diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,37963 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 6323,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473685e-07,
+      "loss": 1.2979,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-07,
+      "loss": 1.2671,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1578947368421055e-07,
+      "loss": 1.2319,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.210526315789474e-07,
+      "loss": 0.1863,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 1.2725,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.315789473684211e-07,
+      "loss": 1.2798,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.368421052631579e-07,
+      "loss": 1.2632,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.421052631578948e-07,
+      "loss": 1.2339,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.473684210526317e-07,
+      "loss": 1.3066,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.3301,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1578947368421053e-06,
+      "loss": 1.2285,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 1.1709,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3684210526315791e-06,
+      "loss": 1.1985,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4736842105263159e-06,
+      "loss": 1.1074,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 1.1548,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6842105263157895e-06,
+      "loss": 1.1421,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7894736842105265e-06,
+      "loss": 1.1194,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8947368421052634e-06,
+      "loss": 1.0896,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0852,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.043,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2105263157894738e-06,
+      "loss": 1.1094,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 1.0791,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4210526315789477e-06,
+      "loss": 1.0481,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5263157894736844e-06,
+      "loss": 1.0579,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 1.0137,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7368421052631583e-06,
+      "loss": 1.0723,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.842105263157895e-06,
+      "loss": 0.998,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9473684210526317e-06,
+      "loss": 1.0159,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.052631578947369e-06,
+      "loss": 1.042,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.9775,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2631578947368423e-06,
+      "loss": 1.0308,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.368421052631579e-06,
+      "loss": 0.9702,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.473684210526316e-06,
+      "loss": 0.9475,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.578947368421053e-06,
+      "loss": 0.96,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 1.0234,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.789473684210527e-06,
+      "loss": 0.9734,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.894736842105263e-06,
+      "loss": 0.9294,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.9688,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.105263157894737e-06,
+      "loss": 0.9299,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.9282,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.315789473684211e-06,
+      "loss": 0.2218,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4210526315789476e-06,
+      "loss": 0.95,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.526315789473685e-06,
+      "loss": 0.8892,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 0.9036,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 1.0,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.842105263157895e-06,
+      "loss": 0.9229,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947368421052632e-06,
+      "loss": 0.9551,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 0.9131,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.157894736842106e-06,
+      "loss": 0.8733,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.0132,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.36842105263158e-06,
+      "loss": 0.958,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4736842105263165e-06,
+      "loss": 0.9053,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.578947368421052e-06,
+      "loss": 0.9302,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 0.9231,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.9961,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8947368421052634e-06,
+      "loss": 0.9312,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.8916,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.105263157894738e-06,
+      "loss": 0.9783,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2105263157894745e-06,
+      "loss": 0.9565,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.8818,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.421052631578948e-06,
+      "loss": 0.9839,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 0.8818,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.631578947368421e-06,
+      "loss": 0.9197,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.736842105263158e-06,
+      "loss": 0.8911,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.8682,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.947368421052632e-06,
+      "loss": 0.8202,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.052631578947369e-06,
+      "loss": 0.9016,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.157894736842106e-06,
+      "loss": 0.8884,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.263157894736843e-06,
+      "loss": 0.2244,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.9414,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.473684210526316e-06,
+      "loss": 0.9465,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 0.8953,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.68421052631579e-06,
+      "loss": 0.9075,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.789473684210526e-06,
+      "loss": 0.9075,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.8706,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8311,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.105263157894736e-06,
+      "loss": 0.8446,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 0.9163,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.315789473684212e-06,
+      "loss": 0.8911,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.907,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.526315789473685e-06,
+      "loss": 0.874,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.631578947368422e-06,
+      "loss": 0.9011,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.736842105263158e-06,
+      "loss": 0.9124,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 0.9226,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.8849,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 0.8812,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.157894736842105e-06,
+      "loss": 0.9329,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.8875,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.36842105263158e-06,
+      "loss": 0.8568,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.8591,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.578947368421054e-06,
+      "loss": 0.8721,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 0.8931,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.789473684210527e-06,
+      "loss": 0.877,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 0.8452,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9683,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 0.9158,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0210526315789476e-05,
+      "loss": 0.792,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0315789473684213e-05,
+      "loss": 0.8545,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0421052631578948e-05,
+      "loss": 0.9009,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.8538,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0631578947368421e-05,
+      "loss": 0.8855,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.073684210526316e-05,
+      "loss": 0.9348,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0842105263157896e-05,
+      "loss": 0.8733,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0947368421052633e-05,
+      "loss": 0.9165,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.105263157894737e-05,
+      "loss": 0.928,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 0.9631,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1263157894736843e-05,
