diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24973 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998797643381027,
+  "global_step": 4158,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.3308,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.322,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.321,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.338,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.2804,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.3527,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.3421,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.248,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.44e-05,
+      "loss": 1.2245,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2601,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.76e-05,
+      "loss": 1.22,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.1947,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.08e-05,
+      "loss": 1.1768,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.2282,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 1.1682,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 0.3205,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 1.1738,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.88e-05,
+      "loss": 1.1004,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.04e-05,
+      "loss": 1.091,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.1552,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 1.1215,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.52e-05,
+      "loss": 1.0603,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.68e-05,
+      "loss": 1.0958,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 1.0418,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.096,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.16e-05,
+      "loss": 1.1166,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.32e-05,
+      "loss": 1.0329,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 1.1345,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.64e-05,
+      "loss": 1.0405,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 1.0841,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96e-05,
+      "loss": 1.0145,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 0.3213,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.28e-05,
+      "loss": 1.0192,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 1.0179,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.0717,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.76e-05,
+      "loss": 1.0473,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.92e-05,
+      "loss": 1.0455,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.08e-05,
+      "loss": 1.0186,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.24e-05,
+      "loss": 1.0277,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.0375,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 0.9798,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 1.0512,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 1.0367,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.04e-05,
+      "loss": 1.0137,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-05,
+      "loss": 0.9828,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.36e-05,
+      "loss": 0.336,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.52e-05,
+      "loss": 1.011,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 1.0411,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 0.9702,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 1.0165,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.16e-05,
+      "loss": 1.0035,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.32e-05,
+      "loss": 1.0234,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.48e-05,
+      "loss": 1.0075,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64e-05,
+      "loss": 0.9499,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.9784,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 0.9924,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 1.0147,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.28e-05,
+      "loss": 0.9892,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.44e-05,
+      "loss": 1.0079,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-05,
+      "loss": 1.0176,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.76e-05,
+      "loss": 1.0118,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.92e-05,
+      "loss": 0.9842,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010080000000000001,
+      "loss": 1.0101,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010240000000000001,
+      "loss": 0.9655,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 0.9454,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001056,
+      "loss": 0.9948,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010720000000000002,
+      "loss": 0.3106,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010880000000000002,
+      "loss": 0.9701,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011040000000000001,
+      "loss": 0.9841,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 0.9346,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001136,
+      "loss": 0.9395,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001152,
+      "loss": 1.0297,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011679999999999999,
+      "loss": 0.9896,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001184,
+      "loss": 0.9483,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012,
+      "loss": 1.0024,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001216,
+      "loss": 1.0062,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001232,
+      "loss": 0.3178,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001248,
+      "loss": 0.9419,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001264,
+      "loss": 0.9837,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 0.968,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001296,
+      "loss": 1.0175,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013120000000000002,
+      "loss": 0.9426,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001328,
+      "loss": 0.9599,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013440000000000001,
+      "loss": 1.0042,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 0.9375,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013759999999999998,
+      "loss": 0.9427,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001392,
+      "loss": 1.0056,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001408,
+      "loss": 0.9596,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001424,
+      "loss": 0.2905,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000144,
+      "loss": 0.9406,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014560000000000002,
+      "loss": 0.9883,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001472,
+      "loss": 0.9791,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001488,
+      "loss": 0.9266,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001504,
+      "loss": 0.9234,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000152,
+      "loss": 0.992,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015360000000000002,
+      "loss": 0.9534,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001552,
+      "loss": 0.986,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015680000000000002,
+      "loss": 0.971,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015840000000000003,
+      "loss": 0.9535,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016,
+      "loss": 0.3167,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016160000000000002,
+      "loss": 0.9471,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001632,
+      "loss": 0.9801,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001648,
+      "loss": 0.9615,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001664,
+      "loss": 0.9311,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000168,
+      "loss": 0.9226,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001696,
+      "loss": 0.9344,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017120000000000001,
+      "loss": 0.9759,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001728,
+      "loss": 0.9618,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001744,
+      "loss": 0.9543,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 0.9387,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001776,
+      "loss": 0.9663,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017920000000000002,
+      "loss": 0.9439,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001808,
+      "loss": 0.9472,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018240000000000002,
+      "loss": 0.979,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 0.953,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001856,
+      "loss": 0.9802,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018720000000000002,
+      "loss": 1.031,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001888,
+      "loss": 0.9786,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001904,
+      "loss": 0.2591,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000192,
+      "loss": 0.9556,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019360000000000002,
+      "loss": 0.9557,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001952,
+      "loss": 0.9443,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001968,
+      "loss": 0.9525,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001984,
+      "loss": 0.95,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9445,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999996966016016,
+      "loss": 0.9822,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999998786406591,
+      "loss": 0.9382,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999972694155198,
+      "loss": 0.9607,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999951456293091,
+      "loss": 0.9585,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999924150492472,
+      "loss": 0.9666,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999890776769915,
+      "loss": 0.9696,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999851335145667,
+      "loss": 0.9459,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999980582564366,
+      "loss": 0.9777,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999975424829151,
+      "loss": 0.9331,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999969660312052,
+      "loss": 0.9807,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999632890165662,
+      "loss": 0.9497,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999563109465598,
+      "loss": 0.9738,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999487261062674,
+      "loss": 0.9236,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999940534500291,
+      "loss": 0.2928,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999317361336013,
+      "loss": 0.3044,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999223310115374,
+      "loss": 0.9406,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999123191398062,
+      "loss": 0.8991,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999901700524483,
+      "loss": 1.0025,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998904751720106,
+      "loss": 0.2916,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999878643089201,
+      "loss": 0.9097,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999866204283234,
+      "loss": 0.9489,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999853158761657,
+      "loss": 0.9587,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998395065323862,
+      "loss": 0.9346,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998252476037057,
+      "loss": 0.972,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999810381984268,
+      "loss": 0.991,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999794909683093,
+      "loss": 0.8947,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997788307095696,
+      "loss": 0.9414,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997621450734544,
+      "loss": 0.9402,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997448527848721,
+      "loss": 0.9544,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997269538543159,
+      "loss": 0.3109,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997084482926462,
+      "loss": 0.9731,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999689336111093,
+      "loss": 0.9797,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996696173212527,
+      "loss": 0.9192,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996492919350912,
+      "loss": 0.9563,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996283599649412,
+      "loss": 0.3077,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996068214235048,
+      "loss": 0.9581,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995846763238512,
+      "loss": 0.949,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999561924679418,
+      "loss": 0.9686,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995385665040112,
+      "loss": 0.9802,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999514601811804,
+      "loss": 0.9513,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999490030617338,
+      "loss": 0.9307,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994648529355233,
+      "loss": 0.9376,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994390687816374,
+      "loss": 0.2999,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994126781713265,
+      "loss": 0.9299,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999385681120604,
+      "loss": 0.9503,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999358077645851,
+      "loss": 0.9011,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993298677638182,
+      "loss": 0.9782,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993010514916232,
+      "loss": 0.9419,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992716288467507,
+      "loss": 0.9016,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999241599847055,
+      "loss": 0.9067,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992109645107576,
+      "loss": 0.9538,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991797228564477,
+      "loss": 0.9528,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991478749030827,
+      "loss": 0.8797,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999115420669988,
+      "loss": 0.9197,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990823601768564,
+      "loss": 0.3249,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999048693443749,
+      "loss": 0.9575,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990144204910953,
+      "loss": 0.9249,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989795413396909,
+      "loss": 0.9642,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989440560107009,
+      "loss": 0.2946,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998907964525658,
+      "loss": 0.9324,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998871266906462,
+      "loss": 0.9571,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998833963175381,
+      "loss": 0.9297,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987960533550506,
+      "loss": 0.9784,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987575374684748,
+      "loss": 0.9289,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987184155390245,
+      "loss": 0.9575,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986786875904393,
+      "loss": 0.9012,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998638353646825,
+      "loss": 0.8941,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985974137326574,
+      "loss": 0.949,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985558678727773,
+      "loss": 0.9698,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998513716092396,
+      "loss": 0.9854,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984709584170899,
+      "loss": 0.935,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984275948728048,
+      "loss": 0.9596,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983836254858537,
+      "loss": 0.9153,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983390502829166,
+      "loss": 0.9754,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998293869291042,
+      "loss": 0.9312,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982480825376454,
+      "loss": 0.9697,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199820169005051,
+      "loss": 0.9527,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981546918577866,
+      "loss": 0.9582,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981070879879938,
+      "loss": 0.9486,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980588784700176,
+      "loss": 0.9416,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980100633331107,
+      "loss": 0.9493,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979606426068943,
+      "loss": 0.9314,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979106163213573,
+      "loss": 1.0036,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978599845068545,
+      "loss": 0.9033,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199780874719411,
+      "loss": 0.9036,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977569044142137,
+      "loss": 0.967,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977044561986243,
+      "loss": 0.9437,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976514025791666,
+      "loss": 0.988,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975977435880338,
+      "loss": 0.9411,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975434792577859,
+      "loss": 0.9499,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974886096213503,
+      "loss": 0.9367,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974331347120218,
+      "loss": 0.2913,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973770545634621,
+      "loss": 0.984,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997320369209701,
+      "loss": 0.9816,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972630786851343,
+      "loss": 0.8787,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972051830245263,
+      "loss": 0.9292,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971466822630074,
+      "loss": 0.904,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970875764360762,
+      "loss": 0.9644,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970278655795975,
+      "loss": 0.9545,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969675497298037,
+      "loss": 0.9652,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969066289232947,
+      "loss": 0.8685,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968451031970364,
+      "loss": 0.9635,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019967829725883632,
+      "loss": 0.9741,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967202371349752,
+      "loss": 0.9051,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966568968749398,
+      "loss": 0.9075,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965929518466925,
+      "loss": 0.3215,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996528402089034,
+      "loss": 0.9347,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964632476411335,
+      "loss": 0.9635,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963974885425266,
+      "loss": 0.9291,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963311248331154,
+      "loss": 0.9353,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962641565531692,
+      "loss": 0.9187,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961965837433243,
+      "loss": 0.9266,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961284064445837,
+      "loss": 0.953,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960596246983164,
+      "loss": 0.9391,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959902385462602,
+      "loss": 0.2938,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995920248030518,
+      "loss": 0.9519,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995849653193559,
+      "loss": 0.937,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995778454078221,
+      "loss": 1.0297,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957066507277063,
+      "loss": 0.2905,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995634243185586,
+      "loss": 0.941,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955612314957963,
+      "loss": 0.8878,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954876157026404,
+      "loss": 0.956,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954133958507883,
+      "loss": 0.9175,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953385719852763,
+      "loss": 0.9055,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952631441515073,
+      "loss": 0.9277,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951871123952505,
+      "loss": 0.9094,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951104767626418,
+      "loss": 0.913,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995033237300184,
+      "loss": 0.9074,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949553940547444,
+      "loss": 0.8961,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948769470735594,
+      "loss": 0.975,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947978964042303,
+      "loss": 0.9517,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947182420947238,
+      "loss": 0.9173,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946379841933744,
+      "loss": 0.9083,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945571227488826,
+      "loss": 0.8929,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944756578103148,
+      "loss": 0.9456,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943935894271034,
+      "loss": 0.9533,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994310917649047,
+      "loss": 0.8889,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942276425263114,
+      "loss": 0.9078,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994143764109427,
+      "loss": 0.2705,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940592824492913,
+      "loss": 0.923,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993974197597167,
+      "loss": 0.9735,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993888509604684,
+      "loss": 0.941,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993802218523837,
+      "loss": 0.9704,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937153244069876,
+      "loss": 0.9695,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936278273068622,
+      "loss": 0.9576,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935397272765544,
+      "loss": 0.96,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934510243695228,
+      "loss": 0.9561,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933617186395917,
+      "loss": 0.9324,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932718101409522,
+      "loss": 0.9303,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199318129892816,
+      "loss": 0.8934,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930901850561373,
+      "loss": 0.9225,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929984685801712,
+      "loss": 0.8767,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929061495559155,
+      "loss": 0.2975,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928132280393887,
+      "loss": 0.9015,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927197040869757,
+      "loss": 0.9371,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926255777554262,
+      "loss": 0.954,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925308491018558,
+      "loss": 0.9874,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924355181837462,
+      "loss": 0.9237,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923395850589425,
+      "loss": 0.924,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992243049785658,
+      "loss": 0.896,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921459124224694,
+      "loss": 0.9327,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920481730283188,
+      "loss": 0.9502,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919498316625154,
+      "loss": 0.98,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918508883847314,
+      "loss": 0.9729,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917513432550057,
+      "loss": 0.9282,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916511963337415,
+      "loss": 0.9712,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915504476817085,
+      "loss": 0.9266,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914490973600398,
+      "loss": 0.9328,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913471454302347,
+      "loss": 0.8708,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912445919541578,
+      "loss": 0.9187,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911414369940377,
+      "loss": 0.9009,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910376806124685,
+      "loss": 0.9389,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909333228724098,
+      "loss": 0.8909,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908283638371844,
+      "loss": 0.8934,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907228035704824,
+      "loss": 0.8998,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906166421363567,
+      "loss": 0.9316,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990509879599226,
+      "loss": 0.9061,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990402516023873,
+      "loss": 0.9683,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019902945514754458,
+      "loss": 0.9013,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901859860194576,
+      "loss": 0.8746,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900768197217845,
+      "loss": 0.9093,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989967052648669,
+      "loss": 0.9564,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898566848667172,
+      "loss": 0.9406,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897457164429,
+      "loss": 0.9322,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896341474445525,
+      "loss": 0.9438,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019895219779393747,
+      "loss": 0.9673,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198940920799543,
+      "loss": 0.9115,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892958376811477,
+      "loss": 0.877,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198918186706532,
+      "loss": 0.3115,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890672962171043,
+      "loss": 0.9103,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889521252060215,
+      "loss": 0.9122,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888363541019572,
+      "loss": 0.9015,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887199829751606,
+      "loss": 0.9384,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886030118962457,
+      "loss": 0.9008,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198848544093619,
+      "loss": 0.9389,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883672701663353,
+      "loss": 0.2853,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882484996583874,
+      "loss": 0.9356,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881291294844154,
+      "loss": 0.9571,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988009159716853,
+      "loss": 0.957,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878885904284975,
+      "loss": 0.912,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877674216925098,
+      "loss": 0.9497,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987645653582415,
+      "loss": 0.9386,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019875232861721011,
+      "loss": 0.912,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874003195358207,
+      "loss": 0.9421,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872767537481895,
+      "loss": 0.9762,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871525888841868,
+      "loss": 0.9127,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870278250191551,
+      "loss": 0.8774,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019869024622288012,
+      "loss": 0.9502,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867765005891945,
+      "loss": 0.9125,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866499401767685,
+      "loss": 0.915,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865227810683192,
+      "loss": 0.9086,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863950233410066,
+      "loss": 0.9047,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862666670723538,
+      "loss": 0.9275,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861377123402466,
+      "loss": 0.9457,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860081592229344,
+      "loss": 0.9201,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198587800779903,
+      "loss": 0.8963,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857472581475088,
+      "loss": 0.8979,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856159103477086,
+      "loss": 0.2868,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854839644793315,
+      "loss": 0.9201,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019853514206224414,
+      "loss": 0.9353,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852182788574662,
+      "loss": 0.8919,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985084539265195,
+      "loss": 0.9018,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849502019267813,
+      "loss": 0.9465,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848152669237399,
+      "loss": 0.3127,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846797343379494,
+      "loss": 0.9352,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845436042516506,
+      "loss": 0.9167,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019844068767474463,
+      "loss": 0.8977,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019842695519083027,
+      "loss": 0.9462,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841316298175481,
+      "loss": 0.9299,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983993110558873,
+      "loss": 0.9654,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838539942163306,
+      "loss": 0.9213,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837142808743359,
+      "loss": 0.2901,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983573970617667,
+      "loss": 0.8707,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019834330635314635,
+      "loss": 0.9728,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983291559701227,
+      "loss": 0.9652,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831494592128225,
+      "loss": 0.8919,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983006762152475,
+      "loss": 0.2988,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019828634686067735,
+      "loss": 0.9165,
+      "step": 363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827195786626675,
+      "loss": 0.9338,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019825750924074695,
+      "loss": 0.885,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982430009928853,
+      "loss": 0.9178,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822843313148532,
+      "loss": 0.9198,
+      "step": 367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982138056653868,
+      "loss": 0.3107,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981991186034656,
+      "loss": 0.9252,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818437195463383,
+      "loss": 0.3047,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816956572783965,
+      "loss": 0.895,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019815469993206746,
+      "loss": 0.9188,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981397745763378,
+      "loss": 0.917,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981247896697073,
+      "loss": 0.9028,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810974522126878,
+      "loss": 0.9415,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019809464124015112,
+      "loss": 0.9007,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807947773551937,
+      "loss": 0.9185,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806425471657476,
+      "loss": 0.3005,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804897219255447,
+      "loss": 0.9401,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019803363017273197,
+      "loss": 0.271,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801822866641672,
+      "loss": 0.9347,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019800276768295428,
+      "loss": 0.8863,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019798724723172638,
+      "loss": 0.9131,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019797166732215076,
+      "loss": 0.8838,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795602796368123,
+      "loss": 0.8836,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794032916580775,
+      "loss": 0.9369,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979245709380562,
+      "loss": 0.9285,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019790875328998877,
+      "loss": 0.9424,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789287623120342,
+      "loss": 0.9062,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978769397713344,
+      "loss": 0.8868,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019786094392005186,
+      "loss": 0.9082,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019784488868706203,
+      "loss": 0.8714,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782877408210716,
+      "loss": 0.9581,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978126001149656,
+      "loss": 0.935,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001977963667954516,
+      "loss": 0.2867,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977800741334155,
+      "loss": 0.9248,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776372213874366,
+      "loss": 0.9235,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774731082135839,
+      "loss": 0.9235,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197730840191218,
+      "loss": 0.9223,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977143102583169,
+      "loss": 0.9147,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019769772103268531,
+      "loss": 0.8844,
+      "step": 401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976810725243896,
+      "loss": 0.9384,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197664364743532,
+      "loss": 0.945,
+      "step": 403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764759770025066,
+      "loss": 0.8962,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976307714047199,
+      "loss": 0.9187,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019761388586714978,
+      "loss": 0.9182,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975969410977864,
+      "loss": 0.9352,
+      "step": 407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757993710691182,
+      "loss": 0.9074,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019756287390484397,
+      "loss": 0.9222,
+      "step": 409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754575150193678,
+      "loss": 0.9234,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019752856990858003,
+      "loss": 0.9361,
+      "step": 411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019751132913519952,
+      "loss": 0.9238,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749402919225684,
+      "loss": 0.9071,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019747667009024954,
+      "loss": 0.9102,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974592518397111,
+      "loss": 0.9437,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019744177445121084,
+      "loss": 0.9366,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019742423793535398,
+      "loss": 0.927,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019740664230278164,
+      "loss": 0.9406,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019738898756417074,
+      "loss": 0.9134,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973712737302342,
+      "loss": 0.9269,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735350081172067,
+      "loss": 0.901,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973356688194147,
+      "loss": 0.9183,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019731777776413673,
+      "loss": 0.8959,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972998276567429,
+      "loss": 0.9244,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019728181850812537,
+      "loss": 0.9676,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197263750329212,
+      "loss": 0.8989,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019724562313096648,
+      "loss": 0.9123,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019722743692438837,
+      "loss": 0.8859,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197209191720513,
+      "loss": 0.9693,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719088753041147,
+      "loss": 0.946,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717252436519074,
+      "loss": 0.3012,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715410223599352,
+      "loss": 0.875,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713562115399828,
+      "loss": 0.915,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019711708113041927,
+      "loss": 0.9799,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970984821765065,
+      "loss": 0.878,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019707982430354585,
+      "loss": 0.8536,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019706110752285878,
+      "loss": 0.9323,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704233184580257,
+      "loss": 0.894,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702349728377026,
+      "loss": 0.8545,
+      "step": 439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970046038481906,
+      "loss": 0.88,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019698565155052807,
+      "loss": 0.889,
+      "step": 441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696664040228288,
+      "loss": 0.9288,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969475704149909,
+      "loss": 0.9286,
+      "step": 443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019692844160022378,
+      "loss": 0.9207,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019690925396958874,
+      "loss": 0.9595,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019689000753472886,
+      "loss": 0.8988,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019687070230732276,
+      "loss": 0.9223,
+      "step": 447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019685133829908485,
+      "loss": 0.9423,
+      "step": 448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968319155217651,
+      "loss": 0.9253,
+      "step": 449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681243398714918,
+      "loss": 0.9536,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019679289370705844,
+      "loss": 0.8813,
+      "step": 451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967732946933499,
+      "loss": 0.8607,
+      "step": 452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967536369579161,
+      "loss": 0.8573,
+      "step": 453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673392051268538,
+      "loss": 0.9418,
+      "step": 454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671414536962152,
+      "loss": 0.9286,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966943115407241,
+      "loss": 0.93,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667441903802816,
+      "loss": 0.8672,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966544678736044,
+      "loss": 0.8622,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019663445805955922,
+      "loss": 0.9061,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019661438960803443,
+      "loss": 0.9252,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965942625312075,
+      "loss": 0.8807,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965740768412915,
+      "loss": 0.8581,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019655383255053502,
+      "loss": 0.9693,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019653352967122228,
+      "loss": 0.8807,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019651316821567294,
+      "loss": 0.8568,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019649274819624234,
+      "loss": 0.8892,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019647226962532117,
+      "loss": 0.9281,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964517325153359,
+      "loss": 0.9243,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019643113687874822,
+      "loss": 0.8671,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019641048272805566,
+      "loss": 0.2939,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196389770075791,
+      "loss": 0.9337,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019636899893452263,
+      "loss": 0.2924,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019634816931685443,
+      "loss": 0.9133,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963272812354257,
+      "loss": 0.8599,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019630633470291132,
+      "loss": 0.9039,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962853297320215,
+      "loss": 0.9308,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019626426633550205,
+      "loss": 0.9126,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019624314452613417,
+      "loss": 0.9095,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001962219643167345,
+      "loss": 0.9363,
+      "step": 479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001962007257201551,
+      "loss": 0.9396,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019617942874928352,
+      "loss": 0.929,
+      "step": 481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961580734170427,
+      "loss": 0.9248,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961366597363909,
+      "loss": 0.8472,
+      "step": 483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019611518772032197,
+      "loss": 0.863,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019609365738186503,
+      "loss": 0.9284,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960720687340846,
+      "loss": 0.9112,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960504217900806,
+      "loss": 0.8887,
+      "step": 487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960287165629884,
+      "loss": 0.9212,
+      "step": 488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600695306597855,
+      "loss": 0.8977,
+      "step": 489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019598513131225713,
+      "loss": 0.9242,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959632513150655,
+      "loss": 0.9004,
+      "step": 491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019594131308768041,
+      "loss": 0.8639,
+      "step": 492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019591931664341385,
+      "loss": 0.915,
+      "step": 493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019589726199561325,
+      "loss": 0.8795,
+      "step": 494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019587514915766124,
+      "loss": 0.9225,
+      "step": 495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019585297814297582,
+      "loss": 0.9176,
+      "step": 496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019583074896501037,
+      "loss": 0.9228,
+      "step": 497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958084616372534,
+      "loss": 0.919,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019578611617322885,
+      "loss": 0.8853,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957637125864958,
+      "loss": 0.9365,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019574125089064873,
+      "loss": 0.912,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571873109931734,
+      "loss": 0.8979,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019569615322616651,
+      "loss": 0.899,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019567351728489647,
+      "loss": 0.9081,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019565082328924262,
+      "loss": 0.8969,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956280712529756,
+      "loss": 0.298,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019560526118990126,
+      "loss": 0.9278,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019558239311386072,
+      "loss": 0.9357,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019555946703873022,
+      "loss": 0.9026,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955364829784212,
+      "loss": 0.8877,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955134409468804,
+      "loss": 0.957,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019549034095808955,
+      "loss": 0.9517,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019546718302606572,
+      "loss": 0.9128,
+      "step": 513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544396716486104,
+      "loss": 0.9057,
+      "step": 514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019542069338856287,
+      "loss": 0.9147,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019539736171129355,
+      "loss": 0.9356,
+      "step": 516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019537397214721078,
+      "loss": 0.8916,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019535052471050723,
+      "loss": 0.3309,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953270194154107,
+      "loss": 0.8883,
+      "step": 519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019530345627618418,
+      "loss": 0.9363,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527983530712569,
+      "loss": 0.8614,
+      "step": 521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019525615652256833,
+      "loss": 0.9019,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019523241993688037,
+      "loss": 0.9488,
+      "step": 523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520862556446505,
+      "loss": 0.9389,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019518477341976068,
+      "loss": 0.9808,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019516086351724075,
+      "loss": 0.9323,
+      "step": 526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019513689587141367,
+      "loss": 0.895,
+      "step": 527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019511287049682295,
+      "loss": 0.9554,
+      "step": 528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019508878740804705,
+      "loss": 0.9039,
+      "step": 529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950646466196996,
+      "loss": 0.8639,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019504044814642907,
+      "loss": 0.9405,
+      "step": 531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019501619200291908,
+      "loss": 0.891,
+      "step": 532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019499187820388815,
+      "loss": 0.9167,
+      "step": 533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019496750676408983,
+      "loss": 0.9013,
+      "step": 534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949430776983126,
+      "loss": 0.9048,
+      "step": 535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019491859102137998,
+      "loss": 0.9562,
+      "step": 536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019489404674815036,
+      "loss": 0.8934,
+      "step": 537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948694448935172,
+      "loss": 0.3261,
+      "step": 538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019484478547240875,
+      "loss": 0.8908,
+      "step": 539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019482006849978832,
+      "loss": 0.9762,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019479529399065405,
+      "loss": 0.9085,
+      "step": 541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019477046196003905,
+      "loss": 0.9176,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019474557242301133,
+      "loss": 0.9504,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019472062539467374,
+      "loss": 0.9302,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019469562089016412,
+      "loss": 0.9105,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946705589246551,
+      "loss": 0.9358,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019464543951335417,
+      "loss": 0.9083,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019462026267150372,
+      "loss": 0.9008,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019459502841438097,
+      "loss": 0.9167,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019456973675729803,
+      "loss": 0.8883,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019454438771560178,
+      "loss": 0.8901,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945189813046739,
+      "loss": 0.9489,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019449351753993095,
+      "loss": 0.8776,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019446799643682425,
+      "loss": 0.8853,
+      "step": 554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019444241801083994,
+      "loss": 0.944,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944167822774989,
+      "loss": 0.8827,
+      "step": 556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019439108925235684,
+      "loss": 0.94,
+      "step": 557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943653389510042,
+      "loss": 0.9743,
+      "step": 558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019433953138906616,
+      "loss": 0.8782,
+      "step": 559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019431366658220268,
+      "loss": 0.9495,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019428774454610843,
+      "loss": 0.922,
+      "step": 561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019426176529651284,
+      "loss": 0.9063,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019423572884918002,
+      "loss": 0.3097,
+      "step": 563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942096352199088,
+      "loss": 0.8924,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019418348442453272,
+      "loss": 0.3059,
+      "step": 565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019415727647892,
+      "loss": 0.9539,
+      "step": 566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019413101139897354,
+      "loss": 0.865,
+      "step": 567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941046892006309,
+      "loss": 0.9104,
+      "step": 568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940783098998643,
+      "loss": 0.9079,
+      "step": 569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940518735126806,
+      "loss": 0.8767,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019402538005512137,
+      "loss": 0.9376,
+      "step": 571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019399882954326272,
+      "loss": 0.8649,
+      "step": 572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019397222199321541,
+      "loss": 0.9177,
+      "step": 573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939455574211248,
+      "loss": 0.8817,
+      "step": 574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939188358431709,
+      "loss": 0.9,
+      "step": 575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938920572755683,
+      "loss": 0.916,
+      "step": 576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019386522173456608,
+      "loss": 0.89,
+      "step": 577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193838329236448,
+      "loss": 0.8799,
+      "step": 578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019381137979753238,
+      "loss": 0.8519,
+      "step": 579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019378437343417195,
+      "loss": 0.851,
+      "step": 580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019375731016275417,
+      "loss": 0.9598,
+      "step": 581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373018999970093,
+      "loss": 0.8698,
+      "step": 582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019370301296146866,
+      "loss": 0.9342,
+      "step": 583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019367577906454828,
+      "loss": 0.8671,
+      "step": 584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019364848832546523,
+      "loss": 0.9145,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936211407607795,
+      "loss": 0.8439,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935937363870854,
+      "loss": 0.9611,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019356627522101188,
+      "loss": 0.335,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019353875727922234,
+      "loss": 0.9012,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019351118257841446,
+      "loss": 0.8799,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934835511353206,
+      "loss": 0.9343,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019345586296670735,
+      "loss": 0.8799,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934281180893758,
+      "loss": 0.8988,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019340031652016152,
+      "loss": 0.9024,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019337245827593438,
+      "loss": 0.911,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001933445433735986,
+      "loss": 0.9006,
+      "step": 596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019331657183009298,
+      "loss": 0.9206,
+      "step": 597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019328854366239048,
+      "loss": 0.9593,
+      "step": 598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019326045888749853,
+      "loss": 0.8788,
+      "step": 599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019323231752245884,
+      "loss": 0.8593,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932041195843476,
+      "loss": 0.9312,
+      "step": 601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019317586509027512,
+      "loss": 0.85,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001931475540573862,
+      "loss": 0.8989,
+      "step": 603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019311918650285982,
+      "loss": 0.9304,
+      "step": 604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019309076244390942,
+      "loss": 0.8888,
+      "step": 605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019306228189778254,
+      "loss": 0.8922,
+      "step": 606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019303374488176112,
+      "loss": 0.9288,
+      "step": 607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019300515141316136,
+      "loss": 0.9173,
+      "step": 608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019297650150933363,
+      "loss": 0.321,
+      "step": 609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929477951876626,
+      "loss": 0.9354,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019291903246556724,
+      "loss": 0.9378,
+      "step": 611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928902133605006,
+      "loss": 0.8875,
+      "step": 612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019286133788995008,
+      "loss": 0.9398,
+      "step": 613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019283240607143718,
+      "loss": 0.916,
+      "step": 614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019280341792251768,
+      "loss": 0.9355,
+      "step": 615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019277437346078145,
+      "loss": 0.9079,
+      "step": 616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019274527270385258,
+      "loss": 0.9011,
+      "step": 617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019271611566938935,
+      "loss": 0.3152,
+      "step": 618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019268690237508412,
+      "loss": 0.9131,
+      "step": 619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019265763283866347,
+      "loss": 0.9118,
+      "step": 620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192628307077888,
+      "loss": 0.8949,
+      "step": 621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019259892511055252,
+      "loss": 0.891,
+      "step": 622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019256948695448592,
+      "loss": 0.8922,
