": 1.2927,
+      "step": 2886
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.687977879105348e-05,
+      "loss": 1.4634,
+      "step": 2887
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.681517713315506e-05,
+      "loss": 1.2805,
+      "step": 2888
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.675059103464112e-05,
+      "loss": 1.3741,
+      "step": 2889
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.668602052579424e-05,
+      "loss": 1.0522,
+      "step": 2890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.662146563688964e-05,
+      "loss": 1.1181,
+      "step": 2891
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.655692639819511e-05,
+      "loss": 1.3785,
+      "step": 2892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.649240283997129e-05,
+      "loss": 1.222,
+      "step": 2893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.642789499247132e-05,
+      "loss": 1.1357,
+      "step": 2894
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.636340288594107e-05,
+      "loss": 1.2735,
+      "step": 2895
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.629892655061895e-05,
+      "loss": 1.2835,
+      "step": 2896
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.623446601673601e-05,
+      "loss": 1.4817,
+      "step": 2897
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.617002131451595e-05,
+      "loss": 1.2247,
+      "step": 2898
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.610559247417488e-05,
+      "loss": 1.2387,
+      "step": 2899
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.604117952592168e-05,
+      "loss": 1.3655,
+      "step": 2900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.597678249995767e-05,
+      "loss": 1.3311,
+      "step": 2901
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.591240142647667e-05,
+      "loss": 1.1248,
+      "step": 2902
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.584803633566509e-05,
+      "loss": 1.3746,
+      "step": 2903
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.578368725770181e-05,
+      "loss": 1.1531,
+      "step": 2904
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.571935422275827e-05,
+      "loss": 1.1658,
+      "step": 2905
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.565503726099829e-05,
+      "loss": 1.0733,
+      "step": 2906
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.559073640257816e-05,
+      "loss": 1.3855,
+      "step": 2907
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.552645167764672e-05,
+      "loss": 1.4104,
+      "step": 2908
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.546218311634513e-05,
+      "loss": 1.2884,
+      "step": 2909
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.539793074880706e-05,
+      "loss": 1.3562,
+      "step": 2910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.533369460515854e-05,
+      "loss": 1.3299,
+      "step": 2911
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.526947471551798e-05,
+      "loss": 1.3106,
+      "step": 2912
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.520527110999624e-05,
+      "loss": 1.4105,
+      "step": 2913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.514108381869643e-05,
+      "loss": 1.3667,
+      "step": 2914
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.507691287171416e-05,
+      "loss": 1.3344,
+      "step": 2915
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.50127582991372e-05,
+      "loss": 1.1111,
+      "step": 2916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.494862013104581e-05,
+      "loss": 1.3257,
+      "step": 2917
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.488449839751248e-05,
+      "loss": 1.327,
+      "step": 2918
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.482039312860194e-05,
+      "loss": 1.3586,
+      "step": 2919
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.475630435437132e-05,
+      "loss": 1.2293,
+      "step": 2920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.469223210486992e-05,
+      "loss": 1.1994,
+      "step": 2921
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.462817641013939e-05,
+      "loss": 1.2901,
+      "step": 2922
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.456413730021346e-05,
+      "loss": 1.2266,
+      "step": 2923
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.450011480511823e-05,
+      "loss": 1.218,
+      "step": 2924
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.443610895487197e-05,
+      "loss": 1.3559,
+      "step": 2925
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.437211977948509e-05,
+      "loss": 1.1896,
+      "step": 2926
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.430814730896033e-05,
+      "loss": 1.432,
+      "step": 2927
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.424419157329238e-05,
+      "loss": 1.1665,
+      "step": 2928
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.418025260246822e-05,
+      "loss": 1.1954,
+      "step": 2929
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.411633042646702e-05,
+      "loss": 1.2171,
+      "step": 2930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.405242507525992e-05,
+      "loss": 1.1322,
+      "step": 2931
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.398853657881027e-05,
+      "loss": 1.1963,
+      "step": 2932
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.392466496707357e-05,
+      "loss": 1.1809,
+      "step": 2933
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.386081026999723e-05,
+      "loss": 1.0655,
+      "step": 2934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.37969725175209e-05,
+      "loss": 1.261,
+      "step": 2935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.37331517395762e-05,
+      "loss": 1.2053,
+      "step": 2936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.366934796608683e-05,
+      "loss": 1.35,
+      "step": 2937
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.360556122696847e-05,
+      "loss": 1.3875,
+      "step": 2938
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.354179155212885e-05,
+      "loss": 1.228,
+      "step": 2939
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.347803897146771e-05,
+      "loss": 1.2942,
+      "step": 2940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.34143035148767e-05,
+      "loss": 1.0451,
+      "step": 2941
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.335058521223956e-05,
+      "loss": 1.2119,
+      "step": 2942
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.328688409343186e-05,
+      "loss": 1.1999,
+      "step": 2943
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.322320018832126e-05,
+      "loss": 1.2732,
+      "step": 2944
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.315953352676718e-05,
+      "loss": 1.2086,
+      "step": 2945
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.309588413862106e-05,
+      "loss": 1.3855,
+      "step": 2946
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.303225205372627e-05,
+      "loss": 1.1001,
+      "step": 2947
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.296863730191796e-05,
+      "loss": 1.3501,
+      "step": 2948
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.290503991302324e-05,
+      "loss": 1.3814,
+      "step": 2949
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.284145991686104e-05,
+      "loss": 1.1211,
+      "step": 2950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.277789734324214e-05,
+      "loss": 1.2971,
+      "step": 2951
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.271435222196916e-05,
+      "loss": 1.2025,
+      "step": 2952
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.26508245828365e-05,
+      "loss": 0.895,
+      "step": 2953
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.25873144556305e-05,
+      "loss": 1.4665,
+      "step": 2954
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.252382187012908e-05,
+      "loss": 1.3075,
+      "step": 2955
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.246034685610208e-05,
+      "loss": 1.3124,
+      "step": 2956
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.239688944331105e-05,
+      "loss": 1.3352,
+      "step": 2957
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.233344966150931e-05,
+      "loss": 1.332,
+      "step": 2958
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.227002754044189e-05,
+      "loss": 1.1784,
+      "step": 2959
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.22066231098456e-05,
+      "loss": 1.2524,
+      "step": 2960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.214323639944881e-05,
+      "loss": 1.377,
+      "step": 2961
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.207986743897177e-05,
+      "loss": 1.4951,
+      "step": 2962
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.201651625812625e-05,
+      "loss": 1.1009,
+      "step": 2963
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.195318288661584e-05,
+      "loss": 1.4171,
+      "step": 2964
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.18898673541356e-05,
+      "loss": 1.3212,
+      "step": 2965
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.182656969037234e-05,
+      "loss": 1.3051,
+      "step": 2966
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.176328992500451e-05,
+      "loss": 1.2647,
+      "step": 2967
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.170002808770201e-05,
+      "loss": 1.513,
+      "step": 2968
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.163678420812661e-05,
+      "loss": 1.0966,
+      "step": 2969
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.157355831593137e-05,
+      "loss": 1.3947,
+      "step": 2970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.151035044076113e-05,
+      "loss": 1.2493,
+      "step": 2971
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.144716061225216e-05,
+      "loss": 1.3021,
+      "step": 2972
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.138398886003227e-05,
+      "loss": 1.2356,
+      "step": 2973
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.13208352137209e-05,
+      "loss": 1.2743,
+      "step": 2974
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.125769970292888e-05,
+      "loss": 1.3897,
+      "step": 2975
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.119458235725866e-05,
+      "loss": 1.2423,
+      "step": 2976
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.113148320630402e-05,
+      "loss": 1.3128,
+      "step": 2977
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.106840227965031e-05,
+      "loss": 1.2303,
+      "step": 2978
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1005339606874357e-05,
+      "loss": 1.1232,
+      "step": 2979
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.094229521754433e-05,
+      "loss": 1.2226,
+      "step": 2980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.087926914121993e-05,
+      "loss": 1.4518,
+      "step": 2981
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.0816261407452225e-05,
+      "loss": 1.28,
+      "step": 2982
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.0753272045783625e-05,
+      "loss": 1.2859,
+      "step": 2983
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.069030108574806e-05,
+      "loss": 1.3395,
+      "step": 2984
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.062734855687069e-05,
+      "loss": 1.0324,
+      "step": 2985
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.0564414488668165e-05,
+      "loss": 1.1789,
+      "step": 2986
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.0501498910648345e-05,
+      "loss": 1.2366,
+      "step": 2987
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.0438601852310515e-05,
+      "loss": 1.49,
+      "step": 2988
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.037572334314529e-05,
+      "loss": 1.2647,
+      "step": 2989
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.031286341263447e-05,
+      "loss": 1.278,
+      "step": 2990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.02500220902513e-05,
+      "loss": 1.4618,
+      "step": 2991
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0187199405460194e-05,
+      "loss": 1.1652,
+      "step": 2992
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.012439538771681e-05,
+      "loss": 1.3825,
+      "step": 2993
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.006161006646817e-05,
+      "loss": 1.3072,
+      "step": 2994
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.99988434711524e-05,
+      "loss": 1.3248,
+      "step": 2995
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9936095631198974e-05,
+      "loss": 1.4385,
+      "step": 2996
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9873366576028424e-05,
+      "loss": 1.1505,
+      "step": 2997
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.98106563350526e-05,
+      "loss": 1.5212,
+      "step": 2998
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9747964937674505e-05,
+      "loss": 1.3484,
+      "step": 2999
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.968529241328822e-05,
+      "loss": 1.4119,
+      "step": 3000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9622638791279095e-05,
+      "loss": 1.0947,
+      "step": 3001
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.956000410102352e-05,
+      "loss": 1.5248,
+      "step": 3002
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.949738837188914e-05,
+      "loss": 1.2024,
+      "step": 3003
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.943479163323453e-05,
+      "loss": 1.4239,
+      "step": 3004
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.937221391440948e-05,
+      "loss": 1.0612,
+      "step": 3005
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.930965524475489e-05,
+      "loss": 1.178,
+      "step": 3006
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.924711565360259e-05,
+      "loss": 1.4742,
+      "step": 3007
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9184595170275614e-05,
+      "loss": 1.5655,
+      "step": 3008
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.912209382408795e-05,
+      "loss": 1.4053,
+      "step": 3009
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9059611644344613e-05,
+      "loss": 1.3196,
+      "step": 3010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.89971486603417e-05,
+      "loss": 1.3448,
+      "step": 3011
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.893470490136621e-05,
+      "loss": 1.2131,
+      "step": 3012
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.887228039669622e-05,
+      "loss": 1.2567,
+      "step": 3013
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.880987517560075e-05,
+      "loss": 1.4001,
+      "step": 3014
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.874748926733971e-05,
+      "loss": 1.4753,
+      "step": 3015
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.868512270116407e-05,
+      "loss": 1.3111,
+      "step": 3016
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8622775506315607e-05,
+      "loss": 1.2261,
+      "step": 3017
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8560447712027176e-05,
+      "loss": 1.4592,
+      "step": 3018
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.849813934752237e-05,
+      "loss": 1.1848,
+      "step": 3019
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.843585044201574e-05,
+      "loss": 1.3157,
+      "step": 3020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.837358102471278e-05,
+      "loss": 1.0973,
+      "step": 3021
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.831133112480968e-05,
+      "loss": 1.1891,
+      "step": 3022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.824910077149371e-05,
+      "loss": 1.4651,
+      "step": 3023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.818688999394271e-05,
+      "loss": 1.425,
+      "step": 3024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.812469882132558e-05,
+      "loss": 1.0332,
+      "step": 3025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.806252728280195e-05,
+      "loss": 1.1484,
+      "step": 3026
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8000375407522065e-05,
+      "loss": 1.3555,
+      "step": 3027
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.7938243224627254e-05,
+      "loss": 1.4091,
+      "step": 3028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.787613076324942e-05,
+      "loss": 1.1907,
+      "step": 3029
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.781403805251124e-05,
+      "loss": 1.1968,
+      "step": 3030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.77519651215262e-05,
+      "loss": 1.4054,
+      "step": 3031
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.768991199939843e-05,
+      "loss": 1.3697,
+      "step": 3032
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.7627878715222814e-05,
+      "loss": 1.3452,
+      "step": 3033
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.756586529808494e-05,
+      "loss": 1.3416,
+      "step": 3034
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.7503871777061044e-05,
+      "loss": 1.2267,
+      "step": 3035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.74418981812181e-05,
+      "loss": 1.1949,
+      "step": 3036
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7379944539613686e-05,
+      "loss": 1.3532,
+      "step": 3037
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7318010881296e-05,
+      "loss": 1.6187,
+      "step": 3038
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.725609723530392e-05,
+      "loss": 1.1819,
+      "step": 3039
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.719420363066702e-05,
+      "loss": 1.1631,
+      "step": 3040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.713233009640526e-05,
+      "loss": 1.1429,
+      "step": 3041
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.707047666152934e-05,
+      "loss": 1.3274,
+      "step": 3042
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.70086433550406e-05,
+      "loss": 1.3586,
+      "step": 3043
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.694683020593072e-05,
+      "loss": 1.3903,
+      "step": 3044
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.688503724318217e-05,
+      "loss": 1.1869,
+      "step": 3045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6823264495767845e-05,
+      "loss": 1.3339,
+      "step": 3046
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.676151199265105e-05,
+      "loss": 1.3361,
+      "step": 3047
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.669977976278583e-05,
+      "loss": 1.3774,
+      "step": 3048
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.663806783511657e-05,
+      "loss": 1.3949,
+      "step": 3049
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.657637623857816e-05,
+      "loss": 1.3569,
+      "step": 3050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.651470500209599e-05,
+      "loss": 1.5296,
+      "step": 3051
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.645305415458586e-05,
+      "loss": 1.3146,
+      "step": 3052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.639142372495406e-05,
+      "loss": 1.4094,
+      "step": 3053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6329813742097227e-05,
+      "loss": 1.5318,
+      "step": 3054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.626822423490257e-05,
+      "loss": 1.381,
+      "step": 3055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.620665523224745e-05,
+      "loss": 1.3726,
+      "step": 3056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6145106762999874e-05,
+      "loss": 1.2277,
+      "step": 3057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6083578856018116e-05,
+      "loss": 1.3261,
+      "step": 3058
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.602207154015064e-05,
+      "loss": 1.3273,
+      "step": 3059
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.596058484423656e-05,
+      "loss": 1.2043,
+      "step": 3060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.589911879710514e-05,
+      "loss": 1.4848,
+      "step": 3061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.583767342757598e-05,
+      "loss": 1.261,
+      "step": 3062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.577624876445899e-05,
+      "loss": 1.3479,
+      "step": 3063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.571484483655442e-05,
+      "loss": 1.3104,
+      "step": 3064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.565346167265273e-05,
+      "loss": 1.3729,
+      "step": 3065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5592099301534686e-05,
+      "loss": 1.4578,
+      "step": 3066
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.553075775197128e-05,
+      "loss": 1.1528,
+      "step": 3067
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5469437052723785e-05,
+      "loss": 1.3021,
+      "step": 3068
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5408137232543636e-05,
+      "loss": 1.2249,
+      "step": 3069
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.534685832017252e-05,
+      "loss": 1.532,
+      "step": 3070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.528560034434229e-05,
+      "loss": 1.3653,
+      "step": 3071
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.52243633337751e-05,
+      "loss": 1.2279,
+      "step": 3072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.516314731718307e-05,
+      "loss": 1.4228,
+      "step": 3073
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.510195232326858e-05,
+      "loss": 1.3818,
+      "step": 3074
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.504077838072428e-05,
+      "loss": 1.4826,
+      "step": 3075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.497962551823266e-05,
+      "loss": 1.1936,
+      "step": 3076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.491849376446663e-05,
+      "loss": 1.2353,
+      "step": 3077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.485738314808905e-05,
+      "loss": 1.3394,
+      "step": 3078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.4796293697752765e-05,
+      "loss": 1.484,
+      "step": 3079
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.4735225442100945e-05,
+      "loss": 1.3231,
+      "step": 3080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.4674178409766654e-05,
+      "loss": 1.4895,
+      "step": 3081
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.4613152629373035e-05,
+      "loss": 1.3658,
+      "step": 3082
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4552148129533286e-05,
+      "loss": 1.2889,
+      "step": 3083
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.449116493885061e-05,
+      "loss": 1.2681,
+      "step": 3084
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.443020308591824e-05,
+      "loss": 1.3142,
+      "step": 3085
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.436926259931936e-05,
+      "loss": 1.2684,
+      "step": 3086
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4308343507627205e-05,
+      "loss": 1.6358,
+      "step": 3087
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4247445839404875e-05,
+      "loss": 1.2432,
+      "step": 3088
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4186569623205605e-05,
+      "loss": 1.1916,
+      "step": 3089
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4125714887572365e-05,
+      "loss": 1.238,
+      "step": 3090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.406488166103814e-05,
+      "loss": 1.2487,
+      "step": 3091
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4004069972125926e-05,
+      "loss": 1.3875,
+      "step": 3092
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.3943279849348395e-05,
+      "loss": 1.201,
+      "step": 3093
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.388251132120834e-05,
+      "loss": 1.4198,
+      "step": 3094
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.3821764416198375e-05,
+      "loss": 1.3096,
+      "step": 3095
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.376103916280075e-05,
+      "loss": 1.3749,
+      "step": 3096
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.3700335589487925e-05,
+      "loss": 1.2329,
+      "step": 3097
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.363965372472195e-05,
+      "loss": 1.5912,
+      "step": 3098
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.3578993596954754e-05,
+      "loss": 1.0337,
+      "step": 3099
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.351835523462808e-05,
+      "loss": 1.2908,
+      "step": 3100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.345773866617348e-05,
+      "loss": 1.3199,
+      "step": 3101
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.339714392001227e-05,
+      "loss": 1.4997,
+      "step": 3102
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.3336571024555506e-05,
+      "loss": 1.4357,
+      "step": 3103
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.327602000820413e-05,
+      "loss": 1.2546,
+      "step": 3104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.321549089934863e-05,
+      "loss": 1.2146,
+      "step": 3105
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3154983726369325e-05,
+      "loss": 1.2546,
+      "step": 3106
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.309449851763633e-05,
+      "loss": 1.1477,
+      "step": 3107
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.303403530150924e-05,
+      "loss": 1.4979,
+      "step": 3108
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.29735941063376e-05,
+      "loss": 1.1709,
+      "step": 3109
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2913174960460446e-05,
+      "loss": 1.2257,
+      "step": 3110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.285277789220653e-05,
+      "loss": 1.3672,
+      "step": 3111