+      "loss": 0.9226,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 0.8716,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1473684210526317e-05,
+      "loss": 0.9102,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.8279,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1684210526315792e-05,
+      "loss": 0.2688,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1789473684210527e-05,
+      "loss": 0.8828,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1894736842105264e-05,
+      "loss": 0.8372,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.2152,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 0.7908,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2210526315789475e-05,
+      "loss": 0.8748,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2315789473684212e-05,
+      "loss": 0.8826,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2421052631578949e-05,
+      "loss": 0.8826,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2526315789473684e-05,
+      "loss": 0.946,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.8997,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2736842105263159e-05,
+      "loss": 0.8218,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2842105263157896e-05,
+      "loss": 0.8687,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2947368421052633e-05,
+      "loss": 0.8999,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.305263157894737e-05,
+      "loss": 0.8492,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3157894736842108e-05,
+      "loss": 0.8398,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 0.8696,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.336842105263158e-05,
+      "loss": 0.8899,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3473684210526316e-05,
+      "loss": 0.866,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3578947368421055e-05,
+      "loss": 0.8743,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.877,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3789473684210528e-05,
+      "loss": 0.9197,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3894736842105265e-05,
+      "loss": 0.9001,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 0.8508,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4105263157894738e-05,
+      "loss": 0.8096,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 0.8921,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4315789473684212e-05,
+      "loss": 0.8896,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4421052631578948e-05,
+      "loss": 0.8679,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4526315789473687e-05,
+      "loss": 0.8652,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4631578947368424e-05,
+      "loss": 0.8523,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.8613,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4842105263157895e-05,
+      "loss": 0.863,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4947368421052632e-05,
+      "loss": 0.8857,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.505263157894737e-05,
+      "loss": 0.8853,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5157894736842107e-05,
+      "loss": 0.9119,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5263157894736846e-05,
+      "loss": 0.8755,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536842105263158e-05,
+      "loss": 0.8672,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5473684210526316e-05,
+      "loss": 0.908,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5578947368421052e-05,
+      "loss": 0.8555,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.568421052631579e-05,
+      "loss": 0.8677,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.8813,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5894736842105266e-05,
+      "loss": 0.9236,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.8853,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6105263157894736e-05,
+      "loss": 0.834,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 0.2408,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 0.9712,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.642105263157895e-05,
+      "loss": 0.8608,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6526315789473686e-05,
+      "loss": 0.8645,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6631578947368423e-05,
+      "loss": 0.9041,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673684210526316e-05,
+      "loss": 0.9232,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.8872,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6947368421052633e-05,
+      "loss": 0.8586,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 0.9077,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7157894736842107e-05,
+      "loss": 0.2216,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7263157894736843e-05,
+      "loss": 0.9087,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 0.8933,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7473684210526317e-05,
+      "loss": 0.8779,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7578947368421054e-05,
+      "loss": 0.2703,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.768421052631579e-05,
+      "loss": 0.895,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7789473684210527e-05,
+      "loss": 0.8525,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.9048,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 0.9158,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.810526315789474e-05,
+      "loss": 0.8865,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8210526315789477e-05,
+      "loss": 0.8696,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.831578947368421e-05,
+      "loss": 0.8623,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.2214,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 0.9087,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8631578947368424e-05,
+      "loss": 0.9028,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.873684210526316e-05,
+      "loss": 0.9092,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8842105263157898e-05,
+      "loss": 0.9187,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.814,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9052631578947368e-05,
+      "loss": 0.9214,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9157894736842108e-05,
+      "loss": 0.8633,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9263157894736845e-05,
+      "loss": 0.8577,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.936842105263158e-05,
+      "loss": 0.9392,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9473684210526318e-05,
+      "loss": 0.2305,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9578947368421055e-05,
+      "loss": 0.8596,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.968421052631579e-05,
+      "loss": 0.853,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9789473684210528e-05,
+      "loss": 0.8896,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9894736842105265e-05,
+      "loss": 0.8733,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8796,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998688030283e-05,
+      "loss": 0.8672,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994752121478e-05,
+      "loss": 0.8333,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998819227461e-05,
+      "loss": 0.2177,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999979008491408e-05,
+      "loss": 0.9114,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996720077428e-05,
+      "loss": 0.2341,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999952769126323e-05,
+      "loss": 0.8784,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999935713551323e-05,
+      "loss": 0.8772,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991603405376e-05,
+      "loss": 0.8367,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999893730638795e-05,
+      "loss": 0.8287,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999868803312275e-05,
+      "loss": 0.84,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999841252080746e-05,
+      "loss": 0.8569,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981107695144e-05,
+      "loss": 0.8928,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999977827793227e-05,
+      "loss": 0.9258,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999742855031844e-05,
+      "loss": 0.8459,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999704808259454e-05,
+      "loss": 0.8806,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999966413762509e-05,
+      "loss": 0.8572,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999620843139417e-05,
+      "loss": 0.9285,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999574924813797e-05,
+      "loss": 0.9175,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999952638266028e-05,