+      "step": 623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925399926275512,
+      "loss": 0.9079,
+      "step": 624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019251044214764536,
+      "loss": 0.3181,
+      "step": 625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924808355326996,
+      "loss": 0.9393,
+      "step": 626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019245117280067904,
+      "loss": 0.9083,
+      "step": 627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019242145396958303,
+      "loss": 0.9574,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019239167905744478,
+      "loss": 0.8171,
+      "step": 629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019236184808233165,
+      "loss": 0.8857,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019233196106234498,
+      "loss": 0.9,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923020180156201,
+      "loss": 0.898,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922720189603264,
+      "loss": 0.9203,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019224196391466714,
+      "loss": 0.9279,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019221185289687966,
+      "loss": 0.925,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019218168592523524,
+      "loss": 0.8994,
+      "step": 636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921514630180391,
+      "loss": 0.8507,
+      "step": 637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019212118419363039,
+      "loss": 0.9051,
+      "step": 638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019209084947038223,
+      "loss": 0.9255,
+      "step": 639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019206045886670158,
+      "loss": 0.8195,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920300124010294,
+      "loss": 0.9005,
+      "step": 641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019199951009184053,
+      "loss": 0.8702,
+      "step": 642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019196895195764362,
+      "loss": 0.8978,
+      "step": 643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001919383380169813,
+      "loss": 0.873,
+      "step": 644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019190766828842994,
+      "loss": 0.3323,
+      "step": 645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918769427905999,
+      "loss": 0.9471,
+      "step": 646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918461615421353,
+      "loss": 0.9071,
+      "step": 647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019181532456171409,
+      "loss": 0.8553,
+      "step": 648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019178443186804804,
+      "loss": 0.8815,
+      "step": 649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019175348347988278,
+      "loss": 0.8734,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019172247941599767,
+      "loss": 0.8939,
+      "step": 651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019169141969520583,
+      "loss": 0.8749,
+      "step": 652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019166030433635426,
+      "loss": 0.8922,
+      "step": 653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019162913335832364,
+      "loss": 0.9158,
+      "step": 654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019159790678002843,
+      "loss": 0.9311,
+      "step": 655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915666246204168,
+      "loss": 0.878,
+      "step": 656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915352868984707,
+      "loss": 0.8822,
+      "step": 657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915038936332057,
+      "loss": 0.8749,
+      "step": 658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019147244484367116,
+      "loss": 0.9144,
+      "step": 659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914409405489501,
+      "loss": 0.9095,
+      "step": 660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019140938076815927,
+      "loss": 0.9015,
+      "step": 661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019137776552044899,
+      "loss": 0.8825,
+      "step": 662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913460948250033,
+      "loss": 0.8844,
+      "step": 663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913143687010399,
+      "loss": 0.8947,
+      "step": 664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019128258716781012,
+      "loss": 0.92,
+      "step": 665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912507502445988,
+      "loss": 0.8388,
+      "step": 666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912188579507246,
+      "loss": 0.8533,
+      "step": 667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019118691030553953,
+      "loss": 0.8913,
+      "step": 668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019115490732842945,
+      "loss": 0.8562,
+      "step": 669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911228490388136,
+      "loss": 0.897,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910907354561448,
+      "loss": 0.9201,
+      "step": 671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019105856659990957,
+      "loss": 0.8993,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019102634248962784,
+      "loss": 0.9137,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019099406314485306,
+      "loss": 0.8947,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019096172858517222,
+      "loss": 0.9242,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909293388302059,
+      "loss": 0.8878,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019089689389960806,
+      "loss": 0.932,
+      "step": 677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019086439381306615,
+      "loss": 0.9461,
+      "step": 678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019083183859030118,
+      "loss": 0.912,
+      "step": 679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907992282510675,
+      "loss": 0.8739,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019076656281515297,
+      "loss": 0.9036,
+      "step": 681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019073384230237892,
+      "loss": 0.9044,
+      "step": 682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070106673259999,
+      "loss": 0.8557,
+      "step": 683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019066823612570431,
+      "loss": 0.8277,
+      "step": 684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019063535050161342,
+      "loss": 0.8907,
+      "step": 685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019060240988028214,
+      "loss": 0.8768,
+      "step": 686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019056941428169881,
+      "loss": 0.9082,
+      "step": 687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019053636372588502,
+      "loss": 0.8347,
+      "step": 688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019050325823289575,
+      "loss": 0.8939,
+      "step": 689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019047009782281927,
+      "loss": 0.8775,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019043688251577727,
+      "loss": 0.8688,
+      "step": 691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019040361233192465,
+      "loss": 0.3211,
+      "step": 692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019037028729144967,
+      "loss": 0.9208,
+      "step": 693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019033690741457386,
+      "loss": 0.912,
+      "step": 694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190303472721552,
+      "loss": 0.8811,
+      "step": 695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019026998323267216,
+      "loss": 0.3185,
+      "step": 696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902364389682557,
+      "loss": 0.9233,
+      "step": 697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019020283994865712,
+      "loss": 0.9076,
+      "step": 698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019016918619426418,
+      "loss": 0.8883,
+      "step": 699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901354777254979,
+      "loss": 0.9029,
+      "step": 700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019010171456281251,
+      "loss": 0.9153,
+      "step": 701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019006789672669535,
+      "loss": 0.8447,
+      "step": 702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190034024237667,
+      "loss": 0.841,
+      "step": 703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019000009711628112,
+      "loss": 0.9075,
+      "step": 704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018996611538312462,
+      "loss": 0.9244,
+      "step": 705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018993207905881746,
+      "loss": 0.9265,
+      "step": 706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018989798816401287,
+      "loss": 0.8997,
+      "step": 707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000189863842719397,
+      "loss": 0.9512,
+      "step": 708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018982964274568927,
+      "loss": 0.9439,
+      "step": 709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018979538826364208,
+      "loss": 0.8883,
+      "step": 710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897610792940409,
+      "loss": 0.8867,
+      "step": 711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897267158577044,
+      "loss": 0.9128,
+      "step": 712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018969229797548412,
+      "loss": 0.9262,
+      "step": 713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018965782566826472,
+      "loss": 0.846,
+      "step": 714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018962329895696396,
+      "loss": 0.8812,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018958871786253244,
+      "loss": 0.836,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018955408240595395,
+      "loss": 0.9302,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018951939260824512,
+      "loss": 0.9161,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894846484904556,
+      "loss": 0.8396,
+      "step": 719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018944985007366804,
+      "loss": 0.8684,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000189414997378998,
+      "loss": 0.8718,
+      "step": 721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018938009042759397,
+      "loss": 0.873,
+      "step": 722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018934512924063742,
+      "loss": 0.9148,
+      "step": 723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018931011383934262,
+      "loss": 0.8611,
+      "step": 724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018927504424495683,
+      "loss": 0.8858,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018923992047876017,
+      "loss": 0.8921,
+      "step": 726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018920474256206568,
+      "loss": 0.9122,
+      "step": 727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891695105162191,
+      "loss": 0.9,
+      "step": 728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018913422436259923,
+      "loss": 0.9181,
+      "step": 729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018909888412261755,
+      "loss": 0.8983,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018906348981771837,
+      "loss": 0.8867,
+      "step": 731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018902804146937887,
+      "loss": 0.9032,
+      "step": 732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018899253909910902,
+      "loss": 0.9177,
+      "step": 733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889569827284515,
+      "loss": 0.9072,
+      "step": 734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018892137237898183,
+      "loss": 0.8933,
+      "step": 735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018888570807230824,
+      "loss": 0.9114,
+      "step": 736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018884998983007177,
+      "loss": 0.8824,
+      "step": 737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018881421767394605,
+      "loss": 0.8971,
+      "step": 738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018877839162563756,
+      "loss": 0.9062,
+      "step": 739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887425117068854,
+      "loss": 0.9218,
+      "step": 740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018870657793946144,
+      "loss": 0.889,
+      "step": 741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018867059034517015,
+      "loss": 0.8923,
+      "step": 742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018863454894584865,
+      "loss": 0.9215,
+      "step": 743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018859845376336683,
+      "loss": 0.81,
+      "step": 744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018856230481962703,
+      "loss": 0.898,
+      "step": 745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885261021365644,
+      "loss": 0.891,
+      "step": 746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018848984573614658,
+      "loss": 0.8779,
+      "step": 747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018845353564037382,
+      "loss": 0.8896,
+      "step": 748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018841717187127895,
+      "loss": 0.8832,
+      "step": 749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018838075445092747,
+      "loss": 0.9338,
+      "step": 750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883442834014173,
+      "loss": 0.9502,
+      "step": 751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018830775874487894,
+      "loss": 0.912,
+      "step": 752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018827118050347548,
+      "loss": 0.8895,
+      "step": 753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018823454869940242,
+      "loss": 0.8887,
+      "step": 754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018819786335488788,
+      "loss": 0.9259,
+      "step": 755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881611244921924,
+      "loss": 0.8989,
+      "step": 756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188124332133609,
+      "loss": 0.907,
+      "step": 757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018808748630146313,
+      "loss": 0.9526,
+      "step": 758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018805058701811277,
+      "loss": 0.8749,
+      "step": 759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018801363430594824,
+      "loss": 0.9293,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018797662818739234,
+      "loss": 0.3213,
+      "step": 761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018793956868490034,
+      "loss": 0.8566,
+      "step": 762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018790245582095972,
+      "loss": 0.8906,
+      "step": 763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878652896180905,
+      "loss": 0.9261,
+      "step": 764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000187828070098845,
+      "loss": 0.3085,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018779079728580792,
+      "loss": 0.852,
+      "step": 766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018775347120159625,
+      "loss": 0.9312,
+      "step": 767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018771609186885938,
+      "loss": 0.8818,
+      "step": 768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018767865931027897,
+      "loss": 0.9335,
+      "step": 769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018764117354856894,
+      "loss": 0.9109,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018760363460647555,
+      "loss": 0.9119,
+      "step": 771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875660425067773,
+      "loss": 0.9097,
+      "step": 772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018752839727228499,
+      "loss": 0.8622,
+      "step": 773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018749069892584156,
+      "loss": 0.8477,
+      "step": 774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018745294749032232,
+      "loss": 0.8787,
+      "step": 775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018741514298863468,
+      "loss": 0.3194,
+      "step": 776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873772854437183,
+      "loss": 0.9139,
+      "step": 777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018733937487854503,
+      "loss": 0.9109,
+      "step": 778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018730141131611882,
+      "loss": 0.8796,
+      "step": 779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018726339477947591,
+      "loss": 0.913,
+      "step": 780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872253252916846,
+      "loss": 0.8725,
+      "step": 781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018718720287584528,
+      "loss": 0.8869,
+      "step": 782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871490275550906,
+      "loss": 0.9063,
+      "step": 783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018711079935258514,
+      "loss": 0.8623,
+      "step": 784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018707251829152565,
+      "loss": 0.9243,
+      "step": 785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018703418439514104,
+      "loss": 0.9474,
+      "step": 786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001869957976866921,
+      "loss": 0.9249,
+      "step": 787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018695735818947184,
+      "loss": 0.9197,
+      "step": 788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018691886592680517,
+      "loss": 0.8816,
+      "step": 789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868803209220491,
+      "loss": 0.8967,
+      "step": 790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868417231985926,
+      "loss": 0.8783,
+      "step": 791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018680307277985664,
+      "loss": 0.9194,
+      "step": 792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867643696892942,
+      "loss": 0.3054,
+      "step": 793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018672561395039015,
+      "loss": 0.8897,
+      "step": 794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018668680558666138,
+      "loss": 0.857,
+      "step": 795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018664794462165666,
+      "loss": 0.328,
+      "step": 796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660903107895672,
+      "loss": 0.9117,
+      "step": 797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018657006498217415,
+      "loss": 0.8652,
+      "step": 798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865310463549535,
+      "loss": 0.8825,
+      "step": 799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018649197522097106,
+      "loss": 0.9029,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018645285160393517,
+      "loss": 0.9415,
+      "step": 801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018641367552758585,
+      "loss": 0.9207,
+      "step": 802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018637444701569502,
+      "loss": 0.8891,
+      "step": 803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018633516609206646,
+      "loss": 0.3538,
+      "step": 804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018629583278053567,
+      "loss": 0.9218,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018625644710496998,
+      "loss": 0.8512,
+      "step": 806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018621700908926848,
+      "loss": 0.351,
+      "step": 807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018617751875736206,
+      "loss": 0.9323,
+      "step": 808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018613797613321333,
+      "loss": 0.9394,
+      "step": 809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018609838124081663,
+      "loss": 0.9508,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018605873410419796,
+      "loss": 0.8403,
+      "step": 811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018601903474741515,
+      "loss": 0.9021,
+      "step": 812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018597928319455758,
+      "loss": 0.9329,
+      "step": 813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859394794697464,
+      "loss": 0.8839,
+      "step": 814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018589962359713438,
+      "loss": 0.9503,
+      "step": 815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018585971560090592,
+      "loss": 0.9199,
+      "step": 816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018581975550527706,
+      "loss": 0.8736,
+      "step": 817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018577974333449548,
+      "loss": 1.0021,
+      "step": 818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018573967911284046,
+      "loss": 0.9228,
+      "step": 819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856995628646228,
+      "loss": 0.9058,
+      "step": 820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856593946141849,
+      "loss": 0.9273,
+      "step": 821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018561917438590078,
+      "loss": 0.9217,
+      "step": 822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855789022041759,
+      "loss": 0.9022,
+      "step": 823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018553857809344728,
+      "loss": 0.8536,
+      "step": 824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018549820207818348,
+      "loss": 0.8829,
+      "step": 825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018545777418288459,
+      "loss": 0.8995,
+      "step": 826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018541729443208204,
+      "loss": 0.925,
+      "step": 827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018537676285033887,
+      "loss": 0.8404,
+      "step": 828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853361794622495,
+      "loss": 0.8837,
+      "step": 829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852955442924398,
+      "loss": 0.9082,
+      "step": 830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018525485736556704,
+      "loss": 0.8781,
+      "step": 831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018521411870631993,
+      "loss": 0.8653,
+      "step": 832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018517332833941857,
+      "loss": 0.3211,
+      "step": 833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851324862896144,
+      "loss": 0.9565,
+      "step": 834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018509159258169028,
+      "loss": 0.8461,
+      "step": 835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018505064724046036,
+      "loss": 0.9067,
+      "step": 836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018500965029077016,
+      "loss": 0.9004,
+      "step": 837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018496860175749643,
+      "loss": 0.8982,
+      "step": 838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849275016655474,
+      "loss": 0.8937,
+      "step": 839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018488635003986238,
+      "loss": 0.8707,
+      "step": 840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848451469054121,
+      "loss": 0.9143,
+      "step": 841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018480389228719843,
+      "loss": 0.8938,
+      "step": 842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018476258621025462,
+      "loss": 0.89,
+      "step": 843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018472122869964502,
+      "loss": 0.8514,
+      "step": 844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846798197804652,
+      "loss": 0.9129,
+      "step": 845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018463835947784201,
+      "loss": 0.8826,
+      "step": 846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018459684781693344,
+      "loss": 0.8985,
+      "step": 847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001845552848229286,
+      "loss": 0.9002,
+      "step": 848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001845136705210478,
+      "loss": 0.8765,
+      "step": 849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018447200493654245,
+      "loss": 0.8738,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001844302880946951,
+      "loss": 0.8636,
+      "step": 851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018438852002081939,
+      "loss": 0.801,
+      "step": 852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001843467007402601,
+      "loss": 0.8925,
+      "step": 853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018430483027839293,
+      "loss": 0.8164,
+      "step": 854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018426290866062486,
+      "loss": 0.89,
+      "step": 855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842209359123937,
+      "loss": 0.9148,
+      "step": 856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018417891205916844,
+      "loss": 0.8847,
+      "step": 857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000184136837126449,
+      "loss": 0.9515,
+      "step": 858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001840947111397663,
+      "loss": 0.9147,
+      "step": 859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001840525341246823,
+      "loss": 0.8929,
+      "step": 860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018401030610678979,
+      "loss": 0.8944,
+      "step": 861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839680271117127,
+      "loss": 0.8919,
+      "step": 862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018392569716510571,
+      "loss": 0.9087,
+      "step": 863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018388331629265453,
+      "loss": 0.8813,
+      "step": 864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018384088452007578,
+      "loss": 0.8555,
+      "step": 865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018379840187311683,
+      "loss": 0.8945,
+      "step": 866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018375586837755608,
+      "loss": 0.8412,
+      "step": 867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018371328405920268,
+      "loss": 0.9717,
+      "step": 868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018367064894389667,
+      "loss": 0.9201,
+      "step": 869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018362796305750894,
+      "loss": 0.9324,
+      "step": 870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835852264259411,
+      "loss": 0.8766,
+      "step": 871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018354243907512562,
+      "loss": 0.9115,
+      "step": 872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018349960103102576,
+      "loss": 0.8704,
+      "step": 873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018345671231963543,
+      "loss": 0.8634,
+      "step": 874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001834137729669794,
+      "loss": 0.9066,
+      "step": 875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833707829991132,
+      "loss": 0.9143,
+      "step": 876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833277424421229,
+      "loss": 0.8946,
+      "step": 877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832846513221254,
+      "loss": 0.8773,
+      "step": 878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018324150966526833,
+      "loss": 0.9094,
+      "step": 879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831983174977298,
+      "loss": 0.9109,
+      "step": 880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018315507484571874,
+      "loss": 0.873,
+      "step": 881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018311178173547465,
+      "loss": 0.8855,
+      "step": 882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018306843819326763,
+      "loss": 0.8807,
+      "step": 883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830250442453984,
+      "loss": 0.9094,
+      "step": 884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829815999181983,
+      "loss": 0.9354,
+      "step": 885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018293810523802918,
+      "loss": 0.9138,
+      "step": 886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828945602312835,
+      "loss": 0.9256,
+      "step": 887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018285096492438424,
+      "loss": 0.2893,
+      "step": 888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018280731934378481,
+      "loss": 0.878,
+      "step": 889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018276362351596928,
+      "loss": 0.9172,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018271987746745213,
+      "loss": 0.8979,
+      "step": 891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018267608122477834,
+      "loss": 0.8685,
+      "step": 892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018263223481452328,
+      "loss": 0.8633,
+      "step": 893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018258833826329283,
+      "loss": 0.8789,
+      "step": 894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825443915977233,
+      "loss": 0.8981,
+      "step": 895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018250039484448135,
+      "loss": 0.8736,
+      "step": 896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018245634803026412,
+      "loss": 0.3241,
+      "step": 897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000182412251181799,
+      "loss": 0.8319,
+      "step": 898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018236810432584392,
+      "loss": 0.8957,
+      "step": 899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018232390748918693,
+      "loss": 0.8831,
+      "step": 900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018227966069864663,
+      "loss": 0.8452,
+      "step": 901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018223536398107176,
+      "loss": 0.9221,
+      "step": 902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018219101736334148,
+      "loss": 0.892,
+      "step": 903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018214662087236513,
+      "loss": 0.8839,
+      "step": 904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018210217453508242,
+      "loss": 0.8773,
+      "step": 905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018205767837846317,
+      "loss": 0.9476,
+      "step": 906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018201313242950754,
+      "loss": 0.8673,
+      "step": 907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018196853671524589,
+      "loss": 0.9126,
+      "step": 908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018192389126273875,
+      "loss": 0.9399,
+      "step": 909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818791960990768,
+      "loss": 0.8708,
+      "step": 910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018183445125138096,
+      "loss": 0.29,
+      "step": 911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018178965674680222,
+      "loss": 0.8672,
+      "step": 912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001817448126125218,
+      "loss": 0.8971,
+      "step": 913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018169991887575094,
+      "loss": 0.8536,
+      "step": 914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018165497556373101,
+      "loss": 0.8544,
+      "step": 915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001816099827037335,
+      "loss": 0.9322,
+      "step": 916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815649403230599,
+      "loss": 0.8705,
+      "step": 917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018151984844904179,
+      "loss": 0.8873,
+      "step": 918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018147470710904077,
+      "loss": 0.9099,
+      "step": 919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814295163304485,
+      "loss": 0.8552,
+      "step": 920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018138427614068653,
+      "loss": 0.8527,
+      "step": 921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018133898656720653,
+      "loss": 0.852,
+      "step": 922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018129364763749006,
+      "loss": 0.8495,
+      "step": 923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018124825937904858,
+      "loss": 0.8959,
+      "step": 924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812028218194236,
+      "loss": 0.9007,
+      "step": 925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018115733498618645,
+      "loss": 0.8823,
+      "step": 926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001811117989069384,
+      "loss": 0.8894,
+      "step": 927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018106621360931065,
+      "loss": 0.8968,
+      "step": 928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810205791209641,
+      "loss": 0.93,
+      "step": 929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018097489546958973,
+      "loss": 0.9412,
+      "step": 930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018092916268290816,
+      "loss": 0.8704,
+      "step": 931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018088338078866992,
+      "loss": 0.8489,
+      "step": 932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018083754981465533,
+      "loss": 0.9359,
+      "step": 933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018079166978867444,
+      "loss": 0.8822,
+      "step": 934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018074574073856717,
+      "loss": 0.9045,
+      "step": 935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018069976269220305,
+      "loss": 0.9545,
+      "step": 936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018065373567748144,
+      "loss": 0.9084,
+      "step": 937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001806076597223314,
+      "loss": 0.8836,
+      "step": 938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018056153485471166,
+      "loss": 0.8931,
+      "step": 939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001805153611026106,
+      "loss": 0.9129,
+      "step": 940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001804691384940464,
+      "loss": 0.8507,
+      "step": 941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018042286705706668,
+      "loss": 0.3402,
+      "step": 942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803765468197489,
+      "loss": 0.9358,
+      "step": 943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018033017781019997,
+      "loss": 0.8844,
+      "step": 944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018028376005655645,
+      "loss": 0.8507,
+      "step": 945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018023729358698453,
+      "loss": 0.9246,
+      "step": 946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801907784296799,
+      "loss": 0.9052,
+      "step": 947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801442146128678,
+      "loss": 0.8987,
+      "step": 948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018009760216480296,
+      "loss": 0.8835,
+      "step": 949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018005094111376975,
+      "loss": 0.8835,
+      "step": 950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018000423148808187,
+      "loss": 0.8775,
+      "step": 951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017995747331608266,
+      "loss": 0.9257,
+      "step": 952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017991066662614475,
+      "loss": 0.8825,
+      "step": 953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798638114466703,
+      "loss": 0.8525,
+      "step": 954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798169078060909,
+      "loss": 0.8461,
+      "step": 955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017976995573286755,
+      "loss": 0.8618,
+      "step": 956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797229552554906,
+      "loss": 0.9292,
+      "step": 957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017967590640247975,
+      "loss": 0.8833,
+      "step": 958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017962880920238415,
+      "loss": 0.8754,
+      "step": 959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795816636837822,
+      "loss": 0.8964,
+      "step": 960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017953446987528166,
+      "loss": 0.2935,
+      "step": 961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017948722780551958,
+      "loss": 0.861,
+      "step": 962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017943993750316225,
+      "loss": 0.8903,
+      "step": 963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017939259899690536,
+      "loss": 0.9197,
+      "step": 964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017934521231547374,
+      "loss": 0.9164,
+      "step": 965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017929777748762143,
+      "loss": 0.8617,
+      "step": 966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017925029454213178,
+      "loss": 0.8906,
+      "step": 967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792027635078173,
+      "loss": 0.8702,
+      "step": 968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017915518441351957,
+      "loss": 0.9227,
+      "step": 969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017910755728810953,
+      "loss": 0.8805,
+      "step": 970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017905988216048715,
+      "loss": 0.8156,
+      "step": 971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017901215905958152,
+      "loss": 0.8837,
+      "step": 972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017896438801435086,
+      "loss": 0.8931,
+      "step": 973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017891656905378255,
+      "loss": 0.8781,
+      "step": 974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017886870220689292,
+      "loss": 0.8766,
+      "step": 975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017882078750272742,
+      "loss": 0.93,
+      "step": 976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017877282497036058,
+      "loss": 0.8647,
+      "step": 977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017872481463889587,
+      "loss": 0.3629,
+      "step": 978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017867675653746585,
+      "loss": 0.9563,
+      "step": 979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017862865069523195,
+      "loss": 0.9233,
+      "step": 980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017858049714138471,
+      "loss": 0.8685,
+      "step": 981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017853229590514353,
+      "loss": 0.8793,
+      "step": 982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017848404701575676,
+      "loss": 0.8926,
+      "step": 983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017843575050250168,
+      "loss": 0.9092,
+      "step": 984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017838740639468444,
+      "loss": 0.861,
+      "step": 985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017833901472164013,
+      "loss": 0.9031,
+      "step": 986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782905755127326,
+      "loss": 0.8717,
+      "step": 987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017824208879735467,
+      "loss": 0.896,
+      "step": 988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781935546049279,
+      "loss": 0.8817,
+      "step": 989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017814497296490264,
+      "loss": 0.8727,
+      "step": 990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017809634390675814,
+      "loss": 0.9109,
+      "step": 991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017804766746000233,
+      "loss": 0.9074,
+      "step": 992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017799894365417193,
+      "loss": 0.9074,
+      "step": 993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017795017251883235,
+      "loss": 0.878,
+      "step": 994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779013540835778,
+      "loss": 0.9002,
+      "step": 995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017785248837803118,
+      "loss": 0.9128,
+      "step": 996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017780357543184397,
+      "loss": 0.8655,
+      "step": 997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017775461527469642,
+      "loss": 0.8996,
+      "step": 998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001777056079362974,
+      "loss": 0.8992,
+      "step": 999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017765655344638442,
+      "loss": 0.9379,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017760745183472354,
+      "loss": 0.8826,
+      "step": 1001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775583031311095,
+      "loss": 0.3223,
+      "step": 1002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017750910736536558,
+      "loss": 0.305,
+      "step": 1003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017745986456734362,
+      "loss": 0.8595,
+      "step": 1004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017741057476692397,
+      "loss": 0.8703,
+      "step": 1005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017736123799401548,
+      "loss": 0.8913,
+      "step": 1006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017731185427855562,
+      "loss": 0.8845,
+      "step": 1007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017726242365051026,
+      "loss": 0.8695,
+      "step": 1008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017721294613987372,
+      "loss": 0.314,
+      "step": 1009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017716342177666878,
+      "loss": 0.8507,
+      "step": 1010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001771138505909467,
+      "loss": 0.8991,
+      "step": 1011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017706423261278713,
+      "loss": 0.9137,
+      "step": 1012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017701456787229804,
+      "loss": 0.8921,
+      "step": 1013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769648563996159,
+      "loss": 0.8782,
+      "step": 1014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017691509822490543,
+      "loss": 0.8873,
+      "step": 1015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017686529337835975,
+      "loss": 0.9005,
+      "step": 1016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768154418902003,
+      "loss": 0.8615,
+      "step": 1017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017676554379067675,
+      "loss": 0.9276,
+      "step": 1018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017671559911006714,
+      "loss": 0.863,
+      "step": 1019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017666560787867774,
+      "loss": 0.8824,
+      "step": 1020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766155701268431,
+      "loss": 0.8898,
+      "step": 1021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765654858849259,
+      "loss": 0.8886,
+      "step": 1022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765153551833171,
+      "loss": 0.8517,
+      "step": 1023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017646517805243594,
+      "loss": 0.9139,
+      "step": 1024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017641495452272966,
+      "loss": 0.8281,
+      "step": 1025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017636468462467377,
+      "loss": 0.8639,
+      "step": 1026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017631436838877189,
+      "loss": 0.3527,
+      "step": 1027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762640058455557,
+      "loss": 0.3315,
+      "step": 1028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762135970255851,
+      "loss": 0.8541,
+      "step": 1029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017616314195944795,
+      "loss": 0.8776,
+      "step": 1030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017611264067776024,
+      "loss": 0.8712,
+      "step": 1031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017606209321116597,
+      "loss": 0.8673,
+      "step": 1032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760114995903372,
+      "loss": 0.8884,
+      "step": 1033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017596085984597398,
+      "loss": 0.8833,
+      "step": 1034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017591017400880432,
+      "loss": 0.9109,
+      "step": 1035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017585944210958426,
+      "loss": 0.891,
+      "step": 1036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017580866417909774,
+      "loss": 0.913,
+      "step": 1037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757578402481566,
+      "loss": 0.9334,
+      "step": 1038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017570697034760072,
+      "loss": 0.8418,
+      "step": 1039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017565605450829775,
+      "loss": 0.8432,
+      "step": 1040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017560509276114323,
+      "loss": 0.9042,
+      "step": 1041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017555408513706064,
+      "loss": 0.8929,
+      "step": 1042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001755030316670012,
+      "loss": 0.8694,
+      "step": 1043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175451932381944,
+      "loss": 0.8973,
+      "step": 1044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017540078731289596,
+      "loss": 0.8785,
+      "step": 1045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001753495964908917,
+      "loss": 0.8518,
+      "step": 1046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017529835994699365,
+      "loss": 0.9092,
+      "step": 1047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017524707771229202,
+      "loss": 0.8726,
+      "step": 1048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017519574981790466,
+      "loss": 0.8647,
+      "step": 1049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751443762949772,
+      "loss": 0.8855,
+      "step": 1050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017509295717468292,
+      "loss": 0.8561,
+      "step": 1051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001750414924882228,
+      "loss": 0.9041,
+      "step": 1052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017498998226682537,
+      "loss": 0.8761,
+      "step": 1053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017493842654174693,
+      "loss": 0.9045,
+      "step": 1054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017488682534427135,
+      "loss": 0.8953,
+      "step": 1055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017483517870571004,
+      "loss": 0.8892,
+      "step": 1056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017478348665740204,
+      "loss": 0.894,
+      "step": 1057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017473174923071384,
+      "loss": 0.8902,
+      "step": 1058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017467996645703963,
+      "loss": 0.8758,
+      "step": 1059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017462813836780102,
+      "loss": 0.8711,
+      "step": 1060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017457626499444708,
+      "loss": 0.8671,
+      "step": 1061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017452434636845444,
+      "loss": 0.9256,
+      "step": 1062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017447238252132715,
+      "loss": 0.932,
+      "step": 1063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744203734845967,
+      "loss": 0.896,
+      "step": 1064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017436831928982203,
+      "loss": 0.8633,
+      "step": 1065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017431621996858945,
+      "loss": 0.8488,
+      "step": 1066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017426407555251266,
+      "loss": 0.8461,
+      "step": 1067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001742118860732327,
+      "loss": 0.921,
+      "step": 1068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017415965156241802,
+      "loss": 0.895,
+      "step": 1069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017410737205176432,
+      "loss": 0.891,
+      "step": 1070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017405504757299467,
+      "loss": 0.8778,
+      "step": 1071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017400267815785938,
+      "loss": 0.9394,
+      "step": 1072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017395026383813602,
+      "loss": 0.8968,
+      "step": 1073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017389780464562946,
+      "loss": 0.8831,
+      "step": 1074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017384530061217177,
+      "loss": 0.8762,
+      "step": 1075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017379275176962223,
+      "loss": 0.8519,
+      "step": 1076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017374015814986725,
+      "loss": 0.3212,
+      "step": 1077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017368751978482058,
+      "loss": 0.8856,
+      "step": 1078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017363483670642294,
+      "loss": 0.3441,
+      "step": 1079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017358210894664227,
+      "loss": 0.8984,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017352933653747357,
+      "loss": 0.8917,
+      "step": 1081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017347651951093903,
+      "loss": 0.8684,
+      "step": 1082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017342365789908785,
+      "loss": 0.8812,
+      "step": 1083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733707517339962,
+      "loss": 0.905,
+      "step": 1084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733178010477675,
+      "loss": 0.9014,
+      "step": 1085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017326480587253194,
+      "loss": 0.8637,
+      "step": 1086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017321176624044687,
+      "loss": 0.8632,
+      "step": 1087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017315868218369657,
+      "loss": 0.8826,
+      "step": 1088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731055537344923,
+      "loss": 0.9185,
+      "step": 1089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017305238092507218,
+      "loss": 0.9099,
+      "step": 1090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017299916378770135,
+      "loss": 0.8756,
+      "step": 1091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017294590235467178,
+      "loss": 0.8869,
+      "step": 1092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017289259665830234,
+      "loss": 0.8847,
+      "step": 1093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017283924673093873,
+      "loss": 0.8654,
+      "step": 1094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017278585260495355,
+      "loss": 0.8461,
+      "step": 1095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017273241431274615,
+      "loss": 0.8757,
+      "step": 1096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017267893188674274,
+      "loss": 0.9297,
+      "step": 1097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726254053593963,
+      "loss": 0.3545,
+      "step": 1098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725718347631865,
+      "loss": 0.8901,
+      "step": 1099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017251822013061987,
+      "loss": 0.8752,
+      "step": 1100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017246456149422955,
+      "loss": 0.9222,
+      "step": 1101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017241085888657548,
+      "loss": 0.8701,
+      "step": 1102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017235711234024414,
+      "loss": 0.8415,
+      "step": 1103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017230332188784885,
+      "loss": 0.8693,
+      "step": 1104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017224948756202948,
+      "loss": 0.8453,
+      "step": 1105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017219560939545246,
+      "loss": 0.8811,
+      "step": 1106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017214168742081092,
+      "loss": 0.8628,
+      "step": 1107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017208772167082457,
+      "loss": 0.937,
+      "step": 1108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017203371217823965,
+      "loss": 0.8427,
+      "step": 1109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719796589758289,
+      "loss": 0.8565,
+      "step": 1110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017192556209639167,
+      "loss": 0.3458,
+      "step": 1111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001718714215727538,
+      "loss": 0.9148,
+      "step": 1112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017181723743776753,
+      "loss": 0.8365,
+      "step": 1113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717630097243116,
+      "loss": 0.9217,
+      "step": 1114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717087384652913,
+      "loss": 0.9442,
+      "step": 1115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716544236936382,
+      "loss": 0.9311,
+      "step": 1116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017160006544231034,
+      "loss": 0.8953,
+      "step": 1117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017154566374429214,
+      "loss": 0.8805,
+      "step": 1118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017149121863259433,
+      "loss": 0.9042,
+      "step": 1119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714367301402541,
+      "loss": 0.8805,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017138219830033485,
+      "loss": 0.9161,
+      "step": 1121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001713276231459263,
+      "loss": 0.8892,
+      "step": 1122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017127300471014454,
+      "loss": 0.909,
+      "step": 1123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017121834302613186,
+      "loss": 0.8687,
+      "step": 1124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017116363812705678,
+      "loss": 0.8761,
+      "step": 1125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017110889004611402,
+      "loss": 0.88,
+      "step": 1126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017105409881652456,
+      "loss": 0.8956,
+      "step": 1127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017099926447153557,
+      "loss": 0.8531,
+      "step": 1128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017094438704442033,
+      "loss": 0.3301,
+      "step": 1129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001708894665684783,
+      "loss": 0.3337,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017083450307703502,
+      "loss": 0.8926,
+      "step": 1131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707794966034422,
+      "loss": 0.834,
+      "step": 1132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017072444718107757,
+      "loss": 0.8783,
+      "step": 1133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017066935484334495,
+      "loss": 0.3431,
+      "step": 1134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017061421962367418,
+      "loss": 0.8461,
+      "step": 1135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017055904155552115,
+      "loss": 0.8964,
+      "step": 1136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017050382067236775,
+      "loss": 0.855,
+      "step": 1137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704485570077218,
+      "loss": 0.9207,
+      "step": 1138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017039325059511713,
+      "loss": 0.8411,
+      "step": 1139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017033790146811351,
+      "loss": 0.8409,