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.279240292989427e-05,
+      "loss": 1.0745,
+      "step": 3112
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.273205010183171e-05,
+      "loss": 1.195,
+      "step": 3113
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.267171943631648e-05,
+      "loss": 1.2647,
+      "step": 3114
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2611410961635885e-05,
+      "loss": 1.2795,
+      "step": 3115
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.255112470606678e-05,
+      "loss": 1.3359,
+      "step": 3116
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.249086069787559e-05,
+      "loss": 1.3078,
+      "step": 3117
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2430618965318356e-05,
+      "loss": 1.2013,
+      "step": 3118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2370399536640644e-05,
+      "loss": 1.2086,
+      "step": 3119
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.231020244007752e-05,
+      "loss": 1.5816,
+      "step": 3120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2250027703853754e-05,
+      "loss": 1.3797,
+      "step": 3121
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.218987535618338e-05,
+      "loss": 1.3985,
+      "step": 3122
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.212974542527006e-05,
+      "loss": 1.2157,
+      "step": 3123
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.206963793930706e-05,
+      "loss": 1.2101,
+      "step": 3124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2009552926476865e-05,
+      "loss": 1.2634,
+      "step": 3125
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.194949041495168e-05,
+      "loss": 1.4164,
+      "step": 3126
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.1889450432893015e-05,
+      "loss": 1.1627,
+      "step": 3127
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.1829433008451755e-05,
+      "loss": 1.3796,
+      "step": 3128
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.1769438169768425e-05,
+      "loss": 1.3366,
+      "step": 3129
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.170946594497278e-05,
+      "loss": 1.3823,
+      "step": 3130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.164951636218402e-05,
+      "loss": 1.2206,
+      "step": 3131
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.158958944951078e-05,
+      "loss": 1.2445,
+      "step": 3132
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.152968523505095e-05,
+      "loss": 1.2997,
+      "step": 3133
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.146980374689192e-05,
+      "loss": 1.3646,
+      "step": 3134
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1409945013110295e-05,
+      "loss": 1.0894,
+      "step": 3135
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.135010906177211e-05,
+      "loss": 1.3259,
+      "step": 3136
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.129029592093262e-05,
+      "loss": 1.5682,
+      "step": 3137
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.123050561863657e-05,
+      "loss": 1.1672,
+      "step": 3138
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1170738182917746e-05,
+      "loss": 1.4875,
+      "step": 3139
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1110993641799324e-05,
+      "loss": 1.1784,
+      "step": 3140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1051272023293894e-05,
+      "loss": 1.4123,
+      "step": 3141
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.099157335540299e-05,
+      "loss": 1.3372,
+      "step": 3142
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.093189766611769e-05,
+      "loss": 1.2334,
+      "step": 3143
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.087224498341815e-05,
+      "loss": 1.4251,
+      "step": 3144
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.081261533527364e-05,
+      "loss": 1.2791,
+      "step": 3145
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.075300874964286e-05,
+      "loss": 1.1525,
+      "step": 3146
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.069342525447354e-05,
+      "loss": 1.2781,
+      "step": 3147
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.063386487770262e-05,
+      "loss": 1.1665,
+      "step": 3148
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.05743276472562e-05,
+      "loss": 1.4394,
+      "step": 3149
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.051481359104956e-05,
+      "loss": 1.4455,
+      "step": 3150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.0455322736987055e-05,
+      "loss": 1.1304,
+      "step": 3151
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.0395855112962164e-05,
+      "loss": 1.2647,
+      "step": 3152
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.033641074685761e-05,
+      "loss": 1.4244,
+      "step": 3153
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0276989666544995e-05,
+      "loss": 1.4171,
+      "step": 3154
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.021759189988511e-05,
+      "loss": 1.1464,
+      "step": 3155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.015821747472793e-05,
+      "loss": 1.294,
+      "step": 3156
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0098866418912215e-05,
+      "loss": 1.0576,
+      "step": 3157
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0039538760266034e-05,
+      "loss": 1.5739,
+      "step": 3158
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.998023452660639e-05,
+      "loss": 1.2991,
+      "step": 3159
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9920953745739174e-05,
+      "loss": 1.3924,
+      "step": 3160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9861696445459506e-05,
+      "loss": 1.2811,
+      "step": 3161
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.980246265355134e-05,
+      "loss": 1.3294,
+      "step": 3162
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9743252397787655e-05,
+      "loss": 1.0992,
+      "step": 3163
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9684065705930396e-05,
+      "loss": 1.3358,
+      "step": 3164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9624902605730464e-05,
+      "loss": 1.4744,
+      "step": 3165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.956576312492769e-05,
+      "loss": 1.5594,
+      "step": 3166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9506647291250815e-05,
+      "loss": 1.4089,
+      "step": 3167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9447555132417536e-05,
+      "loss": 1.48,
+      "step": 3168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.938848667613436e-05,
+      "loss": 1.4905,
+      "step": 3169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9329441950096875e-05,
+      "loss": 1.2882,
+      "step": 3170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9270420981989294e-05,
+      "loss": 1.2867,
+      "step": 3171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9211423799484804e-05,
+      "loss": 1.328,
+      "step": 3172
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.915245043024556e-05,
+      "loss": 1.2349,
+      "step": 3173
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9093500901922275e-05,
+      "loss": 0.9877,
+      "step": 3174
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.903457524215477e-05,
+      "loss": 1.3566,
+      "step": 3175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.897567347857154e-05,
+      "loss": 1.1331,
+      "step": 3176
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.891679563878977e-05,
+      "loss": 1.2695,
+      "step": 3177
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.885794175041567e-05,
+      "loss": 1.2577,
+      "step": 3178
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.879911184104404e-05,
+      "loss": 1.359,
+      "step": 3179
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.874030593825849e-05,
+      "loss": 1.3823,
+      "step": 3180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.868152406963138e-05,
+      "loss": 1.3549,
+      "step": 3181
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8622766262723794e-05,
+      "loss": 1.1773,
+      "step": 3182
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.856403254508553e-05,
+      "loss": 1.2348,
+      "step": 3183
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8505322944255096e-05,
+      "loss": 1.4916,
+      "step": 3184
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8446637487759693e-05,
+      "loss": 1.155,
+      "step": 3185
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.838797620311519e-05,
+      "loss": 1.0662,
+      "step": 3186
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.832933911782616e-05,
+      "loss": 1.1871,
+      "step": 3187
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.827072625938577e-05,
+      "loss": 1.1476,
+      "step": 3188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.821213765527585e-05,
+      "loss": 1.466,
+      "step": 3189
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.815357333296693e-05,
+      "loss": 1.4602,
+      "step": 3190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.809503331991808e-05,
+      "loss": 1.3123,
+      "step": 3191
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.803651764357691e-05,
+      "loss": 1.2451,
+      "step": 3192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7978026331379765e-05,
+      "loss": 1.1872,
+      "step": 3193
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.791955941075148e-05,
+      "loss": 1.352,
+      "step": 3194
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.786111690910547e-05,
+      "loss": 1.2106,
+      "step": 3195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7802698853843706e-05,
+      "loss": 1.5868,
+      "step": 3196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.774430527235667e-05,
+      "loss": 1.2863,
+      "step": 3197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7685936192023415e-05,
+      "loss": 1.21,
+      "step": 3198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7627591640211456e-05,
+      "loss": 1.1649,
+      "step": 3199
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.756927164427685e-05,
+      "loss": 1.2037,
+      "step": 3200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.751097623156408e-05,
+      "loss": 1.4058,
+      "step": 3201
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7452705429406254e-05,
+      "loss": 1.1187,
+      "step": 3202
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.739445926512471e-05,
+      "loss": 1.3147,
+      "step": 3203
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.733623776602939e-05,
+      "loss": 1.1174,
+      "step": 3204
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7278040959418715e-05,
+      "loss": 1.1976,
+      "step": 3205
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.721986887257932e-05,
+      "loss": 1.2507,
+      "step": 3206
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7161721532786485e-05,
+      "loss": 1.2577,
+      "step": 3207
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.710359896730379e-05,
+      "loss": 1.4772,
+      "step": 3208
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.704550120338307e-05,
+      "loss": 1.3075,
+      "step": 3209
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.698742826826477e-05,
+      "loss": 1.5246,
+      "step": 3210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.692938018917754e-05,
+      "loss": 1.404,
+      "step": 3211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6871356993338413e-05,
+      "loss": 1.5429,
+      "step": 3212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.681335870795274e-05,
+      "loss": 1.0312,
+      "step": 3213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.675538536021422e-05,
+      "loss": 1.3581,
+      "step": 3214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.669743697730484e-05,
+      "loss": 1.0553,
+      "step": 3215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.66395135863949e-05,
+      "loss": 1.4645,
+      "step": 3216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.658161521464295e-05,
+      "loss": 1.3485,
+      "step": 3217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.652374188919584e-05,
+      "loss": 1.3448,
+      "step": 3218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6465893637188665e-05,
+      "loss": 1.3903,
+      "step": 3219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.640807048574476e-05,
+      "loss": 0.9903,
+      "step": 3220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6350272461975665e-05,
+      "loss": 1.3177,
+      "step": 3221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6292499592981266e-05,
+      "loss": 1.3216,
+      "step": 3222
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.623475190584942e-05,
+      "loss": 1.3699,
+      "step": 3223
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.617702942765641e-05,
+      "loss": 1.4292,
+      "step": 3224
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.611933218546664e-05,
+      "loss": 1.4288,
+      "step": 3225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.606166020633249e-05,
+      "loss": 1.4416,
+      "step": 3226
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.600401351729479e-05,
+      "loss": 1.1391,
+      "step": 3227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.594639214538231e-05,
+      "loss": 1.4828,
+      "step": 3228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5888796117612045e-05,
+      "loss": 1.3682,
+      "step": 3229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5831225460989046e-05,
+      "loss": 1.4393,
+      "step": 3230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.57736802025065e-05,
+      "loss": 1.2341,
+      "step": 3231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.571616036914569e-05,
+      "loss": 1.2327,
+      "step": 3232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.565866598787593e-05,
+      "loss": 1.2952,
+      "step": 3233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.560119708565475e-05,
+      "loss": 1.2187,
+      "step": 3234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5543753689427506e-05,
+      "loss": 1.1908,
+      "step": 3235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5486335826127714e-05,
+      "loss": 1.2022,
+      "step": 3236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.542894352267703e-05,
+      "loss": 1.4384,
+      "step": 3237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.537157680598486e-05,
+      "loss": 1.4736,
+      "step": 3238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.531423570294888e-05,
+      "loss": 1.1647,
+      "step": 3239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.525692024045465e-05,
+      "loss": 1.3518,
+      "step": 3240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.519963044537557e-05,
+      "loss": 1.2093,
+      "step": 3241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5142366344573264e-05,
+      "loss": 1.4055,
+      "step": 3242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5085127964897135e-05,
+      "loss": 1.2847,
+      "step": 3243
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5027915333184554e-05,
+      "loss": 1.1632,
+      "step": 3244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.497072847626087e-05,
+      "loss": 1.2223,
+      "step": 3245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.4913567420939284e-05,
+      "loss": 1.2247,
+      "step": 3246
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.485643219402095e-05,
+      "loss": 1.3981,
+      "step": 3247
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.479932282229488e-05,
+      "loss": 1.0629,
+      "step": 3248
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.474223933253797e-05,
+      "loss": 1.3645,
+      "step": 3249
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.468518175151495e-05,
+      "loss": 1.1146,
+      "step": 3250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.462815010597855e-05,
+      "loss": 1.2929,
+      "step": 3251
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.457114442266912e-05,
+      "loss": 1.3274,
+      "step": 3252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.451416472831493e-05,
+      "loss": 1.3447,
+      "step": 3253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.445721104963221e-05,
+      "loss": 1.2693,
+      "step": 3254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.44002834133247e-05,
+      "loss": 1.1898,
+      "step": 3255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.434338184608418e-05,
+      "loss": 1.3391,
+      "step": 3256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4286506374590157e-05,
+      "loss": 1.4196,
+      "step": 3257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.422965702550971e-05,
+      "loss": 1.3497,
+      "step": 3258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.417283382549796e-05,
+      "loss": 1.4931,
+      "step": 3259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.411603680119758e-05,
+      "loss": 1.2808,
+      "step": 3260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4059265979239026e-05,
+      "loss": 1.3304,
+      "step": 3261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4002521386240466e-05,
+      "loss": 1.3485,
+      "step": 3262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3945803048807756e-05,
+      "loss": 1.2426,
+      "step": 3263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.388911099353445e-05,
+      "loss": 1.3218,
+      "step": 3264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3832445247001794e-05,
+      "loss": 1.3619,
+      "step": 3265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3775805835778673e-05,
+      "loss": 1.3781,
+      "step": 3266
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3719192786421626e-05,
+      "loss": 1.4291,
+      "step": 3267
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.366260612547487e-05,
+      "loss": 1.4101,
+      "step": 3268
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3606045879470195e-05,
+      "loss": 1.3086,
+      "step": 3269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.354951207492701e-05,
+      "loss": 1.3199,
+      "step": 3270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.349300473835246e-05,
+      "loss": 1.3859,
+      "step": 3271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.343652389624104e-05,
+      "loss": 1.1933,
+      "step": 3272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.338006957507498e-05,
+      "loss": 1.5197,
+      "step": 3273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.332364180132412e-05,
+      "loss": 1.2649,
+      "step": 3274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.326724060144566e-05,
+      "loss": 1.5134,
+      "step": 3275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3210866001884564e-05,
+      "loss": 1.3131,
+      "step": 3276
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.315451802907316e-05,
+      "loss": 1.449,
+      "step": 3277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.309819670943139e-05,
+      "loss": 1.2132,
+      "step": 3278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304190206936662e-05,
+      "loss": 1.2574,
+      "step": 3279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.298563413527378e-05,
+      "loss": 0.9624,
+      "step": 3280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.292939293353522e-05,
+      "loss": 1.21,
+      "step": 3281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.287317849052075e-05,
+      "loss": 1.0894,
+      "step": 3282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2816990832587776e-05,
+      "loss": 1.278,
+      "step": 3283
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.276082998608093e-05,
+      "loss": 1.0844,
+      "step": 3284
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.270469597733236e-05,
+      "loss": 1.3192,
+      "step": 3285
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.264858883266177e-05,
+      "loss": 1.2363,
+      "step": 3286
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.259250857837599e-05,
+      "loss": 1.2364,
+      "step": 3287
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.253645524076952e-05,
+      "loss": 1.4233,
+      "step": 3288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.24804288461241e-05,
+      "loss": 1.4131,
+      "step": 3289
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2424429420708755e-05,
+      "loss": 1.435,
+      "step": 3290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.236845699078007e-05,
+      "loss": 1.5017,
+      "step": 3291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.231251158258183e-05,
+      "loss": 1.382,
+      "step": 3292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2256593222345185e-05,
+      "loss": 1.1788,
+      "step": 3293
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.220070193628861e-05,
+      "loss": 1.1855,
+      "step": 3294
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214483775061787e-05,
+      "loss": 1.3573,
+      "step": 3295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.208900069152605e-05,
+      "loss": 1.347,
+      "step": 3296
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.203319078519348e-05,
+      "loss": 1.2719,
+      "step": 3297
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.197740805778779e-05,
+      "loss": 1.2115,
+      "step": 3298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.192165253546386e-05,
+      "loss": 1.4811,
+      "step": 3299
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1865924244363795e-05,
+      "loss": 1.4014,
+      "step": 3300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.181022321061696e-05,
+      "loss": 1.3086,
+      "step": 3301
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1754549460339885e-05,
+      "loss": 1.1414,
+      "step": 3302
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1698903019636456e-05,
+      "loss": 1.3477,
+      "step": 3303
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.164328391459752e-05,
+      "loss": 1.5661,
+      "step": 3304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.158769217130132e-05,
+      "loss": 1.3979,
+      "step": 3305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1532127815813185e-05,
+      "loss": 1.1589,
+      "step": 3306
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1476590874185504e-05,
+      "loss": 1.3087,
+      "step": 3307
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.142108137245803e-05,
+      "loss": 1.3025,
+      "step": 3308
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1365599336657466e-05,
+      "loss": 1.447,
+      "step": 3309
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.131014479279772e-05,
+      "loss": 1.3679,
+      "step": 3310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.125471776687979e-05,
+      "loss": 1.2814,
+      "step": 3311
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1199318284891786e-05,
+      "loss": 1.0669,
+      "step": 3312
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.114394637280886e-05,
+      "loss": 1.2582,
+      "step": 3313
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.10886020565933e-05,
+      "loss": 1.4612,
+      "step": 3314
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1033285362194396e-05,
+      "loss": 1.2588,
+      "step": 3315
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.097799631554854e-05,
+      "loss": 1.2808,
+      "step": 3316
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.092273494257911e-05,
+      "loss": 1.1189,
+      "step": 3317
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.086750126919655e-05,
+      "loss": 1.2814,
+      "step": 3318
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.081229532129827e-05,
+      "loss": 1.374,
+      "step": 3319
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0757117124768796e-05,
+      "loss": 1.353,
+      "step": 3320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.070196670547947e-05,
+      "loss": 1.0379,
+      "step": 3321
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.064684408928867e-05,
+      "loss": 1.2792,
+      "step": 3322
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.059174930204187e-05,
+      "loss": 1.1562,
+      "step": 3323
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.053668236957134e-05,
+      "loss": 1.34,
+      "step": 3324
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0481643317696336e-05,
+      "loss": 1.3428,
+      "step": 3325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0426632172223046e-05,
+      "loss": 1.4516,
+      "step": 3326
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0371648958944576e-05,
+      "loss": 1.4246,
+      "step": 3327
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.031669370364094e-05,
+      "loss": 1.2081,
+      "step": 3328
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.026176643207903e-05,
+      "loss": 1.3507,
+      "step": 3329
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.020686717001262e-05,
+      "loss": 1.3881,
+      "step": 3330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.015199594318235e-05,
+      "loss": 1.0538,
+      "step": 3331
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.009715277731573e-05,
+      "loss": 1.3187,
+      "step": 3332
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.00423376981271e-05,
+      "loss": 1.3029,
+      "step": 3333
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.99875507313176e-05,
+      "loss": 1.4057,
+      "step": 3334
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.993279190257533e-05,
+      "loss": 1.3538,
+      "step": 3335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.987806123757494e-05,
+      "loss": 1.0792,
+      "step": 3336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.982335876197813e-05,
+      "loss": 1.3723,
+      "step": 3337