+      "loss": 0.9121,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999475216691603e-05,
+      "loss": 0.8477,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999942142692119e-05,
+      "loss": 0.9011,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999365013363162e-05,
+      "loss": 0.8296,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999930597603231e-05,
+      "loss": 0.9089,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999244314944133e-05,
+      "loss": 0.8706,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999180030114805e-05,
+      "loss": 0.2238,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999113121561204e-05,
+      "loss": 0.8748,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999043589300876e-05,
+      "loss": 0.856,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998971433352068e-05,
+      "loss": 0.8682,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998896653733718e-05,
+      "loss": 0.8875,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999881925046544e-05,
+      "loss": 0.915,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998739223567555e-05,
+      "loss": 0.8368,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998656573061054e-05,
+      "loss": 0.7865,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998571298967623e-05,
+      "loss": 0.8481,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999848340130964e-05,
+      "loss": 0.8977,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999839288011017e-05,
+      "loss": 0.8799,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998299735392967e-05,
+      "loss": 0.8743,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998203967182463e-05,
+      "loss": 0.9021,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999810557550379e-05,
+      "loss": 0.781,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998004560382776e-05,
+      "loss": 0.8787,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997900921845913e-05,
+      "loss": 0.9255,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99977946599204e-05,
+      "loss": 0.9204,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997685774634123e-05,
+      "loss": 0.9373,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997574266015653e-05,
+      "loss": 0.8821,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999746013409424e-05,
+      "loss": 0.8967,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999734337889984e-05,
+      "loss": 0.8811,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997224000463084e-05,
+      "loss": 0.9333,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997101998815303e-05,
+      "loss": 0.866,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996977373988505e-05,
+      "loss": 0.8278,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996850126015387e-05,
+      "loss": 0.2501,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996720254929344e-05,
+      "loss": 0.8833,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996587760764453e-05,
+      "loss": 0.8188,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999645264355548e-05,
+      "loss": 0.206,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996314903337875e-05,
+      "loss": 0.8955,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999617454014778e-05,
+      "loss": 0.8398,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999603155402203e-05,
+      "loss": 0.8917,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999588594499814e-05,
+      "loss": 0.8491,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999573771311432e-05,
+      "loss": 0.8337,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995586858409464e-05,
+      "loss": 0.865,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995433380923153e-05,
+      "loss": 0.8669,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995277280695662e-05,
+      "loss": 0.8567,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995118557767947e-05,
+      "loss": 0.8542,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999495721218166e-05,
+      "loss": 0.8738,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994793243979134e-05,
+      "loss": 0.2566,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994626653203397e-05,
+      "loss": 0.8467,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994457439898157e-05,
+      "loss": 0.8262,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994285604107816e-05,
+      "loss": 0.8289,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994111145877463e-05,
+      "loss": 0.9124,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999393406525288e-05,
+      "loss": 0.7944,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999375436228052e-05,
+      "loss": 0.9172,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993572037007545e-05,
+      "loss": 0.8694,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993387089481796e-05,
+      "loss": 0.9043,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993199519751794e-05,
+      "loss": 0.8403,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993009327866767e-05,
+      "loss": 0.8682,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992816513876617e-05,
+      "loss": 0.908,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992621077831936e-05,
+      "loss": 0.8743,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992423019784e-05,
+      "loss": 0.8994,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992222339784785e-05,
+      "loss": 0.9412,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992019037886947e-05,
+      "loss": 0.8923,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999181311414383e-05,
+      "loss": 0.8999,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991604568609465e-05,
+      "loss": 0.9143,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991393401338577e-05,
+      "loss": 0.8887,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991179612386574e-05,
+      "loss": 0.8583,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990963201809555e-05,
+      "loss": 0.803,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99907441696643e-05,
+      "loss": 0.8748,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990522516008285e-05,
+      "loss": 0.8669,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999029824089967e-05,
+      "loss": 0.8674,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99900713443973e-05,
+      "loss": 0.9277,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989841826560715e-05,
+      "loss": 0.8369,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998960968745014e-05,
+      "loss": 0.8962,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989374927126486e-05,
+      "loss": 0.9072,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989137545651353e-05,
+      "loss": 0.8912,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988897543087025e-05,
+      "loss": 0.9104,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998865491949648e-05,
+      "loss": 0.8806,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988409674943382e-05,
+      "loss": 0.8391,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998816180949208e-05,
+      "loss": 0.8181,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998791132320761e-05,
+      "loss": 0.9397,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987658216155707e-05,
+      "loss": 0.9043,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987402488402772e-05,
+      "loss": 0.8412,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987144140015916e-05,
+      "loss": 0.8545,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986883171062925e-05,
+      "loss": 0.8843,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986619581612278e-05,
+      "loss": 0.9089,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986353371733132e-05,
+      "loss": 0.8405,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986084541495344e-05,
+      "loss": 0.8413,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985813090969458e-05,
+      "loss": 0.8024,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985539020226693e-05,
+      "loss": 0.8141,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998526232933897e-05,
+      "loss": 0.8711,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984983018378886e-05,
+      "loss": 0.8538,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998470108741973e-05,
+      "loss": 0.8667,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984416536535485e-05,
+      "loss": 0.8323,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984129365800807e-05,
+      "loss": 0.8889,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983839575291056e-05,
+      "loss": 0.8916,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983547165082267e-05,
+      "loss": 0.9106,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998325213525117e-05,