+      "step": 1140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017028250966029658,
+      "loss": 0.875,
+      "step": 1141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017022707520527794,
+      "loss": 0.9416,
+      "step": 1142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017017159813669503,
+      "loss": 0.8959,
+      "step": 1143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017011607848821115,
+      "loss": 0.8408,
+      "step": 1144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017006051629351546,
+      "loss": 0.864,
+      "step": 1145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017000491158632291,
+      "loss": 0.9065,
+      "step": 1146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016994926440037426,
+      "loss": 0.875,
+      "step": 1147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016989357476943605,
+      "loss": 0.8566,
+      "step": 1148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016983784272730053,
+      "loss": 0.8618,
+      "step": 1149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001697820683077858,
+      "loss": 0.9049,
+      "step": 1150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016972625154473554,
+      "loss": 0.8282,
+      "step": 1151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696703924720192,
+      "loss": 0.8376,
+      "step": 1152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696144911235319,
+      "loss": 0.854,
+      "step": 1153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001695585475331944,
+      "loss": 0.8943,
+      "step": 1154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016950256173495307,
+      "loss": 0.8807,
+      "step": 1155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016944653376277992,
+      "loss": 0.9272,
+      "step": 1156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016939046365067255,
+      "loss": 0.8372,
+      "step": 1157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001693343514326541,
+      "loss": 0.922,
+      "step": 1158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692781971427733,
+      "loss": 0.8457,
+      "step": 1159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016922200081510443,
+      "loss": 0.8975,
+      "step": 1160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016916576248374719,
+      "loss": 0.8797,
+      "step": 1161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016910948218282684,
+      "loss": 0.8969,
+      "step": 1162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690531599464941,
+      "loss": 0.8953,
+      "step": 1163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016899679580892508,
+      "loss": 0.8522,
+      "step": 1164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001689403898043214,
+      "loss": 0.8461,
+      "step": 1165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016888394196691003,
+      "loss": 0.8994,
+      "step": 1166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016882745233094335,
+      "loss": 0.8642,
+      "step": 1167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016877092093069905,
+      "loss": 0.8536,
+      "step": 1168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016871434780048027,
+      "loss": 0.9195,
+      "step": 1169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001686577329746153,
+      "loss": 0.9321,
+      "step": 1170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016860107648745796,
+      "loss": 0.9326,
+      "step": 1171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685443783733871,
+      "loss": 0.958,
+      "step": 1172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016848763866680706,
+      "loss": 0.9261,
+      "step": 1173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016843085740214729,
+      "loss": 0.8564,
+      "step": 1174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016837403461386242,
+      "loss": 0.9187,
+      "step": 1175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016831717033643237,
+      "loss": 0.8734,
+      "step": 1176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682602646043622,
+      "loss": 0.87,
+      "step": 1177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016820331745218215,
+      "loss": 0.88,
+      "step": 1178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016814632891444756,
+      "loss": 0.9214,
+      "step": 1179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016808929902573886,
+      "loss": 0.3232,
+      "step": 1180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016803222782066165,
+      "loss": 0.8874,
+      "step": 1181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679751153338465,
+      "loss": 0.8611,
+      "step": 1182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016791796159994915,
+      "loss": 0.8872,
+      "step": 1183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016786076665365024,
+      "loss": 0.8833,
+      "step": 1184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678035305296555,
+      "loss": 0.3078,
+      "step": 1185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016774625326269568,
+      "loss": 0.8701,
+      "step": 1186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016768893488752637,
+      "loss": 0.8814,
+      "step": 1187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001676315754389282,
+      "loss": 0.8815,
+      "step": 1188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016757417495170672,
+      "loss": 0.8105,
+      "step": 1189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016751673346069231,
+      "loss": 0.8878,
+      "step": 1190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016745925100074036,
+      "loss": 0.8799,
+      "step": 1191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016740172760673096,
+      "loss": 0.8615,
+      "step": 1192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001673441633135692,
+      "loss": 0.8384,
+      "step": 1193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016728655815618484,
+      "loss": 0.8814,
+      "step": 1194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016722891216953257,
+      "loss": 0.9452,
+      "step": 1195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016717122538859172,
+      "loss": 0.8849,
+      "step": 1196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016711349784836647,
+      "loss": 0.854,
+      "step": 1197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016705572958388576,
+      "loss": 0.8662,
+      "step": 1198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669979206302031,
+      "loss": 0.8581,
+      "step": 1199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016694007102239684,
+      "loss": 0.9018,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016688218079556993,
+      "loss": 0.8918,
+      "step": 1201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016682424998484995,
+      "loss": 0.876,
+      "step": 1202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001667662786253891,
+      "loss": 0.8644,
+      "step": 1203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016670826675236434,
+      "loss": 0.9131,
+      "step": 1204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016665021440097694,
+      "loss": 0.9201,
+      "step": 1205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016659212160645298,
+      "loss": 0.8999,
+      "step": 1206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016653398840404292,
+      "loss": 0.8674,
+      "step": 1207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016647581482902188,
+      "loss": 0.8565,
+      "step": 1208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664176009166893,
+      "loss": 0.9007,
+      "step": 1209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016635934670236927,
+      "loss": 0.9054,
+      "step": 1210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016630105222141024,
+      "loss": 0.9135,
+      "step": 1211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662427175091851,
+      "loss": 0.9265,
+      "step": 1212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016618434260109118,
+      "loss": 0.8898,
+      "step": 1213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016612592753255013,
+      "loss": 0.8564,
+      "step": 1214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016606747233900815,
+      "loss": 0.871,
+      "step": 1215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016600897705593558,
+      "loss": 0.84,
+      "step": 1216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016595044171882717,
+      "loss": 0.8198,
+      "step": 1217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000165891866363202,
+      "loss": 0.8432,
+      "step": 1218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016583325102460336,
+      "loss": 0.8973,
+      "step": 1219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016577459573859896,
+      "loss": 0.8915,
+      "step": 1220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001657159005407805,
+      "loss": 0.8641,
+      "step": 1221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016565716546676411,
+      "loss": 0.8246,
+      "step": 1222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016559839055219007,
+      "loss": 0.9333,
+      "step": 1223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553957583272276,
+      "loss": 0.8856,
+      "step": 1224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016548072134405078,
+      "loss": 0.8809,
+      "step": 1225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016542182712188686,
+      "loss": 0.8686,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001653628932019678,
+      "loss": 0.9111,
+      "step": 1227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016530391962005453,
+      "loss": 0.8552,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016524490641193205,
+      "loss": 0.8857,
+      "step": 1229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016518585361340935,
+      "loss": 0.882,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001651267612603195,
+      "loss": 0.874,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016506762938851952,
+      "loss": 0.8493,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016500845803389045,
+      "loss": 0.8822,
+      "step": 1233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016494924723233734,
+      "loss": 0.8508,
+      "step": 1234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016488999701978903,
+      "loss": 0.9195,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648307074321984,
+      "loss": 0.9308,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647713785055422,
+      "loss": 0.89,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016471201027582098,
+      "loss": 0.3402,
+      "step": 1238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016465260277905925,
+      "loss": 0.8984,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016459315605130525,
+      "loss": 0.9138,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016453367012863107,
+      "loss": 0.9158,
+      "step": 1241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001644741450471326,
+      "loss": 0.8932,
+      "step": 1242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016441458084292943,
+      "loss": 0.912,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016435497755216494,
+      "loss": 0.8213,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016429533521100624,
+      "loss": 0.8834,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016423565385564407,
+      "loss": 0.9062,
+      "step": 1246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016417593352229293,
+      "loss": 0.908,
+      "step": 1247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016411617424719093,
+      "loss": 0.9337,
+      "step": 1248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016405637606659975,
+      "loss": 0.8845,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001639965390168048,
+      "loss": 0.8442,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016393666313411495,
+      "loss": 0.8669,
+      "step": 1251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016387674845486276,
+      "loss": 0.8413,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016381679501540418,
+      "loss": 0.9412,
+      "step": 1253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016375680285211886,
+      "loss": 0.9148,
+      "step": 1254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016369677200140978,
+      "loss": 0.8515,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001636367024997035,
+      "loss": 0.8862,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016357659438345,
+      "loss": 0.8763,
+      "step": 1257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016351644768912267,
+      "loss": 0.8578,
+      "step": 1258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016345626245321837,
+      "loss": 0.9217,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016339603871225728,
+      "loss": 0.8609,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016333577650278297,
+      "loss": 0.8681,
+      "step": 1261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001632754758613624,
+      "loss": 0.8521,
+      "step": 1262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016321513682458573,
+      "loss": 0.8798,
+      "step": 1263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631547594290666,
+      "loss": 0.8443,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630943437114417,
+      "loss": 0.8576,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016303388970837114,
+      "loss": 0.9126,
+      "step": 1266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016297339745653826,
+      "loss": 0.9361,
+      "step": 1267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016291286699264955,
+      "loss": 0.8836,
+      "step": 1268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016285229835343463,
+      "loss": 0.3169,
+      "step": 1269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016279169157564643,
+      "loss": 0.8942,
+      "step": 1270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016273104669606094,
+      "loss": 0.8649,
+      "step": 1271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016267036375147725,
+      "loss": 0.8395,
+      "step": 1272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001626096427787176,
+      "loss": 0.8332,
+      "step": 1273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016254888381462727,
+      "loss": 0.8424,
+      "step": 1274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624880868960746,
+      "loss": 0.8904,
+      "step": 1275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162427252059951,
+      "loss": 0.8792,
+      "step": 1276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623663793431708,
+      "loss": 0.3112,
+      "step": 1277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623054687826714,
+      "loss": 0.8641,
+      "step": 1278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016224452041541316,
+      "loss": 0.8763,
+      "step": 1279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016218353427837929,
+      "loss": 0.8714,
+      "step": 1280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016212251040857606,
+      "loss": 0.9269,
+      "step": 1281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001620614488430325,
+      "loss": 0.9004,
+      "step": 1282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016200034961880055,
+      "loss": 0.8572,
+      "step": 1283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619392127729551,
+      "loss": 0.9296,
+      "step": 1284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016187803834259373,
+      "loss": 0.9187,
+      "step": 1285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016181682636483692,
+      "loss": 0.8616,
+      "step": 1286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617555768768279,
+      "loss": 0.8994,
+      "step": 1287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016169428991573268,
+      "loss": 0.8465,
+      "step": 1288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016163296551873995,
+      "loss": 0.837,
+      "step": 1289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016157160372306116,
+      "loss": 0.9333,
+      "step": 1290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016151020456593046,
+      "loss": 0.8949,
+      "step": 1291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614487680846047,
+      "loss": 0.819,
+      "step": 1292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001613872943163633,
+      "loss": 0.8476,
+      "step": 1293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016132578329850836,
+      "loss": 0.844,
+      "step": 1294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016126423506836453,
+      "loss": 0.7886,
+      "step": 1295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016120264966327917,
+      "loss": 0.8897,
+      "step": 1296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000161141027120622,
+      "loss": 0.8547,
+      "step": 1297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016107936747778546,
+      "loss": 0.7929,
+      "step": 1298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016101767077218439,
+      "loss": 0.8203,
+      "step": 1299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016095593704125615,
+      "loss": 0.8099,
+      "step": 1300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608941663224606,
+      "loss": 0.9076,
+      "step": 1301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016083235865327997,
+      "loss": 0.9157,
+      "step": 1302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016077051407121896,
+      "loss": 0.8988,
+      "step": 1303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001607086326138047,
+      "loss": 0.8884,
+      "step": 1304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016064671431858667,
+      "loss": 0.8729,
+      "step": 1305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016058475922313664,
+      "loss": 0.9164,
+      "step": 1306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016052276736504877,
+      "loss": 0.8629,
+      "step": 1307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604607387819396,
+      "loss": 0.8183,
+      "step": 1308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016039867351144778,
+      "loss": 0.8522,
+      "step": 1309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016033657159123435,
+      "loss": 0.8578,
+      "step": 1310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016027443305898254,
+      "loss": 0.8148,
+      "step": 1311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016021225795239785,
+      "loss": 0.8441,
+      "step": 1312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016015004630920792,
+      "loss": 0.8628,
+      "step": 1313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016008779816716257,
+      "loss": 0.8809,
+      "step": 1314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600255135640338,
+      "loss": 0.3335,
+      "step": 1315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015996319253761562,
+      "loss": 0.8258,
+      "step": 1316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015990083512572436,
+      "loss": 0.9093,
+      "step": 1317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015983844136619818,
+      "loss": 0.8578,
+      "step": 1318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597760112968975,
+      "loss": 0.8773,
+      "step": 1319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015971354495570466,
+      "loss": 0.8776,
+      "step": 1320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015965104238052397,
+      "loss": 0.8636,
+      "step": 1321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595885036092819,
+      "loss": 0.8719,
+      "step": 1322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595259286799267,
+      "loss": 0.8763,
+      "step": 1323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015946331763042867,
+      "loss": 0.8502,
+      "step": 1324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015940067049877998,
+      "loss": 0.9329,
+      "step": 1325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015933798732299472,
+      "loss": 0.8825,
+      "step": 1326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015927526814110883,
+      "loss": 0.895,
+      "step": 1327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015921251299118013,
+      "loss": 0.8743,
+      "step": 1328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015914972191128818,
+      "loss": 0.8925,
+      "step": 1329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590868949395345,
+      "loss": 0.3407,
+      "step": 1330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015902403211404223,
+      "loss": 0.8824,
+      "step": 1331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015896113347295635,
+      "loss": 0.7992,
+      "step": 1332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015889819905444354,
+      "loss": 0.9307,
+      "step": 1333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015883522889669226,
+      "loss": 0.8529,
+      "step": 1334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001587722230379125,
+      "loss": 0.9256,
+      "step": 1335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001587091815163361,
+      "loss": 0.8534,
+      "step": 1336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586461043702164,
+      "loss": 0.8915,
+      "step": 1337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015858299163782849,
+      "loss": 0.8194,
+      "step": 1338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015851984335746888,
+      "loss": 0.8533,
+      "step": 1339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015845665956745576,
+      "loss": 0.8954,
+      "step": 1340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001583934403061289,
+      "loss": 0.8246,
+      "step": 1341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015833018561184952,
+      "loss": 0.8512,
+      "step": 1342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015826689552300035,
+      "loss": 0.866,
+      "step": 1343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015820357007798565,
+      "loss": 0.8534,
+      "step": 1344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015814020931523106,
+      "loss": 0.8579,
+      "step": 1345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580768132731837,
+      "loss": 0.9067,
+      "step": 1346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580133819903121,
+      "loss": 0.8222,
+      "step": 1347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001579499155051061,
+      "loss": 0.8863,
+      "step": 1348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015788641385607708,
+      "loss": 0.8684,
+      "step": 1349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015782287708175754,
+      "loss": 0.8339,
+      "step": 1350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015775930522070138,
+      "loss": 0.8272,
+      "step": 1351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015769569831148385,
+      "loss": 0.8342,
+      "step": 1352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015763205639270143,
+      "loss": 0.8907,
+      "step": 1353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001575683795029718,
+      "loss": 0.818,
+      "step": 1354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015750466768093389,
+      "loss": 0.8716,
+      "step": 1355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015744092096524784,
+      "loss": 0.856,
+      "step": 1356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015737713939459496,
+      "loss": 0.9309,
+      "step": 1357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001573133230076777,
+      "loss": 0.8383,
+      "step": 1358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015724947184321962,
+      "loss": 0.8703,
+      "step": 1359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015718558593996545,
+      "loss": 0.3367,
+      "step": 1360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015712166533668088,
+      "loss": 0.9032,
+      "step": 1361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015705771007215277,
+      "loss": 0.8436,
+      "step": 1362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015699372018518898,
+      "loss": 0.8871,
+      "step": 1363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015692969571461836,
+      "loss": 0.885,
+      "step": 1364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015686563669929074,
+      "loss": 0.8406,
+      "step": 1365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015680154317807693,
+      "loss": 0.9171,
+      "step": 1366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015673741518986865,
+      "loss": 0.8395,
+      "step": 1367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015667325277357862,
+      "loss": 0.8853,
+      "step": 1368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015660905596814035,
+      "loss": 0.3317,
+      "step": 1369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015654482481250825,
+      "loss": 0.8802,
+      "step": 1370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015648055934565757,
+      "loss": 0.881,
+      "step": 1371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015641625960658443,
+      "loss": 0.8769,
+      "step": 1372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001563519256343057,
+      "loss": 0.8508,
+      "step": 1373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156287557467859,
+      "loss": 0.8844,
+      "step": 1374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001562231551463027,
+      "loss": 0.9051,
+      "step": 1375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156158718708716,
+      "loss": 0.9137,
+      "step": 1376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001560942481941987,
+      "loss": 0.911,
+      "step": 1377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015602974364187127,
+      "loss": 0.3299,
+      "step": 1378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015596520509087485,
+      "loss": 0.838,
+      "step": 1379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001559006325803713,
+      "loss": 0.8432,
+      "step": 1380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015583602614954294,
+      "loss": 0.8704,
+      "step": 1381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015577138583759277,
+      "loss": 0.8861,
+      "step": 1382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015570671168374438,
+      "loss": 0.8648,
+      "step": 1383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015564200372724177,
+      "loss": 0.8715,
+      "step": 1384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015557726200734952,
+      "loss": 0.8702,
+      "step": 1385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015551248656335274,
+      "loss": 0.8445,
+      "step": 1386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015544767743455693,
+      "loss": 0.8887,
+      "step": 1387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015538283466028807,
+      "loss": 0.8921,
+      "step": 1388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015531795827989256,
+      "loss": 0.8804,
+      "step": 1389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015525304833273714,
+      "loss": 0.8677,
+      "step": 1390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015518810485820904,
+      "loss": 0.8031,
+      "step": 1391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015512312789571565,
+      "loss": 0.8766,
+      "step": 1392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015505811748468487,
+      "loss": 0.3304,
+      "step": 1393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015499307366456473,
+      "loss": 0.8762,
+      "step": 1394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549279964748237,
+      "loss": 0.8827,
+      "step": 1395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015486288595495035,
+      "loss": 0.8982,
+      "step": 1396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015479774214445353,
+      "loss": 0.8979,
+      "step": 1397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015473256508286235,
+      "loss": 0.8462,
+      "step": 1398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000154667354809726,
+      "loss": 0.8582,
+      "step": 1399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015460211136461381,
+      "loss": 0.855,
+      "step": 1400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015453683478711542,
+      "loss": 0.917,
+      "step": 1401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015447152511684036,
+      "loss": 0.3507,
+      "step": 1402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015440618239341837,
+      "loss": 0.8799,
+      "step": 1403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015434080665649918,
+      "loss": 0.8553,
+      "step": 1404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015427539794575257,
+      "loss": 0.9012,
+      "step": 1405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015420995630086834,
+      "loss": 0.853,
+      "step": 1406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015414448176155631,
+      "loss": 0.9219,
+      "step": 1407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015407897436754619,
+      "loss": 0.8499,
+      "step": 1408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015401343415858768,
+      "loss": 0.8914,
+      "step": 1409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001539478611744503,
+      "loss": 0.8276,
+      "step": 1410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015388225545492356,
+      "loss": 0.8683,
+      "step": 1411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015381661703981684,
+      "loss": 0.8811,
+      "step": 1412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015375094596895933,
+      "loss": 0.8611,
+      "step": 1413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015368524228219998,
+      "loss": 0.9088,
+      "step": 1414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015361950601940755,
+      "loss": 0.896,
+      "step": 1415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015355373722047064,
+      "loss": 0.9222,
+      "step": 1416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015348793592529758,
+      "loss": 0.8784,
+      "step": 1417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001534221021738163,
+      "loss": 0.9032,
+      "step": 1418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015335623600597456,
+      "loss": 0.3573,
+      "step": 1419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015329033746173975,
+      "loss": 0.8158,
+      "step": 1420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015322440658109888,
+      "loss": 0.3208,
+      "step": 1421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015315844340405853,
+      "loss": 0.8724,
+      "step": 1422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015309244797064508,
+      "loss": 0.8891,
+      "step": 1423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015302642032090426,
+      "loss": 0.8644,
+      "step": 1424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015296036049490143,
+      "loss": 0.8607,
+      "step": 1425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015289426853272152,
+      "loss": 0.8308,
+      "step": 1426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015282814447446888,
+      "loss": 0.8861,
+      "step": 1427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015276198836026747,
+      "loss": 0.3433,
+      "step": 1428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001526958002302605,
+      "loss": 0.8727,
+      "step": 1429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015262958012461072,
+      "loss": 0.8902,
+      "step": 1430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015256332808350035,
+      "loss": 0.8457,
+      "step": 1431
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015249704414713086,
+      "loss": 0.8746,
+      "step": 1432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015243072835572318,
+      "loss": 0.9231,
+      "step": 1433
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015236438074951747,
+      "loss": 0.8706,
+      "step": 1434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015229800136877327,
+      "loss": 0.8763,
+      "step": 1435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015223159025376935,
+      "loss": 0.8965,
+      "step": 1436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001521651474448038,
+      "loss": 0.3445,
+      "step": 1437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015209867298219388,
+      "loss": 0.8665,
+      "step": 1438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520321669062761,
+      "loss": 0.892,
+      "step": 1439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015196562925740608,
+      "loss": 0.8246,
+      "step": 1440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001518990600759587,
+      "loss": 0.8419,
+      "step": 1441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015183245940232797,
+      "loss": 0.305,
+      "step": 1442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015176582727692686,
+      "loss": 0.8634,
+      "step": 1443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001516991637401876,
+      "loss": 0.9341,
+      "step": 1444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015163246883256138,
+      "loss": 0.8651,
+      "step": 1445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515657425945185,
+      "loss": 0.8824,
+      "step": 1446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514989850665482,
+      "loss": 0.907,
+      "step": 1447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514321962891587,
+      "loss": 0.9154,
+      "step": 1448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015136537630287727,
+      "loss": 0.8833,
+      "step": 1449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015129852514825004,
+      "loss": 0.905,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512316428658421,
+      "loss": 0.7762,
+      "step": 1451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015116472949623733,
+      "loss": 0.8598,
+      "step": 1452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015109778508003865,
+      "loss": 0.8671,
+      "step": 1453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015103080965786766,
+      "loss": 0.8711,
+      "step": 1454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001509638032703648,
+      "loss": 0.8532,
+      "step": 1455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508967659581894,
+      "loss": 0.8377,
+      "step": 1456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015082969776201947,
+      "loss": 0.8395,
+      "step": 1457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015076259872255174,
+      "loss": 0.8277,
+      "step": 1458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015069546888050173,
+      "loss": 0.8738,
+      "step": 1459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001506283082766036,
+      "loss": 0.8927,
+      "step": 1460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015056111695161022,
+      "loss": 0.8786,
+      "step": 1461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000150493894946293,
+      "loss": 0.8538,
+      "step": 1462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015042664230144207,
+      "loss": 0.8867,
+      "step": 1463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015035935905786615,
+      "loss": 0.307,
+      "step": 1464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015029204525639247,
+      "loss": 0.8682,
+      "step": 1465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015022470093786682,
+      "loss": 0.8392,
+      "step": 1466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015015732614315355,
+      "loss": 0.8169,
+      "step": 1467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015008992091313544,
+      "loss": 0.8805,
+      "step": 1468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015002248528871376,
+      "loss": 0.8972,
+      "step": 1469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014995501931080828,
+      "loss": 0.348,
+      "step": 1470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001498875230203571,
+      "loss": 0.8394,
+      "step": 1471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014981999645831675,
+      "loss": 0.8684,
+      "step": 1472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014975243966566213,
+      "loss": 0.8875,
+      "step": 1473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001496848526833865,
+      "loss": 0.9038,
+      "step": 1474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014961723555250144,
+      "loss": 0.8451,
+      "step": 1475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014954958831403677,
+      "loss": 0.8615,
+      "step": 1476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014948191100904067,
+      "loss": 0.8876,
+      "step": 1477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014941420367857943,
+      "loss": 0.8997,
+      "step": 1478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014934646636373774,
+      "loss": 0.8402,
+      "step": 1479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492786991056183,
+      "loss": 0.8408,
+      "step": 1480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492109019453421,
+      "loss": 0.8665,
+      "step": 1481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014914307492404827,
+      "loss": 0.8961,
+      "step": 1482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014907521808289398,
+      "loss": 0.8833,
+      "step": 1483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014900733146305453,
+      "loss": 0.8634,
+      "step": 1484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014893941510572337,
+      "loss": 0.3361,
+      "step": 1485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014887146905211192,
+      "loss": 0.8463,
+      "step": 1486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488034933434496,
+      "loss": 0.8756,
+      "step": 1487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014873548802098383,
+      "loss": 0.8219,
+      "step": 1488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014866745312598002,
+      "loss": 0.8102,
+      "step": 1489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001485993886997216,
+      "loss": 0.8925,
+      "step": 1490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001485312947835098,
+      "loss": 0.8386,
+      "step": 1491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014846317141866376,
+      "loss": 0.8876,
+      "step": 1492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014839501864652056,
+      "loss": 0.8606,
+      "step": 1493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483268365084351,
+      "loss": 0.9063,
+      "step": 1494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014825862504578002,
+      "loss": 0.8574,
+      "step": 1495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014819038429994585,
+      "loss": 0.8728,
+      "step": 1496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014812211431234089,
+      "loss": 0.908,
+      "step": 1497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001480538151243911,
+      "loss": 0.8709,
+      "step": 1498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001479854867775402,
+      "loss": 0.3088,
+      "step": 1499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014791712931324966,
+      "loss": 0.8611,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014784874277299858,
+      "loss": 0.8526,
+      "step": 1501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014778032719828363,
+      "loss": 0.929,
+      "step": 1502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001477118826306192,
+      "loss": 0.8678,
+      "step": 1503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014764340911153725,
+      "loss": 0.8607,
+      "step": 1504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014757490668258725,
+      "loss": 0.869,
+      "step": 1505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014750637538533628,
+      "loss": 0.8615,
+      "step": 1506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014743781526136888,
+      "loss": 0.884,
+      "step": 1507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014736922635228718,
+      "loss": 0.8829,
+      "step": 1508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014730060869971065,
+      "loss": 0.3384,
+      "step": 1509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472319623452763,
+      "loss": 0.8615,
+      "step": 1510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014716328733063846,
+      "loss": 0.8617,
+      "step": 1511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000147094583697469,
+      "loss": 0.8715,
+      "step": 1512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000147025851487457,
+      "loss": 0.8781,
+      "step": 1513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014695709074230895,
+      "loss": 0.8629,
+      "step": 1514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014688830150374866,
+      "loss": 0.9013,
+      "step": 1515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014681948381351724,
+      "loss": 0.8733,
+      "step": 1516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014675063771337296,
+      "loss": 0.3254,
+      "step": 1517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014668176324509152,
+      "loss": 0.843,
+      "step": 1518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466128604504657,
+      "loss": 0.8306,
+      "step": 1519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014654392937130543,
+      "loss": 0.8895,
+      "step": 1520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014647497004943795,
+      "loss": 0.8311,
+      "step": 1521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001464059825267075,
+      "loss": 0.9098,
+      "step": 1522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014633696684497553,
+      "loss": 0.8377,
+      "step": 1523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014626792304612052,
+      "loss": 0.8617,
+      "step": 1524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000146198851172038,
+      "loss": 0.8451,
+      "step": 1525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014612975126464058,
+      "loss": 0.896,
+      "step": 1526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014606062336585787,
+      "loss": 0.861,
+      "step": 1527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014599146751763644,
+      "loss": 0.8497,
+      "step": 1528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014592228376193982,
+      "loss": 0.8781,
+      "step": 1529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458530721407486,
+      "loss": 0.8705,
+      "step": 1530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014578383269606005,
+      "loss": 0.8965,
+      "step": 1531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014571456546988846,
+      "loss": 0.7998,
+      "step": 1532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014564527050426503,
+      "loss": 0.8745,
+      "step": 1533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014557594784123763,
+      "loss": 0.8945,
+      "step": 1534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014550659752287108,
+      "loss": 0.8743,
+      "step": 1535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014543721959124693,
+      "loss": 0.851,
+      "step": 1536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001453678140884635,
+      "loss": 0.8435,
+      "step": 1537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014529838105663582,
+      "loss": 0.3504,
+      "step": 1538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001452289205378956,
+      "loss": 0.8647,
+      "step": 1539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451594325743913,
+      "loss": 0.8174,
+      "step": 1540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000145089917208288,
+      "loss": 0.8658,
+      "step": 1541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014502037448176734,
+      "loss": 0.8747,
+      "step": 1542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014495080443702768,
+      "loss": 0.9182,
+      "step": 1543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014488120711628386,
+      "loss": 0.8134,
+      "step": 1544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014481158256176733,
+      "loss": 0.8567,
+      "step": 1545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014474193081572608,
+      "loss": 0.3065,
+      "step": 1546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446722519204245,
+      "loss": 0.8605,
+      "step": 1547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446025459181436,
+      "loss": 0.841,
+      "step": 1548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001445328128511807,
+      "loss": 0.8587,
+      "step": 1549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001444630527618496,
+      "loss": 0.8737,
+      "step": 1550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014439326569248053,
+      "loss": 0.8525,
+      "step": 1551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014432345168542004,
+      "loss": 0.8343,
+      "step": 1552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014425361078303106,
+      "loss": 0.3099,
+      "step": 1553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001441837430276928,
+      "loss": 0.8366,
+      "step": 1554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014411384846180084,
+      "loss": 0.8616,
+      "step": 1555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014404392712776692,
+      "loss": 0.898,
+      "step": 1556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439739790680191,
+      "loss": 0.3439,
+      "step": 1557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014390400432500166,
+      "loss": 0.8778,
+      "step": 1558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014383400294117507,
+      "loss": 0.8329,
+      "step": 1559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014376397495901588,
+      "loss": 0.8972,
+      "step": 1560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014369392042101685,
+      "loss": 0.8334,
+      "step": 1561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001436238393696869,
+      "loss": 0.8447,
+      "step": 1562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000143553731847551,
+      "loss": 0.8544,
+      "step": 1563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014348359789715008,
+      "loss": 0.9062,
+      "step": 1564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014341343756104128,
+      "loss": 0.9088,
+      "step": 1565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014334325088179764,
+      "loss": 0.8628,
+      "step": 1566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001432730379020082,
+      "loss": 0.8542,
+      "step": 1567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014320279866427796,
+      "loss": 0.8618,
+      "step": 1568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001431325332112279,
+      "loss": 0.8487,
+      "step": 1569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430622415854949,
+      "loss": 0.8293,
+      "step": 1570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001429919238297316,
+      "loss": 0.9071,
+      "step": 1571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014292157998660666,
+      "loss": 0.9395,
+      "step": 1572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014285121009880445,
+      "loss": 0.8462,
+      "step": 1573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014278081420902525,
+      "loss": 0.8935,
+      "step": 1574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014271039235998502,
+      "loss": 0.893,
+      "step": 1575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014263994459441553,
+      "loss": 0.8762,
+      "step": 1576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014256947095506425,
+      "loss": 0.8625,
+      "step": 1577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014249897148469436,
+      "loss": 0.9003,
+      "step": 1578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001424284462260847,
+      "loss": 0.9143,
+      "step": 1579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014235789522202977,
+      "loss": 0.9145,
+      "step": 1580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014228731851533976,
+      "loss": 0.8632,
+      "step": 1581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001422167161488403,
+      "loss": 0.937,
+      "step": 1582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014214608816537274,
+      "loss": 0.8677,
+      "step": 1583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014207543460779387,
+      "loss": 0.8649,
+      "step": 1584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014200475551897607,
+      "loss": 0.8409,
+      "step": 1585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001419340509418072,
+      "loss": 0.8113,
+      "step": 1586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001418633209191905,
+      "loss": 0.8424,
+      "step": 1587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417925654940448,
+      "loss": 0.847,
+      "step": 1588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014172178470930424,
+      "loss": 0.3777,
+      "step": 1589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014165097860791836,
+      "loss": 0.7832,
+      "step": 1590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415801472328521,
+      "loss": 0.8585,
+      "step": 1591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415092906270857,
+      "loss": 0.905,
+      "step": 1592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014143840883361466,
+      "loss": 0.8823,
+      "step": 1593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014136750189544994,
+      "loss": 0.8848,
+      "step": 1594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014129656985561754,
+      "loss": 0.8931,
+      "step": 1595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001412256127571589,
+      "loss": 0.8624,
+      "step": 1596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014115463064313046,
+      "loss": 0.8673,
+      "step": 1597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000141083623556604,
+      "loss": 0.8275,
+      "step": 1598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014101259154066637,
+      "loss": 0.8548,
+      "step": 1599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014094153463841956,
+      "loss": 0.8729,
+      "step": 1600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014087045289298066,
+      "loss": 0.3327,
+      "step": 1601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014079934634748188,
+      "loss": 0.8754,
+      "step": 1602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014072821504507047,
+      "loss": 0.8575,
+      "step": 1603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001406570590289086,
+      "loss": 0.8911,
+      "step": 1604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014058587834217355,
+      "loss": 0.8385,
+      "step": 1605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014051467302805755,
+      "loss": 0.8854,
+      "step": 1606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014044344312976777,
+      "loss": 0.3214,
+      "step": 1607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014037218869052622,
+      "loss": 0.8391,
+      "step": 1608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001403009097535699,
+      "loss": 0.8692,
+      "step": 1609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014022960636215064,
+      "loss": 0.8717,
+      "step": 1610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001401582785595351,
+      "loss": 0.8538,
+      "step": 1611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014008692638900477,
+      "loss": 0.8405,
+      "step": 1612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014001554989385593,
+      "loss": 0.8519,
+      "step": 1613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013994414911739964,
+      "loss": 0.357,
+      "step": 1614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013987272410296157,
+      "loss": 0.8346,
+      "step": 1615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013980127489388226,
+      "loss": 0.8802,
+      "step": 1616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013972980153351684,
+      "loss": 0.892,
+      "step": 1617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013965830406523512,
+      "loss": 0.8703,
+      "step": 1618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001395867825324215,
+      "loss": 0.8725,
+      "step": 1619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013951523697847511,
+      "loss": 0.8102,
+      "step": 1620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394436674468095,
+      "loss": 0.8633,
+      "step": 1621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001393720739808528,
+      "loss": 0.8474,
+      "step": 1622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013930045662404773,
+      "loss": 0.9219,
+      "step": 1623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001392288154198515,
+      "loss": 0.8811,
+      "step": 1624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013915715041173574,
+      "loss": 0.8819,
+      "step": 1625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013908546164318651,
+      "loss": 0.8979,
+      "step": 1626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390137491577044,
+      "loss": 0.886,
+      "step": 1627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013894201299880426,
+      "loss": 0.3118,
+      "step": 1628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013887025321001538,
+      "loss": 0.8325,
+      "step": 1629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013879846983488135,
+      "loss": 0.8167,
+      "step": 1630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013872666291696014,
+      "loss": 0.8823,
+      "step": 1631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013865483249982392,
+      "loss": 0.8721,
+      "step": 1632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013858297862705916,
+      "loss": 0.9074,
+      "step": 1633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013851110134226654,
+      "loss": 0.8494,
+      "step": 1634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000138439200689061,
+      "loss": 0.7958,
+      "step": 1635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013836727671107167,
+      "loss": 0.892,
+      "step": 1636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001382953294519417,
+      "loss": 0.8499,
+      "step": 1637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013822335895532847,
+      "loss": 0.8823,
+      "step": 1638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013815136526490354,
+      "loss": 0.8519,
+      "step": 1639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001380793484243523,
+      "loss": 0.8669,
+      "step": 1640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013800730847737444,
+      "loss": 0.8955,
+      "step": 1641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013793524546768356,
+      "loss": 0.9219,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013786315943900728,
+      "loss": 0.8816,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013779105043508708,
+      "loss": 0.8762,
+      "step": 1644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013771891849967858,
+      "loss": 0.8771,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013764676367655114,
+      "loss": 0.9061,
+      "step": 1646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013757458600948812,
+      "loss": 0.8543,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013750238554228664,
+      "loss": 0.8272,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001374301623187578,
+      "loss": 0.8664,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013735791638272633,
+      "loss": 0.8425,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013728564777803088,
+      "loss": 0.8952,
+      "step": 1651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001372133565485238,
+      "loss": 0.8891,