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.976868450143329e-05,
+      "loss": 1.3179,
+      "step": 3338
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.971403848157544e-05,
+      "loss": 1.2211,
+      "step": 3339
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.965942072802662e-05,
+      "loss": 1.2293,
+      "step": 3340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9604831266395435e-05,
+      "loss": 1.123,
+      "step": 3341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.955027012227727e-05,
+      "loss": 1.3721,
+      "step": 3342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9495737321254237e-05,
+      "loss": 1.4295,
+      "step": 3343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.944123288889515e-05,
+      "loss": 1.1477,
+      "step": 3344
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9386756850755555e-05,
+      "loss": 1.1033,
+      "step": 3345
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.933230923237764e-05,
+      "loss": 1.2437,
+      "step": 3346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.927789005929029e-05,
+      "loss": 1.197,
+      "step": 3347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9223499357009045e-05,
+      "loss": 1.1992,
+      "step": 3348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9169137151036104e-05,
+      "loss": 1.4261,
+      "step": 3349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.911480346686032e-05,
+      "loss": 1.2472,
+      "step": 3350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.906049832995709e-05,
+      "loss": 1.2045,
+      "step": 3351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9006221765788607e-05,
+      "loss": 1.2814,
+      "step": 3352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8951973799803456e-05,
+      "loss": 1.3459,
+      "step": 3353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.88977544574369e-05,
+      "loss": 1.3633,
+      "step": 3354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.884356376411089e-05,
+      "loss": 1.3268,
+      "step": 3355
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.878940174523371e-05,
+      "loss": 1.5748,
+      "step": 3356
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8735268426200444e-05,
+      "loss": 1.3562,
+      "step": 3357
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868116383239259e-05,
+      "loss": 1.3609,
+      "step": 3358
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.86270879891781e-05,
+      "loss": 1.2778,
+      "step": 3359
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8573040921911664e-05,
+      "loss": 1.1038,
+      "step": 3360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8519022655934286e-05,
+      "loss": 1.2615,
+      "step": 3361
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8465033216573574e-05,
+      "loss": 1.3796,
+      "step": 3362
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.841107262914352e-05,
+      "loss": 1.2134,
+      "step": 3363
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.835714091894478e-05,
+      "loss": 1.3729,
+      "step": 3364
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8303238111264215e-05,
+      "loss": 1.3356,
+      "step": 3365
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8249364231375284e-05,
+      "loss": 1.3052,
+      "step": 3366
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.819551930453795e-05,
+      "loss": 1.1071,
+      "step": 3367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.814170335599837e-05,
+      "loss": 1.0656,
+      "step": 3368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.808791641098937e-05,
+      "loss": 1.0755,
+      "step": 3369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.803415849473005e-05,
+      "loss": 1.5256,
+      "step": 3370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.798042963242582e-05,
+      "loss": 1.0363,
+      "step": 3371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.792672984926865e-05,
+      "loss": 1.3437,
+      "step": 3372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7873059170436754e-05,
+      "loss": 1.3938,
+      "step": 3373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7819417621094734e-05,
+      "loss": 1.5714,
+      "step": 3374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.776580522639353e-05,
+      "loss": 1.3552,
+      "step": 3375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.77122220114704e-05,
+      "loss": 1.2488,
+      "step": 3376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7658668001448946e-05,
+      "loss": 1.377,
+      "step": 3377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7605143221439066e-05,
+      "loss": 1.2773,
+      "step": 3378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.755164769653694e-05,
+      "loss": 1.442,
+      "step": 3379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7498181451825054e-05,
+      "loss": 1.2894,
+      "step": 3380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.744474451237213e-05,
+      "loss": 1.379,
+      "step": 3381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.73913369032332e-05,
+      "loss": 1.1994,
+      "step": 3382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.733795864944947e-05,
+      "loss": 1.4349,
+      "step": 3383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.728460977604852e-05,
+      "loss": 1.2833,
+      "step": 3384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.723129030804398e-05,
+      "loss": 1.27,
+      "step": 3385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.717800027043576e-05,
+      "loss": 1.4205,
+      "step": 3386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7124739688210096e-05,
+      "loss": 1.4084,
+      "step": 3387
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.707150858633918e-05,
+      "loss": 1.3387,
+      "step": 3388
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.701830698978158e-05,
+      "loss": 1.4709,
+      "step": 3389
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.696513492348195e-05,
+      "loss": 1.1479,
+      "step": 3390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.691199241237108e-05,
+      "loss": 1.5201,
+      "step": 3391
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.685887948136594e-05,
+      "loss": 1.4072,
+      "step": 3392
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.680579615536961e-05,
+      "loss": 1.039,
+      "step": 3393
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.675274245927129e-05,
+      "loss": 1.3042,
+      "step": 3394
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.669971841794632e-05,
+      "loss": 1.3541,
+      "step": 3395
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6646724056256074e-05,
+      "loss": 1.2415,
+      "step": 3396
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6593759399048067e-05,
+      "loss": 1.2907,
+      "step": 3397
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6540824471155856e-05,
+      "loss": 1.3035,
+      "step": 3398
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.648791929739907e-05,
+      "loss": 1.2704,
+      "step": 3399
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6435043902583343e-05,
+      "loss": 1.4213,
+      "step": 3400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6382198311500507e-05,
+      "loss": 1.1584,
+      "step": 3401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.632938254892818e-05,
+      "loss": 1.3789,
+      "step": 3402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.627659663963012e-05,
+      "loss": 1.5438,
+      "step": 3403
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.62238406083562e-05,
+      "loss": 1.3841,
+      "step": 3404
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6171114479842005e-05,
+      "loss": 1.5508,
+      "step": 3405
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6118418278809384e-05,
+      "loss": 1.1655,
+      "step": 3406
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6065752029966025e-05,
+      "loss": 1.3895,
+      "step": 3407
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.601311575800548e-05,
+      "loss": 1.4335,
+      "step": 3408
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.596050948760743e-05,
+      "loss": 1.2542,
+      "step": 3409
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.590793324343738e-05,
+      "loss": 1.1748,
+      "step": 3410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.58553870501468e-05,
+      "loss": 1.2888,
+      "step": 3411
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5802870932373015e-05,
+      "loss": 1.3672,
+      "step": 3412
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.57503849147393e-05,
+      "loss": 1.2911,
+      "step": 3413
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.56979290218548e-05,
+      "loss": 1.3604,
+      "step": 3414
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5645503278314506e-05,
+      "loss": 1.5821,
+      "step": 3415
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5593107708699405e-05,
+      "loss": 1.363,
+      "step": 3416
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.554074233757608e-05,
+      "loss": 1.1558,
+      "step": 3417
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.548840718949724e-05,
+      "loss": 1.5,
+      "step": 3418
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.543610228900126e-05,
+      "loss": 1.082,
+      "step": 3419
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.538382766061229e-05,
+      "loss": 1.1511,
+      "step": 3420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5331583328840456e-05,
+      "loss": 1.3995,
+      "step": 3421
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.527936931818153e-05,
+      "loss": 1.2266,
+      "step": 3422
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5227185653117154e-05,
+      "loss": 1.3156,
+      "step": 3423
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5175032358114677e-05,
+      "loss": 1.178,
+      "step": 3424
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5122909457627277e-05,
+      "loss": 1.2195,
+      "step": 3425
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.507081697609381e-05,
+      "loss": 1.4673,
+      "step": 3426
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5018754937938914e-05,
+      "loss": 1.2829,
+      "step": 3427
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.496672336757295e-05,
+      "loss": 1.305,
+      "step": 3428
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.491472228939199e-05,
+      "loss": 1.2636,
+      "step": 3429
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4862751727777797e-05,
+      "loss": 1.1179,
+      "step": 3430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.481081170709783e-05,
+      "loss": 1.2319,
+      "step": 3431
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.475890225170521e-05,
+      "loss": 1.0996,
+      "step": 3432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.470702338593884e-05,
+      "loss": 1.5094,
+      "step": 3433
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.46551751341231e-05,
+      "loss": 1.1284,
+      "step": 3434
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.460335752056809e-05,
+      "loss": 1.578,
+      "step": 3435
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.455157056956968e-05,
+      "loss": 1.2368,
+      "step": 3436
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.449981430540911e-05,
+      "loss": 0.9575,
+      "step": 3437
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4448088752353434e-05,
+      "loss": 1.3827,
+      "step": 3438
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.439639393465528e-05,
+      "loss": 1.4017,
+      "step": 3439
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4344729876552685e-05,
+      "loss": 1.362,
+      "step": 3440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.429309660226954e-05,
+      "loss": 1.2146,
+      "step": 3441
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.42414941360151e-05,
+      "loss": 1.5961,
+      "step": 3442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4189922501984254e-05,
+      "loss": 1.5611,
+      "step": 3443
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.41383817243574e-05,
+      "loss": 1.4501,
+      "step": 3444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.408687182730052e-05,
+      "loss": 1.5292,
+      "step": 3445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.403539283496507e-05,
+      "loss": 1.4169,
+      "step": 3446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.398394477148803e-05,
+      "loss": 1.0613,
+      "step": 3447
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.393252766099187e-05,
+      "loss": 1.2219,
+      "step": 3448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.388114152758455e-05,
+      "loss": 1.3048,
+      "step": 3449
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.382978639535959e-05,
+      "loss": 1.0937,
+      "step": 3450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3778462288395784e-05,
+      "loss": 1.397,
+      "step": 3451
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3727169230757524e-05,
+      "loss": 1.2267,
+      "step": 3452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.36759072464947e-05,
+      "loss": 1.7878,
+      "step": 3453
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.36246763596424e-05,
+      "loss": 1.3341,
+      "step": 3454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.357347659422139e-05,
+      "loss": 1.1337,
+      "step": 3455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.352230797423769e-05,
+      "loss": 1.1365,
+      "step": 3456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.347117052368275e-05,
+      "loss": 1.5551,
+      "step": 3457
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3420064266533415e-05,
+      "loss": 1.3962,
+      "step": 3458
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3368989226751914e-05,
+      "loss": 1.4001,
+      "step": 3459
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3317945428285816e-05,
+      "loss": 1.131,
+      "step": 3460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.326693289506805e-05,
+      "loss": 1.2878,
+      "step": 3461
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.321595165101689e-05,
+      "loss": 1.3357,
+      "step": 3462
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.316500172003594e-05,
+      "loss": 1.134,
+      "step": 3463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.31140831260141e-05,
+      "loss": 1.3626,
+      "step": 3464
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.306319589282566e-05,
+      "loss": 1.43,
+      "step": 3465
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.301234004433007e-05,
+      "loss": 1.3857,
+      "step": 3466
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.296151560437214e-05,
+      "loss": 1.3384,
+      "step": 3467
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.291072259678204e-05,
+      "loss": 1.5527,
+      "step": 3468
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.285996104537498e-05,
+      "loss": 1.2651,
+      "step": 3469
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.280923097395165e-05,
+      "loss": 1.4011,
+      "step": 3470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2758532406297895e-05,
+      "loss": 1.3113,
+      "step": 3471
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.270786536618467e-05,
+      "loss": 1.5082,
+      "step": 3472
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2657229877368364e-05,
+      "loss": 1.3489,
+      "step": 3473
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.260662596359043e-05,
+      "loss": 1.1497,
+      "step": 3474
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2556053648577535e-05,
+      "loss": 1.3835,
+      "step": 3475
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.250551295604156e-05,
+      "loss": 1.2874,
+      "step": 3476
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.245500390967954e-05,
+      "loss": 1.3534,
+      "step": 3477
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.240452653317367e-05,
+      "loss": 1.1353,
+      "step": 3478
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2354080850191324e-05,
+      "loss": 1.2747,
+      "step": 3479
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.230366688438497e-05,
+      "loss": 1.445,
+      "step": 3480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2253284659392227e-05,
+      "loss": 1.0387,
+      "step": 3481
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.220293419883589e-05,
+      "loss": 1.271,
+      "step": 3482
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.215261552632374e-05,
+      "loss": 1.3697,
+      "step": 3483
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.210232866544871e-05,
+      "loss": 1.3689,
+      "step": 3484
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.205207363978892e-05,
+      "loss": 1.1836,
+      "step": 3485
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.200185047290734e-05,
+      "loss": 1.2834,
+      "step": 3486
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1951659188352225e-05,
+      "loss": 1.3319,
+      "step": 3487
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1901499809656795e-05,
+      "loss": 1.2274,
+      "step": 3488
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.18513723603392e-05,
+      "loss": 1.2715,
+      "step": 3489
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.180127686390283e-05,
+      "loss": 1.2389,
+      "step": 3490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.175121334383594e-05,
+      "loss": 1.3439,
+      "step": 3491
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.170118182361184e-05,
+      "loss": 1.363,
+      "step": 3492
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1651182326688834e-05,
+      "loss": 1.3735,
+      "step": 3493
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.160121487651022e-05,
+      "loss": 1.0818,
+      "step": 3494
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.155127949650426e-05,
+      "loss": 1.4435,
+      "step": 3495
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.150137621008413e-05,
+      "loss": 1.3472,
+      "step": 3496
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.145150504064813e-05,
+      "loss": 1.3134,
+      "step": 3497
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.140166601157928e-05,
+      "loss": 1.2495,
+      "step": 3498
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.135185914624561e-05,
+      "loss": 1.4976,
+      "step": 3499
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.13020844680002e-05,
+      "loss": 1.4076,
+      "step": 3500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.125234200018081e-05,
+      "loss": 1.5333,
+      "step": 3501
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.120263176611029e-05,
+      "loss": 1.3521,
+      "step": 3502
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1152953789096296e-05,
+      "loss": 1.2312,
+      "step": 3503
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.110330809243134e-05,
+      "loss": 1.2048,
+      "step": 3504
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.105369469939284e-05,
+      "loss": 1.4808,
+      "step": 3505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.100411363324306e-05,
+      "loss": 1.2276,
+      "step": 3506
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.095456491722908e-05,
+      "loss": 1.374,
+      "step": 3507
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.090504857458283e-05,
+      "loss": 1.4973,
+      "step": 3508
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0855564628521084e-05,
+      "loss": 1.3334,
+      "step": 3509
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.080611310224539e-05,
+      "loss": 1.3317,
+      "step": 3510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.075669401894211e-05,
+      "loss": 1.3085,
+      "step": 3511
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0707307401782403e-05,
+      "loss": 1.0892,
+      "step": 3512
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0657953273922156e-05,
+      "loss": 1.4406,
+      "step": 3513
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0608631658502165e-05,
+      "loss": 1.2811,
+      "step": 3514
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.055934257864778e-05,
+      "loss": 1.303,
+      "step": 3515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.05100860574692e-05,
+      "loss": 1.2085,
+      "step": 3516
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0460862118061452e-05,
+      "loss": 1.2036,
+      "step": 3517
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0411670783504053e-05,
+      "loss": 1.3175,
+      "step": 3518
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0362512076861473e-05,
+      "loss": 1.3913,
+      "step": 3519
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0313386021182778e-05,
+      "loss": 1.134,
+      "step": 3520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0264292639501624e-05,
+      "loss": 1.2387,
+      "step": 3521
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0215231954836555e-05,
+      "loss": 1.3041,
+      "step": 3522
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0166203990190634e-05,
+      "loss": 1.4711,
+      "step": 3523
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.011720876855164e-05,
+      "loss": 1.5015,
+      "step": 3524
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.006824631289199e-05,
+      "loss": 1.3089,
+      "step": 3525
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0019316646168726e-05,
+      "loss": 1.4178,
+      "step": 3526
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.997041979132352e-05,
+      "loss": 1.5611,
+      "step": 3527
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9921555771282684e-05,
+      "loss": 1.1065,
+      "step": 3528
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9872724608957094e-05,
+      "loss": 1.0664,
+      "step": 3529
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9823926327242223e-05,
+      "loss": 1.3597,
+      "step": 3530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.977516094901823e-05,
+      "loss": 1.2932,
+      "step": 3531
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9726428497149682e-05,
+      "loss": 1.6662,
+      "step": 3532
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9677728994485787e-05,
+      "loss": 1.335,
+      "step": 3533
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.962906246386038e-05,
+      "loss": 1.4106,
+      "step": 3534
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9580428928091665e-05,
+      "loss": 1.1401,
+      "step": 3535
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.953182840998253e-05,
+      "loss": 1.1568,
+      "step": 3536
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9483260932320366e-05,
+      "loss": 1.2813,
+      "step": 3537
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.94347265178769e-05,
+      "loss": 1.3376,
+      "step": 3538
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.938622518940861e-05,
+      "loss": 1.2585,
+      "step": 3539
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9337756969656283e-05,
+      "loss": 1.206,
+      "step": 3540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 1.4323,
+      "step": 3541
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9240919947185298e-05,
+      "loss": 1.3918,
+      "step": 3542
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9192551189870667e-05,
+      "loss": 1.373,
+      "step": 3543
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.914421563208003e-05,
+      "loss": 1.1711,
+      "step": 3544
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9095913296476472e-05,
+      "loss": 1.0226,
+      "step": 3545
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9047644205707636e-05,
+      "loss": 1.2784,
+      "step": 3546
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.899940838240538e-05,
+      "loss": 1.1784,
+      "step": 3547
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.895120584918606e-05,
+      "loss": 1.3195,
+      "step": 3548
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8903036628650502e-05,
+      "loss": 1.1723,
+      "step": 3549
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8854900743383728e-05,
+      "loss": 1.461,
+      "step": 3550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.880679821595531e-05,
+      "loss": 1.3902,
+      "step": 3551
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.875872906891911e-05,
+      "loss": 1.3913,
+      "step": 3552
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8710693324813253e-05,
+      "loss": 1.4081,
+      "step": 3553
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8662691006160357e-05,
+      "loss": 1.31,
+      "step": 3554
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8614722135467276e-05,
+      "loss": 1.5114,
+      "step": 3555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8566786735225192e-05,
+      "loss": 1.2615,
+      "step": 3556
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8518884827909598e-05,
+      "loss": 1.4832,
+      "step": 3557
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.847101643598028e-05,
+      "loss": 1.2861,
+      "step": 3558
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8423181581881332e-05,
+      "loss": 1.4277,
+      "step": 3559
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8375380288041097e-05,
+      "loss": 0.9642,
+      "step": 3560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8327612576872188e-05,
+      "loss": 1.1893,
+      "step": 3561
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.827987847077145e-05,
+      "loss": 1.3331,
+      "step": 3562
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.823217799212008e-05,
+      "loss": 1.2103,
+      "step": 3563
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8184511163283345e-05,