+      "loss": 0.8356,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982954485875174e-05,
+      "loss": 0.8813,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998265421703238e-05,
+      "loss": 0.9116,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982351328801584e-05,
+      "loss": 0.9023,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998204582126226e-05,
+      "loss": 0.8391,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998173769449456e-05,
+      "loss": 0.8901,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981426948579356e-05,
+      "loss": 0.8384,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981113583598165e-05,
+      "loss": 0.2636,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980797599633223e-05,
+      "loss": 0.8516,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998047899676744e-05,
+      "loss": 0.8521,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998015777508441e-05,
+      "loss": 0.8857,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979833934668435e-05,
+      "loss": 0.823,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997950747560447e-05,
+      "loss": 0.8948,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979178397978186e-05,
+      "loss": 0.848,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997884670187593e-05,
+      "loss": 0.863,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997851238738474e-05,
+      "loss": 0.2547,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978175454592333e-05,
+      "loss": 0.259,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997783590358712e-05,
+      "loss": 0.8899,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977493734458197e-05,
+      "loss": 0.8713,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977148947295343e-05,
+      "loss": 0.8506,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976801542189036e-05,
+      "loss": 0.8645,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976451519230428e-05,
+      "loss": 0.9136,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976098878511366e-05,
+      "loss": 0.8491,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975743620124378e-05,
+      "loss": 0.8851,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997538574416268e-05,
+      "loss": 0.8518,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975025250720183e-05,
+      "loss": 0.8511,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997466213989147e-05,
+      "loss": 0.8347,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974296411771825e-05,
+      "loss": 0.8481,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997392806645721e-05,
+      "loss": 0.8652,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997355710404428e-05,
+      "loss": 0.9294,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997318352463037e-05,
+      "loss": 0.8071,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972807328313506e-05,
+      "loss": 0.8125,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99724285151924e-05,
+      "loss": 0.8708,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972047085366447e-05,
+      "loss": 0.8906,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997166303893574e-05,
+      "loss": 0.9116,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971276376001043e-05,
+      "loss": 0.7809,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970887096663817e-05,
+      "loss": 0.8477,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99704952010262e-05,
+      "loss": 0.8438,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970100689191035e-05,
+      "loss": 0.9016,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996970356126183e-05,
+      "loss": 0.8226,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969303817342796e-05,
+      "loss": 0.7959,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968901457538816e-05,
+      "loss": 0.8562,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996849648195547e-05,
+      "loss": 0.7898,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968088890699027e-05,
+      "loss": 0.2357,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996767868387643e-05,
+      "loss": 0.864,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996726586159531e-05,
+      "loss": 0.8389,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966850423964002e-05,
+      "loss": 0.8865,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99664323710915e-05,
+      "loss": 0.8645,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966011703087513e-05,
+      "loss": 0.9211,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996558842006241e-05,
+      "loss": 0.9165,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965162522127263e-05,
+      "loss": 0.8862,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964734009393824e-05,
+      "loss": 0.9043,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964302881974534e-05,
+      "loss": 0.8584,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963869139982517e-05,
+      "loss": 0.844,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963432783531585e-05,
+      "loss": 0.8618,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996299381273623e-05,
+      "loss": 0.2449,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962552227711645e-05,
+      "loss": 0.8386,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962108028573693e-05,
+      "loss": 0.7914,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961661215438926e-05,
+      "loss": 0.2423,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961211788424594e-05,
+      "loss": 0.8704,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960759747648618e-05,
+      "loss": 0.9229,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996030509322961e-05,
+      "loss": 0.8901,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959847825286867e-05,
+      "loss": 0.8232,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959387943940386e-05,
+      "loss": 0.8655,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995892544931082e-05,
+      "loss": 0.873,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958460341519537e-05,
+      "loss": 0.8765,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957992620688572e-05,
+      "loss": 0.8325,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995752228694065e-05,
+      "loss": 0.8521,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957049340399197e-05,
+      "loss": 0.8715,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956573781188296e-05,
+      "loss": 0.8175,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956095609432738e-05,
+      "loss": 0.8904,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955614825257994e-05,
+      "loss": 0.8782,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955131428790212e-05,
+      "loss": 0.856,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995464542015624e-05,
+      "loss": 0.8857,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99541567994836e-05,
+      "loss": 0.8384,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953665566900503e-05,
+      "loss": 0.8916,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953171722535846e-05,
+      "loss": 0.9028,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995267526651921e-05,
+      "loss": 0.2505,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952176198980863e-05,
+      "loss": 0.8462,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951674520051754e-05,
+      "loss": 0.8765,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995117022986353e-05,
+      "loss": 0.8921,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950663328548502e-05,
+      "loss": 0.8923,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950153816239687e-05,
+      "loss": 0.8083,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949641693070772e-05,
+      "loss": 0.8303,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949126959176138e-05,
+      "loss": 0.8694,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948609614690843e-05,
+      "loss": 0.8743,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948089659750645e-05,
+      "loss": 0.8389,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947567094491965e-05,
+      "loss": 0.8223,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947041919051933e-05,
+      "loss": 0.8618,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946514133568342e-05,
+      "loss": 0.8887,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945983738179685e-05,
+      "loss": 0.7878,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945450733025132e-05,
+      "loss": 0.8174,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944915118244546e-05,
+      "loss": 0.8342,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994437689397846e-05,
+      "loss": 0.9426,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943836060368108e-05,