+      "step": 1652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013714104273807113,
+      "loss": 0.8541,
+      "step": 1653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013706870639055278,
+      "loss": 0.8887,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001369963475498621,
+      "loss": 0.8554,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001369239662599062,
+      "loss": 0.893,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013685156256460588,
+      "loss": 0.8652,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013677913650789552,
+      "loss": 0.8308,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013670668813372286,
+      "loss": 0.8823,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013663421748604947,
+      "loss": 0.8277,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013656172460885024,
+      "loss": 0.8699,
+      "step": 1661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013648920954611362,
+      "loss": 0.8434,
+      "step": 1662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013641667234184156,
+      "loss": 0.3407,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013634411304004937,
+      "loss": 0.854,
+      "step": 1664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013627153168476583,
+      "loss": 0.8726,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013619892832003304,
+      "loss": 0.853,
+      "step": 1666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001361263029899065,
+      "loss": 0.8389,
+      "step": 1667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013605365573845504,
+      "loss": 0.9074,
+      "step": 1668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013598098660976078,
+      "loss": 0.8328,
+      "step": 1669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013590829564791908,
+      "loss": 0.823,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001358355828970386,
+      "loss": 0.8893,
+      "step": 1671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013576284840124126,
+      "loss": 0.8773,
+      "step": 1672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013569009220466206,
+      "loss": 0.8985,
+      "step": 1673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001356173143514492,
+      "loss": 0.8816,
+      "step": 1674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013554451488576415,
+      "loss": 0.8277,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013547169385178128,
+      "loss": 0.8604,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013539885129368825,
+      "loss": 0.8969,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001353259872556856,
+      "loss": 0.8605,
+      "step": 1678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013525310178198705,
+      "loss": 0.8315,
+      "step": 1679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013518019491681933,
+      "loss": 0.8981,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013510726670442198,
+      "loss": 0.8774,
+      "step": 1681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013503431718904765,
+      "loss": 0.3555,
+      "step": 1682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013496134641496189,
+      "loss": 0.8975,
+      "step": 1683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013488835442644314,
+      "loss": 0.8712,
+      "step": 1684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013481534126778266,
+      "loss": 0.7902,
+      "step": 1685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013474230698328464,
+      "loss": 0.831,
+      "step": 1686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013466925161726603,
+      "loss": 0.9186,
+      "step": 1687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001345961752140566,
+      "loss": 0.887,
+      "step": 1688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013452307781799884,
+      "loss": 0.8668,
+      "step": 1689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013444995947344807,
+      "loss": 0.8641,
+      "step": 1690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013437682022477226,
+      "loss": 0.7925,
+      "step": 1691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013430366011635207,
+      "loss": 0.8709,
+      "step": 1692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342304791925808,
+      "loss": 0.867,
+      "step": 1693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001341572774978644,
+      "loss": 0.8688,
+      "step": 1694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013408405507662143,
+      "loss": 0.8377,
+      "step": 1695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013401081197328302,
+      "loss": 0.3379,
+      "step": 1696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013393754823229288,
+      "loss": 0.828,
+      "step": 1697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013386426389810717,
+      "loss": 0.8535,
+      "step": 1698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001337909590151946,
+      "loss": 0.3459,
+      "step": 1699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013371763362803635,
+      "loss": 0.8441,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013364428778112606,
+      "loss": 0.3286,
+      "step": 1701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013357092151896965,
+      "loss": 0.8565,
+      "step": 1702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013349753488608565,
+      "loss": 0.8444,
+      "step": 1703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013342412792700475,
+      "loss": 0.8772,
+      "step": 1704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013335070068627017,
+      "loss": 0.8832,
+      "step": 1705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001332772532084372,
+      "loss": 0.3163,
+      "step": 1706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013320378553807356,
+      "loss": 0.8951,
+      "step": 1707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001331302977197592,
+      "loss": 0.8554,
+      "step": 1708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001330567897980864,
+      "loss": 0.8217,
+      "step": 1709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001329832618176594,
+      "loss": 0.831,
+      "step": 1710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013290971382309477,
+      "loss": 0.8467,
+      "step": 1711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013283614585902123,
+      "loss": 0.8662,
+      "step": 1712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013276255797007956,
+      "loss": 0.8881,
+      "step": 1713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013268895020092267,
+      "loss": 0.8507,
+      "step": 1714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013261532259621552,
+      "loss": 0.8851,
+      "step": 1715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013254167520063509,
+      "loss": 0.8891,
+      "step": 1716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013246800805887042,
+      "loss": 0.9016,
+      "step": 1717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001323943212156224,
+      "loss": 0.8799,
+      "step": 1718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013232061471560408,
+      "loss": 0.8944,
+      "step": 1719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001322468886035403,
+      "loss": 0.8559,
+      "step": 1720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321731429241678,
+      "loss": 0.8534,
+      "step": 1721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013209937772223524,
+      "loss": 0.8973,
+      "step": 1722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001320255930425031,
+      "loss": 0.8291,
+      "step": 1723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001319517889297437,
+      "loss": 0.8608,
+      "step": 1724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001318779654287411,
+      "loss": 0.8314,
+      "step": 1725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013180412258429123,
+      "loss": 0.8283,
+      "step": 1726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013173026044120166,
+      "loss": 0.8829,
+      "step": 1727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013165637904429166,
+      "loss": 0.3124,
+      "step": 1728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001315824784383923,
+      "loss": 0.8585,
+      "step": 1729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013150855866834614,
+      "loss": 0.8458,
+      "step": 1730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013143461977900756,
+      "loss": 0.8853,
+      "step": 1731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001313606618152424,
+      "loss": 0.9118,
+      "step": 1732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001312866848219281,
+      "loss": 0.8511,
+      "step": 1733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013121268884395368,
+      "loss": 0.8835,
+      "step": 1734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013113867392621964,
+      "loss": 0.8774,
+      "step": 1735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000131064640113638,
+      "loss": 0.8806,
+      "step": 1736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013099058745113227,
+      "loss": 0.9117,
+      "step": 1737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013091651598363737,
+      "loss": 0.8882,
+      "step": 1738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013084242575609958,
+      "loss": 0.3604,
+      "step": 1739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013076831681347662,
+      "loss": 0.8726,
+      "step": 1740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013069418920073762,
+      "loss": 0.8506,
+      "step": 1741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013062004296286294,
+      "loss": 0.8471,
+      "step": 1742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013054587814484425,
+      "loss": 0.868,
+      "step": 1743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013047169479168456,
+      "loss": 0.868,
+      "step": 1744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001303974929483981,
+      "loss": 0.8913,
+      "step": 1745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013032327266001027,
+      "loss": 0.8923,
+      "step": 1746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013024903397155772,
+      "loss": 0.8833,
+      "step": 1747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001301747769280883,
+      "loss": 0.3309,
+      "step": 1748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013010050157466087,
+      "loss": 0.8615,
+      "step": 1749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001300262079563455,
+      "loss": 0.8671,
+      "step": 1750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012995189611822333,
+      "loss": 0.8817,
+      "step": 1751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012987756610538656,
+      "loss": 0.8395,
+      "step": 1752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012980321796293836,
+      "loss": 0.8435,
+      "step": 1753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000129728851735993,
+      "loss": 0.3372,
+      "step": 1754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012965446746967562,
+      "loss": 0.8176,
+      "step": 1755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001295800652091224,
+      "loss": 0.8575,
+      "step": 1756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012950564499948034,
+      "loss": 0.8694,
+      "step": 1757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012943120688590741,
+      "loss": 0.8732,
+      "step": 1758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293567509135724,
+      "loss": 0.8506,
+      "step": 1759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012928227712765504,
+      "loss": 0.8554,
+      "step": 1760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012920778557334568,
+      "loss": 0.8848,
+      "step": 1761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012913327629584563,
+      "loss": 0.3374,
+      "step": 1762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012905874934036679,
+      "loss": 0.8535,
+      "step": 1763
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000128984204752132,
+      "loss": 0.8739,
+      "step": 1764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012890964257637456,
+      "loss": 0.8973,
+      "step": 1765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012883506285833865,
+      "loss": 0.8958,
+      "step": 1766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012876046564327896,
+      "loss": 0.8638,
+      "step": 1767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012868585097646084,
+      "loss": 0.8224,
+      "step": 1768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012861121890316024,
+      "loss": 0.8808,
+      "step": 1769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012853656946866366,
+      "loss": 0.903,
+      "step": 1770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012846190271826815,
+      "loss": 0.8926,
+      "step": 1771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012838721869728122,
+      "loss": 0.8388,
+      "step": 1772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012831251745102094,
+      "loss": 0.8266,
+      "step": 1773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012823779902481575,
+      "loss": 0.3257,
+      "step": 1774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012816306346400463,
+      "loss": 0.3422,
+      "step": 1775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012808831081393676,
+      "loss": 0.8379,
+      "step": 1776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012801354111997184,
+      "loss": 0.8747,
+      "step": 1777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012793875442747995,
+      "loss": 0.8703,
+      "step": 1778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012786395078184137,
+      "loss": 0.8598,
+      "step": 1779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012778913022844668,
+      "loss": 0.3367,
+      "step": 1780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012771429281269682,
+      "loss": 0.8594,
+      "step": 1781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012763943858000286,
+      "loss": 0.8595,
+      "step": 1782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001275645675757861,
+      "loss": 0.8975,
+      "step": 1783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012748967984547803,
+      "loss": 0.8596,
+      "step": 1784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001274147754345203,
+      "loss": 0.8857,
+      "step": 1785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012733985438836466,
+      "loss": 0.816,
+      "step": 1786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012726491675247296,
+      "loss": 0.8776,
+      "step": 1787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001271899625723171,
+      "loss": 0.8495,
+      "step": 1788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012711499189337907,
+      "loss": 0.8718,
+      "step": 1789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001270400047611508,
+      "loss": 0.8522,
+      "step": 1790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012696500122113427,
+      "loss": 0.8439,
+      "step": 1791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012688998131884133,
+      "loss": 0.861,
+      "step": 1792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001268149450997939,
+      "loss": 0.9045,
+      "step": 1793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012673989260952374,
+      "loss": 0.7963,
+      "step": 1794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012666482389357234,
+      "loss": 0.8482,
+      "step": 1795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012658973899749122,
+      "loss": 0.8567,
+      "step": 1796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012651463796684167,
+      "loss": 0.3233,
+      "step": 1797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012643952084719472,
+      "loss": 0.843,
+      "step": 1798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263643876841312,
+      "loss": 0.8482,
+      "step": 1799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012628923852324174,
+      "loss": 0.8673,
+      "step": 1800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001262140734101265,
+      "loss": 0.8334,
+      "step": 1801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012613889239039552,
+      "loss": 0.8142,
+      "step": 1802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001260636955096684,
+      "loss": 0.8459,
+      "step": 1803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001259884828135743,
+      "loss": 0.8352,
+      "step": 1804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001259132543477521,
+      "loss": 0.8828,
+      "step": 1805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012583801015785014,
+      "loss": 0.8246,
+      "step": 1806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001257627502895264,
+      "loss": 0.8415,
+      "step": 1807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012568747478844833,
+      "loss": 0.8571,
+      "step": 1808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012561218370029284,
+      "loss": 0.8817,
+      "step": 1809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001255368770707463,
+      "loss": 0.8894,
+      "step": 1810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001254615549455046,
+      "loss": 0.8557,
+      "step": 1811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012538621737027289,
+      "loss": 0.8009,
+      "step": 1812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001253108643907658,
+      "loss": 0.8709,
+      "step": 1813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012523549605270728,
+      "loss": 0.8891,
+      "step": 1814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001251601124018306,
+      "loss": 0.8747,
+      "step": 1815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001250847134838783,
+      "loss": 0.8166,
+      "step": 1816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012500929934460223,
+      "loss": 0.8349,
+      "step": 1817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012493387002976343,
+      "loss": 0.8668,
+      "step": 1818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012485842558513217,
+      "loss": 0.8389,
+      "step": 1819
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012478296605648788,
+      "loss": 0.8584,
+      "step": 1820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001247074914896192,
+      "loss": 0.885,
+      "step": 1821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012463200193032383,
+      "loss": 0.8797,
+      "step": 1822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012455649742440858,
+      "loss": 0.3405,
+      "step": 1823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012448097801768935,
+      "loss": 0.7989,
+      "step": 1824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012440544375599109,
+      "loss": 0.8734,
+      "step": 1825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012432989468514775,
+      "loss": 0.8109,
+      "step": 1826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012425433085100224,
+      "loss": 0.8685,
+      "step": 1827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012417875229940646,
+      "loss": 0.8104,
+      "step": 1828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012410315907622124,
+      "loss": 0.8882,
+      "step": 1829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012402755122731634,
+      "loss": 0.8918,
+      "step": 1830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012395192879857027,
+      "loss": 0.8655,
+      "step": 1831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012387629183587055,
+      "loss": 0.324,
+      "step": 1832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012380064038511345,
+      "loss": 0.8732,
+      "step": 1833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000123724974492204,
+      "loss": 0.8258,
+      "step": 1834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000123649294203056,
+      "loss": 0.8703,
+      "step": 1835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012357359956359206,
+      "loss": 0.8409,
+      "step": 1836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012349789061974343,
+      "loss": 0.9005,
+      "step": 1837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012342216741745,
+      "loss": 0.8829,
+      "step": 1838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012334643000266044,
+      "loss": 0.8567,
+      "step": 1839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012327067842133195,
+      "loss": 0.8376,
+      "step": 1840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012319491271943034,
+      "loss": 0.8228,
+      "step": 1841
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012311913294292998,
+      "loss": 0.8633,
+      "step": 1842
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012304333913781381,
+      "loss": 0.8252,
+      "step": 1843
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001229675313500733,
+      "loss": 0.8876,
+      "step": 1844
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001228917096257083,
+      "loss": 0.8845,
+      "step": 1845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012281587401072723,
+      "loss": 0.8292,
+      "step": 1846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001227400245511469,
+      "loss": 0.84,
+      "step": 1847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012266416129299253,
+      "loss": 0.8768,
+      "step": 1848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001225882842822977,
+      "loss": 0.8436,
+      "step": 1849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012251239356510428,
+      "loss": 0.9032,
+      "step": 1850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001224364891874626,
+      "loss": 0.8366,
+      "step": 1851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012236057119543116,
+      "loss": 0.8529,
+      "step": 1852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001222846396350767,
+      "loss": 0.877,
+      "step": 1853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012220869455247435,
+      "loss": 0.8445,
+      "step": 1854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012213273599370728,
+      "loss": 0.8902,
+      "step": 1855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001220567640048669,
+      "loss": 0.8668,
+      "step": 1856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012198077863205278,
+      "loss": 0.8376,
+      "step": 1857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012190477992137256,
+      "loss": 0.8476,
+      "step": 1858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012182876791894211,
+      "loss": 0.8262,
+      "step": 1859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001217527426708852,
+      "loss": 0.8681,
+      "step": 1860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001216767042233337,
+      "loss": 0.8496,
+      "step": 1861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012160065262242755,
+      "loss": 0.8591,
+      "step": 1862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012152458791431453,
+      "loss": 0.8668,
+      "step": 1863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012144851014515055,
+      "loss": 0.8849,
+      "step": 1864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012137241936109932,
+      "loss": 0.8447,
+      "step": 1865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012129631560833248,
+      "loss": 0.8047,
+      "step": 1866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012122019893302954,
+      "loss": 0.873,
+      "step": 1867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012114406938137786,
+      "loss": 0.8441,
+      "step": 1868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012106792699957263,
+      "loss": 0.8841,
+      "step": 1869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012099177183381678,
+      "loss": 0.843,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012091560393032103,
+      "loss": 0.873,
+      "step": 1871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012083942333530381,
+      "loss": 0.8498,
+      "step": 1872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012076323009499129,
+      "loss": 0.8585,
+      "step": 1873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012068702425561721,
+      "loss": 0.8787,
+      "step": 1874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012061080586342311,
+      "loss": 0.9236,
+      "step": 1875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012053457496465803,
+      "loss": 0.906,
+      "step": 1876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012045833160557867,
+      "loss": 0.8077,
+      "step": 1877
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012038207583244918,
+      "loss": 0.8278,
+      "step": 1878
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012030580769154138,
+      "loss": 0.322,
+      "step": 1879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012022952722913457,
+      "loss": 0.8227,
+      "step": 1880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012015323449151538,
+      "loss": 0.8505,
+      "step": 1881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012007692952497806,
+      "loss": 0.8402,
+      "step": 1882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012000061237582425,
+      "loss": 0.8734,
+      "step": 1883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011992428309036292,
+      "loss": 0.8355,
+      "step": 1884
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001198479417149104,
+      "loss": 0.8595,
+      "step": 1885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011977158829579045,
+      "loss": 0.9425,
+      "step": 1886
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011969522287933406,
+      "loss": 0.8879,
+      "step": 1887
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011961884551187954,
+      "loss": 0.8547,
+      "step": 1888
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011954245623977236,
+      "loss": 0.8884,
+      "step": 1889
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011946605510936537,
+      "loss": 0.8134,
+      "step": 1890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011938964216701851,
+      "loss": 0.8584,
+      "step": 1891
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011931321745909885,
+      "loss": 0.824,
+      "step": 1892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011923678103198073,
+      "loss": 0.8214,
+      "step": 1893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011916033293204549,
+      "loss": 0.8308,
+      "step": 1894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011908387320568164,
+      "loss": 0.8501,
+      "step": 1895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011900740189928461,
+      "loss": 0.8726,
+      "step": 1896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011893091905925702,
+      "loss": 0.8526,
+      "step": 1897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011885442473200839,
+      "loss": 0.826,
+      "step": 1898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011877791896395524,
+      "loss": 0.8632,
+      "step": 1899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011870140180152101,
+      "loss": 0.7872,
+      "step": 1900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011862487329113606,
+      "loss": 0.8908,
+      "step": 1901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011854833347923768,
+      "loss": 0.8232,
+      "step": 1902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011847178241226995,
+      "loss": 0.8866,
+      "step": 1903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001183952201366838,
+      "loss": 0.8772,
+      "step": 1904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011831864669893702,
+      "loss": 0.8691,
+      "step": 1905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011824206214549414,
+      "loss": 0.8527,
+      "step": 1906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011816546652282634,
+      "loss": 0.8571,
+      "step": 1907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011808885987741165,
+      "loss": 0.8883,
+      "step": 1908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011801224225573473,
+      "loss": 0.8067,
+      "step": 1909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011793561370428694,
+      "loss": 0.8707,
+      "step": 1910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011785897426956618,
+      "loss": 0.8088,
+      "step": 1911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011778232399807704,
+      "loss": 0.8386,
+      "step": 1912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011770566293633068,
+      "loss": 0.8582,
+      "step": 1913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011762899113084476,
+      "loss": 0.8635,
+      "step": 1914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011755230862814348,
+      "loss": 0.8792,
+      "step": 1915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011747561547475757,
+      "loss": 0.9071,
+      "step": 1916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011739891171722417,
+      "loss": 0.8474,
+      "step": 1917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011732219740208688,
+      "loss": 0.8414,
+      "step": 1918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011724547257589566,
+      "loss": 0.8657,
+      "step": 1919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011716873728520697,
+      "loss": 0.8497,
+      "step": 1920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011709199157658348,
+      "loss": 0.9123,
+      "step": 1921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011701523549659423,
+      "loss": 0.8545,
+      "step": 1922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011693846909181461,
+      "loss": 0.8856,
+      "step": 1923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001168616924088262,
+      "loss": 0.8531,
+      "step": 1924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011678490549421685,
+      "loss": 0.903,
+      "step": 1925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011670810839458062,
+      "loss": 0.8344,
+      "step": 1926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011663130115651772,
+      "loss": 0.8443,
+      "step": 1927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011655448382663458,
+      "loss": 0.8636,
+      "step": 1928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011647765645154367,
+      "loss": 0.8932,
+      "step": 1929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001164008190778636,
+      "loss": 0.873,
+      "step": 1930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011632397175221907,
+      "loss": 0.8285,
+      "step": 1931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011624711452124077,
+      "loss": 0.8802,
+      "step": 1932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011617024743156541,
+      "loss": 0.9339,
+      "step": 1933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011609337052983574,
+      "loss": 0.8914,
+      "step": 1934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011601648386270036,
+      "loss": 0.8575,
+      "step": 1935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011593958747681389,
+      "loss": 0.8702,
+      "step": 1936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011586268141883679,
+      "loss": 0.854,
+      "step": 1937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001157857657354354,
+      "loss": 0.834,
+      "step": 1938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011570884047328197,
+      "loss": 0.8705,
+      "step": 1939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011563190567905442,
+      "loss": 0.8058,
+      "step": 1940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001155549613994366,
+      "loss": 0.8409,
+      "step": 1941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011547800768111802,
+      "loss": 0.8324,
+      "step": 1942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000115401044570794,
+      "loss": 0.8382,
+      "step": 1943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011532407211516542,
+      "loss": 0.8517,
+      "step": 1944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011524709036093902,
+      "loss": 0.8212,
+      "step": 1945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011517009935482702,
+      "loss": 0.9561,
+      "step": 1946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011509309914354728,
+      "loss": 0.8596,
+      "step": 1947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011501608977382337,
+      "loss": 0.9025,
+      "step": 1948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011493907129238426,
+      "loss": 0.8907,
+      "step": 1949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011486204374596459,
+      "loss": 0.865,
+      "step": 1950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011478500718130433,
+      "loss": 0.9001,
+      "step": 1951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001147079616451491,
+      "loss": 0.7883,
+      "step": 1952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011463090718424987,
+      "loss": 0.8424,
+      "step": 1953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011455384384536302,
+      "loss": 0.8773,
+      "step": 1954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011447677167525034,
+      "loss": 0.8566,
+      "step": 1955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011439969072067898,
+      "loss": 0.8355,
+      "step": 1956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011432260102842143,
+      "loss": 0.8838,
+      "step": 1957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011424550264525544,
+      "loss": 0.8234,
+      "step": 1958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011416839561796408,
+      "loss": 0.8476,
+      "step": 1959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011409127999333566,
+      "loss": 0.877,
+      "step": 1960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011401415581816368,
+      "loss": 0.8583,
+      "step": 1961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011393702313924683,
+      "loss": 0.877,
+      "step": 1962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000113859882003389,
+      "loss": 0.8638,
+      "step": 1963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011378273245739917,
+      "loss": 0.8581,
+      "step": 1964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011370557454809143,
+      "loss": 0.3253,
+      "step": 1965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011362840832228494,
+      "loss": 0.8463,
+      "step": 1966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011355123382680395,
+      "loss": 0.8956,
+      "step": 1967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011347405110847771,
+      "loss": 0.8536,
+      "step": 1968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011339686021414038,
+      "loss": 0.8525,
+      "step": 1969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001133196611906312,
+      "loss": 0.8671,
+      "step": 1970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011324245408479426,
+      "loss": 0.8522,
+      "step": 1971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011316523894347864,
+      "loss": 0.3016,
+      "step": 1972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011308801581353818,
+      "loss": 0.7837,
+      "step": 1973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011301078474183165,
+      "loss": 0.8303,
+      "step": 1974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011293354577522263,
+      "loss": 0.8553,
+      "step": 1975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011285629896057946,
+      "loss": 0.9156,
+      "step": 1976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011277904434477525,
+      "loss": 0.8905,
+      "step": 1977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011270178197468789,
+      "loss": 0.8491,
+      "step": 1978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011262451189719989,
+      "loss": 0.8426,
+      "step": 1979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011254723415919852,
+      "loss": 0.867,
+      "step": 1980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011246994880757566,
+      "loss": 0.885,
+      "step": 1981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011239265588922781,
+      "loss": 0.8175,
+      "step": 1982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011231535545105606,
+      "loss": 0.8382,
+      "step": 1983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001122380475399661,
+      "loss": 0.814,
+      "step": 1984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011216073220286806,
+      "loss": 0.8389,
+      "step": 1985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011208340948667671,
+      "loss": 0.8312,
+      "step": 1986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011200607943831117,
+      "loss": 0.8582,
+      "step": 1987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011192874210469507,
+      "loss": 0.8465,
+      "step": 1988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011185139753275648,
+      "loss": 0.8378,
+      "step": 1989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011177404576942787,
+      "loss": 0.9091,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011169668686164597,
+      "loss": 0.8665,
+      "step": 1991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011161932085635193,
+      "loss": 0.8607,
+      "step": 1992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011154194780049126,
+      "loss": 0.8097,
+      "step": 1993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011146456774101361,
+      "loss": 0.885,
+      "step": 1994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000111387180724873,
+      "loss": 0.8827,
+      "step": 1995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001113097867990276,
+      "loss": 0.8743,
+      "step": 1996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001112323860104398,
+      "loss": 0.8629,
+      "step": 1997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011115497840607614,
+      "loss": 0.9159,
+      "step": 1998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001110775640329073,
+      "loss": 0.834,
+      "step": 1999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001110001429379081,
+      "loss": 0.8282,
+      "step": 2000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011092271516805743,
+      "loss": 0.8293,
+      "step": 2001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011084528077033817,
+      "loss": 0.8666,
+      "step": 2002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011076783979173728,
+      "loss": 0.8419,
+      "step": 2003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001106903922792457,
+      "loss": 0.8429,
+      "step": 2004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011061293827985833,
+      "loss": 0.8418,
+      "step": 2005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000110535477840574,
+      "loss": 0.3664,
+      "step": 2006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011045801100839548,
+      "loss": 0.8299,
+      "step": 2007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011038053783032938,
+      "loss": 0.839,
+      "step": 2008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011030305835338617,
+      "loss": 0.839,
+      "step": 2009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011022557262458013,
+      "loss": 0.8206,
+      "step": 2010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001101480806909294,
+      "loss": 0.3526,
+      "step": 2011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011007058259945584,
+      "loss": 0.8146,
+      "step": 2012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000109993078397185,
+      "loss": 0.853,
+      "step": 2013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010991556813114618,
+      "loss": 0.9118,
+      "step": 2014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001098380518483724,
+      "loss": 0.8483,
+      "step": 2015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010976052959590029,
+      "loss": 0.8683,
+      "step": 2016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010968300142077006,
+      "loss": 0.8568,
+      "step": 2017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001096054673700256,
+      "loss": 0.8737,
+      "step": 2018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001095279274907143,
+      "loss": 0.843,
+      "step": 2019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010945038182988714,
+      "loss": 0.8775,
+      "step": 2020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010937283043459854,
+      "loss": 0.8291,
+      "step": 2021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010929527335190643,
+      "loss": 0.8358,
+      "step": 2022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001092177106288723,
+      "loss": 0.8704,
+      "step": 2023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010914014231256083,
+      "loss": 0.8573,
+      "step": 2024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010906256845004027,
+      "loss": 0.8794,
+      "step": 2025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010898498908838225,
+      "loss": 0.8784,
+      "step": 2026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010890740427466165,
+      "loss": 0.8628,
+      "step": 2027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010882981405595662,
+      "loss": 0.8593,
+      "step": 2028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010875221847934871,
+      "loss": 0.8531,
+      "step": 2029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001086746175919227,
+      "loss": 0.8545,
+      "step": 2030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010859701144076651,
+      "loss": 0.8107,
+      "step": 2031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001085194000729713,
+      "loss": 0.8434,
+      "step": 2032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010844178353563141,
+      "loss": 0.8618,
+      "step": 2033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010836416187584435,
+      "loss": 0.865,
+      "step": 2034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010828653514071063,
+      "loss": 0.818,
+      "step": 2035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010820890337733391,
+      "loss": 0.8338,
+      "step": 2036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010813126663282096,
+      "loss": 0.8248,
+      "step": 2037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010805362495428146,
+      "loss": 0.8495,
+      "step": 2038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010797597838882812,
+      "loss": 0.8314,
+      "step": 2039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010789832698357665,
+      "loss": 0.8905,
+      "step": 2040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010782067078564566,
+      "loss": 0.8522,
+      "step": 2041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001077430098421567,
+      "loss": 0.8312,
+      "step": 2042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010766534420023416,
+      "loss": 0.8726,
+      "step": 2043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010758767390700533,
+      "loss": 0.88,
+      "step": 2044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010750999900960028,
+      "loss": 0.8575,
+      "step": 2045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010743231955515187,
+      "loss": 0.8361,
+      "step": 2046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010735463559079578,
+      "loss": 0.831,
+      "step": 2047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010727694716367037,
+      "loss": 0.8446,
+      "step": 2048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001071992543209167,
+      "loss": 0.8811,
+      "step": 2049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010712155710967861,
+      "loss": 0.8694,
+      "step": 2050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010704385557710246,
+      "loss": 0.8581,
+      "step": 2051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010696614977033732,
+      "loss": 0.8428,
+      "step": 2052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010688843973653482,
+      "loss": 0.3181,
+      "step": 2053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010681072552284912,
+      "loss": 0.8328,
+      "step": 2054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010673300717643701,
+      "loss": 0.8717,
+      "step": 2055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010665528474445774,
+      "loss": 0.8616,
+      "step": 2056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010657755827407299,
+      "loss": 0.8499,
+      "step": 2057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010649982781244694,
+      "loss": 0.8824,
+      "step": 2058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010642209340674621,
+      "loss": 0.8116,
+      "step": 2059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010634435510413976,
+      "loss": 0.8591,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010626661295179894,
+      "loss": 0.3294,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010618886699689747,
+      "loss": 0.8427,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010611111728661136,
+      "loss": 0.8335,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010603336386811882,
+      "loss": 0.8891,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010595560678860042,
+      "loss": 0.8918,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010587784609523889,
+      "loss": 0.8674,
+      "step": 2066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010580008183521921,
+      "loss": 0.8132,
+      "step": 2067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010572231405572843,
+      "loss": 0.3332,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010564454280395582,
+      "loss": 0.8578,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010556676812709272,
+      "loss": 0.8905,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010548899007233253,
+      "loss": 0.3202,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010541120868687076,
+      "loss": 0.7706,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001053334240179049,
+      "loss": 0.8125,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010525563611263443,
+      "loss": 0.8657,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001051778450182608,
+      "loss": 0.8752,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010510005078198739,
+      "loss": 0.8414,
+      "step": 2076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001050222534510195,
+      "loss": 0.8431,
+      "step": 2077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010494445307256433,
+      "loss": 0.9186,
+      "step": 2078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010486664969383086,
+      "loss": 0.863,
+      "step": 2079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010478884336202994,
+      "loss": 0.8231,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010471103412437425,
+      "loss": 0.3262,
+      "step": 2081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010463322202807808,
+      "loss": 0.8658,
+      "step": 2082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010455540712035765,
+      "loss": 0.8504,
+      "step": 2083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010447758944843078,
+      "loss": 0.8547,
+      "step": 2084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010439976905951698,
+      "loss": 0.8517,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001043219460008374,
+      "loss": 0.8556,
+      "step": 2086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010424412031961484,
+      "loss": 0.875,
+      "step": 2087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010416629206307364,
+      "loss": 0.7864,
+      "step": 2088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010408846127843981,
+      "loss": 0.7916,
+      "step": 2089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010401062801294073,
+      "loss": 0.8542,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010393279231380542,
+      "loss": 0.8783,
+      "step": 2091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010385495422826437,
+      "loss": 0.8242,
+      "step": 2092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001037771138035494,
+      "loss": 0.8702,
+      "step": 2093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001036992710868939,
+      "loss": 0.8481,
+      "step": 2094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010362142612553254,
+      "loss": 0.3196,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010354357896670141,
+      "loss": 0.8535,
+      "step": 2096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010346572965763793,
+      "loss": 0.8455,
+      "step": 2097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010338787824558078,
+      "loss": 0.8359,
+      "step": 2098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010331002477776996,
+      "loss": 0.8249,
+      "step": 2099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010323216930144673,
+      "loss": 0.8395,
+      "step": 2100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010315431186385349,
+      "loss": 0.8287,
+      "step": 2101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010307645251223394,
+      "loss": 0.8528,
+      "step": 2102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010299859129383287,
+      "loss": 0.8598,
+      "step": 2103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010292072825589618,
+      "loss": 0.8165,
+      "step": 2104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010284286344567096,
+      "loss": 0.8608,
+      "step": 2105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010276499691040528,
+      "loss": 0.8575,
+      "step": 2106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010268712869734838,
+      "loss": 0.8047,
+      "step": 2107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010260925885375034,
+      "loss": 0.8128,
+      "step": 2108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010253138742686241,
+      "loss": 0.8672,
+      "step": 2109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010245351446393672,
+      "loss": 0.8486,
+      "step": 2110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001023756400122263,
+      "loss": 0.8458,
+      "step": 2111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010229776411898511,
+      "loss": 0.8265,
+      "step": 2112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010221988683146803,
+      "loss": 0.322,
+      "step": 2113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010214200819693074,
+      "loss": 0.9254,
+      "step": 2114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010206412826262972,
+      "loss": 0.8679,
+      "step": 2115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001019862470758223,
+      "loss": 0.8128,
+      "step": 2116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001019083646837665,
+      "loss": 0.3059,
+      "step": 2117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010183048113372114,
+      "loss": 0.8914,
+      "step": 2118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010175259647294569,
+      "loss": 0.8866,
+      "step": 2119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010167471074870028,
+      "loss": 0.8495,
+      "step": 2120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010159682400824577,
+      "loss": 0.8267,
+      "step": 2121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010151893629884359,
+      "loss": 0.8797,
+      "step": 2122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010144104766775572,
+      "loss": 0.8752,
+      "step": 2123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010136315816224473,
+      "loss": 0.8541,
+      "step": 2124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010128526782957375,
+      "loss": 0.8608,
+      "step": 2125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010120737671700635,
+      "loss": 0.8978,
+      "step": 2126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010112948487180665,
+      "loss": 0.8664,
+      "step": 2127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001010515923412391,
+      "loss": 0.8679,
+      "step": 2128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010097369917256876,
+      "loss": 0.8947,
+      "step": 2129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010089580541306083,
+      "loss": 0.823,
+      "step": 2130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010081791110998105,
+      "loss": 0.8681,
+      "step": 2131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010074001631059544,
+      "loss": 0.8793,
+      "step": 2132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001006621210621703,
+      "loss": 0.8457,
+      "step": 2133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010058422541197221,
+      "loss": 0.8676,
+      "step": 2134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010050632940726801,
+      "loss": 0.8318,
+      "step": 2135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010042843309532478,
+      "loss": 0.8804,
+      "step": 2136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010035053652340968,
+      "loss": 0.8188,
+      "step": 2137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010027263973879016,
+      "loss": 0.8251,
+      "step": 2138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001001947427887337,
+      "loss": 0.881,
+      "step": 2139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010011684572050796,
+      "loss": 0.8633,
+      "step": 2140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001000389485813806,
+      "loss": 0.8322,
+      "step": 2141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99610514186194e-05,
+      "loss": 0.8338,
+      "step": 2142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988315427949205e-05,
+      "loss": 0.8131,
+      "step": 2143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980525721126632e-05,
+      "loss": 0.8337,
+      "step": 2144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.972736026120985e-05,
+      "loss": 0.8567,
+      "step": 2145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964946347659034e-05,
+      "loss": 0.8001,
+      "step": 2146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957156690467525e-05,
+      "loss": 0.8691,
+      "step": 2147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949367059273201e-05,
+      "loss": 0.8206,
+      "step": 2148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94157745880278e-05,
+      "loss": 0.8718,
+      "step": 2149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933787893782973e-05,
+      "loss": 0.8573,
+      "step": 2150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925998368940458e-05,
+      "loss": 0.3185,
+      "step": 2151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918208889001898e-05,
+      "loss": 0.848,
+      "step": 2152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910419458693918e-05,
+      "loss": 0.2997,
+      "step": 2153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.902630082743127e-05,
+      "loss": 0.8422,
+      "step": 2154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894840765876089e-05,
+      "loss": 0.86,
+      "step": 2155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887051512819339e-05,
+      "loss": 0.9003,
+      "step": 2156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879262328299363e-05,
+      "loss": 0.8358,
+      "step": 2157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871473217042626e-05,
+      "loss": 0.8317,
+      "step": 2158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863684183775526e-05,
+      "loss": 0.8513,
+      "step": 2159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85589523322443e-05,
+      "loss": 0.8442,
+      "step": 2160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848106370115639e-05,
+      "loss": 0.8424,
+      "step": 2161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840317599175425e-05,
+      "loss": 0.8345,
+      "step": 2162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.832528925129972e-05,