+      "loss": 1.1909,
+      "step": 3564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8136878006610812e-05,
+      "loss": 1.4617,
+      "step": 3565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8089278544436358e-05,
+      "loss": 1.1438,
+      "step": 3566
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8041712799077834e-05,
+      "loss": 1.3017,
+      "step": 3567
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7994180792837522e-05,
+      "loss": 1.2054,
+      "step": 3568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7946682548001757e-05,
+      "loss": 1.2391,
+      "step": 3569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7899218086840984e-05,
+      "loss": 1.4419,
+      "step": 3570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.785178743160999e-05,
+      "loss": 1.1648,
+      "step": 3571
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7804390604547557e-05,
+      "loss": 1.3386,
+      "step": 3572
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7757027627876676e-05,
+      "loss": 1.2797,
+      "step": 3573
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.770969852380445e-05,
+      "loss": 1.2964,
+      "step": 3574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.76624033145221e-05,
+      "loss": 1.4039,
+      "step": 3575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7615142022204942e-05,
+      "loss": 1.1994,
+      "step": 3576
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7567914669012408e-05,
+      "loss": 1.336,
+      "step": 3577
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7520721277088024e-05,
+      "loss": 1.3593,
+      "step": 3578
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7473561868559362e-05,
+      "loss": 1.2449,
+      "step": 3579
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7426436465538074e-05,
+      "loss": 1.3852,
+      "step": 3580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.737934509011989e-05,
+      "loss": 1.3002,
+      "step": 3581
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7332287764384524e-05,
+      "loss": 1.4116,
+      "step": 3582
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.728526451039587e-05,
+      "loss": 1.248,
+      "step": 3583
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7238275350201603e-05,
+      "loss": 1.3429,
+      "step": 3584
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7191320305833666e-05,
+      "loss": 1.5266,
+      "step": 3585
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7144399399307873e-05,
+      "loss": 1.4379,
+      "step": 3586
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.709751265262399e-05,
+      "loss": 1.3093,
+      "step": 3587
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.70506600877659e-05,
+      "loss": 1.4762,
+      "step": 3588
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7003841726701362e-05,
+      "loss": 1.4076,
+      "step": 3589
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.695705759138213e-05,
+      "loss": 1.0254,
+      "step": 3590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6910307703743888e-05,
+      "loss": 1.3441,
+      "step": 3591
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6863592085706303e-05,
+      "loss": 1.3733,
+      "step": 3592
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6816910759172942e-05,
+      "loss": 1.3085,
+      "step": 3593
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6770263746031254e-05,
+      "loss": 1.3318,
+      "step": 3594
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6723651068152768e-05,
+      "loss": 1.0699,
+      "step": 3595
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6677072747392673e-05,
+      "loss": 1.3777,
+      "step": 3596
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6630528805590195e-05,
+      "loss": 1.4771,
+      "step": 3597
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6584019264568495e-05,
+      "loss": 0.9996,
+      "step": 3598
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.653754414613441e-05,
+      "loss": 1.2886,
+      "step": 3599
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6491103472078827e-05,
+      "loss": 1.3104,
+      "step": 3600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.644469726417642e-05,
+      "loss": 1.5153,
+      "step": 3601
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6398325544185596e-05,
+      "loss": 1.2828,
+      "step": 3602
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6351988333848788e-05,
+      "loss": 1.4175,
+      "step": 3603
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.63056856548921e-05,
+      "loss": 1.1846,
+      "step": 3604
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6259417529025497e-05,
+      "loss": 1.5952,
+      "step": 3605
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.621318397794276e-05,
+      "loss": 1.0057,
+      "step": 3606
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.61669850233214e-05,
+      "loss": 1.1642,
+      "step": 3607
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6120820686822778e-05,
+      "loss": 1.4118,
+      "step": 3608
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6074690990091966e-05,
+      "loss": 1.4573,
+      "step": 3609
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6028595954757828e-05,
+      "loss": 1.2678,
+      "step": 3610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5982535602432932e-05,
+      "loss": 1.2757,
+      "step": 3611
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5936509954713718e-05,
+      "loss": 1.3069,
+      "step": 3612
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5890519033180162e-05,
+      "loss": 1.3708,
+      "step": 3613
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5844562859396036e-05,
+      "loss": 1.3864,
+      "step": 3614
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5798641454908944e-05,
+      "loss": 1.1698,
+      "step": 3615
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5752754841249948e-05,
+      "loss": 1.2704,
+      "step": 3616
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5706903039934016e-05,
+      "loss": 1.1397,
+      "step": 3617
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5661086072459707e-05,
+      "loss": 1.4184,
+      "step": 3618
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5615303960309157e-05,
+      "loss": 1.4065,
+      "step": 3619
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.556955672494833e-05,
+      "loss": 1.164,
+      "step": 3620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5523844387826735e-05,
+      "loss": 1.3532,
+      "step": 3621
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5478166970377536e-05,
+      "loss": 1.3529,
+      "step": 3622
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.543252449401753e-05,
+      "loss": 1.4238,
+      "step": 3623
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5386916980147126e-05,
+      "loss": 1.3,
+      "step": 3624
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5341344450150338e-05,
+      "loss": 1.3425,
+      "step": 3625
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5295806925394773e-05,
+      "loss": 1.4764,
+      "step": 3626
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5250304427231654e-05,
+      "loss": 1.3464,
+      "step": 3627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5204836976995738e-05,
+      "loss": 1.1992,
+      "step": 3628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5159404596005364e-05,
+      "loss": 1.3998,
+      "step": 3629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5114007305562504e-05,
+      "loss": 1.1653,
+      "step": 3630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5068645126952504e-05,
+      "loss": 1.162,
+      "step": 3631
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5023318081444424e-05,
+      "loss": 1.1181,
+      "step": 3632
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4978026190290805e-05,
+      "loss": 1.346,
+      "step": 3633
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.493276947472756e-05,
+      "loss": 1.3063,
+      "step": 3634
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4887547955974334e-05,
+      "loss": 1.2299,
+      "step": 3635
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4842361655234135e-05,
+      "loss": 1.0852,
+      "step": 3636
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4797210593693478e-05,
+      "loss": 1.2439,
+      "step": 3637
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4752094792522372e-05,
+      "loss": 1.0423,
+      "step": 3638
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4707014272874274e-05,
+      "loss": 1.2914,
+      "step": 3639
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4661969055886127e-05,
+      "loss": 1.4127,
+      "step": 3640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4616959162678278e-05,
+      "loss": 1.3519,
+      "step": 3641
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.457198461435456e-05,
+      "loss": 1.1925,
+      "step": 3642
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4527045432002182e-05,
+      "loss": 1.4182,
+      "step": 3643
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.448214163669188e-05,
+      "loss": 1.1909,
+      "step": 3644
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4437273249477644e-05,
+      "loss": 1.369,
+      "step": 3645
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.439244029139691e-05,
+      "loss": 1.3913,
+      "step": 3646
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4347642783470648e-05,
+      "loss": 1.3571,
+      "step": 3647
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.430288074670296e-05,
+      "loss": 1.3278,
+      "step": 3648
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4258154202081528e-05,
+      "loss": 1.3604,
+      "step": 3649
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.421346317057731e-05,
+      "loss": 1.2132,
+      "step": 3650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.416880767314452e-05,
+      "loss": 1.2994,
+      "step": 3651
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4124187730720917e-05,
+      "loss": 1.322,
+      "step": 3652
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.407960336422741e-05,
+      "loss": 1.4978,
+      "step": 3653
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4035054594568317e-05,
+      "loss": 1.2828,
+      "step": 3654
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.399054144263123e-05,
+      "loss": 1.2422,
+      "step": 3655
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3946063929287066e-05,
+      "loss": 1.1877,
+      "step": 3656
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3901622075390007e-05,
+      "loss": 1.2523,
+      "step": 3657
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.385721590177754e-05,
+      "loss": 1.5333,
+      "step": 3658
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3812845429270392e-05,
+      "loss": 1.4653,
+      "step": 3659
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3768510678672584e-05,
+      "loss": 1.1061,
+      "step": 3660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3724211670771356e-05,
+      "loss": 1.2005,
+      "step": 3661
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3679948426337227e-05,
+      "loss": 1.2822,
+      "step": 3662
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3635720966123885e-05,
+      "loss": 1.5567,
+      "step": 3663
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3591529310868378e-05,
+      "loss": 1.0216,
+      "step": 3664
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.354737348129077e-05,
+      "loss": 1.1594,
+      "step": 3665
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3503253498094435e-05,
+      "loss": 1.4011,
+      "step": 3666
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3459169381966007e-05,
+      "loss": 1.4096,
+      "step": 3667
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3415121153575137e-05,
+      "loss": 1.0886,
+      "step": 3668
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3371108833574796e-05,
+      "loss": 1.1061,
+      "step": 3669
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.332713244260105e-05,
+      "loss": 1.1637,
+      "step": 3670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3283192001273134e-05,
+      "loss": 1.2998,
+      "step": 3671
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.323928753019342e-05,
+      "loss": 1.2803,
+      "step": 3672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.319541904994742e-05,
+      "loss": 1.2231,
+      "step": 3673
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3151586581103767e-05,
+      "loss": 1.314,
+      "step": 3674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.310779014421419e-05,
+      "loss": 1.2759,
+      "step": 3675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3064029759813633e-05,
+      "loss": 1.1991,
+      "step": 3676
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3020305448419955e-05,
+      "loss": 1.4381,
+      "step": 3677
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2976617230534202e-05,
+      "loss": 1.1805,
+      "step": 3678
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.293296512664057e-05,
+      "loss": 1.1536,
+      "step": 3679
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2889349157206142e-05,
+      "loss": 1.3585,
+      "step": 3680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2845769342681222e-05,
+      "loss": 1.4516,
+      "step": 3681
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2802225703499127e-05,
+      "loss": 1.265,
+      "step": 3682
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2758718260076085e-05,
+      "loss": 1.2824,
+      "step": 3683
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2715247032811536e-05,
+      "loss": 1.2107,
+      "step": 3684
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2671812042087835e-05,
+      "loss": 1.1783,
+      "step": 3685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2628413308270357e-05,
+      "loss": 1.2252,
+      "step": 3686
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.25850508517075e-05,
+      "loss": 1.3225,
+      "step": 3687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.254172469273065e-05,
+      "loss": 1.3457,
+      "step": 3688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2498434851654126e-05,
+      "loss": 1.3758,
+      "step": 3689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.24551813487753e-05,
+      "loss": 1.2742,
+      "step": 3690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2411964204374437e-05,
+      "loss": 1.25,
+      "step": 3691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.236878343871479e-05,
+      "loss": 1.3146,
+      "step": 3692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.232563907204255e-05,
+      "loss": 0.9725,
+      "step": 3693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2282531124586837e-05,
+      "loss": 1.2843,
+      "step": 3694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2239459616559667e-05,
+      "loss": 1.0888,
+      "step": 3695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2196424568156073e-05,
+      "loss": 1.2173,
+      "step": 3696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.215342599955381e-05,
+      "loss": 1.1806,
+      "step": 3697
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.211046393091374e-05,
+      "loss": 0.623,
+      "step": 3698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2067538382379504e-05,
+      "loss": 1.1276,
+      "step": 3699
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2024649374077522e-05,
+      "loss": 1.5488,
+      "step": 3700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1981796926117292e-05,
+      "loss": 1.2557,
+      "step": 3701
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1938981058591023e-05,
+      "loss": 1.0793,
+      "step": 3702
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1896201791573812e-05,
+      "loss": 1.2233,
+      "step": 3703
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.18534591451236e-05,
+      "loss": 1.3374,
+      "step": 3704
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1810753139281147e-05,
+      "loss": 1.3263,
+      "step": 3705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.176808379407005e-05,
+      "loss": 1.4492,
+      "step": 3706
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1725451129496687e-05,
+      "loss": 1.3818,
+      "step": 3707
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1682855165550288e-05,
+      "loss": 1.111,
+      "step": 3708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1640295922202815e-05,
+      "loss": 1.1268,
+      "step": 3709
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.159777341940905e-05,
+      "loss": 1.2413,
+      "step": 3710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1555287677106552e-05,
+      "loss": 1.2755,
+      "step": 3711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.151283871521558e-05,
+      "loss": 1.191,
+      "step": 3712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1470426553639313e-05,
+      "loss": 1.2821,
+      "step": 3713
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1428051212263456e-05,
+      "loss": 1.2282,
+      "step": 3714
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1385712710956563e-05,
+      "loss": 1.288,
+      "step": 3715
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1343411069569997e-05,
+      "loss": 1.3187,
+      "step": 3716
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1301146307937636e-05,
+      "loss": 1.1172,
+      "step": 3717
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1258918445876263e-05,
+      "loss": 1.0801,
+      "step": 3718
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1216727503185276e-05,
+      "loss": 1.6415,
+      "step": 3719
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.117457349964668e-05,
+      "loss": 1.1901,
+      "step": 3720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1132456455025328e-05,
+      "loss": 1.0944,
+      "step": 3721
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.109037638906862e-05,
+      "loss": 1.5033,
+      "step": 3722
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1048333321506686e-05,
+      "loss": 1.2369,
+      "step": 3723
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1006327272052227e-05,
+      "loss": 1.3563,
+      "step": 3724
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.096435826040074e-05,
+      "loss": 1.3599,
+      "step": 3725
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.092242630623016e-05,
+      "loss": 1.3147,
+      "step": 3726
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0880531429201145e-05,
+      "loss": 1.3804,
+      "step": 3727
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0838673648957064e-05,
+      "loss": 1.3876,
+      "step": 3728
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.079685298512366e-05,
+      "loss": 1.2398,
+      "step": 3729
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0755069457309516e-05,
+      "loss": 1.6061,
+      "step": 3730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0713323085105686e-05,
+      "loss": 1.3528,
+      "step": 3731
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0671613888085726e-05,
+      "loss": 1.2526,
+      "step": 3732
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0629941885805937e-05,
+      "loss": 1.5227,
+      "step": 3733
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.058830709780506e-05,
+      "loss": 1.2722,
+      "step": 3734
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0546709543604426e-05,
+      "loss": 1.4851,
+      "step": 3735
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0505149242707888e-05,
+      "loss": 1.4261,
+      "step": 3736
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.046362621460185e-05,
+      "loss": 1.0876,
+      "step": 3737
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0422140478755248e-05,
+      "loss": 1.1449,
+      "step": 3738
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0380692054619498e-05,
+      "loss": 1.3068,
+      "step": 3739
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.033928096162856e-05,
+      "loss": 1.3306,
+      "step": 3740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0297907219198876e-05,
+      "loss": 1.3043,
+      "step": 3741
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.025657084672936e-05,
+      "loss": 1.2759,
+      "step": 3742
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0215271863601413e-05,
+      "loss": 1.5443,
+      "step": 3743
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0174010289178913e-05,
+      "loss": 1.3328,
+      "step": 3744
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0132786142808237e-05,
+      "loss": 1.2881,
+      "step": 3745
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0091599443818122e-05,
+      "loss": 1.3303,
+      "step": 3746
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0050450211519777e-05,
+      "loss": 1.414,
+      "step": 3747
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.000933846520696e-05,
+      "loss": 1.5068,
+      "step": 3748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.996826422415563e-05,
+      "loss": 1.1737,
+      "step": 3749
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.992722750762438e-05,
+      "loss": 1.1493,
+      "step": 3750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9886228334854085e-05,
+      "loss": 1.2078,
+      "step": 3751
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9845266725068058e-05,
+      "loss": 1.2418,
+      "step": 3752
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9804342697471968e-05,
+      "loss": 1.3893,
+      "step": 3753
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.97634562712539e-05,
+      "loss": 1.2871,
+      "step": 3754
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9722607465584296e-05,
+      "loss": 1.2171,
+      "step": 3755
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.968179629961594e-05,
+      "loss": 1.5353,
+      "step": 3756
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9641022792484e-05,
+      "loss": 1.3292,
+      "step": 3757
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9600286963305957e-05,
+      "loss": 1.2506,
+      "step": 3758
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9559588831181653e-05,
+      "loss": 1.3844,
+      "step": 3759
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9518928415193237e-05,
+      "loss": 1.3093,
+      "step": 3760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9478305734405145e-05,
+      "loss": 1.3026,
+      "step": 3761
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9437720807864213e-05,
+      "loss": 1.151,
+      "step": 3762
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.939717365459952e-05,
+      "loss": 1.3395,
+      "step": 3763
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.935666429362234e-05,
+      "loss": 1.0954,
+      "step": 3764
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.93161927439264e-05,
+      "loss": 1.4063,
+      "step": 3765
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9275759024487594e-05,
+      "loss": 1.4461,
+      "step": 3766
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9235363154264108e-05,
+      "loss": 1.0,
+      "step": 3767
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9195005152196356e-05,
+      "loss": 1.12,
+      "step": 3768
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.915468503720702e-05,
+      "loss": 1.5773,
+      "step": 3769
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9114402828201016e-05,
+      "loss": 1.3418,
+      "step": 3770
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.907415854406548e-05,
+      "loss": 1.3533,
+      "step": 3771
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.903395220366978e-05,
+      "loss": 1.3011,
+      "step": 3772
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8993783825865473e-05,
+      "loss": 1.1958,
+      "step": 3773
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8953653429486327e-05,
+      "loss": 1.3881,
+      "step": 3774
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8913561033348305e-05,
+      "loss": 1.1878,
+      "step": 3775
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8873506656249518e-05,
+      "loss": 1.3872,
+      "step": 3776
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.883349031697036e-05,
+      "loss": 1.4133,
+      "step": 3777
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8793512034273242e-05,
+      "loss": 1.4043,
+      "step": 3778
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8753571826902793e-05,
+      "loss": 1.1767,
+      "step": 3779
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.871366971358588e-05,
+      "loss": 1.0733,
+      "step": 3780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8673805713031323e-05,
+      "loss": 1.4714,
+      "step": 3781
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.863397984393025e-05,
+      "loss": 1.0709,
+      "step": 3782
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8594192124955812e-05,
+      "loss": 1.3078,
+      "step": 3783
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8554442574763298e-05,
+      "loss": 1.0961,
+      "step": 3784
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.85147312119901e-05,
+      "loss": 1.3811,
+      "step": 3785
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.847505805525571e-05,
+      "loss": 1.5066,
+      "step": 3786
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8435423123161687e-05,
+      "loss": 1.3066,
+      "step": 3787
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8395826434291697e-05,
+      "loss": 1.1491,
+      "step": 3788
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.835626800721144e-05,
+      "loss": 1.4865,
+      "step": 3789