+      "loss": 0.8882,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943292617555395e-05,
+      "loss": 0.8245,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942746565682924e-05,
+      "loss": 0.8899,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942197904893973e-05,
+      "loss": 0.8435,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941646635332507e-05,
+      "loss": 0.9014,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994109275714317e-05,
+      "loss": 0.2318,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994053627047131e-05,
+      "loss": 0.881,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939977175462935e-05,
+      "loss": 0.8347,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939415472264753e-05,
+      "loss": 0.7668,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938851161024147e-05,
+      "loss": 0.8591,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993828424188919e-05,
+      "loss": 0.8884,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993771471500864e-05,
+      "loss": 0.8279,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937142580531937e-05,
+      "loss": 0.8236,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993656783860921e-05,
+      "loss": 0.8655,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935990489391255e-05,
+      "loss": 0.8577,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935410533029576e-05,
+      "loss": 0.8052,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993482796967635e-05,
+      "loss": 0.2444,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934242799484436e-05,
+      "loss": 0.8457,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933655022607376e-05,
+      "loss": 0.8955,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933064639199403e-05,
+      "loss": 0.834,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932471649415427e-05,
+      "loss": 0.8533,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931876053411047e-05,
+      "loss": 0.9136,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993127785134254e-05,
+      "loss": 0.8154,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993067704336688e-05,
+      "loss": 0.7871,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930073629641704e-05,
+      "loss": 0.855,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929467610325354e-05,
+      "loss": 0.9136,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928858985576838e-05,
+      "loss": 0.8916,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928247755555857e-05,
+      "loss": 0.8197,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99276339204228e-05,
+      "loss": 0.8442,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927017480338727e-05,
+      "loss": 0.8605,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926398435465392e-05,
+      "loss": 0.845,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925776785965225e-05,
+      "loss": 0.7617,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925152532001346e-05,
+      "loss": 0.8489,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924525673737552e-05,
+      "loss": 0.8794,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923896211338335e-05,
+      "loss": 0.8478,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923264144968853e-05,
+      "loss": 0.7905,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922629474794958e-05,
+      "loss": 0.9412,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992199220098319e-05,
+      "loss": 0.863,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921352323700757e-05,
+      "loss": 0.8879,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920709843115564e-05,
+      "loss": 0.9368,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920064759396195e-05,
+      "loss": 0.8765,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919417072711915e-05,
+      "loss": 0.8625,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991876678323267e-05,
+      "loss": 0.8365,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918113891129097e-05,
+      "loss": 0.8601,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917458396572503e-05,
+      "loss": 0.8196,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916800299734896e-05,
+      "loss": 0.8389,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991613960078895e-05,
+      "loss": 0.7252,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991547629990803e-05,
+      "loss": 0.7982,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914810397266186e-05,
+      "loss": 0.8339,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914141893038138e-05,
+      "loss": 0.8926,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913470787399307e-05,
+      "loss": 0.7841,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912797080525778e-05,
+      "loss": 0.8909,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912120772594333e-05,
+      "loss": 0.8259,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991144186378243e-05,
+      "loss": 0.8518,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910760354268213e-05,
+      "loss": 0.8826,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910076244230504e-05,
+      "loss": 0.8593,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990938953384881e-05,
+      "loss": 0.8597,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908700223303318e-05,
+      "loss": 0.918,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908008312774897e-05,
+      "loss": 0.8855,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907313802445106e-05,
+      "loss": 0.2599,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906616692496174e-05,
+      "loss": 0.9131,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905916983111022e-05,
+      "loss": 0.8306,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905214674473253e-05,
+      "loss": 0.8103,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990450976676714e-05,
+      "loss": 0.8594,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903802260177655e-05,
+      "loss": 0.8035,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903092154890434e-05,
+      "loss": 0.8496,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902379451091814e-05,
+      "loss": 0.864,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99016641489688e-05,
+      "loss": 0.7566,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900946248709084e-05,
+      "loss": 0.8066,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900225750501035e-05,
+      "loss": 0.8452,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899502654533712e-05,
+      "loss": 0.8674,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989877696099685e-05,
+      "loss": 0.8506,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898048670080862e-05,
+      "loss": 0.8445,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897317781976856e-05,
+      "loss": 0.8186,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896584296876603e-05,
+      "loss": 0.8694,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989584821497257e-05,
+      "loss": 0.827,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895109536457902e-05,
+      "loss": 0.8313,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989436826152642e-05,
+      "loss": 0.877,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989362439037263e-05,
+      "loss": 0.8218,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892877923191724e-05,
+      "loss": 0.2525,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892128860179568e-05,
+      "loss": 0.8672,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989137720153271e-05,
+      "loss": 0.8344,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989062294744838e-05,
+      "loss": 0.7914,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889866098124492e-05,
+      "loss": 0.8906,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889106653759638e-05,
+      "loss": 0.8506,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888344614553094e-05,
+      "loss": 0.8342,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887579980704807e-05,
+      "loss": 0.8308,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886812752415423e-05,
+      "loss": 0.8748,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988604292988625e-05,
+      "loss": 0.8923,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988527051331929e-05,
+      "loss": 0.8673,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884495502917212e-05,
+      "loss": 0.8552,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883717898883382e-05,
+      "loss": 0.8339,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988293770142184e-05,
+      "loss": 0.8613,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882154910737297e-05,