+      "loss": 0.7926,
+      "step": 2163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824740352705435e-05,
+      "loss": 0.8172,
+      "step": 2164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816951886627888e-05,
+      "loss": 0.8587,
+      "step": 2165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809163531623353e-05,
+      "loss": 0.8777,
+      "step": 2166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80137529241777e-05,
+      "loss": 0.818,
+      "step": 2167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79358717373703e-05,
+      "loss": 0.8372,
+      "step": 2168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785799180306927e-05,
+      "loss": 0.8484,
+      "step": 2169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778011316853199e-05,
+      "loss": 0.8799,
+      "step": 2170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770223588101488e-05,
+      "loss": 0.8655,
+      "step": 2171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762435998777373e-05,
+      "loss": 0.8578,
+      "step": 2172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754648553606329e-05,
+      "loss": 0.8756,
+      "step": 2173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74686125731376e-05,
+      "loss": 0.855,
+      "step": 2174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739074114624964e-05,
+      "loss": 0.868,
+      "step": 2175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731287130265164e-05,
+      "loss": 0.8457,
+      "step": 2176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723500308959471e-05,
+      "loss": 0.8826,
+      "step": 2177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715713655432907e-05,
+      "loss": 0.8667,
+      "step": 2178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707927174410382e-05,
+      "loss": 0.859,
+      "step": 2179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700140870616715e-05,
+      "loss": 0.8035,
+      "step": 2180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692354748776607e-05,
+      "loss": 0.8267,
+      "step": 2181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684568813614652e-05,
+      "loss": 0.8841,
+      "step": 2182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.676783069855327e-05,
+      "loss": 0.8383,
+      "step": 2183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668997522223006e-05,
+      "loss": 0.8477,
+      "step": 2184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661212175441923e-05,
+      "loss": 0.8568,
+      "step": 2185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65342703423621e-05,
+      "loss": 0.8685,
+      "step": 2186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64564210332986e-05,
+      "loss": 0.8334,
+      "step": 2187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63785738744675e-05,
+      "loss": 0.861,
+      "step": 2188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630072891310611e-05,
+      "loss": 0.8152,
+      "step": 2189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622288619645061e-05,
+      "loss": 0.8366,
+      "step": 2190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614504577173565e-05,
+      "loss": 0.8225,
+      "step": 2191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606720768619459e-05,
+      "loss": 0.8892,
+      "step": 2192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598937198705927e-05,
+      "loss": 0.8535,
+      "step": 2193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591153872156022e-05,
+      "loss": 0.8226,
+      "step": 2194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583370793692635e-05,
+      "loss": 0.8355,
+      "step": 2195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57558796803852e-05,
+      "loss": 0.8506,
+      "step": 2196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567805399916259e-05,
+      "loss": 0.9144,
+      "step": 2197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560023094048304e-05,
+      "loss": 0.8451,
+      "step": 2198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552241055156922e-05,
+      "loss": 0.8552,
+      "step": 2199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544459287964236e-05,
+      "loss": 0.8222,
+      "step": 2200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53667779719219e-05,
+      "loss": 0.8929,
+      "step": 2201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528896587562579e-05,
+      "loss": 0.8443,
+      "step": 2202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521115663797006e-05,
+      "loss": 0.8832,
+      "step": 2203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513335030616916e-05,
+      "loss": 0.8929,
+      "step": 2204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505554692743566e-05,
+      "loss": 0.8524,
+      "step": 2205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497774654898052e-05,
+      "loss": 0.8866,
+      "step": 2206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489994921801261e-05,
+      "loss": 0.8769,
+      "step": 2207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482215498173921e-05,
+      "loss": 0.8793,
+      "step": 2208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474436388736558e-05,
+      "loss": 0.8683,
+      "step": 2209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466657598209513e-05,
+      "loss": 0.8763,
+      "step": 2210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458879131312924e-05,
+      "loss": 0.8641,
+      "step": 2211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451100992766748e-05,
+      "loss": 0.8449,
+      "step": 2212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44332318729073e-05,
+      "loss": 0.8611,
+      "step": 2213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435545719604422e-05,
+      "loss": 0.8683,
+      "step": 2214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427768594427158e-05,
+      "loss": 0.8339,
+      "step": 2215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419991816478082e-05,
+      "loss": 0.8603,
+      "step": 2216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41221539047611e-05,
+      "loss": 0.8189,
+      "step": 2217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404439321139961e-05,
+      "loss": 0.3234,
+      "step": 2218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.396663613188118e-05,
+      "loss": 0.8182,
+      "step": 2219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388888271338868e-05,
+      "loss": 0.3398,
+      "step": 2220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381113300310252e-05,
+      "loss": 0.85,
+      "step": 2221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373338704820107e-05,
+      "loss": 0.8233,
+      "step": 2222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365564489586025e-05,
+      "loss": 0.7824,
+      "step": 2223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357790659325381e-05,
+      "loss": 0.8748,
+      "step": 2224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350017218755307e-05,
+      "loss": 0.8183,
+      "step": 2225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342244172592704e-05,
+      "loss": 0.8246,
+      "step": 2226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334471525554229e-05,
+      "loss": 0.8552,
+      "step": 2227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326699282356301e-05,
+      "loss": 0.8891,
+      "step": 2228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318927447715087e-05,
+      "loss": 0.8609,
+      "step": 2229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311156026346522e-05,
+      "loss": 0.888,
+      "step": 2230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303385022966269e-05,
+      "loss": 0.8666,
+      "step": 2231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295614442289756e-05,
+      "loss": 0.8367,
+      "step": 2232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28784428903214e-05,
+      "loss": 0.8711,
+      "step": 2233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28007456790833e-05,
+      "loss": 0.759,
+      "step": 2234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272305283632966e-05,
+      "loss": 0.8204,
+      "step": 2235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264536440920426e-05,
+      "loss": 0.8348,
+      "step": 2236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.256768044484814e-05,
+      "loss": 0.8385,
+      "step": 2237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249000099039976e-05,
+      "loss": 0.8725,
+      "step": 2238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241232609299468e-05,
+      "loss": 0.8947,
+      "step": 2239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233465579976585e-05,
+      "loss": 0.8694,
+      "step": 2240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225699015784331e-05,
+      "loss": 0.8507,
+      "step": 2241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217932921435434e-05,
+      "loss": 0.8544,
+      "step": 2242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210167301642336e-05,
+      "loss": 0.8918,
+      "step": 2243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20240216111719e-05,
+      "loss": 0.8102,
+      "step": 2244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194637504571855e-05,
+      "loss": 0.8421,
+      "step": 2245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186873336717907e-05,
+      "loss": 0.867,
+      "step": 2246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179109662266607e-05,
+      "loss": 0.8267,
+      "step": 2247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17134648592894e-05,
+      "loss": 0.8433,
+      "step": 2248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163583812415568e-05,
+      "loss": 0.8701,
+      "step": 2249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155821646436861e-05,
+      "loss": 0.8517,
+      "step": 2250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148059992702871e-05,
+      "loss": 0.8794,
+      "step": 2251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140298855923353e-05,
+      "loss": 0.836,
+      "step": 2252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132538240807731e-05,
+      "loss": 0.3151,
+      "step": 2253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12477815206513e-05,
+      "loss": 0.8717,
+      "step": 2254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117018594404339e-05,
+      "loss": 0.8291,
+      "step": 2255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109259572533838e-05,
+      "loss": 0.9041,
+      "step": 2256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101501091161776e-05,
+      "loss": 0.8471,
+      "step": 2257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093743154995974e-05,
+      "loss": 0.9353,
+      "step": 2258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085985768743917e-05,
+      "loss": 0.8318,
+      "step": 2259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078228937112774e-05,
+      "loss": 0.8318,
+      "step": 2260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070472664809355e-05,
+      "loss": 0.8337,
+      "step": 2261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06271695654015e-05,
+      "loss": 0.8391,
+      "step": 2262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054961817011287e-05,
+      "loss": 0.8271,
+      "step": 2263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047207250928571e-05,
+      "loss": 0.8314,
+      "step": 2264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039453262997442e-05,
+      "loss": 0.3308,
+      "step": 2265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031699857922996e-05,
+      "loss": 0.3258,
+      "step": 2266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023947040409971e-05,
+      "loss": 0.7972,
+      "step": 2267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016194815162762e-05,
+      "loss": 0.8302,
+      "step": 2268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008443186885382e-05,
+      "loss": 0.8412,
+      "step": 2269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000692160281504e-05,
+      "loss": 0.8034,
+      "step": 2270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992941740054418e-05,
+      "loss": 0.8255,
+      "step": 2271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98519193090706e-05,
+      "loss": 0.8527,
+      "step": 2272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977442737541986e-05,
+      "loss": 0.8661,
+      "step": 2273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969694164661386e-05,
+      "loss": 0.8416,
+      "step": 2274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961946216967064e-05,
+      "loss": 0.8261,
+      "step": 2275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954198899160456e-05,
+      "loss": 0.8524,
+      "step": 2276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9464522159426e-05,
+      "loss": 0.344,
+      "step": 2277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.938706172014168e-05,
+      "loss": 0.8316,
+      "step": 2278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930960772075432e-05,
+      "loss": 0.8385,
+      "step": 2279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923216020826276e-05,
+      "loss": 0.8156,
+      "step": 2280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915471922966184e-05,
+      "loss": 0.8514,
+      "step": 2281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907728483194258e-05,
+      "loss": 0.8644,
+      "step": 2282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.899985706209189e-05,
+      "loss": 0.8472,
+      "step": 2283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892243596709272e-05,
+      "loss": 0.8737,
+      "step": 2284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884502159392387e-05,
+      "loss": 0.8282,
+      "step": 2285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876761398956023e-05,
+      "loss": 0.8983,
+      "step": 2286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869021320097241e-05,
+      "loss": 0.829,
+      "step": 2287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861281927512703e-05,
+      "loss": 0.8591,
+      "step": 2288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853543225898637e-05,
+      "loss": 0.8554,
+      "step": 2289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845805219950878e-05,
+      "loss": 0.8788,
+      "step": 2290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838067914364806e-05,
+      "loss": 0.8466,
+      "step": 2291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830331313835406e-05,
+      "loss": 0.8087,
+      "step": 2292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822595423057217e-05,
+      "loss": 0.8863,
+      "step": 2293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814860246724353e-05,
+      "loss": 0.8847,
+      "step": 2294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807125789530493e-05,
+      "loss": 0.844,
+      "step": 2295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799392056168886e-05,
+      "loss": 0.8882,
+      "step": 2296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791659051332332e-05,
+      "loss": 0.8138,
+      "step": 2297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783926779713196e-05,
+      "loss": 0.8454,
+      "step": 2298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77619524600339e-05,
+      "loss": 0.8283,
+      "step": 2299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768464454894394e-05,
+      "loss": 0.8606,
+      "step": 2300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760734411077219e-05,
+      "loss": 0.8662,
+      "step": 2301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753005119242436e-05,
+      "loss": 0.8777,
+      "step": 2302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745276584080146e-05,
+      "loss": 0.8559,
+      "step": 2303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737548810280012e-05,
+      "loss": 0.3178,
+      "step": 2304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729821802531212e-05,
+      "loss": 0.8676,
+      "step": 2305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722095565522476e-05,
+      "loss": 0.8546,
+      "step": 2306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714370103942054e-05,
+      "loss": 0.8965,
+      "step": 2307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706645422477739e-05,
+      "loss": 0.8979,
+      "step": 2308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698921525816835e-05,
+      "loss": 0.8576,
+      "step": 2309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.691198418646184e-05,
+      "loss": 0.8506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683476105652137e-05,
+      "loss": 0.8262,
+      "step": 2311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675754591520575e-05,
+      "loss": 0.8556,
+      "step": 2312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.668033880936882e-05,
+      "loss": 0.88,
+      "step": 2313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660313978585964e-05,
+      "loss": 0.8607,
+      "step": 2314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652594889152233e-05,
+      "loss": 0.7922,
+      "step": 2315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644876617319608e-05,
+      "loss": 0.8275,
+      "step": 2316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637159167771506e-05,
+      "loss": 0.3318,
+      "step": 2317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62944254519086e-05,
+      "loss": 0.8343,
+      "step": 2318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621726754260084e-05,
+      "loss": 0.8134,
+      "step": 2319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614011799661103e-05,
+      "loss": 0.8498,
+      "step": 2320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606297686075315e-05,
+      "loss": 0.7965,
+      "step": 2321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598584418183633e-05,
+      "loss": 0.8359,
+      "step": 2322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590872000666433e-05,
+      "loss": 0.8598,
+      "step": 2323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583160438203593e-05,
+      "loss": 0.8323,
+      "step": 2324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575449735474454e-05,
+      "loss": 0.3116,
+      "step": 2325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567739897157858e-05,
+      "loss": 0.8153,
+      "step": 2326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560030927932102e-05,
+      "loss": 0.8519,
+      "step": 2327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552322832474968e-05,
+      "loss": 0.8865,
+      "step": 2328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544615615463699e-05,
+      "loss": 0.8056,
+      "step": 2329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536909281575016e-05,
+      "loss": 0.3281,
+      "step": 2330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52920383548509e-05,
+      "loss": 0.8077,
+      "step": 2331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521499281869568e-05,
+      "loss": 0.8476,
+      "step": 2332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513795625403545e-05,
+      "loss": 0.8976,
+      "step": 2333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506092870761576e-05,
+      "loss": 0.844,
+      "step": 2334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498391022617663e-05,
+      "loss": 0.3281,
+      "step": 2335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490690085645272e-05,
+      "loss": 0.8637,
+      "step": 2336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482990064517302e-05,
+      "loss": 0.3043,
+      "step": 2337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475290963906102e-05,
+      "loss": 0.3304,
+      "step": 2338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467592788483456e-05,
+      "loss": 0.8037,
+      "step": 2339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459895542920601e-05,
+      "loss": 0.877,
+      "step": 2340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452199231888197e-05,
+      "loss": 0.9036,
+      "step": 2341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444503860056341e-05,
+      "loss": 0.8244,
+      "step": 2342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436809432094557e-05,
+      "loss": 0.8365,
+      "step": 2343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429115952671806e-05,
+      "loss": 0.847,
+      "step": 2344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42142342645646e-05,
+      "loss": 0.7832,
+      "step": 2345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413731858116325e-05,
+      "loss": 0.8207,
+      "step": 2346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406041252318612e-05,
+      "loss": 0.8672,
+      "step": 2347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398351613729966e-05,
+      "loss": 0.8307,
+      "step": 2348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390662947016427e-05,
+      "loss": 0.8581,
+      "step": 2349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38297525684346e-05,
+      "loss": 0.3602,
+      "step": 2350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375288547875923e-05,
+      "loss": 0.827,
+      "step": 2351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367602824778096e-05,
+      "loss": 0.8666,
+      "step": 2352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359918092213639e-05,
+      "loss": 0.8223,
+      "step": 2353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352234354845634e-05,
+      "loss": 0.8601,
+      "step": 2354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344551617336543e-05,
+      "loss": 0.8149,
+      "step": 2355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33686988434823e-05,
+      "loss": 0.8312,
+      "step": 2356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329189160541938e-05,
+      "loss": 0.88,
+      "step": 2357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321509450578316e-05,
+      "loss": 0.3263,
+      "step": 2358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313830759117381e-05,
+      "loss": 0.8587,
+      "step": 2359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306153090818541e-05,
+      "loss": 0.8486,
+      "step": 2360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298476450340576e-05,
+      "loss": 0.8538,
+      "step": 2361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290800842341655e-05,
+      "loss": 0.808,
+      "step": 2362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283126271479304e-05,
+      "loss": 0.8796,
+      "step": 2363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275452742410435e-05,
+      "loss": 0.8373,
+      "step": 2364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267780259791313e-05,
+      "loss": 0.8823,
+      "step": 2365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260108828277584e-05,
+      "loss": 0.8566,
+      "step": 2366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252438452524244e-05,
+      "loss": 0.8391,
+      "step": 2367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244769137185654e-05,
+      "loss": 0.8442,
+      "step": 2368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237100886915524e-05,
+      "loss": 0.8425,
+      "step": 2369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229433706366933e-05,
+      "loss": 0.8377,
+      "step": 2370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221767600192297e-05,
+      "loss": 0.8191,
+      "step": 2371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214102573043385e-05,
+      "loss": 0.874,
+      "step": 2372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206438629571307e-05,
+      "loss": 0.8368,
+      "step": 2373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19877577442653e-05,
+      "loss": 0.853,
+      "step": 2374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191114012258836e-05,
+      "loss": 0.8917,
+      "step": 2375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18345334771737e-05,
+      "loss": 0.8662,
+      "step": 2376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17579378545059e-05,
+      "loss": 0.8564,
+      "step": 2377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1681353301063e-05,
+      "loss": 0.8759,
+      "step": 2378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.160477986331619e-05,
+      "loss": 0.9096,
+      "step": 2379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152821758773008e-05,
+      "loss": 0.8713,
+      "step": 2380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145166652076235e-05,
+      "loss": 0.7853,
+      "step": 2381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137512670886397e-05,
+      "loss": 0.8351,
+      "step": 2382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1298598198479e-05,
+      "loss": 0.8508,
+      "step": 2383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122208103604478e-05,
+      "loss": 0.841,
+      "step": 2384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114557526799162e-05,
+      "loss": 0.8421,
+      "step": 2385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106908094074299e-05,
+      "loss": 0.8681,
+      "step": 2386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099259810071538e-05,
+      "loss": 0.8515,
+      "step": 2387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09161267943184e-05,
+      "loss": 0.8157,
+      "step": 2388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08396670679545e-05,
+      "loss": 0.8754,
+      "step": 2389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07632189680193e-05,
+      "loss": 0.8253,
+      "step": 2390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.068678254090114e-05,
+      "loss": 0.3813,
+      "step": 2391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061035783298152e-05,
+      "loss": 0.8505,
+      "step": 2392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053394489063464e-05,
+      "loss": 0.8093,
+      "step": 2393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045754376022765e-05,
+      "loss": 0.3403,
+      "step": 2394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038115448812049e-05,
+      "loss": 0.8554,
+      "step": 2395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030477712066596e-05,
+      "loss": 0.8643,
+      "step": 2396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022841170420953e-05,
+      "loss": 0.8118,
+      "step": 2397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01520582850896e-05,
+      "loss": 0.85,
+      "step": 2398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00757169096371e-05,
+      "loss": 0.8828,
+      "step": 2399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999938762417578e-05,
+      "loss": 0.8245,
+      "step": 2400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992307047502192e-05,
+      "loss": 0.8066,
+      "step": 2401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984676550848464e-05,
+      "loss": 0.8418,
+      "step": 2402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977047277086547e-05,
+      "loss": 0.8394,
+      "step": 2403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.969419230845863e-05,
+      "loss": 0.8556,
+      "step": 2404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96179241675508e-05,
+      "loss": 0.7905,
+      "step": 2405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954166839442137e-05,
+      "loss": 0.8406,
+      "step": 2406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946542503534196e-05,
+      "loss": 0.8325,
+      "step": 2407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93891941365769e-05,
+      "loss": 0.8192,
+      "step": 2408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931297574438278e-05,
+      "loss": 0.8624,
+      "step": 2409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923676990500875e-05,
+      "loss": 0.8099,
+      "step": 2410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91605766646962e-05,
+      "loss": 0.8744,
+      "step": 2411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908439606967898e-05,
+      "loss": 0.8105,
+      "step": 2412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900822816618321e-05,
+      "loss": 0.866,
+      "step": 2413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89320730004274e-05,
+      "loss": 0.8131,
+      "step": 2414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885593061862214e-05,
+      "loss": 0.2975,
+      "step": 2415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.877980106697049e-05,
+      "loss": 0.8637,
+      "step": 2416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870368439166755e-05,
+      "loss": 0.8247,
+      "step": 2417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862758063890072e-05,
+      "loss": 0.8366,
+      "step": 2418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855148985484946e-05,
+      "loss": 0.8617,
+      "step": 2419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847541208568549e-05,
+      "loss": 0.7915,
+      "step": 2420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839934737757249e-05,
+      "loss": 0.8362,
+      "step": 2421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832329577666633e-05,
+      "loss": 0.3407,
+      "step": 2422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824725732911482e-05,
+      "loss": 0.828,
+      "step": 2423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81712320810579e-05,
+      "loss": 0.3267,
+      "step": 2424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809522007862742e-05,
+      "loss": 0.8831,
+      "step": 2425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801922136794725e-05,
+      "loss": 0.8209,
+      "step": 2426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79432359951331e-05,
+      "loss": 0.3156,
+      "step": 2427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786726400629273e-05,
+      "loss": 0.8069,
+      "step": 2428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779130544752564e-05,
+      "loss": 0.8617,
+      "step": 2429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771536036492328e-05,
+      "loss": 0.8956,
+      "step": 2430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763942880456883e-05,
+      "loss": 0.8077,
+      "step": 2431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75635108125374e-05,
+      "loss": 0.9017,
+      "step": 2432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74876064348957e-05,
+      "loss": 0.3153,
+      "step": 2433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741171571770234e-05,
+      "loss": 0.799,
+      "step": 2434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733583870700745e-05,
+      "loss": 0.8007,
+      "step": 2435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725997544885313e-05,
+      "loss": 0.8653,
+      "step": 2436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718412598927278e-05,
+      "loss": 0.8505,
+      "step": 2437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710829037429173e-05,
+      "loss": 0.8653,
+      "step": 2438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703246864992674e-05,
+      "loss": 0.8653,
+      "step": 2439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695666086218622e-05,
+      "loss": 0.8267,
+      "step": 2440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688086705707003e-05,
+      "loss": 0.8462,
+      "step": 2441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.680508728056969e-05,
+      "loss": 0.8662,
+      "step": 2442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672932157866806e-05,
+      "loss": 0.8805,
+      "step": 2443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.665356999733958e-05,
+      "loss": 0.8296,
+      "step": 2444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657783258254999e-05,
+      "loss": 0.8513,
+      "step": 2445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650210938025661e-05,
+      "loss": 0.8391,
+      "step": 2446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642640043640795e-05,
+      "loss": 0.8273,
+      "step": 2447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635070579694403e-05,
+      "loss": 0.837,
+      "step": 2448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627502550779601e-05,
+      "loss": 0.8243,
+      "step": 2449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619935961488658e-05,
+      "loss": 0.8595,
+      "step": 2450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612370816412944e-05,
+      "loss": 0.8631,
+      "step": 2451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604807120142975e-05,
+      "loss": 0.8333,
+      "step": 2452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597244877268368e-05,
+      "loss": 0.8475,
+      "step": 2453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589684092377876e-05,
+      "loss": 0.8261,
+      "step": 2454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582124770059355e-05,
+      "loss": 0.8415,
+      "step": 2455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57456691489978e-05,
+      "loss": 0.8189,
+      "step": 2456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567010531485224e-05,
+      "loss": 0.8788,
+      "step": 2457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559455624400894e-05,
+      "loss": 0.8464,
+      "step": 2458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551902198231065e-05,
+      "loss": 0.8384,
+      "step": 2459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544350257559146e-05,
+      "loss": 0.8516,
+      "step": 2460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53679980696762e-05,
+      "loss": 0.838,
+      "step": 2461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.529250851038083e-05,
+      "loss": 0.3564,
+      "step": 2462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52170339435121e-05,
+      "loss": 0.8328,
+      "step": 2463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514157441486785e-05,
+      "loss": 0.8464,
+      "step": 2464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506612997023657e-05,
+      "loss": 0.882,
+      "step": 2465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499070065539778e-05,
+      "loss": 0.8304,
+      "step": 2466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491528651612168e-05,
+      "loss": 0.8037,
+      "step": 2467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.483988759816941e-05,
+      "loss": 0.3376,
+      "step": 2468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476450394729273e-05,
+      "loss": 0.832,
+      "step": 2469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468913560923423e-05,
+      "loss": 0.9135,
+      "step": 2470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461378262972711e-05,
+      "loss": 0.8456,
+      "step": 2471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453844505449544e-05,
+      "loss": 0.809,
+      "step": 2472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44631229292537e-05,
+      "loss": 0.8398,
+      "step": 2473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.43878162997072e-05,
+      "loss": 0.8327,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431252521155166e-05,
+      "loss": 0.8374,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423724971047359e-05,
+      "loss": 0.8239,
+      "step": 2476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416198984214987e-05,
+      "loss": 0.8505,
+      "step": 2477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408674565224792e-05,
+      "loss": 0.8638,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40115171864257e-05,
+      "loss": 0.8412,
+      "step": 2479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393630449033164e-05,
+      "loss": 0.8385,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386110760960446e-05,
+      "loss": 0.8599,
+      "step": 2481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378592658987349e-05,
+      "loss": 0.8216,
+      "step": 2482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.371076147675829e-05,
+      "loss": 0.7918,
+      "step": 2483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363561231586881e-05,
+      "loss": 0.836,
+      "step": 2484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356047915280529e-05,
+      "loss": 0.866,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348536203315837e-05,
+      "loss": 0.8088,
+      "step": 2486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34102610025088e-05,
+      "loss": 0.8609,
+      "step": 2487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33351761064277e-05,
+      "loss": 0.8399,
+      "step": 2488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326010739047629e-05,
+      "loss": 0.8161,
+      "step": 2489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318505490020608e-05,
+      "loss": 0.8123,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311001868115865e-05,
+      "loss": 0.3325,
+      "step": 2491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303499877886577e-05,
+      "loss": 0.8475,
+      "step": 2492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295999523884921e-05,
+      "loss": 0.8433,
+      "step": 2493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288500810662096e-05,
+      "loss": 0.8146,
+      "step": 2494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28100374276829e-05,
+      "loss": 0.8208,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273508324752706e-05,
+      "loss": 0.8424,
+      "step": 2496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266014561163532e-05,
+      "loss": 0.8519,
+      "step": 2497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258522456547972e-05,
+      "loss": 0.822,
+      "step": 2498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251032015452195e-05,
+      "loss": 0.8757,
+      "step": 2499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243543242421392e-05,
+      "loss": 0.829,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236056141999714e-05,
+      "loss": 0.8483,
+      "step": 2501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228570718730319e-05,
+      "loss": 0.8257,
+      "step": 2502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.22108697715533e-05,
+      "loss": 0.8449,
+      "step": 2503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213604921815866e-05,
+      "loss": 0.8577,
+      "step": 2504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206124557252006e-05,
+      "loss": 0.8322,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198645888002817e-05,
+      "loss": 0.8115,
+      "step": 2506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191168918606326e-05,
+      "loss": 0.8395,
+      "step": 2507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183693653599541e-05,
+      "loss": 0.8418,
+      "step": 2508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176220097518424e-05,
+      "loss": 0.8795,
+      "step": 2509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168748254897908e-05,
+      "loss": 0.8442,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161278130271876e-05,
+      "loss": 0.817,
+      "step": 2511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153809728173187e-05,
+      "loss": 0.3572,
+      "step": 2512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146343053133635e-05,
+      "loss": 0.821,
+      "step": 2513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138878109683978e-05,
+      "loss": 0.8573,
+      "step": 2514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131414902353917e-05,
+      "loss": 0.8731,
+      "step": 2515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123953435672106e-05,
+      "loss": 0.8111,
+      "step": 2516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116493714166136e-05,
+      "loss": 0.8794,
+      "step": 2517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109035742362544e-05,
+      "loss": 0.8656,
+      "step": 2518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101579524786802e-05,
+      "loss": 0.8295,
+      "step": 2519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094125065963322e-05,
+      "loss": 0.8094,
+      "step": 2520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086672370415439e-05,
+      "loss": 0.8712,
+      "step": 2521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079221442665434e-05,
+      "loss": 0.8071,
+      "step": 2522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071772287234497e-05,
+      "loss": 0.7981,
+      "step": 2523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064324908642759e-05,
+      "loss": 0.8359,
+      "step": 2524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05687931140926e-05,
+      "loss": 0.8398,
+      "step": 2525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049435500051968e-05,
+      "loss": 0.8545,
+      "step": 2526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041993479087763e-05,
+      "loss": 0.8611,
+      "step": 2527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03455325303244e-05,
+      "loss": 0.8546,
+      "step": 2528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027114826400701e-05,
+      "loss": 0.8307,
+      "step": 2529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019678203706163e-05,
+      "loss": 0.8118,
+      "step": 2530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012243389461347e-05,
+      "loss": 0.8723,
+      "step": 2531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004810388177669e-05,
+      "loss": 0.8387,
+      "step": 2532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99737920436545e-05,
+      "loss": 0.8939,
+      "step": 2533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989949842533916e-05,
+      "loss": 0.3143,
+      "step": 2534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982522307191172e-05,
+      "loss": 0.8244,
+      "step": 2535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975096602844229e-05,
+      "loss": 0.8356,
+      "step": 2536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967672733998974e-05,
+      "loss": 0.3229,
+      "step": 2537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960250705160192e-05,
+      "loss": 0.8619,
+      "step": 2538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952830520831544e-05,
+      "loss": 0.9006,
+      "step": 2539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.945412185515576e-05,
+      "loss": 0.8522,
+      "step": 2540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937995703713707e-05,
+      "loss": 0.8233,
+      "step": 2541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.930581079926241e-05,
+      "loss": 0.7954,
+      "step": 2542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923168318652336e-05,
+      "loss": 0.8414,
+      "step": 2543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915757424390045e-05,
+      "loss": 0.8251,
+      "step": 2544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908348401636265e-05,
+      "loss": 0.9158,
+      "step": 2545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900941254886775e-05,
+      "loss": 0.8361,
+      "step": 2546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8935359886362e-05,
+      "loss": 0.8587,
+      "step": 2547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886132607378038e-05,
+      "loss": 0.826,
+      "step": 2548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878731115604634e-05,
+      "loss": 0.8623,
+      "step": 2549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871331517807193e-05,
+      "loss": 0.8494,
+      "step": 2550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86393381847576e-05,
+      "loss": 0.8522,
+      "step": 2551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.856538022099243e-05,
+      "loss": 0.7794,
+      "step": 2552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849144133165384e-05,
+      "loss": 0.8045,
+      "step": 2553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841752156160775e-05,
+      "loss": 0.855,
+      "step": 2554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834362095570834e-05,
+      "loss": 0.871,
+      "step": 2555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826973955879839e-05,
+      "loss": 0.3433,
+      "step": 2556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819587741570878e-05,
+      "loss": 0.8753,
+      "step": 2557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812203457125892e-05,
+      "loss": 0.8043,
+      "step": 2558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.804821107025632e-05,
+      "loss": 0.3513,
+      "step": 2559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797440695749691e-05,
+      "loss": 0.7929,
+      "step": 2560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790062227776477e-05,
+      "loss": 0.8531,
+      "step": 2561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782685707583223e-05,
+      "loss": 0.8776,
+      "step": 2562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.775311139645971e-05,
+      "loss": 0.8302,
+      "step": 2563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767938528439593e-05,
+      "loss": 0.8591,
+      "step": 2564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760567878437758e-05,
+      "loss": 0.834,
+      "step": 2565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753199194112962e-05,
+      "loss": 0.8593,
+      "step": 2566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745832479936491e-05,
+      "loss": 0.826,
+      "step": 2567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738467740378452e-05,
+      "loss": 0.3033,
+      "step": 2568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731104979907732e-05,
+      "loss": 0.8114,
+      "step": 2569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723744202992045e-05,
+      "loss": 0.8362,
+      "step": 2570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716385414097879e-05,
+      "loss": 0.7754,
+      "step": 2571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709028617690525e-05,
+      "loss": 0.8577,
+      "step": 2572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701673818234061e-05,
+      "loss": 0.8174,
+      "step": 2573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694321020191363e-05,
+      "loss": 0.795,
+      "step": 2574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686970228024077e-05,
+      "loss": 0.8517,
+      "step": 2575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679621446192646e-05,
+      "loss": 0.8503,
+      "step": 2576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672274679156282e-05,
+      "loss": 0.8382,
+      "step": 2577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664929931372987e-05,
+      "loss": 0.8276,
+      "step": 2578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657587207299523e-05,
+      "loss": 0.8591,
+      "step": 2579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650246511391436e-05,
+      "loss": 0.8532,
+      "step": 2580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.642907848103033e-05,
+      "loss": 0.8138,
+      "step": 2581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635571221887398e-05,
+      "loss": 0.8425,
+      "step": 2582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628236637196365e-05,
+      "loss": 0.8654,
+      "step": 2583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620904098480541e-05,
+      "loss": 0.3216,
+      "step": 2584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613573610189285e-05,
+      "loss": 0.8625,
+      "step": 2585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606245176770715e-05,
+      "loss": 0.8454,
+      "step": 2586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598918802671697e-05,
+      "loss": 0.8276,
+      "step": 2587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591594492337858e-05,
+      "loss": 0.8187,
+      "step": 2588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584272250213562e-05,
+      "loss": 0.757,
+      "step": 2589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576952080741924e-05,
+      "loss": 0.8389,
+      "step": 2590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569633988364794e-05,
+      "loss": 0.825,
+      "step": 2591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562317977522774e-05,
+      "loss": 0.8571,
+      "step": 2592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.555004052655193e-05,
+      "loss": 0.8383,
+      "step": 2593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54769221820012e-05,
+      "loss": 0.8274,
+      "step": 2594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540382478594342e-05,
+      "loss": 0.8175,
+      "step": 2595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.533074838273401e-05,
+      "loss": 0.8588,
+      "step": 2596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525769301671539e-05,
+      "loss": 0.8133,
+      "step": 2597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518465873221738e-05,
+      "loss": 0.8949,
+      "step": 2598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.511164557355687e-05,
+      "loss": 0.8134,
+      "step": 2599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503865358503811e-05,
+      "loss": 0.8522,
+      "step": 2600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496568281095235e-05,
+      "loss": 0.8317,
+      "step": 2601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489273329557804e-05,
+      "loss": 0.8604,
+      "step": 2602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.481980508318068e-05,
+      "loss": 0.8536,
+      "step": 2603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474689821801295e-05,
+      "loss": 0.8244,
+      "step": 2604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467401274431439e-05,
+      "loss": 0.8413,
+      "step": 2605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460114870631179e-05,
+      "loss": 0.8363,
+      "step": 2606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452830614821872e-05,
+      "loss": 0.8369,
+      "step": 2607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44554851142359e-05,
+      "loss": 0.8407,
+      "step": 2608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438268564855077e-05,
+      "loss": 0.8283,
+      "step": 2609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430990779533796e-05,
+      "loss": 0.3331,
+      "step": 2610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423715159875874e-05,
+      "loss": 0.8778,
+      "step": 2611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41644171029614e-05,
+      "loss": 0.2965,
+      "step": 2612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409170435208091e-05,
+      "loss": 0.8237,
+      "step": 2613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401901339023923e-05,
+      "loss": 0.8135,
+      "step": 2614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394634426154497e-05,
+      "loss": 0.8515,
+      "step": 2615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387369701009351e-05,
+      "loss": 0.8267,
+      "step": 2616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380107167996696e-05,
+      "loss": 0.8324,
+      "step": 2617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372846831523418e-05,
+      "loss": 0.8502,
+      "step": 2618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365588695995062e-05,
+      "loss": 0.8409,
+      "step": 2619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358332765815846e-05,
+      "loss": 0.8067,
+      "step": 2620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351079045388637e-05,
+      "loss": 0.8486,
+      "step": 2621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34382753911498e-05,
+      "loss": 0.8211,
+      "step": 2622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336578251395056e-05,
+      "loss": 0.8184,
+      "step": 2623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329331186627717e-05,
+      "loss": 0.8331,
+      "step": 2624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32208634921045e-05,
+      "loss": 0.845,
+      "step": 2625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314843743539411e-05,
+      "loss": 0.8783,
+      "step": 2626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307603374009378e-05,
+      "loss": 0.8441,
+      "step": 2627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300365245013792e-05,
+      "loss": 0.8436,
+      "step": 2628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293129360944725e-05,
+      "loss": 0.8252,
+      "step": 2629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285895726192888e-05,
+      "loss": 0.7865,
+      "step": 2630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278664345147623e-05,
+      "loss": 0.8351,
+      "step": 2631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271435222196916e-05,
+      "loss": 0.8648,
+      "step": 2632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26420836172737e-05,
+      "loss": 0.8785,
+      "step": 2633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256983768124225e-05,
+      "loss": 0.8156,
+      "step": 2634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249761445771334e-05,
+      "loss": 0.8517,
+      "step": 2635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242541399051189e-05,
+      "loss": 0.8686,
+      "step": 2636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235323632344886e-05,
+      "loss": 0.8239,
+      "step": 2637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228108150032145e-05,
+      "loss": 0.331,
+      "step": 2638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22089495649129e-05,
+      "loss": 0.848,
+      "step": 2639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213684056099276e-05,
+      "loss": 0.8253,
+      "step": 2640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206475453231644e-05,
+      "loss": 0.8727,
+      "step": 2641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199269152262556e-05,
+      "loss": 0.8493,
+      "step": 2642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192065157564769e-05,
+      "loss": 0.8543,
+      "step": 2643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18486347350965e-05,
+      "loss": 0.8325,
+      "step": 2644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177664104467151e-05,
+      "loss": 0.8276,
+      "step": 2645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170467054805833e-05,
+      "loss": 0.8056,
+      "step": 2646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163272328892835e-05,
+      "loss": 0.8929,
+      "step": 2647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156079931093898e-05,
+      "loss": 0.8463,
+      "step": 2648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148889865773344e-05,
+      "loss": 0.8629,
+      "step": 2649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141702137294086e-05,
+      "loss": 0.8168,
+      "step": 2650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13451675001761e-05,
+      "loss": 0.8179,
+      "step": 2651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127333708303989e-05,
+      "loss": 0.8626,
+      "step": 2652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120153016511863e-05,
+      "loss": 0.8614,
+      "step": 2653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112974678998465e-05,
+      "loss": 0.825,
+      "step": 2654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105798700119576e-05,
+      "loss": 0.8123,
+      "step": 2655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098625084229563e-05,
+      "loss": 0.8158,
+      "step": 2656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091453835681347e-05,
+      "loss": 0.8576,
+      "step": 2657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084284958826428e-05,
+      "loss": 0.8746,
+      "step": 2658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07711845801485e-05,
+      "loss": 0.8182,
+      "step": 2659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0699543375952275e-05,
+      "loss": 0.8353,
+      "step": 2660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0627926019147194e-05,
+      "loss": 0.8769,
+      "step": 2661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055633255319053e-05,
+      "loss": 0.3489,
+      "step": 2662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048476302152488e-05,
+      "loss": 0.8075,
+      "step": 2663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041321746757849e-05,
+      "loss": 0.8347,
+      "step": 2664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034169593476488e-05,
+      "loss": 0.3153,