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8316747860468708e-05,
+      "loss": 1.4353,
+      "step": 3790
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8277266012593343e-05,
+      "loss": 1.1335,
+      "step": 3791
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.82378224820972e-05,
+      "loss": 1.0836,
+      "step": 3792
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8198417287474168e-05,
+      "loss": 1.4331,
+      "step": 3793
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8159050447200254e-05,
+      "loss": 1.5855,
+      "step": 3794
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.811972197973334e-05,
+      "loss": 1.3104,
+      "step": 3795
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8080431903513374e-05,
+      "loss": 1.4097,
+      "step": 3796
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8041180236962395e-05,
+      "loss": 1.5653,
+      "step": 3797
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8001966998484253e-05,
+      "loss": 1.4251,
+      "step": 3798
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.796279220646495e-05,
+      "loss": 1.3631,
+      "step": 3799
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7923655879272393e-05,
+      "loss": 1.2462,
+      "step": 3800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.788455803525637e-05,
+      "loss": 1.3161,
+      "step": 3801
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7845498692748798e-05,
+      "loss": 1.3513,
+      "step": 3802
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.780647787006341e-05,
+      "loss": 1.3211,
+      "step": 3803
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7767495585495918e-05,
+      "loss": 1.3755,
+      "step": 3804
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.772855185732397e-05,
+      "loss": 1.243,
+      "step": 3805
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.768964670380714e-05,
+      "loss": 1.4364,
+      "step": 3806
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7650780143186897e-05,
+      "loss": 1.1604,
+      "step": 3807
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7611952193686598e-05,
+      "loss": 1.5062,
+      "step": 3808
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7573162873511596e-05,
+      "loss": 1.2465,
+      "step": 3809
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.753441220084897e-05,
+      "loss": 1.2576,
+      "step": 3810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7495700193867836e-05,
+      "loss": 1.2609,
+      "step": 3811
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7457026870719116e-05,
+      "loss": 1.376,
+      "step": 3812
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.741839224953551e-05,
+      "loss": 1.2913,
+      "step": 3813
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.737979634843172e-05,
+      "loss": 1.271,
+      "step": 3814
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.734123918550422e-05,
+      "loss": 1.1598,
+      "step": 3815
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7302720778831316e-05,
+      "loss": 1.4488,
+      "step": 3816
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7264241146473147e-05,
+      "loss": 1.4011,
+      "step": 3817
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7225800306471696e-05,
+      "loss": 1.1821,
+      "step": 3818
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7187398276850708e-05,
+      "loss": 1.3271,
+      "step": 3819
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7149035075615794e-05,
+      "loss": 1.2728,
+      "step": 3820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7110710720754307e-05,
+      "loss": 1.2479,
+      "step": 3821
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.707242523023542e-05,
+      "loss": 1.5403,
+      "step": 3822
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7034178622010056e-05,
+      "loss": 0.9868,
+      "step": 3823
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6995970914010928e-05,
+      "loss": 1.3235,
+      "step": 3824
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.695780212415249e-05,
+      "loss": 1.3476,
+      "step": 3825
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.691967227033101e-05,
+      "loss": 1.2581,
+      "step": 3826
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6881581370424393e-05,
+      "loss": 1.221,
+      "step": 3827
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.684352944229235e-05,
+      "loss": 1.1976,
+      "step": 3828
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6805516503776353e-05,
+      "loss": 1.2566,
+      "step": 3829
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.676754257269948e-05,
+      "loss": 1.3571,
+      "step": 3830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6729607666866643e-05,
+      "loss": 1.1726,
+      "step": 3831
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.669171180406439e-05,
+      "loss": 1.3294,
+      "step": 3832
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6653855002060926e-05,
+      "loss": 1.1981,
+      "step": 3833
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6616037278606246e-05,
+      "loss": 1.15,
+      "step": 3834
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6578258651431944e-05,
+      "loss": 1.2389,
+      "step": 3835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.654051913825131e-05,
+      "loss": 1.3709,
+      "step": 3836
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6502818756759276e-05,
+      "loss": 1.2488,
+      "step": 3837
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6465157524632436e-05,
+      "loss": 1.3588,
+      "step": 3838
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6427535459529052e-05,
+      "loss": 1.1904,
+      "step": 3839
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6389952579088975e-05,
+      "loss": 1.5486,
+      "step": 3840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6352408900933734e-05,
+      "loss": 1.2718,
+      "step": 3841
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.631490444266641e-05,
+      "loss": 1.466,
+      "step": 3842
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6277439221871815e-05,
+      "loss": 1.3209,
+      "step": 3843
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6240013256116216e-05,
+      "loss": 1.223,
+      "step": 3844
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6202626562947544e-05,
+      "loss": 1.275,
+      "step": 3845
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6165279159895387e-05,
+      "loss": 1.192,
+      "step": 3846
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6127971064470747e-05,
+      "loss": 1.2689,
+      "step": 3847
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6090702294166347e-05,
+      "loss": 1.3017,
+      "step": 3848
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6053472866456444e-05,
+      "loss": 1.3849,
+      "step": 3849
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.601628279879671e-05,
+      "loss": 1.1136,
+      "step": 3850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5979132108624574e-05,
+      "loss": 1.4385,
+      "step": 3851
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.594202081335885e-05,
+      "loss": 1.417,
+      "step": 3852
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5904948930399955e-05,
+      "loss": 1.4248,
+      "step": 3853
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5867916477129774e-05,
+      "loss": 1.3549,
+      "step": 3854
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5830923470911752e-05,
+      "loss": 1.5187,
+      "step": 3855
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5793969929090812e-05,
+      "loss": 1.358,
+      "step": 3856
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5757055868993353e-05,
+      "loss": 1.6123,
+      "step": 3857
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5720181307927373e-05,
+      "loss": 1.2998,
+      "step": 3858
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5683346263182175e-05,
+      "loss": 1.4343,
+      "step": 3859
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5646550752028644e-05,
+      "loss": 1.1686,
+      "step": 3860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5609794791719156e-05,
+      "loss": 1.3949,
+      "step": 3861
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5573078399487428e-05,
+      "loss": 1.3118,
+      "step": 3862
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5536401592548754e-05,
+      "loss": 1.3508,
+      "step": 3863
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.549976438809978e-05,
+      "loss": 1.3108,
+      "step": 3864
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.546316680331861e-05,
+      "loss": 1.5029,
+      "step": 3865
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5426608855364778e-05,
+      "loss": 1.2539,
+      "step": 3866
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5390090561379224e-05,
+      "loss": 1.5544,
+      "step": 3867
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.53536119384843e-05,
+      "loss": 1.2499,
+      "step": 3868
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5317173003783756e-05,
+      "loss": 1.2537,
+      "step": 3869
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5280773774362732e-05,
+      "loss": 1.1193,
+      "step": 3870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5244414267287754e-05,
+      "loss": 1.3486,
+      "step": 3871
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5208094499606718e-05,
+      "loss": 1.1086,
+      "step": 3872
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5171814488348902e-05,
+      "loss": 1.3328,
+      "step": 3873
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5135574250524897e-05,
+      "loss": 1.3529,
+      "step": 3874
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.509937380312676e-05,
+      "loss": 1.2355,
+      "step": 3875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5063213163127743e-05,
+      "loss": 1.258,
+      "step": 3876
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5027092347482497e-05,
+      "loss": 1.1544,
+      "step": 3877
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4991011373127084e-05,
+      "loss": 1.3303,
+      "step": 3878
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4954970256978718e-05,
+      "loss": 1.2575,
+      "step": 3879
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4918969015936069e-05,
+      "loss": 1.44,
+      "step": 3880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4883007666879079e-05,
+      "loss": 1.5821,
+      "step": 3881
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4847086226668872e-05,
+      "loss": 1.2538,
+      "step": 3882
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.481120471214803e-05,
+      "loss": 1.3004,
+      "step": 3883
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.477536314014033e-05,
+      "loss": 1.2887,
+      "step": 3884
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4739561527450796e-05,
+      "loss": 1.2431,
+      "step": 3885
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4703799890865778e-05,
+      "loss": 1.1207,
+      "step": 3886
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.466807824715284e-05,
+      "loss": 1.3305,
+      "step": 3887
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4632396613060795e-05,
+      "loss": 1.2246,
+      "step": 3888
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4596755005319729e-05,
+      "loss": 1.2067,
+      "step": 3889
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4561153440640929e-05,
+      "loss": 1.1662,
+      "step": 3890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4525591935716898e-05,
+      "loss": 1.4889,
+      "step": 3891
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4490070507221443e-05,
+      "loss": 1.2466,
+      "step": 3892
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4454589171809451e-05,
+      "loss": 1.1982,
+      "step": 3893
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4419147946117073e-05,
+      "loss": 1.333,
+      "step": 3894
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4383746846761692e-05,
+      "loss": 1.2763,
+      "step": 3895
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.434838589034183e-05,
+      "loss": 1.1565,
+      "step": 3896
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4313065093437183e-05,
+      "loss": 1.3147,
+      "step": 3897
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4277784472608635e-05,
+      "loss": 1.0337,
+      "step": 3898
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4242544044398231e-05,
+      "loss": 1.2028,
+      "step": 3899
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4207343825329168e-05,
+      "loss": 1.4463,
+      "step": 3900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4172183831905784e-05,
+      "loss": 1.4094,
+      "step": 3901
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4137064080613571e-05,
+      "loss": 1.3119,
+      "step": 3902
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4101984587919137e-05,
+      "loss": 1.1834,
+      "step": 3903
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4066945370270223e-05,
+      "loss": 1.214,
+      "step": 3904
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4031946444095701e-05,
+      "loss": 1.2083,
+      "step": 3905
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3996987825805486e-05,
+      "loss": 1.3696,
+      "step": 3906
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3962069531790722e-05,
+      "loss": 1.5248,
+      "step": 3907
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3927191578423504e-05,
+      "loss": 1.18,
+      "step": 3908
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3892353982057071e-05,
+      "loss": 1.1453,
+      "step": 3909
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3857556759025813e-05,
+      "loss": 1.2116,
+      "step": 3910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3822799925645036e-05,
+      "loss": 1.4207,
+      "step": 3911
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3788083498211247e-05,
+      "loss": 1.3322,
+      "step": 3912
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3753407493001968e-05,
+      "loss": 1.2395,
+      "step": 3913
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3718771926275687e-05,
+      "loss": 1.4002,
+      "step": 3914
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3684176814272054e-05,
+      "loss": 1.3005,
+      "step": 3915
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3649622173211696e-05,
+      "loss": 1.0755,
+      "step": 3916
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.361510801929624e-05,
+      "loss": 1.4111,
+      "step": 3917
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3580634368708366e-05,
+      "loss": 1.4074,
+      "step": 3918
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3546201237611766e-05,
+      "loss": 1.1082,
+      "step": 3919
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3511808642151102e-05,
+      "loss": 1.3041,
+      "step": 3920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3477456598452042e-05,
+      "loss": 1.2216,
+      "step": 3921
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.344314512262126e-05,
+      "loss": 1.4058,
+      "step": 3922
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3408874230746371e-05,
+      "loss": 1.2418,
+      "step": 3923
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3374643938896047e-05,
+      "loss": 1.3891,
+      "step": 3924
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3340454263119805e-05,
+      "loss": 1.1752,
+      "step": 3925
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3306305219448156e-05,
+      "loss": 1.4306,
+      "step": 3926
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3272196823892668e-05,
+      "loss": 1.0562,
+      "step": 3927
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3238129092445661e-05,
+      "loss": 1.5187,
+      "step": 3928
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3204102041080546e-05,
+      "loss": 1.2197,
+      "step": 3929
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3170115685751627e-05,
+      "loss": 1.2996,
+      "step": 3930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3136170042394025e-05,
+      "loss": 1.3535,
+      "step": 3931
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3102265126923907e-05,
+      "loss": 1.3318,
+      "step": 3932
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.306840095523828e-05,
+      "loss": 1.2535,
+      "step": 3933
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3034577543215054e-05,
+      "loss": 1.0449,
+      "step": 3934
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3000794906713032e-05,
+      "loss": 1.17,
+      "step": 3935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2967053061571877e-05,
+      "loss": 1.3514,
+      "step": 3936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2933352023612166e-05,
+      "loss": 1.3889,
+      "step": 3937
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2899691808635283e-05,
+      "loss": 1.2946,
+      "step": 3938
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2866072432423593e-05,
+      "loss": 1.3369,
+      "step": 3939
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.283249391074015e-05,
+      "loss": 1.0556,
+      "step": 3940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2798956259328942e-05,
+      "loss": 1.3548,
+      "step": 3941
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2765459493914843e-05,
+      "loss": 1.2684,
+      "step": 3942
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2732003630203415e-05,
+      "loss": 1.3328,
+      "step": 3943
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2698588683881186e-05,
+      "loss": 1.3305,
+      "step": 3944
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2665214670615432e-05,
+      "loss": 1.3767,
+      "step": 3945
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2631881606054196e-05,
+      "loss": 1.3009,
+      "step": 3946
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2598589505826431e-05,
+      "loss": 1.4248,
+      "step": 3947
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2565338385541792e-05,
+      "loss": 1.4252,
+      "step": 3948
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2532128260790743e-05,
+      "loss": 1.3805,
+      "step": 3949
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.249895914714454e-05,
+      "loss": 1.2624,
+      "step": 3950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.246583106015521e-05,
+      "loss": 1.3787,
+      "step": 3951
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2432744015355524e-05,
+      "loss": 1.4193,
+      "step": 3952
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.239969802825901e-05,
+      "loss": 1.5175,
+      "step": 3953
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2366693114359984e-05,
+      "loss": 1.3542,
+      "step": 3954
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2333729289133444e-05,
+      "loss": 1.3369,
+      "step": 3955
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2300806568035206e-05,
+      "loss": 1.286,
+      "step": 3956
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.226792496650172e-05,
+      "loss": 1.3059,
+      "step": 3957
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2235084499950178e-05,
+      "loss": 1.2128,
+      "step": 3958
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2202285183778583e-05,
+      "loss": 1.5455,
+      "step": 3959
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.216952703336548e-05,
+      "loss": 1.1659,
+      "step": 3960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.213681006407027e-05,
+      "loss": 1.3764,
+      "step": 3961
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2104134291232961e-05,
+      "loss": 1.2411,
+      "step": 3962
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2071499730174197e-05,
+      "loss": 1.4159,
+      "step": 3963
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2038906396195427e-05,
+      "loss": 1.5442,
+      "step": 3964
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2006354304578692e-05,
+      "loss": 1.1292,
+      "step": 3965
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1973843470586699e-05,
+      "loss": 1.2563,
+      "step": 3966
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1941373909462816e-05,
+      "loss": 1.5493,
+      "step": 3967
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.190894563643108e-05,
+      "loss": 1.1622,
+      "step": 3968
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1876558666696125e-05,
+      "loss": 1.3485,
+      "step": 3969
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1844213015443262e-05,
+      "loss": 1.3058,
+      "step": 3970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1811908697838403e-05,
+      "loss": 1.2382,
+      "step": 3971
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.177964572902811e-05,
+      "loss": 1.3328,
+      "step": 3972
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1747424124139517e-05,
+      "loss": 1.2447,
+      "step": 3973
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1715243898280392e-05,
+      "loss": 1.4658,
+      "step": 3974
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1683105066539068e-05,
+      "loss": 1.4642,
+      "step": 3975
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1651007643984557e-05,
+      "loss": 1.2916,
+      "step": 3976
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.161895164566632e-05,
+      "loss": 1.2319,
+      "step": 3977
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1586937086614525e-05,
+      "loss": 1.3185,
+      "step": 3978
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1554963981839851e-05,
+      "loss": 1.5992,
+      "step": 3979
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1523032346333495e-05,
+      "loss": 1.2046,
+      "step": 3980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1491142195067306e-05,
+      "loss": 1.5655,
+      "step": 3981
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1459293542993622e-05,
+      "loss": 1.0577,
+      "step": 3982
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.142748640504533e-05,
+      "loss": 1.3195,
+      "step": 3983
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1395720796135867e-05,
+      "loss": 1.2173,
+      "step": 3984
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1363996731159188e-05,
+      "loss": 1.2933,
+      "step": 3985
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.133231422498976e-05,
+      "loss": 1.178,
+      "step": 3986
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1300673292482567e-05,
+      "loss": 1.2587,
+      "step": 3987
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1269073948473174e-05,
+      "loss": 1.0108,
+      "step": 3988
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.123751620777751e-05,
+      "loss": 1.3767,
+      "step": 3989
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1206000085192069e-05,
+      "loss": 1.2319,
+      "step": 3990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.11745255954939e-05,
+      "loss": 1.4054,
+      "step": 3991
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1143092753440387e-05,
+      "loss": 1.2386,
+      "step": 3992
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1111701573769495e-05,
+      "loss": 1.4923,
+      "step": 3993
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1080352071199652e-05,
+      "loss": 1.331,
+      "step": 3994
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1049044260429642e-05,
+      "loss": 1.2133,
+      "step": 3995
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1017778156138836e-05,
+      "loss": 1.5025,
+      "step": 3996
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0986553772986962e-05,
+      "loss": 1.0303,
+      "step": 3997
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0955371125614223e-05,
+      "loss": 1.0437,
+      "step": 3998
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.092423022864123e-05,
+      "loss": 1.2,
+      "step": 3999
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.089313109666904e-05,
+      "loss": 1.3838,
+      "step": 4000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0862073744279111e-05,
+      "loss": 1.2578,
+      "step": 4001
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0831058186033338e-05,
+      "loss": 1.2271,
+      "step": 4002
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0800084436473978e-05,
+      "loss": 1.4623,
+      "step": 4003
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0769152510123704e-05,
+      "loss": 1.4072,
+      "step": 4004
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0738262421485634e-05,
+      "loss": 1.1971,
+      "step": 4005
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0707414185043163e-05,
+      "loss": 1.3987,
+      "step": 4006
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0676607815260121e-05,
+      "loss": 1.4341,
+      "step": 4007
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0645843326580763e-05,
+      "loss": 1.3205,
+      "step": 4008
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0615120733429584e-05,
+      "loss": 1.3654,
+      "step": 4009
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0584440050211541e-05,
+      "loss": 1.2875,
+      "step": 4010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0553801291311905e-05,
+      "loss": 1.4102,
+      "step": 4011
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0523204471096237e-05,
+      "loss": 1.2083,
+      "step": 4012
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0492649603910531e-05,
+      "loss": 1.4174,
+      "step": 4013
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.046213670408106e-05,
+      "loss": 1.4022,
+      "step": 4014
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0431665785914402e-05,