+      "loss": 0.8477,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881369527035163e-05,
+      "loss": 0.2676,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880581550521504e-05,
+      "loss": 0.8169,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879790981403095e-05,
+      "loss": 0.8674,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878997819887363e-05,
+      "loss": 0.8494,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878202066182443e-05,
+      "loss": 0.8494,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877403720497123e-05,
+      "loss": 0.8328,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987660278304089e-05,
+      "loss": 0.8,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875799254023908e-05,
+      "loss": 0.8488,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987499313365701e-05,
+      "loss": 0.8767,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874184422151727e-05,
+      "loss": 0.8347,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873373119720248e-05,
+      "loss": 0.8275,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872559226575467e-05,
+      "loss": 0.8398,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871742742930938e-05,
+      "loss": 0.9346,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870923669000895e-05,
+      "loss": 0.8159,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870102005000273e-05,
+      "loss": 0.8263,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986927775114466e-05,
+      "loss": 0.8433,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986845090765034e-05,
+      "loss": 0.8267,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986762147473427e-05,
+      "loss": 0.9047,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866789452614086e-05,
+      "loss": 0.9048,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986595484150811e-05,
+      "loss": 0.8369,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865117641635335e-05,
+      "loss": 0.8232,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864277853215443e-05,
+      "loss": 0.8369,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863435476468785e-05,
+      "loss": 0.865,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862590511616395e-05,
+      "loss": 0.79,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861742958879987e-05,
+      "loss": 0.8521,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860892818481953e-05,
+      "loss": 0.2679,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986004009064537e-05,
+      "loss": 0.8625,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859184775593983e-05,
+      "loss": 0.8948,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858326873552224e-05,
+      "loss": 0.8713,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98574663847452e-05,
+      "loss": 0.2545,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856603309398697e-05,
+      "loss": 0.8215,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855737647739185e-05,
+      "loss": 0.9026,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854869399993807e-05,
+      "loss": 0.7954,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985399856639038e-05,
+      "loss": 0.8887,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853125147157412e-05,
+      "loss": 0.8407,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985224914252408e-05,
+      "loss": 0.9255,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985137055272025e-05,
+      "loss": 0.874,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850489377976444e-05,
+      "loss": 0.8774,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984960561852389e-05,
+      "loss": 0.8376,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848719274594475e-05,
+      "loss": 0.8516,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984783034642077e-05,
+      "loss": 0.8416,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846938834236026e-05,
+      "loss": 0.8672,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984604473827417e-05,
+      "loss": 0.8124,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984514805876981e-05,
+      "loss": 0.8838,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844248795958226e-05,
+      "loss": 0.8395,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984334695007538e-05,
+      "loss": 0.8481,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842442521357907e-05,
+      "loss": 0.8528,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984153551004313e-05,
+      "loss": 0.8457,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840625916369044e-05,
+      "loss": 0.8425,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839713740574314e-05,
+      "loss": 0.8271,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983879898289829e-05,
+      "loss": 0.7646,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837881643581007e-05,
+      "loss": 0.8179,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836961722863162e-05,
+      "loss": 0.7937,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983603922098614e-05,
+      "loss": 0.9062,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835114138191996e-05,
+      "loss": 0.8828,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983418647472347e-05,
+      "loss": 0.8623,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833256230823978e-05,
+      "loss": 0.8435,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98323234067376e-05,
+      "loss": 0.8562,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831388002709115e-05,
+      "loss": 0.2468,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983045001898396e-05,
+      "loss": 0.9236,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829509455808255e-05,
+      "loss": 0.8354,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828566313428806e-05,
+      "loss": 0.8235,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982762059209308e-05,
+      "loss": 0.9265,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826672292049236e-05,
+      "loss": 0.8547,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825721413546096e-05,
+      "loss": 0.8684,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824767956833168e-05,
+      "loss": 0.8063,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982381192216063e-05,
+      "loss": 0.8398,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822853309779343e-05,
+      "loss": 0.884,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982189211994084e-05,
+      "loss": 0.8268,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982092835289733e-05,
+      "loss": 0.9043,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819962008901702e-05,
+      "loss": 0.8348,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818993088207516e-05,
+      "loss": 0.8557,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818021591069016e-05,
+      "loss": 0.8733,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817047517741112e-05,
+      "loss": 0.8423,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816070868479394e-05,
+      "loss": 0.8647,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815091643540137e-05,
+      "loss": 0.8229,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814109843180277e-05,
+      "loss": 0.8936,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981312546765743e-05,
+      "loss": 0.8345,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812138517229895e-05,
+      "loss": 0.8506,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811148992156642e-05,
+      "loss": 0.8488,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810156892697312e-05,
+      "loss": 0.8718,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980916221911223e-05,
+      "loss": 0.2433,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808164971662393e-05,
+      "loss": 0.8451,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807165150609468e-05,
+      "loss": 0.8755,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980616275621581e-05,
+      "loss": 0.8657,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980515778874443e-05,
+      "loss": 0.8477,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804150248459032e-05,
+      "loss": 0.8589,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980314013562399e-05,
+      "loss": 0.8867,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980212745050435e-05,
+      "loss": 0.8506,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801112193365833e-05,
+      "loss": 0.8368,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980009436447484e-05,
+      "loss": 0.8489,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799073964098438e-05,
+      "loss": 0.8667,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979805099250437e-05,
+      "loss": 0.8171,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979702544996107e-05,