+      "step": 2665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027019846648316e-05,
+      "loss": 0.8449,
+      "step": 2666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019872510611775e-05,
+      "loss": 0.8281,
+      "step": 2667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012727589703844e-05,
+      "loss": 0.8659,
+      "step": 2668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005585088260038e-05,
+      "loss": 0.8472,
+      "step": 2669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9984450106144075e-05,
+      "loss": 0.8926,
+      "step": 2670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991307361099521e-05,
+      "loss": 0.8109,
+      "step": 2671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9841721440464916e-05,
+      "loss": 0.8207,
+      "step": 2672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977039363784937e-05,
+      "loss": 0.8271,
+      "step": 2673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9699090246430146e-05,
+      "loss": 0.8312,
+      "step": 2674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962781130947379e-05,
+      "loss": 0.8472,
+      "step": 2675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955655687023226e-05,
+      "loss": 0.7722,
+      "step": 2676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9485326971942445e-05,
+      "loss": 0.8622,
+      "step": 2677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941412165782645e-05,
+      "loss": 0.8464,
+      "step": 2678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9342940971091385e-05,
+      "loss": 0.8436,
+      "step": 2679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9271784954929556e-05,
+      "loss": 0.8501,
+      "step": 2680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.92006536525181e-05,
+      "loss": 0.8555,
+      "step": 2681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.912954710701936e-05,
+      "loss": 0.7874,
+      "step": 2682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9058465361580464e-05,
+      "loss": 0.805,
+      "step": 2683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898740845933366e-05,
+      "loss": 0.8342,
+      "step": 2684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891637644339602e-05,
+      "loss": 0.8398,
+      "step": 2685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884536935686955e-05,
+      "loss": 0.8586,
+      "step": 2686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87743872428411e-05,
+      "loss": 0.8062,
+      "step": 2687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8703430144382465e-05,
+      "loss": 0.7907,
+      "step": 2688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863249810455007e-05,
+      "loss": 0.8287,
+      "step": 2689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8561591166385345e-05,
+      "loss": 0.8382,
+      "step": 2690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849070937291434e-05,
+      "loss": 0.858,
+      "step": 2691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8419852767147943e-05,
+      "loss": 0.872,
+      "step": 2692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834902139208166e-05,
+      "loss": 0.8271,
+      "step": 2693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827821529069579e-05,
+      "loss": 0.7966,
+      "step": 2694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82074345059552e-05,
+      "loss": 0.8092,
+      "step": 2695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813667908080951e-05,
+      "loss": 0.8575,
+      "step": 2696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806594905819283e-05,
+      "loss": 0.8413,
+      "step": 2697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7995244481023936e-05,
+      "loss": 0.7733,
+      "step": 2698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792456539220612e-05,
+      "loss": 0.8091,
+      "step": 2699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785391183462727e-05,
+      "loss": 0.8115,
+      "step": 2700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778328385115972e-05,
+      "loss": 0.3157,
+      "step": 2701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771268148466027e-05,
+      "loss": 0.8006,
+      "step": 2702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764210477797021e-05,
+      "loss": 0.8381,
+      "step": 2703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757155377391532e-05,
+      "loss": 0.8262,
+      "step": 2704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7501028515305653e-05,
+      "loss": 0.8371,
+      "step": 2705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.743052904493578e-05,
+      "loss": 0.8447,
+      "step": 2706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7360055405584464e-05,
+      "loss": 0.8439,
+      "step": 2707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7289607640014985e-05,
+      "loss": 0.8404,
+      "step": 2708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721918579097477e-05,
+      "loss": 0.8464,
+      "step": 2709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7148789901195575e-05,
+      "loss": 0.8354,
+      "step": 2710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7078420013393367e-05,
+      "loss": 0.7938,
+      "step": 2711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700807617026842e-05,
+      "loss": 0.8285,
+      "step": 2712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6937758414505114e-05,
+      "loss": 0.8111,
+      "step": 2713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6867466788772094e-05,
+      "loss": 0.812,
+      "step": 2714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679720133572206e-05,
+      "loss": 0.8443,
+      "step": 2715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672696209799184e-05,
+      "loss": 0.8296,
+      "step": 2716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665674911820238e-05,
+      "loss": 0.855,
+      "step": 2717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6586562438958746e-05,
+      "loss": 0.8542,
+      "step": 2718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651640210284992e-05,
+      "loss": 0.8382,
+      "step": 2719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644626815244903e-05,
+      "loss": 0.8742,
+      "step": 2720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637616063031308e-05,
+      "loss": 0.3077,
+      "step": 2721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630607957898315e-05,
+      "loss": 0.8509,
+      "step": 2722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623602504098415e-05,
+      "loss": 0.7795,
+      "step": 2723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6165997058824984e-05,
+      "loss": 0.8523,
+      "step": 2724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6095995674998326e-05,
+      "loss": 0.8962,
+      "step": 2725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60260209319809e-05,
+      "loss": 0.83,
+      "step": 2726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595607287223308e-05,
+      "loss": 0.8432,
+      "step": 2727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5886151538199185e-05,
+      "loss": 0.7941,
+      "step": 2728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581625697230717e-05,
+      "loss": 0.8347,
+      "step": 2729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5746389216968975e-05,
+      "loss": 0.8746,
+      "step": 2730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567654831457998e-05,
+      "loss": 0.8317,
+      "step": 2731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56067343075195e-05,
+      "loss": 0.3567,
+      "step": 2732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5536947238150396e-05,
+      "loss": 0.8406,
+      "step": 2733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546718714881933e-05,
+      "loss": 0.8374,
+      "step": 2734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539745408185639e-05,
+      "loss": 0.8879,
+      "step": 2735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532774807957548e-05,
+      "loss": 0.816,
+      "step": 2736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525806918427394e-05,
+      "loss": 0.7937,
+      "step": 2737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518841743823269e-05,
+      "loss": 0.8169,
+      "step": 2738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.511879288371615e-05,
+      "loss": 0.8104,
+      "step": 2739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5049195562972365e-05,
+      "loss": 0.8027,
+      "step": 2740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497962551823266e-05,
+      "loss": 0.7764,
+      "step": 2741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491008279171204e-05,
+      "loss": 0.8138,
+      "step": 2742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484056742560868e-05,
+      "loss": 0.779,
+      "step": 2743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477107946210438e-05,
+      "loss": 0.795,
+      "step": 2744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470161894336418e-05,
+      "loss": 0.8244,
+      "step": 2745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463218591153651e-05,
+      "loss": 0.8569,
+      "step": 2746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456278040875305e-05,
+      "loss": 0.3388,
+      "step": 2747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449340247712893e-05,
+      "loss": 0.7785,
+      "step": 2748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4424052158762375e-05,
+      "loss": 0.8314,
+      "step": 2749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4354729495735013e-05,
+      "loss": 0.8543,
+      "step": 2750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4285434530111525e-05,
+      "loss": 0.8221,
+      "step": 2751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421616730394e-05,
+      "loss": 0.8328,
+      "step": 2752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414692785925142e-05,
+      "loss": 0.7858,
+      "step": 2753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.407771623806017e-05,
+      "loss": 0.8815,
+      "step": 2754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400853248236357e-05,
+      "loss": 0.8236,
+      "step": 2755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393937663414216e-05,
+      "loss": 0.8076,
+      "step": 2756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387024873535942e-05,
+      "loss": 0.8739,
+      "step": 2757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3801148827962014e-05,
+      "loss": 0.8482,
+      "step": 2758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373207695387952e-05,
+      "loss": 0.8305,
+      "step": 2759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36630331550245e-05,
+      "loss": 0.8055,
+      "step": 2760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359401747329249e-05,
+      "loss": 0.8701,
+      "step": 2761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352502995056208e-05,
+      "loss": 0.835,
+      "step": 2762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3456070628694555e-05,
+      "loss": 0.8894,
+      "step": 2763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3387139549534325e-05,
+      "loss": 0.8472,
+      "step": 2764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331823675490846e-05,
+      "loss": 0.8349,
+      "step": 2765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3249362286627027e-05,
+      "loss": 0.8246,
+      "step": 2766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318051618648281e-05,
+      "loss": 0.8236,
+      "step": 2767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311169849625137e-05,
+      "loss": 0.3165,
+      "step": 2768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304290925769106e-05,
+      "loss": 0.8086,
+      "step": 2769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2974148512543034e-05,
+      "loss": 0.8059,
+      "step": 2770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2905416302531e-05,
+      "loss": 0.8329,
+      "step": 2771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283671266936152e-05,
+      "loss": 0.8161,
+      "step": 2772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2768037654723734e-05,
+      "loss": 0.7899,
+      "step": 2773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269939130028938e-05,
+      "loss": 0.8792,
+      "step": 2774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263077364771284e-05,
+      "loss": 0.3184,
+      "step": 2775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256218473863114e-05,
+      "loss": 0.8903,
+      "step": 2776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2493624614663736e-05,
+      "loss": 0.3129,
+      "step": 2777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242509331741278e-05,
+      "loss": 0.8572,
+      "step": 2778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235659088846275e-05,
+      "loss": 0.8354,
+      "step": 2779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2288117369380794e-05,
+      "loss": 0.8464,
+      "step": 2780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221967280171638e-05,
+      "loss": 0.3211,
+      "step": 2781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215125722700146e-05,
+      "loss": 0.8214,
+      "step": 2782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2082870686750326e-05,
+      "loss": 0.8438,
+      "step": 2783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2014513222459804e-05,
+      "loss": 0.8502,
+      "step": 2784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1946184875608915e-05,
+      "loss": 0.8356,
+      "step": 2785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1877885687659144e-05,
+      "loss": 0.8035,
+      "step": 2786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180961570005414e-05,
+      "loss": 0.3349,
+      "step": 2787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174137495422e-05,
+      "loss": 0.8203,
+      "step": 2788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167316349156495e-05,
+      "loss": 0.7911,
+      "step": 2789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1604981353479466e-05,
+      "loss": 0.8455,
+      "step": 2790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1536828581336236e-05,
+      "loss": 0.8042,
+      "step": 2791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1468705216490235e-05,
+      "loss": 0.8247,
+      "step": 2792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140061130027839e-05,
+      "loss": 0.8721,
+      "step": 2793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133254687401997e-05,
+      "loss": 0.3049,
+      "step": 2794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1264511979016205e-05,
+      "loss": 0.8345,
+      "step": 2795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119650665655046e-05,
+      "loss": 0.8562,
+      "step": 2796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1128530947888095e-05,
+      "loss": 0.858,
+      "step": 2797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106058489427663e-05,
+      "loss": 0.8076,
+      "step": 2798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0992668536945444e-05,
+      "loss": 0.813,
+      "step": 2799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0924781917106044e-05,
+      "loss": 0.8338,
+      "step": 2800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0856925075951736e-05,
+      "loss": 0.8969,
+      "step": 2801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0789098054657904e-05,
+      "loss": 0.8511,
+      "step": 2802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072130089438172e-05,
+      "loss": 0.8188,
+      "step": 2803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0653533636262306e-05,
+      "loss": 0.8587,
+      "step": 2804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058579632142057e-05,
+      "loss": 0.8372,
+      "step": 2805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051808899095937e-05,
+      "loss": 0.842,
+      "step": 2806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045041168596322e-05,
+      "loss": 0.8433,
+      "step": 2807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038276444749857e-05,
+      "loss": 0.8684,
+      "step": 2808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.031514731661347e-05,
+      "loss": 0.816,
+      "step": 2809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0247560334337864e-05,
+      "loss": 0.8463,
+      "step": 2810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0180003541683274e-05,
+      "loss": 0.8324,
+      "step": 2811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011247697964294e-05,
+      "loss": 0.8595,
+      "step": 2812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004498068919172e-05,
+      "loss": 0.335,
+      "step": 2813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9977514711286246e-05,
+      "loss": 0.7826,
+      "step": 2814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991007908686455e-05,
+      "loss": 0.7974,
+      "step": 2815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984267385684646e-05,
+      "loss": 0.8839,
+      "step": 2816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977529906213318e-05,
+      "loss": 0.8461,
+      "step": 2817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970795474360757e-05,
+      "loss": 0.8004,
+      "step": 2818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964064094213385e-05,
+      "loss": 0.8236,
+      "step": 2819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957335769855794e-05,
+      "loss": 0.8022,
+      "step": 2820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950610505370701e-05,
+      "loss": 0.8564,
+      "step": 2821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.943888304838982e-05,
+      "loss": 0.8797,
+      "step": 2822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9371691723396375e-05,
+      "loss": 0.8256,
+      "step": 2823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.930453111949826e-05,
+      "loss": 0.848,
+      "step": 2824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923740127744827e-05,
+      "loss": 0.817,
+      "step": 2825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917030223798057e-05,
+      "loss": 0.3166,
+      "step": 2826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91032340418106e-05,
+      "loss": 0.8371,
+      "step": 2827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9036196729635216e-05,
+      "loss": 0.8669,
+      "step": 2828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896919034213237e-05,
+      "loss": 0.8191,
+      "step": 2829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890221491996138e-05,
+      "loss": 0.3252,
+      "step": 2830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8835270503762655e-05,
+      "loss": 0.8754,
+      "step": 2831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8768357134157924e-05,
+      "loss": 0.8319,
+      "step": 2832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8701474851749976e-05,
+      "loss": 0.8274,
+      "step": 2833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863462369712275e-05,
+      "loss": 0.8842,
+      "step": 2834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85678037108413e-05,
+      "loss": 0.8358,
+      "step": 2835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850101493345183e-05,
+      "loss": 0.8902,
+      "step": 2836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843425740548149e-05,
+      "loss": 0.8307,
+      "step": 2837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8367531167438605e-05,
+      "loss": 0.8212,
+      "step": 2838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8300836259812424e-05,
+      "loss": 0.8353,
+      "step": 2839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823417272307318e-05,
+      "loss": 0.8099,
+      "step": 2840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816754059767206e-05,
+      "loss": 0.85,
+      "step": 2841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810093992404131e-05,
+      "loss": 0.8243,
+      "step": 2842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8034370742593914e-05,
+      "loss": 0.8111,
+      "step": 2843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796783309372393e-05,
+      "loss": 0.8784,
+      "step": 2844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790132701780612e-05,
+      "loss": 0.8347,
+      "step": 2845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783485255519621e-05,
+      "loss": 0.8585,
+      "step": 2846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7768409746230666e-05,
+      "loss": 0.8481,
+      "step": 2847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7701998631226773e-05,
+      "loss": 0.7746,
+      "step": 2848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7635619250482544e-05,
+      "loss": 0.8082,
+      "step": 2849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756927164427685e-05,
+      "loss": 0.8529,
+      "step": 2850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7502955852869126e-05,
+      "loss": 0.3361,
+      "step": 2851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743667191649965e-05,
+      "loss": 0.8495,
+      "step": 2852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737041987538926e-05,
+      "loss": 0.8076,
+      "step": 2853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730419976973952e-05,
+      "loss": 0.8071,
+      "step": 2854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723801163973256e-05,
+      "loss": 0.8034,
+      "step": 2855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7171855525531114e-05,
+      "loss": 0.8252,
+      "step": 2856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710573146727848e-05,
+      "loss": 0.7847,
+      "step": 2857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703963950509858e-05,
+      "loss": 0.8208,
+      "step": 2858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697357967909574e-05,
+      "loss": 0.8183,
+      "step": 2859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690755202935493e-05,
+      "loss": 0.8081,
+      "step": 2860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684155659594147e-05,
+      "loss": 0.8385,
+      "step": 2861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6775593418901175e-05,
+      "loss": 0.8547,
+      "step": 2862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6709662538260267e-05,
+      "loss": 0.8408,
+      "step": 2863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664376399402546e-05,
+      "loss": 0.3276,
+      "step": 2864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657789782618369e-05,
+      "loss": 0.857,
+      "step": 2865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6512064074702444e-05,
+      "loss": 0.8247,
+      "step": 2866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6446262779529326e-05,
+      "loss": 0.795,
+      "step": 2867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638049398059244e-05,
+      "loss": 0.8398,
+      "step": 2868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631475771780005e-05,
+      "loss": 0.8592,
+      "step": 2869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624905403104069e-05,
+      "loss": 0.8537,
+      "step": 2870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618338296018314e-05,
+      "loss": 0.8144,
+      "step": 2871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611774454507645e-05,
+      "loss": 0.8329,
+      "step": 2872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6052138825549715e-05,
+      "loss": 0.824,
+      "step": 2873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5986565841412364e-05,
+      "loss": 0.796,
+      "step": 2874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5921025632453804e-05,
+      "loss": 0.8632,
+      "step": 2875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5855518238443696e-05,
+      "loss": 0.8377,
+      "step": 2876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579004369913167e-05,
+      "loss": 0.8081,
+      "step": 2877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5724602054247464e-05,
+      "loss": 0.8549,
+      "step": 2878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565919334350084e-05,
+      "loss": 0.804,
+      "step": 2879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5593817606581656e-05,
+      "loss": 0.8624,
+      "step": 2880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552847488315963e-05,
+      "loss": 0.8169,
+      "step": 2881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546316521288459e-05,
+      "loss": 0.8611,
+      "step": 2882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53978886353862e-05,
+      "loss": 0.8291,
+      "step": 2883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5332645190274056e-05,
+      "loss": 0.799,
+      "step": 2884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5267434917137676e-05,
+      "loss": 0.8102,
+      "step": 2885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5202257855546484e-05,
+      "loss": 0.7886,
+      "step": 2886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5137114045049654e-05,
+      "loss": 0.8197,
+      "step": 2887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507200352517631e-05,
+      "loss": 0.3443,
+      "step": 2888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500692633543524e-05,
+      "loss": 0.877,
+      "step": 2889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4941882515315135e-05,
+      "loss": 0.8455,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487687210428436e-05,
+      "loss": 0.8267,
+      "step": 2891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481189514179101e-05,
+      "loss": 0.3516,
+      "step": 2892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474695166726285e-05,
+      "loss": 0.8744,
+      "step": 2893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468204172010747e-05,
+      "loss": 0.8333,
+      "step": 2894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461716533971193e-05,
+      "loss": 0.7974,
+      "step": 2895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455232256544308e-05,
+      "loss": 0.8038,
+      "step": 2896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448751343664725e-05,
+      "loss": 0.8079,
+      "step": 2897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442273799265052e-05,
+      "loss": 0.8142,
+      "step": 2898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435799627275825e-05,
+      "loss": 0.8512,
+      "step": 2899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4293288316255653e-05,
+      "loss": 0.8528,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4228614162407214e-05,
+      "loss": 0.828,
+      "step": 2901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416397385045707e-05,
+      "loss": 0.8983,
+      "step": 2902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4099367419628704e-05,
+      "loss": 0.848,
+      "step": 2903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4034794909125156e-05,
+      "loss": 0.871,
+      "step": 2904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397025635812877e-05,
+      "loss": 0.7842,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390575180580134e-05,
+      "loss": 0.326,
+      "step": 2906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3841281291283996e-05,
+      "loss": 0.828,
+      "step": 2907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3776844853697305e-05,
+      "loss": 0.7734,
+      "step": 2908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371244253214101e-05,
+      "loss": 0.8069,
+      "step": 2909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3648074365694314e-05,
+      "loss": 0.7975,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358374039341554e-05,
+      "loss": 0.7928,
+      "step": 2911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3519440654342416e-05,
+      "loss": 0.8359,
+      "step": 2912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345517518749177e-05,
+      "loss": 0.8483,
+      "step": 2913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339094403185968e-05,
+      "loss": 0.8628,
+      "step": 2914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332674722642138e-05,
+      "loss": 0.8202,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326258481013136e-05,
+      "loss": 0.8039,
+      "step": 2916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319845682192307e-05,
+      "loss": 0.8471,
+      "step": 2917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313436330070928e-05,
+      "loss": 0.8513,
+      "step": 2918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307030428538163e-05,
+      "loss": 0.8533,
+      "step": 2919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300627981481106e-05,
+      "loss": 0.8629,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2942289927847236e-05,
+      "loss": 0.7872,
+      "step": 2921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287833466331915e-05,
+      "loss": 0.7982,
+      "step": 2922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281441406003457e-05,
+      "loss": 0.8324,
+      "step": 2923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275052815678039e-05,
+      "loss": 0.8571,
+      "step": 2924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268667699232229e-05,
+      "loss": 0.8384,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262286060540505e-05,
+      "loss": 0.8531,
+      "step": 2926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255907903475218e-05,
+      "loss": 0.8308,
+      "step": 2927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249533231906614e-05,
+      "loss": 0.8259,
+      "step": 2928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243162049702821e-05,
+      "loss": 0.8321,
+      "step": 2929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236794360729859e-05,
+      "loss": 0.8599,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2304301688516135e-05,
+      "loss": 0.8043,
+      "step": 2931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224069477929863e-05,
+      "loss": 0.8136,
+      "step": 2932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217712291824247e-05,
+      "loss": 0.8862,
+      "step": 2933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211358614392294e-05,
+      "loss": 0.8354,
+      "step": 2934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2050084494893896e-05,
+      "loss": 0.8178,
+      "step": 2935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198661800968794e-05,
+      "loss": 0.8497,
+      "step": 2936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192318672681631e-05,
+      "loss": 0.8274,
+      "step": 2937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185979068476896e-05,
+      "loss": 0.8261,
+      "step": 2938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1796429922014355e-05,
+      "loss": 0.8214,
+      "step": 2939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173310447699965e-05,
+      "loss": 0.7824,
+      "step": 2940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166981438815051e-05,
+      "loss": 0.8142,
+      "step": 2941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1606559693871125e-05,
+      "loss": 0.8576,
+      "step": 2942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154334043254424e-05,
+      "loss": 0.8076,
+      "step": 2943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148015664253116e-05,
+      "loss": 0.8343,
+      "step": 2944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141700836217153e-05,
+      "loss": 0.8602,
+      "step": 2945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13538956297836e-05,
+      "loss": 0.8077,
+      "step": 2946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1290818483663894e-05,
+      "loss": 0.829,
+      "step": 2947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1227776962087506e-05,
+      "loss": 0.8479,
+      "step": 2948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116477110330779e-05,
+      "loss": 0.3196,
+      "step": 2949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110180094555649e-05,
+      "loss": 0.8297,
+      "step": 2950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103886652704367e-05,
+      "loss": 0.8213,
+      "step": 2951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09759678859578e-05,
+      "loss": 0.3376,
+      "step": 2952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091310506046551e-05,
+      "loss": 0.3389,
+      "step": 2953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0850278088711825e-05,
+      "loss": 0.7846,
+      "step": 2954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078748700881988e-05,
+      "loss": 0.7849,
+      "step": 2955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0724731858891176e-05,
+      "loss": 0.8592,
+      "step": 2956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0662012677005304e-05,
+      "loss": 0.8022,
+      "step": 2957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059932950122005e-05,
+      "loss": 0.7988,
+      "step": 2958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053668236957134e-05,
+      "loss": 0.8359,
+      "step": 2959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0474071320073314e-05,
+      "loss": 0.8661,
+      "step": 2960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0411496390718104e-05,
+      "loss": 0.8974,
+      "step": 2961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0348957619476027e-05,
+      "loss": 0.8442,
+      "step": 2962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0286455044295376e-05,
+      "loss": 0.8204,
+      "step": 2963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0223988703102534e-05,
+      "loss": 0.8729,
+      "step": 2964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016155863380182e-05,
+      "loss": 0.8398,
+      "step": 2965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0099164874275674e-05,
+      "loss": 0.7963,
+      "step": 2966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0036807462384364e-05,
+      "loss": 0.8026,
+      "step": 2967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997448643596624e-05,
+      "loss": 0.3059,
+      "step": 2968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9912201832837424e-05,
+      "loss": 0.8995,
+      "step": 2969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984995369079209e-05,
+      "loss": 0.8259,
+      "step": 2970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978774204760216e-05,
+      "loss": 0.8137,
+      "step": 2971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972556694101749e-05,
+      "loss": 0.3336,
+      "step": 2972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966342840876568e-05,
+      "loss": 0.7836,
+      "step": 2973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960132648855226e-05,
+      "loss": 0.3596,
+      "step": 2974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953926121806042e-05,
+      "loss": 0.8172,
+      "step": 2975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947723263495122e-05,
+      "loss": 0.851,
+      "step": 2976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9415240776863346e-05,
+      "loss": 0.8331,
+      "step": 2977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9353285681413334e-05,
+      "loss": 0.8194,
+      "step": 2978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92913673861953e-05,
+      "loss": 0.8355,
+      "step": 2979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922948592878105e-05,
+      "loss": 0.8121,
+      "step": 2980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916764134672004e-05,
+      "loss": 0.8405,
+      "step": 2981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910583367753943e-05,
+      "loss": 0.8347,
+      "step": 2982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9044062958743844e-05,
+      "loss": 0.8045,
+      "step": 2983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898232922781562e-05,
+      "loss": 0.8108,
+      "step": 2984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892063252221455e-05,
+      "loss": 0.8398,
+      "step": 2985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885897287937802e-05,
+      "loss": 0.8244,
+      "step": 2986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879735033672085e-05,
+      "loss": 0.8673,
+      "step": 2987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8735764931635476e-05,
+      "loss": 0.8032,
+      "step": 2988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867421670149165e-05,
+      "loss": 0.8084,
+      "step": 2989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861270568363672e-05,
+      "loss": 0.8213,
+      "step": 2990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855123191539529e-05,
+      "loss": 0.8263,
+      "step": 2991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8489795434069533e-05,
+      "loss": 0.8181,
+      "step": 2992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8428396276938874e-05,
+      "loss": 0.7839,
+      "step": 2993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83670344812601e-05,
+      "loss": 0.8421,
+      "step": 2994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8305710084267346e-05,
+      "loss": 0.8114,
+      "step": 2995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8244423123172115e-05,
+      "loss": 0.8204,
+      "step": 2996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8183173635163075e-05,
+      "loss": 0.867,
+      "step": 2997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812196165740628e-05,
+      "loss": 0.8209,
+      "step": 2998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806078722704489e-05,
+      "loss": 0.7828,
+      "step": 2999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799965038119945e-05,
+      "loss": 0.8331,
+      "step": 3000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.793855115696754e-05,
+      "loss": 0.8668,
+      "step": 3001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.787748959142398e-05,
+      "loss": 0.8525,
+      "step": 3002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78164657216207e-05,
+      "loss": 0.8342,
+      "step": 3003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775547958458686e-05,
+      "loss": 0.8301,
+      "step": 3004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769453121732858e-05,
+      "loss": 0.859,
+      "step": 3005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763362065682919e-05,
+      "loss": 0.8298,
+      "step": 3006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7572747940049026e-05,
+      "loss": 0.8132,
+      "step": 3007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7511913103925425e-05,
+      "loss": 0.8086,
+      "step": 3008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7451116185372735e-05,
+      "loss": 0.8624,
+      "step": 3009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.739035722128243e-05,
+      "loss": 0.8206,
+      "step": 3010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.732963624852275e-05,
+      "loss": 0.7998,
+      "step": 3011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726895330393908e-05,
+      "loss": 0.8392,
+      "step": 3012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7208308424353555e-05,
+      "loss": 0.7744,
+      "step": 3013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.714770164656537e-05,
+      "loss": 0.8252,
+      "step": 3014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7087133007350494e-05,
+      "loss": 0.7704,
+      "step": 3015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702660254346176e-05,
+      "loss": 0.8256,
+      "step": 3016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6966110291628855e-05,
+      "loss": 0.8105,
+      "step": 3017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6905656288558336e-05,
+      "loss": 0.8237,
+      "step": 3018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6845240570933424e-05,
+      "loss": 0.8253,
+      "step": 3019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.678486317541426e-05,
+      "loss": 0.8537,
+      "step": 3020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.672452413863759e-05,
+      "loss": 0.815,
+      "step": 3021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666422349721701e-05,
+      "loss": 0.8277,
+      "step": 3022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660396128774273e-05,
+      "loss": 0.85,
+      "step": 3023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6543737546781656e-05,
+      "loss": 0.8465,
+      "step": 3024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.648355231087732e-05,
+      "loss": 0.8297,
+      "step": 3025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642340561655002e-05,
+      "loss": 0.8279,
+      "step": 3026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636329750029649e-05,
+      "loss": 0.8526,
+      "step": 3027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630322799859023e-05,
+      "loss": 0.8633,
+      "step": 3028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.624319714788117e-05,
+      "loss": 0.8044,
+      "step": 3029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618320498459584e-05,
+      "loss": 0.8514,
+      "step": 3030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.612325154513726e-05,
+      "loss": 0.813,
+      "step": 3031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6063336865885064e-05,
+      "loss": 0.802,
+      "step": 3032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60034609831952e-05,
+      "loss": 0.868,
+      "step": 3033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594362393340025e-05,
+      "loss": 0.8078,
+      "step": 3034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588382575280906e-05,
+      "loss": 0.8313,
+      "step": 3035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.582406647770706e-05,
+      "loss": 0.8106,
+      "step": 3036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5764346144355934e-05,
+      "loss": 0.795,
+      "step": 3037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57046647889938e-05,
+      "loss": 0.8331,
+      "step": 3038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5645022447835076e-05,
+      "loss": 0.8069,
+      "step": 3039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558541915707061e-05,
+      "loss": 0.8753,
+      "step": 3040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.552585495286742e-05,
+      "loss": 0.8512,
+      "step": 3041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5466329871368945e-05,
+      "loss": 0.8217,
+      "step": 3042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5406843948694735e-05,
+      "loss": 0.8037,
+      "step": 3043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534739722094075e-05,
+      "loss": 0.8642,
+      "step": 3044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528798972417902e-05,
+      "loss": 0.813,
+      "step": 3045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522862149445783e-05,
+      "loss": 0.8257,
+      "step": 3046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5169292567801584e-05,
+      "loss": 0.8013,
+      "step": 3047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5110002980210975e-05,
+      "loss": 0.8489,
+      "step": 3048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5050752767662655e-05,
+      "loss": 0.8237,
+      "step": 3049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4991541966109534e-05,
+      "loss": 0.8417,
+      "step": 3050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49323706114805e-05,
+      "loss": 0.8568,
+      "step": 3051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487323873968055e-05,
+      "loss": 0.8137,
+      "step": 3052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4814146386590664e-05,
+      "loss": 0.8359,
+      "step": 3053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475509358806797e-05,
+      "loss": 0.8309,
+      "step": 3054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469608037994545e-05,
+      "loss": 0.8499,
+      "step": 3055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4637106798032204e-05,
+      "loss": 0.8232,
+      "step": 3056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4578172878113124e-05,
+      "loss": 0.7704,
+      "step": 3057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.451927865594922e-05,
+      "loss": 0.7888,
+      "step": 3058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446042416727726e-05,
+      "loss": 0.8891,
+      "step": 3059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4401609447809965e-05,
+      "loss": 0.8193,
+      "step": 3060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.434283453323589e-05,
+      "loss": 0.8339,
+      "step": 3061
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.428409945921953e-05,
+      "loss": 0.7951,
+      "step": 3062
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422540426140107e-05,
+      "loss": 0.7961,
+      "step": 3063
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4166748975396634e-05,
+      "loss": 0.8066,
+      "step": 3064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4108133636798004e-05,
+      "loss": 0.8324,
+      "step": 3065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404955828117288e-05,
+      "loss": 0.8606,
+      "step": 3066
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399102294406444e-05,
+      "loss": 0.8573,
+      "step": 3067
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393252766099187e-05,
+      "loss": 0.8469,
+      "step": 3068
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387407246744985e-05,
+      "loss": 0.8217,
+      "step": 3069
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.381565739890886e-05,
+      "loss": 0.8266,
+      "step": 3070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3757282490814915e-05,
+      "loss": 0.3054,
+      "step": 3071
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369894777858977e-05,
+      "loss": 0.8137,
+      "step": 3072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.364065329763075e-05,
+      "loss": 0.8338,
+      "step": 3073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.358239908331072e-05,
+      "loss": 0.7762,
+      "step": 3074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352418517097814e-05,
+      "loss": 0.8125,
+      "step": 3075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.346601159595708e-05,
+      "loss": 0.8222,
+      "step": 3076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340787839354702e-05,
+      "loss": 0.848,
+      "step": 3077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3349785599023065e-05,
+      "loss": 0.8412,
+      "step": 3078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329173324763567e-05,
+      "loss": 0.819,
+      "step": 3079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323372137461088e-05,
+      "loss": 0.797,
+      "step": 3080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3175750015150084e-05,
+      "loss": 0.7482,
+      "step": 3081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311781920443011e-05,
+      "loss": 0.8388,
+      "step": 3082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.305992897760317e-05,
+      "loss": 0.8457,
+      "step": 3083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300207936979692e-05,
+      "loss": 0.8351,
+      "step": 3084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.294427041611425e-05,
+      "loss": 0.8683,
+      "step": 3085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288650215163354e-05,
+      "loss": 0.8281,
+      "step": 3086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2828774611408285e-05,
+      "loss": 0.8112,
+      "step": 3087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.277108783046748e-05,
+      "loss": 0.8142,
+      "step": 3088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2713441843815174e-05,
+      "loss": 0.7936,
+      "step": 3089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.265583668643083e-05,
+      "loss": 0.8083,
+      "step": 3090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2598272393269035e-05,
+      "loss": 0.8645,
+      "step": 3091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2540748999259665e-05,
+      "loss": 0.8451,
+      "step": 3092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2483266539307675e-05,
+      "loss": 0.8571,
+      "step": 3093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2425825048293304e-05,
+      "loss": 0.3177,
+      "step": 3094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236842456107182e-05,
+      "loss": 0.3221,
+      "step": 3095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231106511247366e-05,
+      "loss": 0.8863,
+      "step": 3096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2253746737304333e-05,
+      "loss": 0.8372,
+      "step": 3097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2196469470344505e-05,
+      "loss": 0.8306,
+      "step": 3098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213923334634976e-05,
+      "loss": 0.8054,
+      "step": 3099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2082038400050875e-05,
+      "loss": 0.7837,
+      "step": 3100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202488466615349e-05,
+      "loss": 0.7879,
+      "step": 3101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196777217933836e-05,
+      "loss": 0.8035,
+      "step": 3102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.191070097426115e-05,
+      "loss": 0.8401,
+      "step": 3103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185367108555247e-05,
+      "loss": 0.7964,
+      "step": 3104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1796682547817845e-05,
+      "loss": 0.8498,
+      "step": 3105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173973539563781e-05,
+      "loss": 0.3167,
+      "step": 3106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168282966356764e-05,
+      "loss": 0.8304,
+      "step": 3107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1625965386137604e-05,
+      "loss": 0.326,
+      "step": 3108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156914259785275e-05,
+      "loss": 0.8397,
+      "step": 3109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1512361333192966e-05,
+      "loss": 0.8014,
+      "step": 3110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1455621626612885e-05,
+      "loss": 0.8278,
+      "step": 3111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139892351254208e-05,
+      "loss": 0.8555,
+      "step": 3112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1342267025384686e-05,
+      "loss": 0.34,
+      "step": 3113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1285652199519756e-05,
+      "loss": 0.8071,
+      "step": 3114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122907906930093e-05,
+      "loss": 0.8458,
+      "step": 3115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.117254766905665e-05,
+      "loss": 0.8537,
+      "step": 3116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111605803308997e-05,
+      "loss": 0.8302,
+      "step": 3117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1059610195678615e-05,
+      "loss": 0.8729,
+      "step": 3118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.100320419107492e-05,
+      "loss": 0.8342,
+      "step": 3119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.094684005350592e-05,
+      "loss": 0.8129,
+      "step": 3120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089051781717315e-05,
+      "loss": 0.8096,
+      "step": 3121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083423751625282e-05,
+      "loss": 0.7584,
+      "step": 3122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0777999184895556e-05,
+      "loss": 0.8166,
+      "step": 3123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.072180285722669e-05,
+      "loss": 0.8047,
+      "step": 3124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066564856734592e-05,
+      "loss": 0.8663,
+      "step": 3125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060953634932749e-05,
+      "loss": 0.8475,
+      "step": 3126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055346623722009e-05,
+      "loss": 0.8242,
+      "step": 3127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.049743826504695e-05,
+      "loss": 0.8339,
+      "step": 3128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.04414524668056e-05,
+      "loss": 0.339,
+      "step": 3129
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0385508876468095e-05,
+      "loss": 0.8246,
+      "step": 3130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0329607527980807e-05,
+      "loss": 0.3246,
+      "step": 3131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0273748455264484e-05,
+      "loss": 0.832,
+      "step": 3132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021793169221421e-05,
+      "loss": 0.8063,
+      "step": 3133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0162157272699486e-05,
+      "loss": 0.7835,
+      "step": 3134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0106425230563973e-05,
+      "loss": 0.8366,
+      "step": 3135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0050735599625766e-05,
+      "loss": 0.8367,
+      "step": 3136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9995088413677096e-05,
+      "loss": 0.81,
+      "step": 3137
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993948370648455e-05,
+      "loss": 0.8445,
+      "step": 3138
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.988392151178887e-05,
+      "loss": 0.8238,
+      "step": 3139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9828401863305e-05,
+      "loss": 0.864,
+      "step": 3140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977292479472207e-05,
+      "loss": 0.8466,
+      "step": 3141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9717490339703435e-05,
+      "loss": 0.8841,
+      "step": 3142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9662098531886485e-05,
+      "loss": 0.8221,
+      "step": 3143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9606749404882873e-05,
+      "loss": 0.8293,
+      "step": 3144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9551442992278187e-05,
+      "loss": 0.8615,
+      "step": 3145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9496179327632246e-05,
+      "loss": 0.8252,
+      "step": 3146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944095844447885e-05,
+      "loss": 0.8263,
+      "step": 3147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9385780376325833e-05,
+      "loss": 0.8484,
+      "step": 3148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9330645156655057e-05,
+      "loss": 0.8559,
+      "step": 3149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9275552818922448e-05,
+      "loss": 0.8858,
+      "step": 3150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9220503396557797e-05,
+      "loss": 0.8582,
+      "step": 3151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9165496922964985e-05,
+      "loss": 0.8158,
+      "step": 3152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9110533431521723e-05,
+      "loss": 0.8484,
+      "step": 3153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9055612955579704e-05,
+      "loss": 0.8335,
+      "step": 3154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.900073552846444e-05,
+      "loss": 0.8167,
+      "step": 3155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8945901183475454e-05,
+      "loss": 0.7723,
+      "step": 3156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8891109953885987e-05,
+      "loss": 0.8266,
+      "step": 3157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8836361872943242e-05,
+      "loss": 0.3239,
+      "step": 3158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878165697386812e-05,
+      "loss": 0.8622,
+      "step": 3159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.872699528985543e-05,
+      "loss": 0.8611,
+      "step": 3160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.867237685407369e-05,
+      "loss": 0.8714,
+      "step": 3161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.861780169966519e-05,
+      "loss": 0.8459,
+      "step": 3162
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8563269859745922e-05,
+      "loss": 0.8403,
+      "step": 3163
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850878136740569e-05,
+      "loss": 0.8306,
+      "step": 3164
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845433625570788e-05,
+      "loss": 0.8029,