+      "loss": 1.3079,
+      "step": 4015
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0401236863697494e-05,
+      "loss": 1.7153,
+      "step": 4016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0370849951697537e-05,
+      "loss": 1.393,
+      "step": 4017
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0340505064162076e-05,
+      "loss": 1.0931,
+      "step": 4018
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.031020221531892e-05,
+      "loss": 1.3765,
+      "step": 4019
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0279941419376182e-05,
+      "loss": 1.0296,
+      "step": 4020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0249722690522268e-05,
+      "loss": 1.1054,
+      "step": 4021
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0219546042925843e-05,
+      "loss": 1.2219,
+      "step": 4022
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0189411490735846e-05,
+      "loss": 1.3441,
+      "step": 4023
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.015931904808146e-05,
+      "loss": 1.2702,
+      "step": 4024
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0129268729072205e-05,
+      "loss": 1.1955,
+      "step": 4025
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0099260547797729e-05,
+      "loss": 1.3033,
+      "step": 4026
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0069294518327976e-05,
+      "loss": 1.4755,
+      "step": 4027
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0039370654713199e-05,
+      "loss": 1.4929,
+      "step": 4028
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0009488970983727e-05,
+      "loss": 1.2055,
+      "step": 4029
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.979649481150277e-06,
+      "loss": 0.9926,
+      "step": 4030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.94985219920368e-06,
+      "loss": 1.2381,
+      "step": 4031
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.920097139114992e-06,
+      "loss": 1.4373,
+      "step": 4032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.890384314835488e-06,
+      "loss": 1.2233,
+      "step": 4033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.860713740296634e-06,
+      "loss": 1.2898,
+      "step": 4034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.831085429410102e-06,
+      "loss": 1.41,
+      "step": 4035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.801499396067693e-06,
+      "loss": 1.2161,
+      "step": 4036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.771955654141496e-06,
+      "loss": 1.2447,
+      "step": 4037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.742454217483632e-06,
+      "loss": 1.3202,
+      "step": 4038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.712995099926459e-06,
+      "loss": 1.2657,
+      "step": 4039
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.683578315282549e-06,
+      "loss": 1.1596,
+      "step": 4040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.654203877344482e-06,
+      "loss": 1.2966,
+      "step": 4041
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.624871799885127e-06,
+      "loss": 1.114,
+      "step": 4042
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.595582096657441e-06,
+      "loss": 1.4058,
+      "step": 4043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.566334781394426e-06,
+      "loss": 1.4747,
+      "step": 4044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.537129867809347e-06,
+      "loss": 1.0505,
+      "step": 4045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.507967369595506e-06,
+      "loss": 1.3956,
+      "step": 4046
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.478847300426352e-06,
+      "loss": 1.5116,
+      "step": 4047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.449769673955411e-06,
+      "loss": 0.9317,
+      "step": 4048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.420734503816331e-06,
+      "loss": 1.4447,
+      "step": 4049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.391741803622845e-06,
+      "loss": 1.4177,
+      "step": 4050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.362791586968767e-06,
+      "loss": 1.1777,
+      "step": 4051
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.333883867428017e-06,
+      "loss": 1.3757,
+      "step": 4052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.305018658554565e-06,
+      "loss": 1.0227,
+      "step": 4053
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.27619597388245e-06,
+      "loss": 1.1268,
+      "step": 4054
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.247415826925788e-06,
+      "loss": 1.453,
+      "step": 4055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.218678231178702e-06,
+      "loss": 1.3213,
+      "step": 4056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.189983200115481e-06,
+      "loss": 1.186,
+      "step": 4057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.161330747190321e-06,
+      "loss": 1.3184,
+      "step": 4058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.13272088583751e-06,
+      "loss": 1.3458,
+      "step": 4059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.104153629471413e-06,
+      "loss": 1.4533,
+      "step": 4060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.075628991486307e-06,
+      "loss": 1.2984,
+      "step": 4061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.047146985256604e-06,
+      "loss": 1.2007,
+      "step": 4062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.018707624136669e-06,
+      "loss": 1.4009,
+      "step": 4063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.990310921460865e-06,
+      "loss": 1.3691,
+      "step": 4064
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.961956890543588e-06,
+      "loss": 1.1572,
+      "step": 4065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.933645544679182e-06,
+      "loss": 1.1902,
+      "step": 4066
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.905376897142026e-06,
+      "loss": 1.4608,
+      "step": 4067
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.87715096118642e-06,
+      "loss": 1.5668,
+      "step": 4068
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.848967750046732e-06,
+      "loss": 1.223,
+      "step": 4069
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.820827276937182e-06,
+      "loss": 1.2876,
+      "step": 4070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.792729555052026e-06,
+      "loss": 1.256,
+      "step": 4071
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.764674597565481e-06,
+      "loss": 1.4745,
+      "step": 4072
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.736662417631636e-06,
+      "loss": 1.2233,
+      "step": 4073
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.708693028384617e-06,
+      "loss": 1.3665,
+      "step": 4074
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.68076644293847e-06,
+      "loss": 1.2802,
+      "step": 4075
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.652882674387064e-06,
+      "loss": 1.0851,
+      "step": 4076
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.625041735804352e-06,
+      "loss": 1.4054,
+      "step": 4077
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.59724364024409e-06,
+      "loss": 1.1187,
+      "step": 4078
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.56948840074e-06,
+      "loss": 1.3978,
+      "step": 4079
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.541776030305693e-06,
+      "loss": 1.4267,
+      "step": 4080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.514106541934685e-06,
+      "loss": 0.875,
+      "step": 4081
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.486479948600368e-06,
+      "loss": 1.2197,
+      "step": 4082
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.458896263256055e-06,
+      "loss": 1.2126,
+      "step": 4083
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.431355498834915e-06,
+      "loss": 1.1845,
+      "step": 4084
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.403857668249993e-06,
+      "loss": 1.2334,
+      "step": 4085
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.376402784394221e-06,
+      "loss": 1.2243,
+      "step": 4086
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.348990860140382e-06,
+      "loss": 1.2247,
+      "step": 4087
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.321621908341093e-06,
+      "loss": 1.1688,
+      "step": 4088
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.294295941828911e-06,
+      "loss": 1.5978,
+      "step": 4089
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.2670129734161e-06,
+      "loss": 1.1036,
+      "step": 4090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.239773015894903e-06,
+      "loss": 1.2569,
+      "step": 4091
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.212576082037304e-06,
+      "loss": 1.135,
+      "step": 4092
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.185422184595127e-06,
+      "loss": 1.3138,
+      "step": 4093
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.15831133630005e-06,
+      "loss": 1.141,
+      "step": 4094
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.131243549863566e-06,
+      "loss": 1.3222,
+      "step": 4095
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.10421883797694e-06,
+      "loss": 1.294,
+      "step": 4096
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.077237213311272e-06,
+      "loss": 1.155,
+      "step": 4097
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.050298688517444e-06,
+      "loss": 1.216,
+      "step": 4098
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.023403276226126e-06,
+      "loss": 1.2217,
+      "step": 4099
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.996550989047813e-06,
+      "loss": 1.2325,
+      "step": 4100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.96974183957272e-06,
+      "loss": 1.2897,
+      "step": 4101
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.942975840370892e-06,
+      "loss": 1.3194,
+      "step": 4102
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.91625300399209e-06,
+      "loss": 1.2306,
+      "step": 4103
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.889573342965884e-06,
+      "loss": 1.4601,
+      "step": 4104
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.862936869801551e-06,
+      "loss": 1.1739,
+      "step": 4105
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.836343596988216e-06,
+      "loss": 1.1883,
+      "step": 4106
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.809793536994603e-06,
+      "loss": 1.0954,
+      "step": 4107
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.783286702269278e-06,
+      "loss": 1.0699,
+      "step": 4108
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75682310524054e-06,
+      "loss": 1.2561,
+      "step": 4109
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.730402758316334e-06,
+      "loss": 1.2218,
+      "step": 4110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.70402567388444e-06,
+      "loss": 1.5597,
+      "step": 4111
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.677691864312287e-06,
+      "loss": 1.2793,
+      "step": 4112
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.651401341946962e-06,
+      "loss": 1.543,
+      "step": 4113
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.625154119115385e-06,
+      "loss": 0.9852,
+      "step": 4114
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.598950208124067e-06,
+      "loss": 1.3472,
+      "step": 4115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.572789621259268e-06,
+      "loss": 1.2297,
+      "step": 4116
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.546672370786889e-06,
+      "loss": 1.2572,
+      "step": 4117
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.520598468952578e-06,
+      "loss": 1.336,
+      "step": 4118
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.494567927981588e-06,
+      "loss": 1.3818,
+      "step": 4119
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.468580760078869e-06,
+      "loss": 1.3598,
+      "step": 4120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.442636977429074e-06,
+      "loss": 1.1291,
+      "step": 4121
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.416736592196416e-06,
+      "loss": 1.3058,
+      "step": 4122
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.390879616524881e-06,
+      "loss": 1.3059,
+      "step": 4123
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.3650660625380285e-06,
+      "loss": 1.3425,
+      "step": 4124
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.339295942339031e-06,
+      "loss": 1.2404,
+      "step": 4125
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.313569268010789e-06,
+      "loss": 1.5169,
+      "step": 4126
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.287886051615766e-06,
+      "loss": 1.4374,
+      "step": 4127
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.262246305196052e-06,
+      "loss": 1.1479,
+      "step": 4128
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.236650040773374e-06,
+      "loss": 1.3792,
+      "step": 4129
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.211097270349066e-06,
+      "loss": 1.3963,
+      "step": 4130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.185588005904076e-06,
+      "loss": 1.0611,
+      "step": 4131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.160122259398938e-06,
+      "loss": 1.4262,
+      "step": 4132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.13470004277379e-06,
+      "loss": 1.3963,
+      "step": 4133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.109321367948352e-06,
+      "loss": 1.1655,
+      "step": 4134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.0839862468219495e-06,
+      "loss": 1.3686,
+      "step": 4135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.058694691273471e-06,
+      "loss": 1.2818,
+      "step": 4136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.033446713161362e-06,
+      "loss": 1.2804,
+      "step": 4137
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.0082423243236995e-06,
+      "loss": 1.2908,
+      "step": 4138
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.983081536578029e-06,
+      "loss": 1.4158,
+      "step": 4139
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.957964361721503e-06,
+      "loss": 1.0373,
+      "step": 4140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.932890811530879e-06,
+      "loss": 1.3419,
+      "step": 4141
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.907860897762331e-06,
+      "loss": 1.2767,
+      "step": 4142
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.882874632151704e-06,
+      "loss": 1.2151,
+      "step": 4143
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.857932026414304e-06,
+      "loss": 1.2449,
+      "step": 4144
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.833033092244978e-06,
+      "loss": 1.4335,
+      "step": 4145
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.808177841318097e-06,
+      "loss": 1.5725,
+      "step": 4146
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.783366285287573e-06,
+      "loss": 1.2311,
+      "step": 4147
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.758598435786789e-06,
+      "loss": 1.314,
+      "step": 4148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.733874304428689e-06,
+      "loss": 1.1489,
+      "step": 4149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.709193902805655e-06,
+      "loss": 1.3699,
+      "step": 4150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.68455724248962e-06,
+      "loss": 1.3413,
+      "step": 4151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.659964335031987e-06,
+      "loss": 1.2503,
+      "step": 4152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.635415191963634e-06,
+      "loss": 1.203,
+      "step": 4153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.610909824794931e-06,
+      "loss": 1.2766,
+      "step": 4154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.586448245015753e-06,
+      "loss": 1.3545,
+      "step": 4155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.56203046409537e-06,
+      "loss": 1.4185,
+      "step": 4156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.537656493482558e-06,
+      "loss": 1.2339,
+      "step": 4157
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.513326344605608e-06,
+      "loss": 1.1641,
+      "step": 4158
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.489040028872151e-06,
+      "loss": 1.2812,
+      "step": 4159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.464797557669344e-06,
+      "loss": 1.2317,
+      "step": 4160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.440598942363796e-06,
+      "loss": 1.1993,
+      "step": 4161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.416444194301463e-06,
+      "loss": 1.4849,
+      "step": 4162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.392333324807842e-06,
+      "loss": 1.2594,
+      "step": 4163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.36826634518779e-06,
+      "loss": 1.3444,
+      "step": 4164
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.3442432667256135e-06,
+      "loss": 1.2563,
+      "step": 4165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.320264100685014e-06,
+      "loss": 1.2642,
+      "step": 4166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.296328858309119e-06,
+      "loss": 1.3199,
+      "step": 4167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.272437550820464e-06,
+      "loss": 1.2845,
+      "step": 4168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.2485901894209555e-06,
+      "loss": 1.4264,
+      "step": 4169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.22478678529197e-06,
+      "loss": 1.3918,
+      "step": 4170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.201027349594147e-06,
+      "loss": 1.3063,
+      "step": 4171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.177311893467641e-06,
+      "loss": 1.1472,
+      "step": 4172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.153640428031937e-06,
+      "loss": 1.1753,
+      "step": 4173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.130012964385834e-06,
+      "loss": 1.4065,
+      "step": 4174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.106429513607592e-06,
+      "loss": 1.3117,
+      "step": 4175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.082890086754789e-06,
+      "loss": 1.4846,
+      "step": 4176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.059394694864362e-06,
+      "loss": 1.5226,
+      "step": 4177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.035943348952611e-06,
+      "loss": 1.4718,
+      "step": 4178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.0125360600151835e-06,
+      "loss": 1.1874,
+      "step": 4179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.989172839027057e-06,
+      "loss": 1.176,
+      "step": 4180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.9658536969425584e-06,
+      "loss": 1.2126,
+      "step": 4181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.942578644695351e-06,
+      "loss": 1.1625,
+      "step": 4182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.91934769319843e-06,
+      "loss": 1.2154,
+      "step": 4183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.896160853344079e-06,
+      "loss": 1.3551,
+      "step": 4184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.873018136003938e-06,
+      "loss": 1.5373,
+      "step": 4185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.8499195520289265e-06,
+      "loss": 1.2948,
+      "step": 4186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.826865112249347e-06,
+      "loss": 1.2172,
+      "step": 4187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.803854827474675e-06,
+      "loss": 1.1287,
+      "step": 4188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.78088870849377e-06,
+      "loss": 1.2911,
+      "step": 4189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.757966766074818e-06,
+      "loss": 1.3431,
+      "step": 4190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.735089010965167e-06,
+      "loss": 1.5076,
+      "step": 4191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.71225545389158e-06,
+      "loss": 1.3136,
+      "step": 4192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.6894661055600375e-06,
+      "loss": 1.1789,
+      "step": 4193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.6667209766557375e-06,
+      "loss": 1.3642,
+      "step": 4194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.644020077843249e-06,
+      "loss": 1.1664,
+      "step": 4195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.621363419766335e-06,
+      "loss": 1.1074,
+      "step": 4196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.598751013048043e-06,
+      "loss": 1.3399,
+      "step": 4197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.576182868290669e-06,
+      "loss": 1.2435,
+      "step": 4198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.553658996075728e-06,
+      "loss": 1.3817,
+      "step": 4199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.5311794069640155e-06,
+      "loss": 1.2611,
+      "step": 4200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.508744111495534e-06,
+      "loss": 1.5055,
+      "step": 4201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.486353120189569e-06,
+      "loss": 1.2295,
+      "step": 4202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.464006443544545e-06,
+      "loss": 1.0514,
+      "step": 4203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.441704092038191e-06,
+      "loss": 1.1517,
+      "step": 4204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.419446076127433e-06,
+      "loss": 1.2563,
+      "step": 4205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.397232406248342e-06,
+      "loss": 1.0924,
+      "step": 4206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.375063092816313e-06,
+      "loss": 1.4546,
+      "step": 4207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.352938146225861e-06,
+      "loss": 1.3916,
+      "step": 4208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.330857576850712e-06,
+      "loss": 1.373,
+      "step": 4209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.308821395043806e-06,
+      "loss": 1.2975,
+      "step": 4210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2868296111372565e-06,
+      "loss": 1.2624,
+      "step": 4211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.264882235442359e-06,
+      "loss": 1.4948,
+      "step": 4212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2429792782495825e-06,
+      "loss": 1.2224,
+      "step": 4213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.221120749828601e-06,
+      "loss": 1.1842,
+      "step": 4214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.199306660428205e-06,
+      "loss": 1.212,
+      "step": 4215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.177537020276402e-06,
+      "loss": 1.3711,
+      "step": 4216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.155811839580316e-06,
+      "loss": 1.3647,
+      "step": 4217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.134131128526243e-06,
+      "loss": 1.2279,
+      "step": 4218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.1124948972796515e-06,
+      "loss": 1.3909,
+      "step": 4219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.090903155985116e-06,
+      "loss": 1.134,
+      "step": 4220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.06935591476635e-06,
+      "loss": 1.2622,
+      "step": 4221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.04785318372627e-06,
+      "loss": 1.3484,
+      "step": 4222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.026394972946813e-06,
+      "loss": 1.2959,
+      "step": 4223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.00498129248913e-06,
+      "loss": 1.3631,
+      "step": 4224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.983612152393491e-06,
+      "loss": 1.2851,
+      "step": 4225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.962287562679191e-06,
+      "loss": 1.2116,
+      "step": 4226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.941007533344744e-06,
+      "loss": 1.1274,
+      "step": 4227
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.919772074367734e-06,
+      "loss": 1.278,
+      "step": 4228
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.898581195704821e-06,
+      "loss": 1.446,
+      "step": 4229
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.877434907291789e-06,
+      "loss": 1.2588,
+      "step": 4230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.856333219043508e-06,
+      "loss": 1.3733,
+      "step": 4231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.835276140853939e-06,
+      "loss": 1.3433,
+      "step": 4232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.814263682596132e-06,
+      "loss": 1.328,
+      "step": 4233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.793295854122193e-06,
+      "loss": 1.3636,
+      "step": 4234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.772372665263325e-06,
+      "loss": 1.2517,
+      "step": 4235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.751494125829814e-06,
+      "loss": 1.2506,
+      "step": 4236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.730660245610951e-06,
+      "loss": 1.2772,
+      "step": 4237
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.709871034375146e-06,
+      "loss": 1.2323,
+      "step": 4238
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.689126501869878e-06,
+      "loss": 1.2338,
+      "step": 4239
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6684266578215894e-06,
+      "loss": 1.2701,
+      "step": 4240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.647771511935873e-06,
+      "loss": 1.3169,