+      "loss": 0.7793,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979599733673763e-05,
+      "loss": 0.8276,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794966653103815e-05,
+      "loss": 0.2787,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793933399330075e-05,
+      "loss": 0.2364,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792897575687528e-05,
+      "loss": 0.8135,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791859182447967e-05,
+      "loss": 0.7432,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790818219883864e-05,
+      "loss": 0.7935,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789774688268358e-05,
+      "loss": 0.8977,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978872858787526e-05,
+      "loss": 0.8796,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978767991897907e-05,
+      "loss": 0.827,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978662868185495e-05,
+      "loss": 0.8518,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785574876778737e-05,
+      "loss": 0.8728,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784518504026944e-05,
+      "loss": 0.8584,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783459563876754e-05,
+      "loss": 0.8833,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782398056606026e-05,
+      "loss": 0.8672,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97813339824933e-05,
+      "loss": 0.8098,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780267341817775e-05,
+      "loss": 0.8452,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977919813485933e-05,
+      "loss": 0.8071,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778126361898527e-05,
+      "loss": 0.8125,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977705202321659e-05,
+      "loss": 0.8369,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775975119095414e-05,
+      "loss": 0.9146,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774895649817577e-05,
+      "loss": 0.8458,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977381361566632e-05,
+      "loss": 0.8586,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977272901692557e-05,
+      "loss": 0.8669,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771641853879912e-05,
+      "loss": 0.7994,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770552126814614e-05,
+      "loss": 0.7804,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769459836015616e-05,
+      "loss": 0.8103,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768364981769523e-05,
+      "loss": 0.8557,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767267564363622e-05,
+      "loss": 0.8726,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766167584085867e-05,
+      "loss": 0.8362,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976506504122489e-05,
+      "loss": 0.845,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763959936069988e-05,
+      "loss": 0.8472,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976285226891113e-05,
+      "loss": 0.8411,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976174204003897e-05,
+      "loss": 0.9062,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976062924974482e-05,
+      "loss": 0.8269,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975951389832067e-05,
+      "loss": 0.832,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975839598605918e-05,
+      "loss": 0.7842,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757275513253688e-05,
+      "loss": 0.9214,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756152480198195e-05,
+      "loss": 0.8518,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755026887187382e-05,
+      "loss": 0.833,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753898734516596e-05,
+      "loss": 0.285,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975276802248185e-05,
+      "loss": 0.8877,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975163475137985e-05,
+      "loss": 0.7888,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750498921507948e-05,
+      "loss": 0.7721,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749360533164186e-05,
+      "loss": 0.7886,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748219586647264e-05,
+      "loss": 0.8828,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747076082256566e-05,
+      "loss": 0.8904,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745930020292133e-05,
+      "loss": 0.8357,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744781401054697e-05,
+      "loss": 0.8906,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743630224845635e-05,
+      "loss": 0.8276,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742476491967016e-05,
+      "loss": 0.8442,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974132020272157e-05,
+      "loss": 0.8528,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740161357412702e-05,
+      "loss": 0.7712,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738999956344486e-05,
+      "loss": 0.8252,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737835999821667e-05,
+      "loss": 0.907,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736669488149657e-05,
+      "loss": 0.8684,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735500421634547e-05,
+      "loss": 0.8333,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734328800583085e-05,
+      "loss": 0.8655,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733154625302704e-05,
+      "loss": 0.812,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97319778961015e-05,
+      "loss": 0.8904,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730798613288234e-05,
+      "loss": 0.8254,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729616777172352e-05,
+      "loss": 0.8469,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972843238806395e-05,
+      "loss": 0.8755,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727245446273816e-05,
+      "loss": 0.8325,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726055952113382e-05,
+      "loss": 0.8396,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972486390589478e-05,
+      "loss": 0.8092,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723669307930785e-05,
+      "loss": 0.79,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722472158534854e-05,
+      "loss": 0.8394,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721272458021117e-05,
+      "loss": 0.8652,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972007020670436e-05,
+      "loss": 0.9036,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971886540490005e-05,
+      "loss": 0.7856,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717658052924324e-05,
+      "loss": 0.8806,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971644815109398e-05,
+      "loss": 0.8181,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715235699726487e-05,
+      "loss": 0.7998,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714020699139988e-05,
+      "loss": 0.8223,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971280314965329e-05,
+      "loss": 0.8362,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971158305158587e-05,
+      "loss": 0.877,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971036040525788e-05,
+      "loss": 0.8137,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970913521099013e-05,
+      "loss": 0.8816,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707907469104103e-05,
+      "loss": 0.8525,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970667717992195e-05,
+      "loss": 0.8401,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705444343766492e-05,
+      "loss": 0.8538,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704208960961225e-05,
+      "loss": 0.8308,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702971031830297e-05,
+      "loss": 0.7632,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970173055669854e-05,
+      "loss": 0.8843,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700487535891444e-05,
+      "loss": 0.8833,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699241969735165e-05,
+      "loss": 0.2219,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969799385855654e-05,
+      "loss": 0.8044,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696743202683063e-05,
+      "loss": 0.8711,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695490002442898e-05,
+      "loss": 0.864,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694234258164876e-05,
+      "loss": 0.8257,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692975970178498e-05,
+      "loss": 0.7969,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691715138813935e-05,
+      "loss": 0.8547,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690451764402013e-05,
+      "loss": 0.7955,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968918584727424e-05,