+      "step": 3165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.839993455768968e-05,
+      "loss": 0.8429,
+      "step": 3166
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8345576306361798e-05,
+      "loss": 0.7997,
+      "step": 3167
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8291261534708702e-05,
+      "loss": 0.8359,
+      "step": 3168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82369902756884e-05,
+      "loss": 0.801,
+      "step": 3169
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.818276256223251e-05,
+      "loss": 0.7653,
+      "step": 3170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812857842724621e-05,
+      "loss": 0.8017,
+      "step": 3171
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.807443790360833e-05,
+      "loss": 0.8221,
+      "step": 3172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802034102417108e-05,
+      "loss": 0.8345,
+      "step": 3173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7966287821760362e-05,
+      "loss": 0.8515,
+      "step": 3174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7912278329175435e-05,
+      "loss": 0.8641,
+      "step": 3175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7858312579189095e-05,
+      "loss": 0.832,
+      "step": 3176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7804390604547557e-05,
+      "loss": 0.824,
+      "step": 3177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7750512437970567e-05,
+      "loss": 0.8819,
+      "step": 3178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769667811215114e-05,
+      "loss": 0.8188,
+      "step": 3179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7642887659755867e-05,
+      "loss": 0.8301,
+      "step": 3180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7589141113424534e-05,
+      "loss": 0.8554,
+      "step": 3181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.753543850577045e-05,
+      "loss": 0.8216,
+      "step": 3182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7481779869380154e-05,
+      "loss": 0.8121,
+      "step": 3183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742816523681352e-05,
+      "loss": 0.81,
+      "step": 3184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7374594640603724e-05,
+      "loss": 0.8516,
+      "step": 3185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7321068113257274e-05,
+      "loss": 0.3235,
+      "step": 3186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7267585687253848e-05,
+      "loss": 0.8173,
+      "step": 3187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7214147395046473e-05,
+      "loss": 0.8386,
+      "step": 3188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7160753269061267e-05,
+      "loss": 0.7964,
+      "step": 3189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7107403341697667e-05,
+      "loss": 0.848,
+      "step": 3190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7054097645328223e-05,
+      "loss": 0.8244,
+      "step": 3191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7000836212298663e-05,
+      "loss": 0.8357,
+      "step": 3192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.694761907492781e-05,
+      "loss": 0.818,
+      "step": 3193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6894446265507712e-05,
+      "loss": 0.8084,
+      "step": 3194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684131781630341e-05,
+      "loss": 0.8748,
+      "step": 3195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678823375955314e-05,
+      "loss": 0.3311,
+      "step": 3196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.673519412746809e-05,
+      "loss": 0.8234,
+      "step": 3197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668219895223255e-05,
+      "loss": 0.857,
+      "step": 3198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6629248266003792e-05,
+      "loss": 0.8027,
+      "step": 3199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6576342100912188e-05,
+      "loss": 0.8171,
+      "step": 3200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652348048906096e-05,
+      "loss": 0.8116,
+      "step": 3201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6470663462526423e-05,
+      "loss": 0.8192,
+      "step": 3202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6417891053357724e-05,
+      "loss": 0.8144,
+      "step": 3203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6365163293577054e-05,
+      "loss": 0.8008,
+      "step": 3204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6312480215179424e-05,
+      "loss": 0.8128,
+      "step": 3205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625984185013275e-05,
+      "loss": 0.8324,
+      "step": 3206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620724823037779e-05,
+      "loss": 0.8191,
+      "step": 3207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615469938782825e-05,
+      "loss": 0.8342,
+      "step": 3208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.610219535437054e-05,
+      "loss": 0.8459,
+      "step": 3209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6049736161863992e-05,
+      "loss": 0.8085,
+      "step": 3210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.599732184214062e-05,
+      "loss": 0.8321,
+      "step": 3211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594495242700533e-05,
+      "loss": 0.3412,
+      "step": 3212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5892627948235682e-05,
+      "loss": 0.8135,
+      "step": 3213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5840348437581996e-05,
+      "loss": 0.8515,
+      "step": 3214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578811392676729e-05,
+      "loss": 0.8125,
+      "step": 3215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5735924447487358e-05,
+      "loss": 0.3066,
+      "step": 3216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.568378003141054e-05,
+      "loss": 0.8163,
+      "step": 3217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5631680710177962e-05,
+      "loss": 0.8023,
+      "step": 3218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557962651540331e-05,
+      "loss": 0.7929,
+      "step": 3219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5527617478672883e-05,
+      "loss": 0.814,
+      "step": 3220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5475653631545582e-05,
+      "loss": 0.8326,
+      "step": 3221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5423735005552963e-05,
+      "loss": 0.8485,
+      "step": 3222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5371861632199012e-05,
+      "loss": 0.7832,
+      "step": 3223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5320033542960386e-05,
+      "loss": 0.8169,
+      "step": 3224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526825076928615e-05,
+      "loss": 0.841,
+      "step": 3225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521651334259799e-05,
+      "loss": 0.8247,
+      "step": 3226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5164821294289963e-05,
+      "loss": 0.8627,
+      "step": 3227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5113174655728656e-05,
+      "loss": 0.8153,
+      "step": 3228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506157345825305e-05,
+      "loss": 0.8765,
+      "step": 3229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501001773317464e-05,
+      "loss": 0.8095,
+      "step": 3230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.495850751177722e-05,
+      "loss": 0.7773,
+      "step": 3231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4907042825317085e-05,
+      "loss": 0.8695,
+      "step": 3232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485562370502279e-05,
+      "loss": 0.8253,
+      "step": 3233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480425018209537e-05,
+      "loss": 0.8615,
+      "step": 3234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475292228770799e-05,
+      "loss": 0.8684,
+      "step": 3235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.470164005300636e-05,
+      "loss": 0.8619,
+      "step": 3236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465040350910831e-05,
+      "loss": 0.8368,
+      "step": 3237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4599212687104056e-05,
+      "loss": 0.8344,
+      "step": 3238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4548067618055994e-05,
+      "loss": 0.8344,
+      "step": 3239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4496968332998815e-05,
+      "loss": 0.8533,
+      "step": 3240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444591486293939e-05,
+      "loss": 0.837,
+      "step": 3241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43949072388568e-05,
+      "loss": 0.7821,
+      "step": 3242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434394549170228e-05,
+      "loss": 0.3132,
+      "step": 3243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4293029652399303e-05,
+      "loss": 0.8575,
+      "step": 3244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4242159751843386e-05,
+      "loss": 0.8266,
+      "step": 3245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4191335820902285e-05,
+      "loss": 0.8458,
+      "step": 3246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4140557890415738e-05,
+      "loss": 0.8028,
+      "step": 3247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408982599119568e-05,
+      "loss": 0.8456,
+      "step": 3248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.403914015402604e-05,
+      "loss": 0.815,
+      "step": 3249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398850040966283e-05,
+      "loss": 0.8176,
+      "step": 3250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3937906788834052e-05,
+      "loss": 0.8367,
+      "step": 3251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38873593222398e-05,
+      "loss": 0.8398,
+      "step": 3252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.383685804055207e-05,
+      "loss": 0.8986,
+      "step": 3253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3786402974414924e-05,
+      "loss": 0.8342,
+      "step": 3254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3735994154444286e-05,
+      "loss": 0.8187,
+      "step": 3255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3685631611228154e-05,
+      "loss": 0.8748,
+      "step": 3256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.363531537532624e-05,
+      "loss": 0.781,
+      "step": 3257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3585045477270352e-05,
+      "loss": 0.8259,
+      "step": 3258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3534821947564057e-05,
+      "loss": 0.8725,
+      "step": 3259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348464481668289e-05,
+      "loss": 0.8453,
+      "step": 3260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3434514115074114e-05,
+      "loss": 0.8176,
+      "step": 3261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3384429873156933e-05,
+      "loss": 0.8536,
+      "step": 3262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3334392121322267e-05,
+      "loss": 0.8496,
+      "step": 3263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.328440088993289e-05,
+      "loss": 0.7829,
+      "step": 3264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3234456209323263e-05,
+      "loss": 0.8478,
+      "step": 3265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3184558109799726e-05,
+      "loss": 0.8281,
+      "step": 3266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3134706621640235e-05,
+      "loss": 0.8336,
+      "step": 3267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308490177509457e-05,
+      "loss": 0.8355,
+      "step": 3268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3035143600384092e-05,
+      "loss": 0.7949,
+      "step": 3269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2985432127701946e-05,
+      "loss": 0.8605,
+      "step": 3270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2935767387212894e-05,
+      "loss": 0.7968,
+      "step": 3271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2886149409053313e-05,
+      "loss": 0.7999,
+      "step": 3272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283657822333123e-05,
+      "loss": 0.8422,
+      "step": 3273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2787053860126317e-05,
+      "loss": 0.8343,
+      "step": 3274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2737576349489753e-05,
+      "loss": 0.3267,
+      "step": 3275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2688145721444385e-05,
+      "loss": 0.8778,
+      "step": 3276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263876200598454e-05,
+      "loss": 0.8315,
+      "step": 3277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.258942523307609e-05,
+      "loss": 0.778,
+      "step": 3278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2540135432656407e-05,
+      "loss": 0.7812,
+      "step": 3279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249089263463443e-05,
+      "loss": 0.8529,
+      "step": 3280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2441696868890495e-05,
+      "loss": 0.8374,
+      "step": 3281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239254816527647e-05,
+      "loss": 0.8095,
+      "step": 3282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23434465536156e-05,
+      "loss": 0.8719,
+      "step": 3283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229439206370262e-05,
+      "loss": 0.8461,
+      "step": 3284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2245384725303608e-05,
+      "loss": 0.3382,
+      "step": 3285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2196424568156073e-05,
+      "loss": 0.8203,
+      "step": 3286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214751162196884e-05,
+      "loss": 0.8281,
+      "step": 3287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.20986459164222e-05,
+      "loss": 0.8235,
+      "step": 3288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.204982748116764e-05,
+      "loss": 0.8461,
+      "step": 3289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2001056345828097e-05,
+      "loss": 0.8089,
+      "step": 3290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1952332539997667e-05,
+      "loss": 0.8162,
+      "step": 3291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1903656093241866e-05,
+      "loss": 0.8613,
+      "step": 3292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1855027035097376e-05,
+      "loss": 0.8594,
+      "step": 3293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180644539507215e-05,
+      "loss": 0.8796,
+      "step": 3294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1757911202645342e-05,
+      "loss": 0.8308,
+      "step": 3295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1709424487267417e-05,
+      "loss": 0.8058,
+      "step": 3296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1660985278359892e-05,
+      "loss": 0.7955,
+      "step": 3297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1612593605315566e-05,
+      "loss": 0.8261,
+      "step": 3298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1564249497498347e-05,
+      "loss": 0.823,
+      "step": 3299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151595298424327e-05,
+      "loss": 0.8022,
+      "step": 3300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1467704094856477e-05,
+      "loss": 0.8373,
+      "step": 3301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1419502858615302e-05,
+      "loss": 0.8221,
+      "step": 3302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1371349304768052e-05,
+      "loss": 0.3179,
+      "step": 3303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1323243462534183e-05,
+      "loss": 0.8011,
+      "step": 3304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1275185361104123e-05,
+      "loss": 0.8178,
+      "step": 3305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1227175029639425e-05,
+      "loss": 0.8631,
+      "step": 3306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1179212497272584e-05,
+      "loss": 0.843,
+      "step": 3307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1131297793107107e-05,
+      "loss": 0.8181,
+      "step": 3308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1083430946217464e-05,
+      "loss": 0.8134,
+      "step": 3309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103561198564914e-05,
+      "loss": 0.8443,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098784094041848e-05,
+      "loss": 0.8034,
+      "step": 3311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0940117839512873e-05,
+      "loss": 0.8112,
+      "step": 3312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.089244271189047e-05,
+      "loss": 0.8197,
+      "step": 3313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0844815586480437e-05,
+      "loss": 0.8372,
+      "step": 3314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0797236492182748e-05,
+      "loss": 0.887,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.074970545786824e-05,
+      "loss": 0.8628,
+      "step": 3316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0702222512378568e-05,
+      "loss": 0.7745,
+      "step": 3317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0654787684526277e-05,
+      "loss": 0.8516,
+      "step": 3318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0607401003094616e-05,
+      "loss": 0.7672,
+      "step": 3319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056006249683774e-05,
+      "loss": 0.8146,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0512772194480457e-05,
+      "loss": 0.8084,
+      "step": 3321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046553012471838e-05,
+      "loss": 0.8434,
+      "step": 3322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0418336316217822e-05,
+      "loss": 0.872,
+      "step": 3323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037119079761588e-05,
+      "loss": 0.816,
+      "step": 3324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0324093597520255e-05,
+      "loss": 0.8006,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0277044744509434e-05,
+      "loss": 0.7973,
+      "step": 3326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023004426713244e-05,
+      "loss": 0.7871,
+      "step": 3327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0183092193909092e-05,
+      "loss": 0.8168,
+      "step": 3328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0136188553329717e-05,
+      "loss": 0.8853,
+      "step": 3329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008933337385529e-05,
+      "loss": 0.3627,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004252668391736e-05,
+      "loss": 0.8335,
+      "step": 3331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9995768511918133e-05,
+      "loss": 0.8099,
+      "step": 3332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9949058886230254e-05,
+      "loss": 0.8438,
+      "step": 3333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990239783519705e-05,
+      "loss": 0.7923,
+      "step": 3334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985578538713222e-05,
+      "loss": 0.8118,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9809221570320115e-05,
+      "loss": 0.8477,
+      "step": 3336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9762706413015487e-05,
+      "loss": 0.8097,
+      "step": 3337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971623994344357e-05,
+      "loss": 0.8036,
+      "step": 3338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9669822189800046e-05,
+      "loss": 0.826,
+      "step": 3339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962345318025113e-05,
+      "loss": 0.8339,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9577132942933306e-05,
+      "loss": 0.858,
+      "step": 3341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9530861505953623e-05,
+      "loss": 0.8259,
+      "step": 3342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9484638897389405e-05,
+      "loss": 0.8157,
+      "step": 3343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9438465145288374e-05,
+      "loss": 0.8008,
+      "step": 3344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9392340277668608e-05,
+      "loss": 0.8376,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9346264322518572e-05,
+      "loss": 0.8606,
+      "step": 3346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9300237307796953e-05,
+      "loss": 0.8405,
+      "step": 3347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925425926143285e-05,
+      "loss": 0.8315,
+      "step": 3348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920833021132553e-05,
+      "loss": 0.806,
+      "step": 3349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916245018534467e-05,
+      "loss": 0.7964,
+      "step": 3350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911661921133009e-05,
+      "loss": 0.7767,
+      "step": 3351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.907083731709186e-05,
+      "loss": 0.8538,
+      "step": 3352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9025104530410286e-05,
+      "loss": 0.8642,
+      "step": 3353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8979420879035903e-05,
+      "loss": 0.8426,
+      "step": 3354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.893378639068938e-05,
+      "loss": 0.8244,
+      "step": 3355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8888201093061598e-05,
+      "loss": 0.808,
+      "step": 3356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.884266501381354e-05,
+      "loss": 0.8061,
+      "step": 3357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8797178180576413e-05,
+      "loss": 0.8322,
+      "step": 3358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875174062095143e-05,
+      "loss": 0.853,
+      "step": 3359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8706352362509972e-05,
+      "loss": 0.8566,
+      "step": 3360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.866101343279346e-05,
+      "loss": 0.8621,
+      "step": 3361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8615723859313473e-05,
+      "loss": 0.8284,
+      "step": 3362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.857048366955151e-05,
+      "loss": 0.8112,
+      "step": 3363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8525292890959234e-05,
+      "loss": 0.8051,
+      "step": 3364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8480151550958234e-05,
+      "loss": 0.8227,
+      "step": 3365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8435059676940136e-05,
+      "loss": 0.8213,
+      "step": 3366
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8390017296266526e-05,
+      "loss": 0.8579,
+      "step": 3367
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.834502443626901e-05,
+      "loss": 0.7833,
+      "step": 3368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830008112424907e-05,
+      "loss": 0.8442,
+      "step": 3369
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8255187387478212e-05,
+      "loss": 0.763,
+      "step": 3370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.821034325319777e-05,
+      "loss": 0.8316,
+      "step": 3371
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8165548748619054e-05,
+      "loss": 0.8886,
+      "step": 3372
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8120803900923222e-05,
+      "loss": 0.3035,
+      "step": 3373
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.807610873726129e-05,
+      "loss": 0.8585,
+      "step": 3374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8031463284754114e-05,
+      "loss": 0.802,
+      "step": 3375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.798686757049246e-05,
+      "loss": 0.8174,
+      "step": 3376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794232162153683e-05,
+      "loss": 0.8079,
+      "step": 3377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7897825464917606e-05,
+      "loss": 0.8093,
+      "step": 3378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.785337912763485e-05,
+      "loss": 0.831,
+      "step": 3379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7808982636658534e-05,
+      "loss": 0.3537,
+      "step": 3380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776463601892825e-05,
+      "loss": 0.8016,
+      "step": 3381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7720339301353405e-05,
+      "loss": 0.8176,
+      "step": 3382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7676092510813068e-05,
+      "loss": 0.8827,
+      "step": 3383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7631895674156117e-05,
+      "loss": 0.8393,
+      "step": 3384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7587748818200988e-05,
+      "loss": 0.7852,
+      "step": 3385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7543651969735897e-05,
+      "loss": 0.8199,
+      "step": 3386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7499605155518662e-05,
+      "loss": 0.8251,
+      "step": 3387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7455608402276734e-05,
+      "loss": 0.3293,
+      "step": 3388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7411661736707175e-05,
+      "loss": 0.784,
+      "step": 3389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7367765185476747e-05,
+      "loss": 0.8871,
+      "step": 3390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732391877522167e-05,
+      "loss": 0.8439,
+      "step": 3391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728012253254787e-05,
+      "loss": 0.8492,
+      "step": 3392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7236376484030704e-05,
+      "loss": 0.8254,
+      "step": 3393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.719268065621519e-05,
+      "loss": 0.7879,
+      "step": 3394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7149035075615794e-05,
+      "loss": 0.8321,
+      "step": 3395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7105439768716512e-05,
+      "loss": 0.8194,
+      "step": 3396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706189476197081e-05,
+      "loss": 0.8424,
+      "step": 3397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7018400081801712e-05,
+      "loss": 0.7816,
+      "step": 3398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6974955754601584e-05,
+      "loss": 0.8306,
+      "step": 3399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.693156180673239e-05,
+      "loss": 0.8039,
+      "step": 3400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6888218264525345e-05,
+      "loss": 0.802,
+      "step": 3401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684492515428129e-05,
+      "loss": 0.8234,
+      "step": 3402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6801682502270212e-05,
+      "loss": 0.8082,
+      "step": 3403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.67584903347317e-05,
+      "loss": 0.8763,
+      "step": 3404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6715348677874586e-05,
+      "loss": 0.837,
+      "step": 3405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6672257557877114e-05,
+      "loss": 0.823,
+      "step": 3406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.662921700088681e-05,
+      "loss": 0.7932,
+      "step": 3407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6586227033020584e-05,
+      "loss": 0.8475,
+      "step": 3408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6543287680364584e-05,
+      "loss": 0.7873,
+      "step": 3409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6500398968974284e-05,
+      "loss": 0.8547,
+      "step": 3410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645756092487438e-05,
+      "loss": 0.8505,
+      "step": 3411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6414773574058918e-05,
+      "loss": 0.854,
+      "step": 3412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6372036942491064e-05,
+      "loss": 0.8384,
+      "step": 3413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6329351056103325e-05,
+      "loss": 0.8431,
+      "step": 3414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6286715940797325e-05,
+      "loss": 0.8023,
+      "step": 3415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6244131622443948e-05,
+      "loss": 0.8992,
+      "step": 3416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620159812688319e-05,
+      "loss": 0.7743,
+      "step": 3417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6159115479924257e-05,
+      "loss": 0.83,
+      "step": 3418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.611668370734546e-05,
+      "loss": 0.8441,
+      "step": 3419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6074302834894296e-05,
+      "loss": 0.8551,
+      "step": 3420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6031972888287306e-05,
+      "loss": 0.8319,
+      "step": 3421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59896938932102e-05,
+      "loss": 0.8348,
+      "step": 3422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5947465875317712e-05,
+      "loss": 0.822,
+      "step": 3423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5905288860233724e-05,
+      "loss": 0.7983,
+      "step": 3424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5863162873551018e-05,
+      "loss": 0.8184,
+      "step": 3425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.582108794083158e-05,
+      "loss": 0.8042,
+      "step": 3426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577906408760631e-05,
+      "loss": 0.7973,
+      "step": 3427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5737091339375176e-05,
+      "loss": 0.8139,
+      "step": 3428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.569516972160706e-05,
+      "loss": 0.8259,
+      "step": 3429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565329925973994e-05,
+      "loss": 0.7954,
+      "step": 3430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.561147997918062e-05,
+      "loss": 0.8402,
+      "step": 3431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556971190530493e-05,
+      "loss": 0.8423,
+      "step": 3432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5527995063457558e-05,
+      "loss": 0.8432,
+      "step": 3433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5486329478952222e-05,
+      "loss": 0.8063,
+      "step": 3434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.54447151770714e-05,
+      "loss": 0.8277,
+      "step": 3435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5403152183066572e-05,
+      "loss": 0.7882,
+      "step": 3436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.536164052215797e-05,
+      "loss": 0.8487,
+      "step": 3437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5320180219534807e-05,
+      "loss": 0.3314,
+      "step": 3438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.527877130035502e-05,
+      "loss": 0.7903,
+      "step": 3439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5237413789745414e-05,
+      "loss": 0.3008,
+      "step": 3440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5196107712801577e-05,
+      "loss": 0.8178,
+      "step": 3441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.515485309458794e-05,
+      "loss": 0.8507,
+      "step": 3442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5113649960137632e-05,
+      "loss": 0.8518,
+      "step": 3443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5072498334452622e-05,
+      "loss": 0.8184,
+      "step": 3444
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5031398242503547e-05,
+      "loss": 0.8403,
+      "step": 3445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4990349709229889e-05,
+      "loss": 0.8767,
+      "step": 3446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4949352759539658e-05,
+      "loss": 0.8025,
+      "step": 3447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4908407418309744e-05,
+      "loss": 0.8228,
+      "step": 3448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4867513710385606e-05,
+      "loss": 0.8337,
+      "step": 3449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.482667166058146e-05,
+      "loss": 0.8257,
+      "step": 3450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4785881293680082e-05,
+      "loss": 0.8247,
+      "step": 3451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4745142634432985e-05,
+      "loss": 0.8096,
+      "step": 3452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4704455707560238e-05,
+      "loss": 0.8696,
+      "step": 3453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4663820537750538e-05,
+      "loss": 0.8208,
+      "step": 3454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462323714966114e-05,
+      "loss": 0.8169,
+      "step": 3455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4582705567917975e-05,
+      "loss": 0.8055,
+      "step": 3456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454222581711543e-05,
+      "loss": 0.8907,
+      "step": 3457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4501797921816518e-05,
+      "loss": 0.8543,
+      "step": 3458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4461421906552718e-05,
+      "loss": 0.8598,
+      "step": 3459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4421097795824124e-05,
+      "loss": 0.8255,
+      "step": 3460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4380825614099247e-05,
+      "loss": 0.8279,
+      "step": 3461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.434060538581512e-05,
+      "loss": 0.8013,
+      "step": 3462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4300437135377221e-05,
+      "loss": 0.8368,
+      "step": 3463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.426032088715955e-05,
+      "loss": 0.8089,
+      "step": 3464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4220256665504506e-05,
+      "loss": 0.7887,
+      "step": 3465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.418024449472294e-05,
+      "loss": 0.8508,
+      "step": 3466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4140284399094105e-05,
+      "loss": 0.8184,
+      "step": 3467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4100376402865655e-05,
+      "loss": 0.8091,
+      "step": 3468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4060520530253618e-05,
+      "loss": 0.8012,
+      "step": 3469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4020716805442447e-05,
+      "loss": 0.7991,
+      "step": 3470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3980965252584877e-05,
+      "loss": 0.8172,
+      "step": 3471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3941265895802057e-05,
+      "loss": 0.7864,
+      "step": 3472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3901618759183388e-05,
+      "loss": 0.7848,
+      "step": 3473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3862023866786677e-05,
+      "loss": 0.8452,
+      "step": 3474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3822481242637953e-05,
+      "loss": 0.83,
+      "step": 3475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3782990910731552e-05,
+      "loss": 0.8149,
+      "step": 3476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3743552895030054e-05,
+      "loss": 0.8086,
+      "step": 3477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370416721946437e-05,
+      "loss": 0.8069,
+      "step": 3478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3664833907933561e-05,
+      "loss": 0.8472,
+      "step": 3479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.362555298430499e-05,
+      "loss": 0.8529,
+      "step": 3480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3586324472414157e-05,
+      "loss": 0.7935,
+      "step": 3481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3547148396064846e-05,
+      "loss": 0.7994,
+      "step": 3482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.350802477902895e-05,
+      "loss": 0.873,
+      "step": 3483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3468953645046544e-05,
+      "loss": 0.8218,
+      "step": 3484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3429935017825857e-05,
+      "loss": 0.8301,
+      "step": 3485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3390968921043302e-05,
+      "loss": 0.8288,
+      "step": 3486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3352055378343342e-05,
+      "loss": 0.8409,
+      "step": 3487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3313194413338636e-05,
+      "loss": 0.8708,
+      "step": 3488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3274386049609877e-05,
+      "loss": 0.7859,
+      "step": 3489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3235630310705838e-05,
+      "loss": 0.8675,
+      "step": 3490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3196927220143373e-05,
+      "loss": 0.8081,
+      "step": 3491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3158276801407431e-05,
+      "loss": 0.8289,
+      "step": 3492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3119679077950909e-05,
+      "loss": 0.8455,
+      "step": 3493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3081134073194845e-05,
+      "loss": 0.7864,
+      "step": 3494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3042641810528155e-05,
+      "loss": 0.8012,
+      "step": 3495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3004202313307889e-05,
+      "loss": 0.7893,
+      "step": 3496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2965815604858989e-05,
+      "loss": 0.8038,
+      "step": 3497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2927481708474365e-05,
+      "loss": 0.7723,
+      "step": 3498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.288920064741489e-05,
+      "loss": 0.8012,
+      "step": 3499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2850972444909437e-05,
+      "loss": 0.8574,
+      "step": 3500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2812797124154719e-05,
+      "loss": 0.818,
+      "step": 3501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2774674708315426e-05,
+      "loss": 0.814,
+      "step": 3502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2736605220524089e-05,
+      "loss": 0.7771,
+      "step": 3503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2698588683881186e-05,
+      "loss": 0.8451,
+      "step": 3504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2660625121455006e-05,
+      "loss": 0.8333,
+      "step": 3505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2622714556281722e-05,
+      "loss": 0.8327,
+      "step": 3506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2584857011365325e-05,
+      "loss": 0.837,
+      "step": 3507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2547052509677693e-05,
+      "loss": 0.8364,
+      "step": 3508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2509301074158431e-05,
+      "loss": 0.3045,
+      "step": 3509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2471602727715038e-05,
+      "loss": 0.7774,
+      "step": 3510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2433957493222713e-05,
+      "loss": 0.8436,
+      "step": 3511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2396365393524478e-05,
+      "loss": 0.8879,
+      "step": 3512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2358826451431072e-05,
+      "loss": 0.8353,
+      "step": 3513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2321340689721061e-05,
+      "loss": 0.8713,
+      "step": 3514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.228390813114062e-05,
+      "loss": 0.8104,
+      "step": 3515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2246528798403756e-05,
+      "loss": 0.8284,
+      "step": 3516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2209202714192091e-05,
+      "loss": 0.8483,
+      "step": 3517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.217192990115502e-05,
+      "loss": 0.8489,
+      "step": 3518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213471038190953e-05,
+      "loss": 0.8684,
+      "step": 3519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2097544179040321e-05,
+      "loss": 0.823,
+      "step": 3520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2060431315099696e-05,
+      "loss": 0.8576,
+      "step": 3521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2023371812607665e-05,
+      "loss": 0.8104,
+      "step": 3522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1986365694051782e-05,
+      "loss": 0.7647,
+      "step": 3523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1949412981887264e-05,
+      "loss": 0.8479,
+      "step": 3524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1912513698536876e-05,
+      "loss": 0.7787,
+      "step": 3525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1875667866391027e-05,
+      "loss": 0.8552,
+      "step": 3526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1838875507807612e-05,
+      "loss": 0.8084,
+      "step": 3527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1802136645112128e-05,
+      "loss": 0.8212,
+      "step": 3528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1765451300597573e-05,
+      "loss": 0.8224,
+      "step": 3529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1728819496524546e-05,
+      "loss": 0.7819,
+      "step": 3530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.169224125512105e-05,
+      "loss": 0.8385,
+      "step": 3531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1655716598582712e-05,
+      "loss": 0.8276,
+      "step": 3532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1619245549072544e-05,
+      "loss": 0.8356,
+      "step": 3533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1582828128721046e-05,
+      "loss": 0.8164,
+      "step": 3534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1546464359626197e-05,
+      "loss": 0.3355,
+      "step": 3535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1510154263853446e-05,
+      "loss": 0.8192,
+      "step": 3536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.147389786343559e-05,
+      "loss": 0.8599,
+      "step": 3537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1437695180372954e-05,
+      "loss": 0.8209,
+      "step": 3538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1401546236633165e-05,
+      "loss": 0.829,
+      "step": 3539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1365451054151332e-05,
+      "loss": 0.8755,
+      "step": 3540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1329409654829858e-05,
+      "loss": 0.771,
+      "step": 3541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1293422060538572e-05,
+      "loss": 0.8064,
+      "step": 3542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1257488293114593e-05,
+      "loss": 0.7974,
+      "step": 3543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1221608374362458e-05,
+      "loss": 0.8158,
+      "step": 3544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.118578232605395e-05,
+      "loss": 0.8303,
+      "step": 3545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1150010169928249e-05,
+      "loss": 0.804,
+      "step": 3546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1114291927691734e-05,
+      "loss": 0.8599,
+      "step": 3547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1078627621018167e-05,
+      "loss": 0.8107,
+      "step": 3548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1043017271548506e-05,
+      "loss": 0.3373,
+      "step": 3549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1007460900891008e-05,
+      "loss": 0.8761,
+      "step": 3550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.097195853062113e-05,
+      "loss": 0.8269,
+      "step": 3551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0936510182281656e-05,
+      "loss": 0.8308,
+      "step": 3552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0901115877382472e-05,
+      "loss": 0.3331,
+      "step": 3553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0865775637400777e-05,
+      "loss": 0.8093,
+      "step": 3554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0830489483780892e-05,
+      "loss": 0.8447,
+      "step": 3555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0795257437934347e-05,
+      "loss": 0.8522,
+      "step": 3556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0760079521239818e-05,
+      "loss": 0.8294,
+      "step": 3557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0724955755043187e-05,
+      "loss": 0.7961,
+      "step": 3558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0689886160657392e-05,
+      "loss": 0.8754,
+      "step": 3559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0654870759362611e-05,
+      "loss": 0.8078,
+      "step": 3560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0619909572406029e-05,
+      "loss": 0.809,
+      "step": 3561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0585002621002005e-05,
+      "loss": 0.8264,
+      "step": 3562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0550149926331975e-05,
+      "loss": 0.8236,
+      "step": 3563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0515351509544425e-05,
+      "loss": 0.8487,
+      "step": 3564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0480607391754905e-05,
+      "loss": 0.3042,
+      "step": 3565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0445917594046072e-05,
+      "loss": 0.7414,
+      "step": 3566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0411282137467549e-05,
+      "loss": 0.8382,
+      "step": 3567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0376701043036063e-05,
+      "loss": 0.8329,
+      "step": 3568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0342174331735266e-05,
+      "loss": 0.8717,
+      "step": 3569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.030770202451592e-05,
+      "loss": 0.8306,
+      "step": 3570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0273284142295624e-05,
+      "loss": 0.8086,
+      "step": 3571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.02389207059591e-05,
+      "loss": 0.8124,
+      "step": 3572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0204611736357927e-05,
+      "loss": 0.7895,
+      "step": 3573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0170357254310737e-05,
+      "loss": 0.8151,
+      "step": 3574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0136157280602976e-05,
+      "loss": 0.8259,
+      "step": 3575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0102011835987135e-05,
+      "loss": 0.8412,
+      "step": 3576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006792094118254e-05,
+      "loss": 0.8005,
+      "step": 3577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0033884616875422e-05,
+      "loss": 0.7837,
+      "step": 3578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.999902883718914e-06,
+      "loss": 0.8016,
+      "step": 3579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.965975762333046e-06,
+      "loss": 0.8031,
+      "step": 3580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93210327330465e-06,
+      "loss": 0.7944,
+      "step": 3581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.898285437187493e-06,
+      "loss": 0.8192,
+      "step": 3582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.864522274502076e-06,
+      "loss": 0.8098,
+      "step": 3583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.830813805735828e-06,
+      "loss": 0.7931,
+      "step": 3584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.797160051342925e-06,
+      "loss": 0.854,
+      "step": 3585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.76356103174434e-06,
+      "loss": 0.8021,
+      "step": 3586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.730016767327844e-06,
+      "loss": 0.7886,
+      "step": 3587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69652727844802e-06,
+      "loss": 0.7705,
+      "step": 3588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.663092585426148e-06,
+      "loss": 0.8067,
+      "step": 3589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.629712708550331e-06,
+      "loss": 0.8,
+      "step": 3590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.596387668075336e-06,
+      "loss": 0.8258,
+      "step": 3591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.56311748422275e-06,
+      "loss": 0.7469,
+      "step": 3592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.529902177180728e-06,
+      "loss": 0.8339,
+      "step": 3593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.496741767104278e-06,
+      "loss": 0.769,
+      "step": 3594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.463636274114983e-06,
+      "loss": 0.7578,
+      "step": 3595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.4305857183012e-06,
+      "loss": 0.8533,
+      "step": 3596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.397590119717847e-06,
+      "loss": 0.8587,
+      "step": 3597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.364649498386601e-06,
+      "loss": 0.8178,
+      "step": 3598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.331763874295707e-06,
+      "loss": 0.8657,
+      "step": 3599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.298933267400045e-06,
+      "loss": 0.8325,
+      "step": 3600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.266157697621103e-06,
+      "loss": 0.8487,
+      "step": 3601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.233437184847038e-06,
+      "loss": 0.8504,
+      "step": 3602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.200771748932513e-06,
+      "loss": 0.8332,
+      "step": 3603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.168161409698838e-06,
+      "loss": 0.8447,
+      "step": 3604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.135606186933843e-06,
+      "loss": 0.313,
+      "step": 3605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.103106100391957e-06,
+      "loss": 0.8177,
+      "step": 3606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.070661169794103e-06,
+      "loss": 0.8172,
+      "step": 3607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.038271414827781e-06,
+      "loss": 0.8147,
+      "step": 3608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.005936855146957e-06,
+      "loss": 0.8144,
+      "step": 3609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.97365751037218e-06,
+      "loss": 0.8361,
+      "step": 3610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.941433400090416e-06,
+      "loss": 0.7807,
+      "step": 3611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.90926454385519e-06,
+      "loss": 0.8422,
+      "step": 3612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87715096118642e-06,
+      "loss": 0.8137,
+      "step": 3613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.845092671570588e-06,
+      "loss": 0.8064,
+      "step": 3614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.813089694460474e-06,
+      "loss": 0.8173,
+      "step": 3615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.781142049275448e-06,
+      "loss": 0.7913,
+      "step": 3616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7492497554012e-06,
+      "loss": 0.8535,
+      "step": 3617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.71741283218992e-06,
+      "loss": 0.8378,
+      "step": 3618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.685631298960096e-06,
+      "loss": 0.3302,
+      "step": 3619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.65390517499669e-06,
+      "loss": 0.838,
+      "step": 3620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.622234479551028e-06,
+      "loss": 0.825,
+      "step": 3621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590619231840757e-06,
+      "loss": 0.8374,
+      "step": 3622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.5590594510499e-06,
+      "loss": 0.7865,
+      "step": 3623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52755515632887e-06,
+      "loss": 0.8542,
+      "step": 3624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.49610636679432e-06,
+      "loss": 0.7971,
+      "step": 3625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.464713101529343e-06,
+      "loss": 0.7957,
+      "step": 3626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.433375379583197e-06,
+      "loss": 0.7978,
+      "step": 3627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.402093219971585e-06,
+      "loss": 0.8439,
+      "step": 3628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.370866641676367e-06,
+      "loss": 0.8051,
+      "step": 3629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.339695663645763e-06,
+      "loss": 0.8368,
+      "step": 3630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.308580304794178e-06,
+      "loss": 0.802,
+      "step": 3631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.277520584002374e-06,
+      "loss": 0.8525,
+      "step": 3632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.24651652011723e-06,
+      "loss": 0.8134,
+      "step": 3633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.215568131951957e-06,
+      "loss": 0.8329,
+      "step": 3634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.184675438285928e-06,
+      "loss": 0.7749,
+      "step": 3635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.153838457864715e-06,
+      "loss": 0.8305,
+      "step": 3636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.123057209400098e-06,
+      "loss": 0.818,
+      "step": 3637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.092331711570067e-06,
+      "loss": 0.8043,
+      "step": 3638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.06166198301872e-06,
+      "loss": 0.8461,
+      "step": 3639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.031048042356392e-06,
+      "loss": 0.8433,
+      "step": 3640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.00048990815947e-06,
+      "loss": 0.7911,
+      "step": 3641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.969987598970596e-06,
+      "loss": 0.8571,
+      "step": 3642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.939541133298433e-06,
+      "loss": 0.7941,
+      "step": 3643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.909150529617803e-06,
+      "loss": 0.3194,
+      "step": 3644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87881580636961e-06,
+      "loss": 0.8644,
+      "step": 3645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.848536981960908e-06,
+      "loss": 0.8172,
+      "step": 3646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.818314074764755e-06,
+      "loss": 0.8227,
+      "step": 3647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.788147103120347e-06,
+      "loss": 0.8244,
+      "step": 3648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.758036085332877e-06,
+      "loss": 0.8335,
+      "step": 3649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.727981039673627e-06,
+      "loss": 0.7978,
+      "step": 3650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.69798198437991e-06,
+      "loss": 0.8187,
+      "step": 3651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668038937655041e-06,
+      "loss": 0.3367,
+      "step": 3652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.638151917668356e-06,
+      "loss": 0.3032,
+      "step": 3653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.608320942555236e-06,
+      "loss": 0.8209,
+      "step": 3654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.578546030416988e-06,
+      "loss": 0.8532,
+      "step": 3655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5488271993209625e-06,
+      "loss": 0.8084,
+      "step": 3656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.519164467300432e-06,
+      "loss": 0.2949,
+      "step": 3657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.489557852354656e-06,
+      "loss": 0.7829,
+      "step": 3658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.460007372448819e-06,
+      "loss": 0.8391,
+      "step": 3659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.430513045514076e-06,
+      "loss": 0.7922,
+      "step": 3660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.401074889447468e-06,
+      "loss": 0.7991,
+      "step": 3661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.371692922112005e-06,
+      "loss": 0.8037,
+      "step": 3662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.342367161336539e-06,
+      "loss": 0.824,
+      "step": 3663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.313097624915877e-06,