+      "step": 4241
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.627161073897313e-06,
+      "loss": 1.5377,
+      "step": 4242
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6065953533694915e-06,
+      "loss": 1.1965,
+      "step": 4243
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.586074359995119e-06,
+      "loss": 1.1569,
+      "step": 4244
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.565598103395863e-06,
+      "loss": 1.0439,
+      "step": 4245
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.545166593172434e-06,
+      "loss": 1.5593,
+      "step": 4246
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5247798389045696e-06,
+      "loss": 1.0474,
+      "step": 4247
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.504437850151022e-06,
+      "loss": 1.2302,
+      "step": 4248
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.484140636449541e-06,
+      "loss": 1.2317,
+      "step": 4249
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.46388820731688e-06,
+      "loss": 0.9073,
+      "step": 4250
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.443680572248843e-06,
+      "loss": 1.5629,
+      "step": 4251
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.423517740720162e-06,
+      "loss": 0.9906,
+      "step": 4252
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.403399722184598e-06,
+      "loss": 1.2791,
+      "step": 4253
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.383326526074916e-06,
+      "loss": 1.2396,
+      "step": 4254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.363298161802809e-06,
+      "loss": 1.3154,
+      "step": 4255
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3433146387590215e-06,
+      "loss": 1.1992,
+      "step": 4256
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.323375966313225e-06,
+      "loss": 1.2906,
+      "step": 4257
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.303482153814064e-06,
+      "loss": 1.6495,
+      "step": 4258
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.283633210589188e-06,
+      "loss": 1.2534,
+      "step": 4259
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.263829145945153e-06,
+      "loss": 1.4159,
+      "step": 4260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.244069969167508e-06,
+      "loss": 1.3761,
+      "step": 4261
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.224355689520765e-06,
+      "loss": 1.2271,
+      "step": 4262
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.20468631624834e-06,
+      "loss": 1.2746,
+      "step": 4263
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.185061858572648e-06,
+      "loss": 1.2516,
+      "step": 4264
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.165482325695003e-06,
+      "loss": 1.4578,
+      "step": 4265
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.145947726795662e-06,
+      "loss": 1.3044,
+      "step": 4266
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.126458071033834e-06,
+      "loss": 1.1608,
+      "step": 4267
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.107013367547674e-06,
+      "loss": 1.513,
+      "step": 4268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0876136254541785e-06,
+      "loss": 1.087,
+      "step": 4269
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.068258853849327e-06,
+      "loss": 1.1372,
+      "step": 4270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.048949061808027e-06,
+      "loss": 1.4,
+      "step": 4271
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.029684258384036e-06,
+      "loss": 1.2407,
+      "step": 4272
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.010464452610085e-06,
+      "loss": 1.337,
+      "step": 4273
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.991289653497765e-06,
+      "loss": 1.3598,
+      "step": 4274
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.97215987003754e-06,
+      "loss": 1.2482,
+      "step": 4275
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.953075111198845e-06,
+      "loss": 1.5075,
+      "step": 4276
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.934035385929946e-06,
+      "loss": 1.3812,
+      "step": 4277
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.915040703157991e-06,
+      "loss": 1.3154,
+      "step": 4278
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8960910717890546e-06,
+      "loss": 1.1626,
+      "step": 4279
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.877186500708041e-06,
+      "loss": 1.1825,
+      "step": 4280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.858326998778761e-06,
+      "loss": 1.2062,
+      "step": 4281
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.839512574843862e-06,
+      "loss": 1.2962,
+      "step": 4282
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.820743237724878e-06,
+      "loss": 1.2831,
+      "step": 4283
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8020189962221895e-06,
+      "loss": 1.3449,
+      "step": 4284
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.783339859115065e-06,
+      "loss": 1.4524,
+      "step": 4285
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.764705835161586e-06,
+      "loss": 1.5441,
+      "step": 4286
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.746116933098687e-06,
+      "loss": 1.0414,
+      "step": 4287
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.727573161642184e-06,
+      "loss": 1.2612,
+      "step": 4288
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.709074529486667e-06,
+      "loss": 1.1848,
+      "step": 4289
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.69062104530562e-06,
+      "loss": 1.0897,
+      "step": 4290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6722127177513555e-06,
+      "loss": 1.1965,
+      "step": 4291
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.653849555454947e-06,
+      "loss": 1.3235,
+      "step": 4292
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6355315670263847e-06,
+      "loss": 1.6648,
+      "step": 4293
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6172587610544205e-06,
+      "loss": 1.325,
+      "step": 4294
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.599031146106624e-06,
+      "loss": 1.397,
+      "step": 4295
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5808487307293915e-06,
+      "loss": 1.2235,
+      "step": 4296
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5627115234479256e-06,
+      "loss": 1.2653,
+      "step": 4297
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5446195327662133e-06,
+      "loss": 1.2167,
+      "step": 4298
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.526572767167069e-06,
+      "loss": 1.4174,
+      "step": 4299
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5085712351121015e-06,
+      "loss": 1.1991,
+      "step": 4300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.49061494504167e-06,
+      "loss": 1.2554,
+      "step": 4301
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.472703905374952e-06,
+      "loss": 1.2018,
+      "step": 4302
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4548381245099403e-06,
+      "loss": 1.2736,
+      "step": 4303
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4370176108233232e-06,
+      "loss": 1.3045,
+      "step": 4304
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.419242372670639e-06,
+      "loss": 1.3557,
+      "step": 4305
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.401512418386199e-06,
+      "loss": 1.2213,
+      "step": 4306
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.383827756282998e-06,
+      "loss": 1.4471,
+      "step": 4307
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3661883946528914e-06,
+      "loss": 1.2537,
+      "step": 4308
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3485943417664536e-06,
+      "loss": 1.3526,
+      "step": 4309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3310456058730065e-06,
+      "loss": 1.2929,
+      "step": 4310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3135421952006362e-06,
+      "loss": 1.049,
+      "step": 4311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2960841179561884e-06,
+      "loss": 1.2974,
+      "step": 4312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2786713823252155e-06,
+      "loss": 1.4983,
+      "step": 4313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2613039964720647e-06,
+      "loss": 1.052,
+      "step": 4314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2439819685397667e-06,
+      "loss": 1.4658,
+      "step": 4315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.226705306650113e-06,
+      "loss": 1.3368,
+      "step": 4316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2094740189036464e-06,
+      "loss": 1.2115,
+      "step": 4317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1922881133795825e-06,
+      "loss": 1.1024,
+      "step": 4318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1751475981358746e-06,
+      "loss": 1.0474,
+      "step": 4319
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1580524812092504e-06,
+      "loss": 1.4188,
+      "step": 4320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1410027706150537e-06,
+      "loss": 1.3029,
+      "step": 4321
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1239984743474227e-06,
+      "loss": 1.217,
+      "step": 4322
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1070396003791804e-06,
+      "loss": 1.4578,
+      "step": 4323
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.090126156661799e-06,
+      "loss": 1.3138,
+      "step": 4324
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0732581511255356e-06,
+      "loss": 1.2307,
+      "step": 4325
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0564355916792744e-06,
+      "loss": 1.088,
+      "step": 4326
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.039658486210639e-06,
+      "loss": 1.382,
+      "step": 4327
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0229268425859046e-06,
+      "loss": 1.2537,
+      "step": 4328
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.00624066865004e-06,
+      "loss": 1.1606,
+      "step": 4329
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.98959997222672e-06,
+      "loss": 1.2774,
+      "step": 4330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.973004761118259e-06,
+      "loss": 1.1234,
+      "step": 4331
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9564550431056657e-06,
+      "loss": 1.2988,
+      "step": 4332
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.939950825948623e-06,
+      "loss": 1.3507,
+      "step": 4333
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9234921173854514e-06,
+      "loss": 1.2539,
+      "step": 4334
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9070789251331775e-06,
+      "loss": 1.358,
+      "step": 4335
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.890711256887446e-06,
+      "loss": 1.2766,
+      "step": 4336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8743891203225848e-06,
+      "loss": 1.5089,
+      "step": 4337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8581125230915605e-06,
+      "loss": 1.1418,
+      "step": 4338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8418814728259803e-06,
+      "loss": 1.092,
+      "step": 4339
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8256959771361114e-06,
+      "loss": 1.5246,
+      "step": 4340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8095560436108616e-06,
+      "loss": 1.3378,
+      "step": 4341
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.793461679817766e-06,
+      "loss": 1.2595,
+      "step": 4342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7774128933029887e-06,
+      "loss": 1.2966,
+      "step": 4343
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7614096915913324e-06,
+      "loss": 1.4475,
+      "step": 4344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7454520821862395e-06,
+      "loss": 1.4799,
+      "step": 4345
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7295400725697474e-06,
+      "loss": 1.1172,
+      "step": 4346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7136736702025433e-06,
+      "loss": 1.3849,
+      "step": 4347
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.697852882523899e-06,
+      "loss": 1.0842,
+      "step": 4348
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6820777169517366e-06,
+      "loss": 1.0013,
+      "step": 4349
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6663481808825385e-06,
+      "loss": 1.2099,
+      "step": 4350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6506642816914396e-06,
+      "loss": 1.1939,
+      "step": 4351
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.635026026732157e-06,
+      "loss": 1.5344,
+      "step": 4352
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.619433423336992e-06,
+      "loss": 1.1439,
+      "step": 4353
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.603886478816875e-06,
+      "loss": 1.4016,
+      "step": 4354
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5883852004613074e-06,
+      "loss": 1.2602,
+      "step": 4355
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5729295955383537e-06,
+      "loss": 1.2183,
+      "step": 4356
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.557519671294728e-06,
+      "loss": 1.2318,
+      "step": 4357
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5421554349556733e-06,
+      "loss": 1.2698,
+      "step": 4358
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5268368937250153e-06,
+      "loss": 1.4675,
+      "step": 4359
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5115640547851983e-06,
+      "loss": 1.3611,
+      "step": 4360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4963369252971715e-06,
+      "loss": 1.0836,
+      "step": 4361
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4811555124005014e-06,
+      "loss": 1.6161,
+      "step": 4362
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.466019823213317e-06,
+      "loss": 1.4552,
+      "step": 4363
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4509298648322743e-06,
+      "loss": 1.2259,
+      "step": 4364
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.435885644332614e-06,
+      "loss": 1.0668,
+      "step": 4365
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4208871687681378e-06,
+      "loss": 1.3845,
+      "step": 4366
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4059344451711875e-06,
+      "loss": 1.3047,
+      "step": 4367
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3910274805526435e-06,
+      "loss": 1.2742,
+      "step": 4368
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3761662819019594e-06,
+      "loss": 1.2059,
+      "step": 4369
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3613508561870833e-06,
+      "loss": 1.3303,
+      "step": 4370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3465812103545594e-06,
+      "loss": 1.2312,
+      "step": 4371
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.331857351329436e-06,
+      "loss": 1.1929,
+      "step": 4372
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3171792860152695e-06,
+      "loss": 1.3456,
+      "step": 4373
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.302547021294199e-06,
+      "loss": 1.3752,
+      "step": 4374
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2879605640268586e-06,
+      "loss": 1.2568,
+      "step": 4375
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.273419921052411e-06,
+      "loss": 1.157,
+      "step": 4376
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2589250991885137e-06,
+      "loss": 1.4995,
+      "step": 4377
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2444761052313856e-06,
+      "loss": 1.4443,
+      "step": 4378
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.230072945955719e-06,
+      "loss": 1.2969,
+      "step": 4379
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2157156281147227e-06,
+      "loss": 1.1803,
+      "step": 4380
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2014041584401457e-06,
+      "loss": 1.3488,
+      "step": 4381
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1871385436421974e-06,
+      "loss": 1.3056,
+      "step": 4382
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.172918790409595e-06,
+      "loss": 1.3203,
+      "step": 4383
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1587449054095933e-06,
+      "loss": 1.1488,
+      "step": 4384
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.144616895287865e-06,
+      "loss": 1.2511,
+      "step": 4385
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.130534766668646e-06,
+      "loss": 1.3642,
+      "step": 4386
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.116498526154631e-06,
+      "loss": 1.3486,
+      "step": 4387
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1025081803269897e-06,
+      "loss": 1.3727,
+      "step": 4388
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0885637357453857e-06,
+      "loss": 1.1427,
+      "step": 4389
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.074665198947956e-06,
+      "loss": 1.3323,
+      "step": 4390
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.06081257645131e-06,
+      "loss": 1.094,
+      "step": 4391
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0470058747505516e-06,
+      "loss": 1.1674,
+      "step": 4392
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0332451003192255e-06,
+      "loss": 1.3181,
+      "step": 4393
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.019530259609348e-06,
+      "loss": 1.2722,
+      "step": 4394
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0058613590513974e-06,
+      "loss": 1.2889,
+      "step": 4395
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9922384050543364e-06,
+      "loss": 1.4102,
+      "step": 4396
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9786614040055552e-06,
+      "loss": 1.2599,
+      "step": 4397
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.965130362270917e-06,
+      "loss": 1.4581,
+      "step": 4398
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9516452861947255e-06,
+      "loss": 1.4419,
+      "step": 4399
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.938206182099711e-06,
+      "loss": 1.3713,
+      "step": 4400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9248130562871227e-06,
+      "loss": 1.4838,
+      "step": 4401
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9114659150365477e-06,
+      "loss": 1.4036,
+      "step": 4402
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8981647646061029e-06,
+      "loss": 1.2411,
+      "step": 4403
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8849096112323106e-06,
+      "loss": 1.2925,
+      "step": 4404
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.871700461130077e-06,
+      "loss": 1.0454,
+      "step": 4405
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8585373204928257e-06,
+      "loss": 1.2573,
+      "step": 4406
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8454201954923534e-06,
+      "loss": 1.2394,
+      "step": 4407
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8323490922788843e-06,
+      "loss": 1.0796,
+      "step": 4408
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8193240169810943e-06,
+      "loss": 1.2932,
+      "step": 4409
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8063449757060314e-06,
+      "loss": 1.32,
+      "step": 4410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7934119745391942e-06,
+      "loss": 1.3657,
+      "step": 4411
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7805250195444878e-06,
+      "loss": 1.2458,
+      "step": 4412
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7676841167642232e-06,
+      "loss": 1.3014,
+      "step": 4413
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.754889272219129e-06,
+      "loss": 1.3249,
+      "step": 4414
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7421404919083173e-06,
+      "loss": 1.2774,
+      "step": 4415
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7294377818093288e-06,
+      "loss": 1.4059,
+      "step": 4416
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.716781147878066e-06,
+      "loss": 1.3853,
+      "step": 4417
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.704170596048893e-06,
+      "loss": 1.3069,
+      "step": 4418
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6916061322344801e-06,
+      "loss": 1.2154,
+      "step": 4419
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6790877623259593e-06,
+      "loss": 1.2845,
+      "step": 4420
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6666154921928244e-06,
+      "loss": 1.2194,
+      "step": 4421
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6541893276829422e-06,
+      "loss": 1.2273,
+      "step": 4422
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6418092746225856e-06,
+      "loss": 1.2651,
+      "step": 4423
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6294753388163885e-06,
+      "loss": 1.1458,
+      "step": 4424
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6171875260473811e-06,
+      "loss": 1.1749,
+      "step": 4425
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6049458420769436e-06,
+      "loss": 1.3305,
+      "step": 4426
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5927502926448512e-06,
+      "loss": 1.1431,
+      "step": 4427
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5806008834692299e-06,
+      "loss": 1.2357,
+      "step": 4428
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5684976202465784e-06,
+      "loss": 1.3016,
+      "step": 4429
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5564405086517798e-06,
+      "loss": 1.4226,
+      "step": 4430
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5444295543380227e-06,
+      "loss": 1.4267,
+      "step": 4431
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5324647629369137e-06,
+      "loss": 1.1371,
+      "step": 4432
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5205461400583876e-06,
+      "loss": 1.3268,
+      "step": 4433
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5086736912907295e-06,
+      "loss": 1.3566,
+      "step": 4434
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4968474222005758e-06,
+      "loss": 1.2578,
+      "step": 4435
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.485067338332946e-06,
+      "loss": 1.3279,
+      "step": 4436
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4733334452111226e-06,
+      "loss": 1.2296,
+      "step": 4437
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4616457483368263e-06,
+      "loss": 1.4869,
+      "step": 4438
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4500042531900515e-06,
+      "loss": 1.509,
+      "step": 4439
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4384089652291543e-06,
+      "loss": 1.328,
+      "step": 4440
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4268598898908304e-06,
+      "loss": 1.2784,
+      "step": 4441
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4153570325900923e-06,
+      "loss": 1.0727,
+      "step": 4442
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4039003987203037e-06,
+      "loss": 1.4714,
+      "step": 4443
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3924899936531232e-06,
+      "loss": 1.4147,
+      "step": 4444
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3811258227385715e-06,
+      "loss": 1.2633,
+      "step": 4445
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.369807891304964e-06,
+      "loss": 1.3898,
+      "step": 4446
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3585362046589334e-06,
+      "loss": 1.1521,
+      "step": 4447
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3473107680854636e-06,
+      "loss": 1.3846,
+      "step": 4448
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3361315868478108e-06,
+      "loss": 1.2375,
+      "step": 4449
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3249986661875935e-06,
+      "loss": 1.2046,
+      "step": 4450
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3139120113246583e-06,
+      "loss": 1.3386,
+      "step": 4451
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3028716274572584e-06,
+      "loss": 1.3492,
+      "step": 4452
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2918775197618972e-06,
+      "loss": 1.1563,
+      "step": 4453
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2809296933933622e-06,
+      "loss": 1.488,
+      "step": 4454
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2700281534847814e-06,
+      "loss": 1.6463,
+      "step": 4455
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2591729051475765e-06,
+      "loss": 1.1711,
+      "step": 4456
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.248363953471432e-06,
+      "loss": 1.3141,
+      "step": 4457
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2376013035243606e-06,
+      "loss": 1.3494,
+      "step": 4458
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2268849603526477e-06,
+      "loss": 1.1288,
+      "step": 4459
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2162149289808634e-06,
+      "loss": 1.4005,
+      "step": 4460
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.205591214411883e-06,
+      "loss": 1.2866,
+      "step": 4461
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1950138216268447e-06,
+      "loss": 1.1919,
+      "step": 4462
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1844827555851813e-06,
+      "loss": 1.4811,
+      "step": 4463
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1739980212245982e-06,
+      "loss": 1.4864,
+      "step": 4464
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1635596234610747e-06,
+      "loss": 1.2174,
+      "step": 4465
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1531675671888619e-06,
+      "loss": 1.1799,
+      "step": 4466
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1428218572805071e-06,
+      "loss": 1.4985,
+      "step": 4467
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1325224985867855e-06,
+      "loss": 1.4382,
+      "step": 4468
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1222694959367674e-06,
+      "loss": 1.1,
+      "step": 4469
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1120628541378074e-06,
+      "loss": 1.338,
+      "step": 4470
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1019025779754666e-06,
+      "loss": 1.3892,
+      "step": 4471
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0917886722136228e-06,
+      "loss": 1.4784,
+      "step": 4472
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0817211415943828e-06,
+      "loss": 1.1591,
+      "step": 4473
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0716999908381042e-06,
+      "loss": 1.4801,
+      "step": 4474
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0617252246434062e-06,
+      "loss": 1.3999,
+      "step": 4475
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0517968476871808e-06,
+      "loss": 1.309,
+      "step": 4476
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0419148646245492e-06,
+      "loss": 1.157,
+      "step": 4477
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0320792800888602e-06,
+      "loss": 1.2237,
+      "step": 4478
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0222900986917473e-06,
+      "loss": 1.4386,
+      "step": 4479
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0125473250230722e-06,
+      "loss": 1.356,
+      "step": 4480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0028509636509142e-06,
+      "loss": 1.2174,
+      "step": 4481
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.932010191216368e-07,
+      "loss": 0.9998,
+      "step": 4482
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.835974959597871e-07,
+      "loss": 1.4075,
+      "step": 4483
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.740403986681967e-07,
+      "loss": 1.3816,
+      "step": 4484
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.645297317279035e-07,
+      "loss": 1.1276,
+      "step": 4485
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.550654995981622e-07,
+      "loss": 1.318,
+      "step": 4486
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.456477067164904e-07,
+      "loss": 1.3087,
+      "step": 4487
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.362763574985999e-07,
+      "loss": 1.1235,
+      "step": 4488
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.26951456338443e-07,
+      "loss": 1.303,
+      "step": 4489
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.176730076081997e-07,
+      "loss": 1.2182,
+      "step": 4490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.084410156582568e-07,
+      "loss": 1.281,
+      "step": 4491
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.992554848172185e-07,
+      "loss": 1.2216,
+      "step": 4492
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.901164193919065e-07,
+      "loss": 1.2776,
+      "step": 4493
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.810238236673707e-07,
+      "loss": 1.1624,
+      "step": 4494
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.719777019068564e-07,
+      "loss": 1.1161,
+      "step": 4495
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.629780583518154e-07,
+      "loss": 1.5291,
+      "step": 4496
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.540248972219278e-07,
+      "loss": 1.1373,
+      "step": 4497
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.451182227150467e-07,
+      "loss": 1.2032,
+      "step": 4498
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.362580390072761e-07,
+      "loss": 1.0459,
+      "step": 4499
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.274443502528817e-07,
+      "loss": 1.33,
+      "step": 4500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.186771605843247e-07,
+      "loss": 1.2401,
+      "step": 4501
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.099564741123166e-07,
+      "loss": 1.3303,
+      "step": 4502
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.012822949256982e-07,
+      "loss": 1.0933,
+      "step": 4503
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.926546270915602e-07,
+      "loss": 1.408,
+      "step": 4504
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.840734746551559e-07,
+      "loss": 1.4107,
+      "step": 4505
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.755388416399112e-07,
+      "loss": 1.2326,
+      "step": 4506
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.670507320474918e-07,
+      "loss": 1.2403,
+      "step": 4507
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.586091498577141e-07,
+      "loss": 1.3617,
+      "step": 4508
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.502140990285899e-07,
+      "loss": 1.4379,
+      "step": 4509
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.418655834962928e-07,
+      "loss": 1.5402,
+      "step": 4510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.33563607175225e-07,
+      "loss": 1.3225,
+      "step": 4511
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.25308173957917e-07,
+      "loss": 1.443,
+      "step": 4512
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.17099287715095e-07,
+      "loss": 1.3992,
+      "step": 4513
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.089369522956801e-07,
+      "loss": 1.4185,
+      "step": 4514
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.008211715267443e-07,
+      "loss": 1.4999,
+      "step": 4515
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.927519492135326e-07,
+      "loss": 1.1108,
+      "step": 4516
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.84729289139463e-07,
+      "loss": 1.1041,
+      "step": 4517
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.767531950661377e-07,
+      "loss": 1.0964,
+      "step": 4518
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.688236707332984e-07,
+      "loss": 1.1857,
+      "step": 4519
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.609407198588713e-07,
+      "loss": 1.2521,
+      "step": 4520
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.531043461389441e-07,
+      "loss": 1.1066,
+      "step": 4521
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.453145532477556e-07,
+      "loss": 1.3674,
+      "step": 4522
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.375713448377063e-07,
+      "loss": 1.4668,
+      "step": 4523
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.298747245393699e-07,
+      "loss": 1.1265,
+      "step": 4524
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.222246959614486e-07,
+      "loss": 1.3459,
+      "step": 4525
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.146212626908288e-07,
+      "loss": 1.4079,
+      "step": 4526
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.070644282925253e-07,
+      "loss": 1.4249,
+      "step": 4527
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995541963097262e-07,
+      "loss": 1.1716,
+      "step": 4528
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.920905702637481e-07,
+      "loss": 1.2097,
+      "step": 4529
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.846735536540693e-07,
+      "loss": 1.4215,
+      "step": 4530
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.773031499583081e-07,
+      "loss": 1.0043,
+      "step": 4531
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.699793626322336e-07,
+      "loss": 1.3089,
+      "step": 4532
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.627021951097545e-07,
+      "loss": 1.3081,
+      "step": 4533
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.554716508029189e-07,
+      "loss": 1.3767,
+      "step": 4534
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.482877331018932e-07,
+      "loss": 1.3087,
+      "step": 4535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.411504453750383e-07,
+      "loss": 1.3111,
+      "step": 4536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.340597909688105e-07,
+      "loss": 1.2318,
+      "step": 4537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.270157732077951e-07,
+      "loss": 1.3036,
+      "step": 4538
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.20018395394728e-07,
+      "loss": 0.9073,
+      "step": 4539
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.130676608104845e-07,
+      "loss": 1.3855,
+      "step": 4540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.061635727140468e-07,
+      "loss": 1.3705,
+      "step": 4541
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.993061343425476e-07,
+      "loss": 1.3318,
+      "step": 4542
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.924953489112261e-07,
+      "loss": 1.362,
+      "step": 4543
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.857312196134722e-07,
+      "loss": 1.1437,
+      "step": 4544
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.790137496207714e-07,
+      "loss": 1.3356,
+      "step": 4545
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7234294208274854e-07,
+      "loss": 1.2778,
+      "step": 4546
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.657188001271684e-07,
+      "loss": 1.345,
+      "step": 4547
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5914132685987986e-07,
+      "loss": 1.2097,
+      "step": 4548
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.526105253648605e-07,
+      "loss": 1.1253,
+      "step": 4549
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.461263987042275e-07,
+      "loss": 1.3377,
+      "step": 4550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.396889499181822e-07,
+      "loss": 1.2603,
+      "step": 4551
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.332981820250659e-07,
+      "loss": 1.2444,
+      "step": 4552
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.269540980213038e-07,
+      "loss": 1.0845,
+      "step": 4553
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2065670088147215e-07,
+      "loss": 1.2303,
+      "step": 4554
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1440599355822006e-07,
+      "loss": 1.4125,
+      "step": 4555
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.082019789823255e-07,
+      "loss": 1.2668,
+      "step": 4556
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.020446600626615e-07,
+      "loss": 1.2459,
+      "step": 4557
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.959340396862188e-07,
+      "loss": 1.3744,
+      "step": 4558
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.8987012071808324e-07,
+      "loss": 1.2674,
+      "step": 4559
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.838529060014584e-07,
+      "loss": 1.1468,
+      "step": 4560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7788239835762076e-07,
+      "loss": 1.2024,
+      "step": 4561
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.719586005859865e-07,
+      "loss": 1.5132,
+      "step": 4562
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.66081515464034e-07,
+      "loss": 1.2316,
+      "step": 4563
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6025114574734785e-07,
+      "loss": 1.1534,
+      "step": 4564
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.544674941696413e-07,
+      "loss": 1.0892,
+      "step": 4565
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4873056344267853e-07,
+      "loss": 1.4179,
+      "step": 4566
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4304035625634113e-07,
+      "loss": 1.1715,
+      "step": 4567
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3739687527860606e-07,
+      "loss": 1.0122,
+      "step": 4568
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3180012315552347e-07,
+      "loss": 1.5935,
+      "step": 4569
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.262501025112608e-07,
+      "loss": 1.4111,
+      "step": 4570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2074681594804755e-07,
+      "loss": 1.3512,
+      "step": 4571
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.152902660462087e-07,
+      "loss": 1.1158,
+      "step": 4572
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0988045536417544e-07,
+      "loss": 0.9677,
+      "step": 4573
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.045173864384299e-07,
+      "loss": 1.237,
+      "step": 4574
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9920106178357167e-07,
+      "loss": 1.33,
+      "step": 4575
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9393148389226246e-07,
+      "loss": 1.3406,
+      "step": 4576
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8870865523525915e-07,
+      "loss": 1.3105,
+      "step": 4577
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8353257826136957e-07,
+      "loss": 1.2966,
+      "step": 4578
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.784032553975302e-07,
+      "loss": 1.4354,
+      "step": 4579
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7332068904871723e-07,
+      "loss": 1.3735,
+      "step": 4580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.682848815980021e-07,
+      "loss": 1.4002,
+      "step": 4581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.632958354065074e-07,
+      "loss": 1.3446,
+      "step": 4582
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5835355281347287e-07,
+      "loss": 1.4493,
+      "step": 4583
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5345803613617825e-07,
+      "loss": 1.1835,
+      "step": 4584
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4860928766998745e-07,
+      "loss": 1.2216,
+      "step": 4585
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4380730968833756e-07,
+      "loss": 1.0652,
+      "step": 4586
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3905210444273853e-07,
+      "loss": 1.0931,
+      "step": 4587
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3434367416275138e-07,
+      "loss": 1.403,
+      "step": 4588
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2968202105603242e-07,
+      "loss": 1.3699,
+      "step": 4589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2506714730828883e-07,
+      "loss": 1.3104,
+      "step": 4590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2049905508331192e-07,
+      "loss": 1.2193,
+      "step": 4591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.159777465229218e-07,
+      "loss": 1.2779,
+      "step": 4592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.115032237470449e-07,
+      "loss": 1.1091,
+      "step": 4593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0707548885363638e-07,
+      "loss": 1.4437,
+      "step": 4594
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0269454391874666e-07,
+      "loss": 1.3129,
+      "step": 4595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9836039099646597e-07,
+      "loss": 1.2777,
+      "step": 4596
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9407303211895765e-07,
+      "loss": 1.468,
+      "step": 4597
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8983246929642485e-07,
+      "loss": 1.2579,
+      "step": 4598
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.856387045171437e-07,
+      "loss": 1.1262,
+      "step": 4599
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.814917397474636e-07,
+      "loss": 1.342,
+      "step": 4600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7739157693175136e-07,
+      "loss": 1.2913,
+      "step": 4601
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.733382179924692e-07,
+      "loss": 1.0402,
+      "step": 4602
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6933166483010798e-07,
+      "loss": 1.2409,
+      "step": 4603
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.653719193232206e-07,
+      "loss": 1.3736,
+      "step": 4604
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6145898332842192e-07,
+      "loss": 1.33,
+      "step": 4605
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5759285868037764e-07,
+      "loss": 1.2197,
+      "step": 4606
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5377354719178227e-07,
+      "loss": 1.3538,
+      "step": 4607
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5000105065341441e-07,
+      "loss": 1.2481,
+      "step": 4608
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4627537083408139e-07,
+      "loss": 1.3252,
+      "step": 4609
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4259650948064142e-07,
+      "loss": 1.5145,
+      "step": 4610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.389644683180036e-07,
+      "loss": 1.4008,
+      "step": 4611
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.353792490491279e-07,
+      "loss": 1.2821,
+      "step": 4612
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3184085335501416e-07,
+      "loss": 1.1922,
+      "step": 4613
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2834928289472416e-07,
+      "loss": 1.2458,
+      "step": 4614
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2490453930533718e-07,
+      "loss": 1.2758,
+      "step": 4615
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2150662420199466e-07,
+      "loss": 1.2208,
+      "step": 4616
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.181555391778777e-07,
+      "loss": 1.2821,
+      "step": 4617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1485128580422944e-07,
+      "loss": 1.3178,
+      "step": 4618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.115938656302884e-07,
+      "loss": 1.5424,
+      "step": 4619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.083832801833773e-07,
+      "loss": 1.3165,
+      "step": 4620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0521953096885861e-07,
+      "loss": 1.353,
+      "step": 4621
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0210261947009025e-07,
+      "loss": 1.5373,
+      "step": 4622
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.90325471485254e-08,
+      "loss": 1.2166,
+      "step": 4623
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.600931544362369e-08,
+      "loss": 1.412,
+      "step": 4624
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.303292577288459e-08,
+      "loss": 1.2731,
+      "step": 4625
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.010337953185843e-08,
+      "loss": 1.2071,
+      "step": 4626
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.722067809410206e-08,
+      "loss": 1.3117,
+      "step": 4627
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.438482281125649e-08,
+      "loss": 1.3378,
+      "step": 4628
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.15958150129581e-08,
+      "loss": 1.3593,
+      "step": 4629
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.88536560068831e-08,
+      "loss": 1.5035,
+      "step": 4630
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.615834707875858e-08,
+      "loss": 1.3309,
+      "step": 4631
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.350988949232917e-08,
+      "loss": 1.4354,
+      "step": 4632
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.090828448939046e-08,
+      "loss": 1.059,
+      "step": 4633
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.835353328974447e-08,
+      "loss": 1.1914,
+      "step": 4634
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.584563709124414e-08,
+      "loss": 1.2785,
+      "step": 4635
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.338459706977107e-08,
+      "loss": 1.3441,
+      "step": 4636
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.097041437923557e-08,
+      "loss": 1.4605,
+      "step": 4637
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.860309015156551e-08,
+      "loss": 1.4163,
+      "step": 4638
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.6282625496739684e-08,
+      "loss": 1.3523,
+      "step": 4639
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.400902150274334e-08,
+      "loss": 1.4996,
+      "step": 4640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.1782279235623734e-08,
+      "loss": 1.352,
+      "step": 4641
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.960239973941238e-08,
+      "loss": 1.1489,
+      "step": 4642
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.7469384036213924e-08,
+      "loss": 1.5116,
+      "step": 4643
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.538323312610615e-08,
+      "loss": 1.3582,
+      "step": 4644
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.334394798725105e-08,
+      "loss": 1.34,
+      "step": 4645
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.135152957579491e-08,
+      "loss": 1.3935,
+      "step": 4646
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.940597882593489e-08,
+      "loss": 1.2639,
+      "step": 4647
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7507296649863524e-08,
+      "loss": 1.3993,
+      "step": 4648
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.565548393783535e-08,
+      "loss": 1.1489,
+      "step": 4649
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.38505415581003e-08,
+      "loss": 1.1735,
+      "step": 4650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2092470356948066e-08,
+      "loss": 1.1538,
+      "step": 4651
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.038127115868594e-08,
+      "loss": 1.2982,
+      "step": 4652
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.87169447656388e-08,
+      "loss": 1.4465,
+      "step": 4653
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7099491958171298e-08,
+      "loss": 1.197,
+      "step": 4654
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.552891349464348e-08,
+      "loss": 1.3542,
+      "step": 4655
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.400521011146628e-08,
+      "loss": 1.1965,
+      "step": 4656
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2528382523057113e-08,
+      "loss": 1.1821,
+      "step": 4657
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1098431421862076e-08,
+      "loss": 1.4328,
+      "step": 4658
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9715357478322648e-08,
+      "loss": 1.3733,
+      "step": 4659
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.837916134094231e-08,
+      "loss": 1.2707,
+      "step": 4660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.70898436362088e-08,
+      "loss": 1.3045,
+      "step": 4661
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5847404968660773e-08,
+      "loss": 1.1169,
+      "step": 4662
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4651845920832241e-08,
+      "loss": 1.491,
+      "step": 4663
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3503167053285915e-08,
+      "loss": 1.3458,
+      "step": 4664
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2401368904613186e-08,
+      "loss": 1.2251,
+      "step": 4665
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1346451991389728e-08,
+      "loss": 1.1974,
+      "step": 4666
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0338416808264305e-08,
+      "loss": 1.2853,
+      "step": 4667
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.37726382784776e-09,
+      "loss": 1.3771,
+      "step": 4668
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.462993500824023e-09,
+      "loss": 1.6266,
+      "step": 4669
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.595606255839104e-09,
+      "loss": 1.2458,
+      "step": 4670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.775102499612107e-09,
+      "loss": 1.4458,
+      "step": 4671
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.001482616835308e-09,
+      "loss": 1.2566,
+      "step": 4672
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.274746970240773e-09,
+      "loss": 1.2934,
+      "step": 4673
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.594895900578155e-09,
+      "loss": 1.227,
+      "step": 4674
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.961929726603586e-09,
+      "loss": 1.2194,
+      "step": 4675
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3758487450907816e-09,
+      "loss": 1.3814,
+      "step": 4676
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.836653230853248e-09,
+      "loss": 1.2371,
+      "step": 4677
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3443434366887673e-09,
+      "loss": 1.1802,
+      "step": 4678
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8989195934349113e-09,
+      "loss": 1.4825,
+      "step": 4679
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.500381909935733e-09,
+      "loss": 1.367,
+      "step": 4680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1487305730528698e-09,
+      "loss": 1.456,
+      "step": 4681
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.439657476655427e-10,
+      "loss": 1.3109,
+      "step": 4682
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.860875766705576e-10,
+      "loss": 1.2148,
+      "step": 4683
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7509618097120167e-10,
+      "loss": 1.1803,
+      "step": 4684
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1099165949944878e-10,
+      "loss": 1.3814,
+      "step": 4685
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.377408921595887e-11,
+      "loss": 1.1762,
+      "step": 4686
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.344352505456726e-11,
+      "loss": 1.3766,
+      "step": 4687
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 1.4355,
+      "step": 4688
+    },
+    {
+      "epoch": 2.0,
+      "step": 4688,
+      "total_flos": 1.3162175477630632e+19,
+      "train_loss": 1.326163134172101,
+      "train_runtime": 114803.0567,
+      "train_samples_per_second": 0.082,
+      "train_steps_per_second": 0.041
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 4688,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 0,
+  "total_flos": 1.3162175477630632e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}