+      "loss": 0.8204,
+      "step": 3664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.283884330610658e-06,
+      "loss": 0.7832,
+      "step": 3665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.254727296147435e-06,
+      "loss": 0.7966,
+      "step": 3666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.225626539218566e-06,
+      "loss": 0.7848,
+      "step": 3667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196582077482339e-06,
+      "loss": 0.8023,
+      "step": 3668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.16759392856281e-06,
+      "loss": 0.8203,
+      "step": 3669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.138662110049932e-06,
+      "loss": 0.8561,
+      "step": 3670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109786639499383e-06,
+      "loss": 0.8121,
+      "step": 3671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.08096753443277e-06,
+      "loss": 0.8361,
+      "step": 3672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.052204812337393e-06,
+      "loss": 0.8137,
+      "step": 3673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.023498490666391e-06,
+      "loss": 0.8208,
+      "step": 3674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9948485868386535e-06,
+      "loss": 0.8501,
+      "step": 3675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.966255118238873e-06,
+      "loss": 0.8482,
+      "step": 3676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.937718102217461e-06,
+      "loss": 0.7939,
+      "step": 3677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.909237556090598e-06,
+      "loss": 0.8834,
+      "step": 3678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.880813497140182e-06,
+      "loss": 0.8601,
+      "step": 3679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.852445942613839e-06,
+      "loss": 0.8258,
+      "step": 3680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8241349097249e-06,
+      "loss": 0.772,
+      "step": 3681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.795880415652433e-06,
+      "loss": 0.8295,
+      "step": 3682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7676824775411485e-06,
+      "loss": 0.8043,
+      "step": 3683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.739541112501491e-06,
+      "loss": 0.8142,
+      "step": 3684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.711456337609523e-06,
+      "loss": 0.8352,
+      "step": 3685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.683428169907024e-06,
+      "loss": 0.8034,
+      "step": 3686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.65545662640139e-06,
+      "loss": 0.7496,
+      "step": 3687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.627541724065656e-06,
+      "loss": 0.8253,
+      "step": 3688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.599683479838481e-06,
+      "loss": 0.8433,
+      "step": 3689
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.571881910624189e-06,
+      "loss": 0.3063,
+      "step": 3690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.544137033292652e-06,
+      "loss": 0.8588,
+      "step": 3691
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.516448864679403e-06,
+      "loss": 0.8718,
+      "step": 3692
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.488817421585513e-06,
+      "loss": 0.8506,
+      "step": 3693
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.461242720777671e-06,
+      "loss": 0.8512,
+      "step": 3694
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4337247789881035e-06,
+      "loss": 0.832,
+      "step": 3695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.406263612914609e-06,
+      "loss": 0.8303,
+      "step": 3696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.378859239220525e-06,
+      "loss": 0.8454,
+      "step": 3697
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.351511674534772e-06,
+      "loss": 0.8351,
+      "step": 3698
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.32422093545173e-06,
+      "loss": 0.7977,
+      "step": 3699
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.296987038531355e-06,
+      "loss": 0.772,
+      "step": 3700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.269810000299081e-06,
+      "loss": 0.8482,
+      "step": 3701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.242689837245841e-06,
+      "loss": 0.8253,
+      "step": 3702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2156265658280585e-06,
+      "loss": 0.8119,
+      "step": 3703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.188620202467665e-06,
+      "loss": 0.8418,
+      "step": 3704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.161670763551997e-06,
+      "loss": 0.3292,
+      "step": 3705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.134778265433938e-06,
+      "loss": 0.8287,
+      "step": 3706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.107942724431714e-06,
+      "loss": 0.8068,
+      "step": 3707
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0811641568291e-06,
+      "loss": 0.8269,
+      "step": 3708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054442578875208e-06,
+      "loss": 0.8509,
+      "step": 3709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.027778006784624e-06,
+      "loss": 0.8071,
+      "step": 3710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.001170456737304e-06,
+      "loss": 0.8219,
+      "step": 3711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.974619944878646e-06,
+      "loss": 0.8347,
+      "step": 3712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.948126487319394e-06,
+      "loss": 0.7837,
+      "step": 3713
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921690100135713e-06,
+      "loss": 0.8394,
+      "step": 3714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.895310799369102e-06,
+      "loss": 0.857,
+      "step": 3715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86898860102647e-06,
+      "loss": 0.841,
+      "step": 3716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8427235210800066e-06,
+      "loss": 0.8432,
+      "step": 3717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.816515575467296e-06,
+      "loss": 0.7965,
+      "step": 3718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.790364780091206e-06,
+      "loss": 0.7765,
+      "step": 3719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.76427115082e-06,
+      "loss": 0.8024,
+      "step": 3720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7382347034871665e-06,
+      "loss": 0.8516,
+      "step": 3721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.71225545389158e-06,
+      "loss": 0.7974,
+      "step": 3722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.686333417797352e-06,
+      "loss": 0.8096,
+      "step": 3723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.660468610933867e-06,
+      "loss": 0.8716,
+      "step": 3724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.634661048995815e-06,
+      "loss": 0.8696,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.60891074764317e-06,
+      "loss": 0.8551,
+      "step": 3726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5832177225010995e-06,
+      "loss": 0.8086,
+      "step": 3727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.557581989160076e-06,
+      "loss": 0.8188,
+      "step": 3728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5320035631757475e-06,
+      "loss": 0.8457,
+      "step": 3729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.506482460069062e-06,
+      "loss": 0.8347,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.481018695326123e-06,
+      "loss": 0.7634,
+      "step": 3731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.455612284398249e-06,
+      "loss": 0.8431,
+      "step": 3732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4302632427019715e-06,
+      "loss": 0.8168,
+      "step": 3733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.404971585619034e-06,
+      "loss": 0.8487,
+      "step": 3734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379737328496293e-06,
+      "loss": 0.8194,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3545604866458545e-06,
+      "loss": 0.8874,
+      "step": 3736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3294410753449205e-06,
+      "loss": 0.8397,
+      "step": 3737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3043791098358995e-06,
+      "loss": 0.8447,
+      "step": 3738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279374605326259e-06,
+      "loss": 0.8285,
+      "step": 3739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254427576988697e-06,
+      "loss": 0.8326,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229538039960957e-06,
+      "loss": 0.8327,
+      "step": 3741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.204706009345972e-06,
+      "loss": 0.8779,
+      "step": 3742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.179931500211699e-06,
+      "loss": 0.8333,
+      "step": 3743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.155214527591257e-06,
+      "loss": 0.7215,
+      "step": 3744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.130555106482815e-06,
+      "loss": 0.815,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.105953251849637e-06,
+      "loss": 0.8262,
+      "step": 3746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0814089786200285e-06,
+      "loss": 0.7984,
+      "step": 3747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0569223016874165e-06,
+      "loss": 0.7807,
+      "step": 3748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.032493235910185e-06,
+      "loss": 0.7429,
+      "step": 3749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0081217961118645e-06,
+      "loss": 0.8238,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.983807997080925e-06,
+      "loss": 0.8134,
+      "step": 3751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.959551853570932e-06,
+      "loss": 0.8102,
+      "step": 3752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935353380300422e-06,
+      "loss": 0.8194,
+      "step": 3753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.911212591952963e-06,
+      "loss": 0.8108,
+      "step": 3754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.887129503177079e-06,
+      "loss": 0.7885,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.863104128586349e-06,
+      "loss": 0.7793,
+      "step": 3756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.839136482759255e-06,
+      "loss": 0.8219,
+      "step": 3757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8152265802393185e-06,
+      "loss": 0.8261,
+      "step": 3758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7913744355349785e-06,
+      "loss": 0.8337,
+      "step": 3759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.767580063119659e-06,
+      "loss": 0.8319,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.743843477431664e-06,
+      "loss": 0.7919,
+      "step": 3761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.720164692874329e-06,
+      "loss": 0.8174,
+      "step": 3762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.696543723815816e-06,
+      "loss": 0.8197,
+      "step": 3763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6729805845893035e-06,
+      "loss": 0.8006,
+      "step": 3764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.649475289492777e-06,
+      "loss": 0.7855,
+      "step": 3765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.626027852789227e-06,
+      "loss": 0.7805,
+      "step": 3766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6026382887064515e-06,
+      "loss": 0.8422,
+      "step": 3767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.579306611437173e-06,
+      "loss": 0.775,
+      "step": 3768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556032835138957e-06,
+      "loss": 0.8419,
+      "step": 3769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.532816973934284e-06,
+      "loss": 0.816,
+      "step": 3770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.509659041910452e-06,
+      "loss": 0.7695,
+      "step": 3771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.486559053119621e-06,
+      "loss": 0.8535,
+      "step": 3772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.463517021578789e-06,
+      "loss": 0.8464,
+      "step": 3773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.440532961269794e-06,
+      "loss": 0.8199,
+      "step": 3774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.417606886139292e-06,
+      "loss": 0.7905,
+      "step": 3775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.394738810098742e-06,
+      "loss": 0.8357,
+      "step": 3776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.371928747024412e-06,
+      "loss": 0.8222,
+      "step": 3777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.349176710757397e-06,
+      "loss": 0.7514,
+      "step": 3778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.326482715103541e-06,
+      "loss": 0.7849,
+      "step": 3779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.303846773833498e-06,
+      "loss": 0.8131,
+      "step": 3780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2812689006826775e-06,
+      "loss": 0.786,
+      "step": 3781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.25874910935129e-06,
+      "loss": 0.818,
+      "step": 3782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.236287413504214e-06,
+      "loss": 0.7918,
+      "step": 3783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.213883826771192e-06,
+      "loss": 0.8251,
+      "step": 3784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.191538362746606e-06,
+      "loss": 0.8331,
+      "step": 3785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.16925103498964e-06,
+      "loss": 0.8397,
+      "step": 3786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1470218570241585e-06,
+      "loss": 0.8588,
+      "step": 3787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.124850842338779e-06,
+      "loss": 0.8294,
+      "step": 3788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.102738004386775e-06,
+      "loss": 0.8453,
+      "step": 3789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.080683356586157e-06,
+      "loss": 0.8179,
+      "step": 3790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.058686912319598e-06,
+      "loss": 0.8354,
+      "step": 3791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.036748684934499e-06,
+      "loss": 0.8646,
+      "step": 3792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.014868687742879e-06,
+      "loss": 0.81,
+      "step": 3793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.993046934021482e-06,
+      "loss": 0.7959,
+      "step": 3794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.97128343701163e-06,
+      "loss": 0.8602,
+      "step": 3795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949578209919402e-06,
+      "loss": 0.8167,
+      "step": 3796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9279312659154185e-06,
+      "loss": 0.8067,
+      "step": 3797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.906342618134995e-06,
+      "loss": 0.8258,
+      "step": 3798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88481227967804e-06,
+      "loss": 0.8579,
+      "step": 3799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.863340263609106e-06,
+      "loss": 0.7699,
+      "step": 3800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.841926582957322e-06,
+      "loss": 0.8535,
+      "step": 3801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8205712507164755e-06,
+      "loss": 0.8037,
+      "step": 3802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.799274279844889e-06,
+      "loss": 0.7616,
+      "step": 3803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7780356832655127e-06,
+      "loss": 0.7833,
+      "step": 3804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.756855473865828e-06,
+      "loss": 0.8217,
+      "step": 3805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7357336644979535e-06,
+      "loss": 0.8482,
+      "step": 3806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.714670267978504e-06,
+      "loss": 0.3375,
+      "step": 3807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.693665297088722e-06,
+      "loss": 0.7733,
+      "step": 3808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6727187645743124e-06,
+      "loss": 0.84,
+      "step": 3809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.651830683145596e-06,
+      "loss": 0.8176,
+      "step": 3810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6310010654773797e-06,
+      "loss": 0.8052,
+      "step": 3811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.610229924209019e-06,
+      "loss": 0.8324,
+      "step": 3812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.589517271944343e-06,
+      "loss": 0.8276,
+      "step": 3813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5688631212517642e-06,
+      "loss": 0.7932,
+      "step": 3814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5482674846641363e-06,
+      "loss": 0.7901,
+      "step": 3815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5277303746788283e-06,
+      "loss": 0.7791,
+      "step": 3816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.507251803757683e-06,
+      "loss": 0.8299,
+      "step": 3817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4868317843270605e-06,
+      "loss": 0.8208,
+      "step": 3818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4664703287777267e-06,
+      "loss": 0.8286,
+      "step": 3819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4461674494649763e-06,
+      "loss": 0.7856,
+      "step": 3820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4259231587085105e-06,
+      "loss": 0.842,
+      "step": 3821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4057374687925135e-06,
+      "loss": 0.8491,
+      "step": 3822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.385610391965588e-06,
+      "loss": 0.8456,
+      "step": 3823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.365541940440786e-06,
+      "loss": 0.7695,
+      "step": 3824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3455321263955786e-06,
+      "loss": 0.7977,
+      "step": 3825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.325580961971864e-06,
+      "loss": 0.8121,
+      "step": 3826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.305688459275924e-06,
+      "loss": 0.8208,
+      "step": 3827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.285854630378493e-06,
+      "loss": 0.8132,
+      "step": 3828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2660794873146437e-06,
+      "loss": 0.8112,
+      "step": 3829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2463630420839e-06,
+      "loss": 0.7484,
+      "step": 3830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.226705306650113e-06,
+      "loss": 0.8442,
+      "step": 3831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.207106292941553e-06,
+      "loss": 0.8445,
+      "step": 3832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187566012850829e-06,
+      "loss": 0.7964,
+      "step": 3833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168084478234934e-06,
+      "loss": 0.8995,
+      "step": 3834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1486617009151674e-06,
+      "loss": 0.7736,
+      "step": 3835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1292976926772354e-06,
+      "loss": 0.8056,
+      "step": 3836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.10999246527115e-06,
+      "loss": 0.8844,
+      "step": 3837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0907460304112733e-06,
+      "loss": 0.7867,
+      "step": 3838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.071558399776253e-06,
+      "loss": 0.8152,
+      "step": 3839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052429585009109e-06,
+      "loss": 0.8516,
+      "step": 3840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0333595977171336e-06,
+      "loss": 0.8422,
+      "step": 3841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0143484494719265e-06,
+      "loss": 0.8463,
+      "step": 3842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.995396151809404e-06,
+      "loss": 0.8535,
+      "step": 3843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9765027162297563e-06,
+      "loss": 0.8284,
+      "step": 3844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.957668154197446e-06,
+      "loss": 0.8376,
+      "step": 3845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.938892477141242e-06,
+      "loss": 0.8437,
+      "step": 3846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9201756964541748e-06,
+      "loss": 0.8043,
+      "step": 3847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9015178234934937e-06,
+      "loss": 0.8281,
+      "step": 3848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.882918869580753e-06,
+      "loss": 0.8016,
+      "step": 3849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8643788460017586e-06,
+      "loss": 0.7901,
+      "step": 3850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8458977640065e-06,
+      "loss": 0.8383,
+      "step": 3851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8274756348092624e-06,
+      "loss": 0.8235,
+      "step": 3852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8091124695885263e-06,
+      "loss": 0.8184,
+      "step": 3853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7908082794870114e-06,
+      "loss": 0.8463,
+      "step": 3854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.772563075611645e-06,
+      "loss": 0.8021,
+      "step": 3855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.754376869033537e-06,
+      "loss": 0.8531,
+      "step": 3856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.736249670788027e-06,
+      "loss": 0.8341,
+      "step": 3857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7181814918746496e-06,
+      "loss": 0.8341,
+      "step": 3858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7001723432571013e-06,
+      "loss": 0.8151,
+      "step": 3859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6822222358632966e-06,
+      "loss": 0.8342,
+      "step": 3860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664331180585289e-06,
+      "loss": 0.8157,
+      "step": 3861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6464991882793278e-06,
+      "loss": 0.7516,
+      "step": 3862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6287262697658023e-06,
+      "loss": 0.8249,
+      "step": 3863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.611012435829252e-06,
+      "loss": 0.3169,
+      "step": 3864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5933576972183793e-06,
+      "loss": 0.8368,
+      "step": 3865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5757620646460367e-06,
+      "loss": 0.8111,
+      "step": 3866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.558225548789173e-06,
+      "loss": 0.812,
+      "step": 3867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5407481602889084e-06,
+      "loss": 0.8469,
+      "step": 3868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5233299097504716e-06,
+      "loss": 0.8094,
+      "step": 3869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5059708077431853e-06,
+      "loss": 0.8291,
+      "step": 3870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4886708648004908e-06,
+      "loss": 0.8063,
+      "step": 3871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4714300914199684e-06,
+      "loss": 0.7965,
+      "step": 3872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4542484980632273e-06,
+      "loss": 0.8436,
+      "step": 3873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4371260951560394e-06,
+      "loss": 0.848,
+      "step": 3874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.420062893088193e-06,
+      "loss": 0.8732,
+      "step": 3875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.403058902213606e-06,
+      "loss": 0.8015,
+      "step": 3876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3861141328502368e-06,
+      "loss": 0.7441,
+      "step": 3877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3692285952801263e-06,
+      "loss": 0.8514,
+      "step": 3878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3524022997493346e-06,
+      "loss": 0.822,
+      "step": 3879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3356352564680383e-06,
+      "loss": 0.8538,
+      "step": 3880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3189274756104107e-06,
+      "loss": 0.8302,
+      "step": 3881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.302278967314686e-06,
+      "loss": 0.7911,
+      "step": 3882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2856897416831168e-06,
+      "loss": 0.7999,
+      "step": 3883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2691598087819953e-06,
+      "loss": 0.3455,
+      "step": 3884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2526891786416427e-06,
+      "loss": 0.7615,
+      "step": 3885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.236277861256375e-06,
+      "loss": 0.7816,
+      "step": 3886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2199258665845156e-06,
+      "loss": 0.8641,
+      "step": 3887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2036332045484167e-06,
+      "loss": 0.8324,
+      "step": 3888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1873998850344136e-06,
+      "loss": 0.8258,
+      "step": 3889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1712259178928383e-06,
+      "loss": 0.741,
+      "step": 3890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.155111312937985e-06,
+      "loss": 0.8441,
+      "step": 3891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.139056079948154e-06,
+      "loss": 0.8846,
+      "step": 3892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1230602286656076e-06,
+      "loss": 0.8463,
+      "step": 3893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1071237687965815e-06,
+      "loss": 0.834,
+      "step": 3894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0912467100112522e-06,
+      "loss": 0.8153,
+      "step": 3895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.07542906194379e-06,
+      "loss": 0.8212,
+      "step": 3896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0596708341922843e-06,
+      "loss": 0.854,
+      "step": 3897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0439720363187754e-06,
+      "loss": 0.7958,
+      "step": 3898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.028332677849254e-06,
+      "loss": 0.8255,
+      "step": 3899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.012752768273618e-06,
+      "loss": 0.3141,
+      "step": 3900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9972323170457054e-06,
+      "loss": 0.8073,
+      "step": 3901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.981771333583293e-06,
+      "loss": 0.8127,
+      "step": 3902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9663698272680333e-06,
+      "loss": 0.8281,
+      "step": 3903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.951027807445538e-06,
+      "loss": 0.8404,
+      "step": 3904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935745283425272e-06,
+      "loss": 0.8431,
+      "step": 3905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.92052226448064e-06,
+      "loss": 0.834,
+      "step": 3906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9053587598489077e-06,
+      "loss": 0.8728,
+      "step": 3907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8902547787312486e-06,
+      "loss": 0.8318,
+      "step": 3908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8752103302926982e-06,
+      "loss": 0.823,
+      "step": 3909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8602254236621985e-06,
+      "loss": 0.8115,
+      "step": 3910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8453000679325205e-06,
+      "loss": 0.7889,
+      "step": 3911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.830434272160353e-06,
+      "loss": 0.8224,
+      "step": 3912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8156280453661912e-06,
+      "loss": 0.827,
+      "step": 3913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.800881396534404e-06,
+      "loss": 0.8478,
+      "step": 3914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7861943346132115e-06,
+      "loss": 0.8545,
+      "step": 3915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7715668685146958e-06,
+      "loss": 0.7986,
+      "step": 3916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7569990071147236e-06,
+      "loss": 0.8495,
+      "step": 3917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7424907592530571e-06,
+      "loss": 0.8487,
+      "step": 3918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.728042133733243e-06,
+      "loss": 0.8145,
+      "step": 3919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7136531393226575e-06,
+      "loss": 0.8455,
+      "step": 3920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6993237847525046e-06,
+      "loss": 0.8301,
+      "step": 3921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6850540787177739e-06,
+      "loss": 0.8221,
+      "step": 3922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6708440298772944e-06,
+      "loss": 0.851,
+      "step": 3923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6566936468536798e-06,
+      "loss": 0.7809,
+      "step": 3924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6426029382333174e-06,
+      "loss": 0.8144,
+      "step": 3925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6285719125664234e-06,
+      "loss": 0.8722,
+      "step": 3926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.614600578366965e-06,
+      "loss": 0.8355,
+      "step": 3927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.600688944112727e-06,
+      "loss": 0.8673,
+      "step": 3928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5868370182452019e-06,
+      "loss": 0.815,
+      "step": 3929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5730448091697435e-06,
+      "loss": 0.8365,
+      "step": 3930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5593123252553798e-06,
+      "loss": 0.8083,
+      "step": 3931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5456395748349672e-06,
+      "loss": 0.7568,
+      "step": 3932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5320265662050581e-06,
+      "loss": 0.8549,
+      "step": 3933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5184733076260116e-06,
+      "loss": 0.8202,
+      "step": 3934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5049798073218934e-06,
+      "loss": 0.8497,
+      "step": 3935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4915460734805097e-06,
+      "loss": 0.8586,
+      "step": 3936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4781721142533956e-06,
+      "loss": 0.8354,
+      "step": 3937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.464857937755859e-06,
+      "loss": 0.8413,
+      "step": 3938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4516035520668603e-06,
+      "loss": 0.7601,
+      "step": 3939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4384089652291543e-06,
+      "loss": 0.8215,
+      "step": 3940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.425274185249148e-06,
+      "loss": 0.8412,
+      "step": 3941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4121992200969992e-06,
+      "loss": 0.8262,
+      "step": 3942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3991840777065612e-06,
+      "loss": 0.787,
+      "step": 3943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3862287659753614e-06,
+      "loss": 0.8257,
+      "step": 3944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3733332927646337e-06,
+      "loss": 0.7767,
+      "step": 3945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3604976658993519e-06,
+      "loss": 0.806,
+      "step": 3946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3477218931680858e-06,
+      "loss": 0.8024,
+      "step": 3947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3350059823231674e-06,
+      "loss": 0.338,
+      "step": 3948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3223499410805474e-06,
+      "loss": 0.7796,
+      "step": 3949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3097537771198931e-06,
+      "loss": 0.8168,
+      "step": 3950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.297217498084502e-06,
+      "loss": 0.3174,
+      "step": 3951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2847411115813447e-06,
+      "loss": 0.7921,
+      "step": 3952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.272324625181065e-06,
+      "loss": 0.8184,
+      "step": 3953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2599680464179364e-06,
+      "loss": 0.794,
+      "step": 3954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2476713827898945e-06,
+      "loss": 0.7703,
+      "step": 3955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.235434641758526e-06,
+      "loss": 0.9018,
+      "step": 3956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.223257830749036e-06,
+      "loss": 0.7861,
+      "step": 3957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2111409571502697e-06,
+      "loss": 0.8594,
+      "step": 3958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.199084028314712e-06,
+      "loss": 0.8396,
+      "step": 3959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1870870515584664e-06,
+      "loss": 0.7665,
+      "step": 3960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1751500341612653e-06,
+      "loss": 0.8092,
+      "step": 3961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1632729833664592e-06,
+      "loss": 0.8209,
+      "step": 3962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1514559063809828e-06,
+      "loss": 0.7874,
+      "step": 3963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1396988103754226e-06,
+      "loss": 0.8537,
+      "step": 3964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1280017024839384e-06,
+      "loss": 0.7992,
+      "step": 3965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1163645898042974e-06,
+      "loss": 0.7696,
+      "step": 3966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1047874793978507e-06,
+      "loss": 0.8499,
+      "step": 3967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0932703782895793e-06,
+      "loss": 0.8002,
+      "step": 3968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0818132934679925e-06,
+      "loss": 0.8259,
+      "step": 3969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0704162318852406e-06,
+      "loss": 0.8442,
+      "step": 3970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0590792004570027e-06,
+      "loss": 0.8129,
+      "step": 3971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0478022060625647e-06,
+      "loss": 0.8248,
+      "step": 3972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.036585255544764e-06,
+      "loss": 0.7662,
+      "step": 3973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.025428355710012e-06,
+      "loss": 0.878,
+      "step": 3974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0143315133282816e-06,
+      "loss": 0.8005,
+      "step": 3975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.003294735133098e-06,
+      "loss": 0.821,
+      "step": 3976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.923180278215483e-07,
+      "loss": 0.8528,
+      "step": 3977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.814013980542603e-07,
+      "loss": 0.8227,
+      "step": 3978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.70544852455413e-07,
+      "loss": 0.8001,
+      "step": 3979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.597483976127252e-07,
+      "loss": 0.8217,
+      "step": 3980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.490120400774349e-07,
+      "loss": 0.8256,
+      "step": 3981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.383357863643527e-07,
+      "loss": 0.8302,
+      "step": 3982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.277196429517743e-07,
+      "loss": 0.7964,
+      "step": 3983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.171636162815578e-07,
+      "loss": 0.8596,
+      "step": 3984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.066677127590461e-07,
+      "loss": 0.8555,
+      "step": 3985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.962319387531449e-07,
+      "loss": 0.7948,
+      "step": 3986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.858563005962328e-07,
+      "loss": 0.8293,
+      "step": 3987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.755408045842295e-07,
+      "loss": 0.8016,
+      "step": 3988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.652854569765167e-07,
+      "loss": 0.8392,
+      "step": 3989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.550902639960279e-07,
+      "loss": 0.8046,
+      "step": 3990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.449552318291698e-07,
+      "loss": 0.3133,
+      "step": 3991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.348803666258564e-07,
+      "loss": 0.814,
+      "step": 3992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.248656744994642e-07,
+      "loss": 0.3436,
+      "step": 3993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.149111615268878e-07,
+      "loss": 0.8086,
+      "step": 3994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.050168337484843e-07,
+      "loss": 0.8925,
+      "step": 3995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.951826971681064e-07,
+      "loss": 0.875,
+      "step": 3996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.854087577530811e-07,
+      "loss": 0.8428,
+      "step": 3997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.756950214342084e-07,
+      "loss": 0.8844,
+      "step": 3998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.660414941057292e-07,
+      "loss": 0.3365,
+      "step": 3999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.564481816254021e-07,
+      "loss": 0.8513,
+      "step": 4000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.46915089814404e-07,
+      "loss": 0.8791,
+      "step": 4001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.374422244573742e-07,
+      "loss": 0.3459,
+      "step": 4002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.280295913024371e-07,
+      "loss": 0.8085,
+      "step": 4003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.186771960611349e-07,
+      "loss": 0.7687,
+      "step": 4004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.093850444084615e-07,
+      "loss": 0.8021,
+      "step": 4005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.001531419828955e-07,
+      "loss": 0.8108,
+      "step": 4006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.909814943863002e-07,
+      "loss": 0.3095,
+      "step": 4007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.818701071840128e-07,
+      "loss": 0.7946,
+      "step": 4008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.728189859047995e-07,
+      "loss": 0.8061,
+      "step": 4009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.638281360408339e-07,
+      "loss": 0.8756,
+      "step": 4010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.548975630477405e-07,
+      "loss": 0.8328,
+      "step": 4011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.460272723445737e-07,
+      "loss": 0.8098,
+      "step": 4012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.372172693137834e-07,
+      "loss": 0.8203,
+      "step": 4013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.284675593012601e-07,
+      "loss": 0.8096,
+      "step": 4014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.197781476163012e-07,
+      "loss": 0.8347,
+      "step": 4015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.111490395316111e-07,
+      "loss": 0.8045,
+      "step": 4016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.025802402833013e-07,
+      "loss": 0.8153,
+      "step": 4017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.940717550709018e-07,
+      "loss": 0.8294,
+      "step": 4018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.856235890573158e-07,
+      "loss": 0.8672,
+      "step": 4019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.772357473688761e-07,
+      "loss": 0.7992,
+      "step": 4020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.68908235095289e-07,
+      "loss": 0.8209,
+      "step": 4021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.606410572896903e-07,
+      "loss": 0.8235,
+      "step": 4022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.524342189685449e-07,
+      "loss": 0.7996,
+      "step": 4023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.442877251117473e-07,
+      "loss": 0.8391,
+      "step": 4024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.362015806625653e-07,
+      "loss": 0.8362,
+      "step": 4025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.281757905276408e-07,
+      "loss": 0.332,
+      "step": 4026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.202103595770002e-07,
+      "loss": 0.7963,
+      "step": 4027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.123052926440552e-07,
+      "loss": 0.8036,
+      "step": 4028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.044605945255465e-07,
+      "loss": 0.8504,
+      "step": 4029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.96676269981633e-07,
+      "loss": 0.7636,
+      "step": 4030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.88952323735814e-07,
+      "loss": 0.844,
+      "step": 4031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.812887604749628e-07,
+      "loss": 0.8222,
+      "step": 4032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7368558484928163e-07,
+      "loss": 0.8025,
+      "step": 4033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.66142801472369e-07,
+      "loss": 0.8181,
+      "step": 4034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5866041492116376e-07,
+      "loss": 0.8251,
+      "step": 4035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.512384297359562e-07,
+      "loss": 0.8106,
+      "step": 4036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.438768504203661e-07,
+      "loss": 0.8043,
+      "step": 4037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3657568144139793e-07,
+      "loss": 0.8446,
+      "step": 4038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.293349272293634e-07,
+      "loss": 0.7746,
+      "step": 4039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2215459217792577e-07,
+      "loss": 0.8749,
+      "step": 4040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.150346806440997e-07,
+      "loss": 0.3209,
+      "step": 4041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0797519694822927e-07,
+      "loss": 0.8494,
+      "step": 4042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.009761453739769e-07,
+      "loss": 0.812,
+      "step": 4043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.940375301683452e-07,
+      "loss": 0.7993,
+      "step": 4044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8715935554165525e-07,
+      "loss": 0.8457,
+      "step": 4045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.803416256675796e-07,
+      "loss": 0.8371,
+      "step": 4046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7358434468308667e-07,
+      "loss": 0.8334,
+      "step": 4047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6688751668847443e-07,
+      "loss": 0.8123,
+      "step": 4048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6025114574734785e-07,
+      "loss": 0.8089,
+      "step": 4049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5367523588665244e-07,
+      "loss": 0.8002,
+      "step": 4050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4715979109660737e-07,
+      "loss": 0.8238,
+      "step": 4051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.407048153307835e-07,
+      "loss": 0.8444,
+      "step": 4052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.343103125060365e-07,
+      "loss": 0.7979,
+      "step": 4053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.279762865025182e-07,
+      "loss": 0.8181,
+      "step": 4054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2170274116369856e-07,
+      "loss": 0.8145,
+      "step": 4055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1548968029635473e-07,
+      "loss": 0.7511,
+      "step": 4056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0933710767053757e-07,
+      "loss": 0.8416,
+      "step": 4057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.032450270196274e-07,
+      "loss": 0.8094,
+      "step": 4058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9721344204026725e-07,
+      "loss": 0.8203,
+      "step": 4059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9124235639240715e-07,
+      "loss": 0.7871,
+      "step": 4060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.853317736992711e-07,
+      "loss": 0.8043,
+      "step": 4061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.794816975473902e-07,
+      "loss": 0.8565,
+      "step": 4062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7369213148658036e-07,
+      "loss": 0.8261,
+      "step": 4063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6796307902992035e-07,
+      "loss": 0.7934,
+      "step": 4064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.62294543653796e-07,
+      "loss": 0.8087,
+      "step": 4065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5668652879783374e-07,
+      "loss": 0.8373,
+      "step": 4066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5113903786497804e-07,
+      "loss": 0.8316,
+      "step": 4067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4565207422141413e-07,
+      "loss": 0.8323,
+      "step": 4068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4022564119662307e-07,
+      "loss": 0.3177,
+      "step": 4069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3485974208334872e-07,
+      "loss": 0.8007,
+      "step": 4070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2955438013758657e-07,
+      "loss": 0.8255,
+      "step": 4071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2430955857863922e-07,
+      "loss": 0.8014,
+      "step": 4072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.191252805890165e-07,
+      "loss": 0.8306,
+      "step": 4073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.140015493145575e-07,
+      "loss": 0.8349,
+      "step": 4074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0893836786429754e-07,
+      "loss": 0.8243,
+      "step": 4075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0393573931055676e-07,
+      "loss": 0.8232,
+      "step": 4076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9899366668894025e-07,
+      "loss": 0.8114,
+      "step": 4077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9411215299826036e-07,
+      "loss": 0.8819,
+      "step": 4078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8929120120061429e-07,
+      "loss": 0.8084,
+      "step": 4079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8453081422132866e-07,
+      "loss": 0.8274,
+      "step": 4080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.79830994949004e-07,
+      "loss": 0.8017,
+      "step": 4081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7519174623547019e-07,
+      "loss": 0.8505,
+      "step": 4082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.706130708958087e-07,
+      "loss": 0.8234,
+      "step": 4083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6609497170834155e-07,
+      "loss": 0.8694,
+      "step": 4084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6163745141464238e-07,
+      "loss": 0.8157,
+      "step": 4085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5724051271951423e-07,
+      "loss": 0.8116,
+      "step": 4086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.529041582910118e-07,
+      "loss": 0.8428,
+      "step": 4087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4862839076041913e-07,
+      "loss": 0.8002,
+      "step": 4088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4441321272226082e-07,
+      "loss": 0.822,
+      "step": 4089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.402586267342909e-07,
+      "loss": 0.8621,
+      "step": 4090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.361646353174928e-07,
+      "loss": 0.8692,
+      "step": 4091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3213124095609043e-07,
+      "loss": 0.828,
+      "step": 4092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2815844609753712e-07,
+      "loss": 0.8391,
+      "step": 4093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2424625315252682e-07,
+      "loss": 0.7656,
+      "step": 4094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2039466449492721e-07,
+      "loss": 0.8428,
+      "step": 4095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1660368246191322e-07,
+      "loss": 0.7995,
+      "step": 4096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1287330935381146e-07,
+      "loss": 0.8526,
+      "step": 4097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0920354743421124e-07,
+      "loss": 0.8378,
+      "step": 4098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0559439892990907e-07,
+      "loss": 0.8074,
+      "step": 4099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0204586603093092e-07,
+      "loss": 0.8317,
+      "step": 4100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.855795089049879e-08,
+      "loss": 0.337,
+      "step": 4101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.513065562509749e-08,
+      "loss": 0.7639,
+      "step": 4102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.176398231437456e-08,
+      "loss": 0.8158,
+      "step": 4103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.845793300121808e-08,
+      "loss": 0.834,
+      "step": 4104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.521250969173445e-08,
+      "loss": 0.7817,
+      "step": 4105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.202771435523726e-08,
+      "loss": 0.7855,
+      "step": 4106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.890354892424734e-08,
+      "loss": 0.8619,
+      "step": 4107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.58400152945038e-08,
+      "loss": 0.8379,
+      "step": 4108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.283711532494186e-08,
+      "loss": 0.8664,
+      "step": 4109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.989485083771507e-08,
+      "loss": 0.8047,
+      "step": 4110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.701322361818419e-08,
+      "loss": 0.8174,
+      "step": 4111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.419223541489494e-08,
+      "loss": 0.7864,
+      "step": 4112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.143188793963361e-08,
+      "loss": 0.8474,
+      "step": 4113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.873218286736037e-08,
+      "loss": 0.8212,
+      "step": 4114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.60931218362426e-08,
+      "loss": 0.8431,
+      "step": 4115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3514706447665984e-08,
+      "loss": 0.7871,
+      "step": 4116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0996938266201223e-08,
+      "loss": 0.8152,
+      "step": 4117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8539818819626215e-08,
+      "loss": 0.8493,
+      "step": 4118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.614334959889277e-08,
+      "loss": 0.877,
+      "step": 4119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3807532058193216e-08,
+      "loss": 0.819,
+      "step": 4120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.153236761488266e-08,
+      "loss": 0.8378,
+      "step": 4121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.931785764953455e-08,
+      "loss": 0.8489,
+      "step": 4122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.716400350588511e-08,
+      "loss": 0.7888,
+      "step": 4123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.50708064909111e-08,
+      "loss": 0.8494,
+      "step": 4124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.303826787474096e-08,
+      "loss": 0.8002,
+      "step": 4125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1066388890721444e-08,
+      "loss": 0.8163,
+      "step": 4126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9155170735384317e-08,
+      "loss": 0.8606,
+      "step": 4127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.730461456843525e-08,
+      "loss": 0.8316,
+      "step": 4128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.551472151279821e-08,
+      "loss": 0.8003,
+      "step": 4129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3785492654582187e-08,
+      "loss": 0.3065,
+      "step": 4130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2116929043058952e-08,
+      "loss": 0.8465,
+      "step": 4131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0509031690718604e-08,
+      "loss": 0.3394,
+      "step": 4132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8961801573225134e-08,
+      "loss": 0.8708,
+      "step": 4133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7475239629427542e-08,
+      "loss": 0.8604,
+      "step": 4134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6049346761382033e-08,
+      "loss": 0.816,
+      "step": 4135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4684123834307618e-08,
+      "loss": 0.3259,
+      "step": 4136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3379571676608304e-08,
+      "loss": 0.8099,
+      "step": 4137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2135691079895317e-08,
+      "loss": 0.3196,
+      "step": 4138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0952482798942676e-08,
+      "loss": 0.8092,
+      "step": 4139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.829947551731612e-09,
+      "loss": 0.7855,
+      "step": 4140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.76808601939505e-09,
+      "loss": 0.8716,
+      "step": 4141
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.766898846273129e-09,
+      "loss": 0.8377,
+      "step": 4142
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.82638663987989e-09,
+      "loss": 0.7783,
+      "step": 4143
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.946549970925475e-09,
+      "loss": 0.8231,
+      "step": 4144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.127389373282832e-09,
+      "loss": 0.8989,
+      "step": 4145
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.368905344032115e-09,
+      "loss": 0.8482,
+      "step": 4146
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.671098343394075e-09,
+      "loss": 0.8591,
+      "step": 4147
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0339687948077733e-09,
+      "loss": 0.8141,
+      "step": 4148
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.457517084886174e-09,
+      "loss": 0.8006,
+      "step": 4149
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9417435634050407e-09,
+      "loss": 0.7972,
+      "step": 4150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4866485433473466e-09,
+      "loss": 0.8707,
+      "step": 4151
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0922323008588643e-09,
+      "loss": 0.8241,
+      "step": 4152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.584950752703712e-10,
+      "loss": 0.8168,
+      "step": 4153
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854370690976496e-10,
+      "loss": 0.822,
+      "step": 4154
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7305844801928103e-10,
+      "loss": 0.8324,
+      "step": 4155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.213593409099545e-10,
+      "loss": 0.8469,
+      "step": 4156
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.033983982936306e-11,
+      "loss": 0.789,
+      "step": 4157
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5957,
+      "step": 4158
+    },
+    {
+      "epoch": 1.0,
+      "step": 4158,
+      "total_flos": 8667931872591872.0,
+      "train_loss": 0.8304108607118207,
+      "train_runtime": 102034.3599,
+      "train_samples_per_second": 6.52,
+      "train_steps_per_second": 0.041
+    }
+  ],
+  "max_steps": 4158,
+  "num_train_epochs": 1,
+  "total_flos": 8667931872591872.0,
+  "trial_name": null,
+  "trial_params": null
+}