+      "loss": 0.9287,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687917387762784e-05,
+      "loss": 0.8904,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968664638620048e-05,
+      "loss": 0.8389,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685372842920832e-05,
+      "loss": 0.9048,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684096758258012e-05,
+      "loss": 0.9153,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968281813254685e-05,
+      "loss": 0.2607,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681536966122855e-05,
+      "loss": 0.8435,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96802532593222e-05,
+      "loss": 0.8557,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678967012481717e-05,
+      "loss": 0.8048,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967767822593891e-05,
+      "loss": 0.8733,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676386900031953e-05,
+      "loss": 0.8661,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967509303509967e-05,
+      "loss": 0.8613,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673796631481582e-05,
+      "loss": 0.842,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672497689517843e-05,
+      "loss": 0.8604,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967119620954929e-05,
+      "loss": 0.885,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966989219191743e-05,
+      "loss": 0.2734,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668585636964425e-05,
+      "loss": 0.8201,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667276545033103e-05,
+      "loss": 0.2449,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665964916466968e-05,
+      "loss": 0.8083,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664650751610178e-05,
+      "loss": 0.8857,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966333405080757e-05,
+      "loss": 0.9033,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662014814404632e-05,
+      "loss": 0.8914,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966069304274752e-05,
+      "loss": 0.8037,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965936873618307e-05,
+      "loss": 0.8433,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658041895058763e-05,
+      "loss": 0.8757,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656712519722758e-05,
+      "loss": 0.8049,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655380610523873e-05,
+      "loss": 0.8684,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965404616781159e-05,
+      "loss": 0.8678,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965270919193607e-05,
+      "loss": 0.8292,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651369683248117e-05,
+      "loss": 0.8528,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965002764209921e-05,
+      "loss": 0.9241,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96486830688415e-05,
+      "loss": 0.824,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647335963827786e-05,
+      "loss": 0.8135,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964598632741155e-05,
+      "loss": 0.8367,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644634159946917e-05,
+      "loss": 0.8748,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643279461788696e-05,
+      "loss": 0.8486,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964192223329235e-05,
+      "loss": 0.9067,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640562474814008e-05,
+      "loss": 0.8049,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639200186710456e-05,
+      "loss": 0.8849,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637835369339155e-05,
+      "loss": 0.8708,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963646802305823e-05,
+      "loss": 0.8234,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963509814822646e-05,
+      "loss": 0.8113,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633725745203283e-05,
+      "loss": 0.8457,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963235081434882e-05,
+      "loss": 0.8154,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630973356023847e-05,
+      "loss": 0.8447,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629593370589793e-05,
+      "loss": 0.9077,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962821085840876e-05,
+      "loss": 0.8634,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962682581984351e-05,
+      "loss": 0.8215,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625438255257472e-05,
+      "loss": 0.8577,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624048165014732e-05,
+      "loss": 0.8606,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622655549480042e-05,
+      "loss": 0.8251,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621260409018815e-05,
+      "loss": 0.7711,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961986274399713e-05,
+      "loss": 0.7805,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961846255478172e-05,
+      "loss": 0.8286,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617059841739994e-05,
+      "loss": 0.8171,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961565460524001e-05,
+      "loss": 0.8511,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614246845650496e-05,
+      "loss": 0.858,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612836563340835e-05,
+      "loss": 0.8416,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611423758681085e-05,
+      "loss": 0.8701,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961000843204195e-05,
+      "loss": 0.8577,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608590583794806e-05,
+      "loss": 0.8459,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607170214311688e-05,
+      "loss": 0.8118,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960574732396529e-05,
+      "loss": 0.8279,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604321913128977e-05,
+      "loss": 0.8159,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960289398217676e-05,
+      "loss": 0.8389,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960146353148332e-05,
+      "loss": 0.8987,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600030561424002e-05,
+      "loss": 0.7943,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598595072374807e-05,
+      "loss": 0.8413,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95971570647124e-05,
+      "loss": 0.8547,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595716538814104e-05,
+      "loss": 0.8708,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594273495057906e-05,
+      "loss": 0.8806,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959282793382245e-05,
+      "loss": 0.8174,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591379855487044e-05,
+      "loss": 0.8562,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589929260431653e-05,
+      "loss": 0.877,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958847614903691e-05,
+      "loss": 0.7639,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587020521684094e-05,
+      "loss": 0.7593,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585562378755157e-05,
+      "loss": 0.8401,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584101720632708e-05,
+      "loss": 0.7767,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582638547700013e-05,
+      "loss": 0.8943,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581172860341005e-05,
+      "loss": 0.8184,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957970465894026e-05,
+      "loss": 0.8694,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957823394388304e-05,
+      "loss": 0.8131,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576760715555242e-05,
+      "loss": 0.7971,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957528497434343e-05,
+      "loss": 0.845,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573806720634837e-05,
+      "loss": 0.7902,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957232595481734e-05,
+      "loss": 0.802,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570842677279495e-05,
+      "loss": 0.8335,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569356888410492e-05,
+      "loss": 0.874,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95678685886002e-05,
+      "loss": 0.2932,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566377778239137e-05,
+      "loss": 0.8357,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564884457718488e-05,
+      "loss": 0.8938,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563388627430084e-05,
+      "loss": 0.8198,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561890287766426e-05,
+      "loss": 0.8347,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560389439120667e-05,
+      "loss": 0.8611,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558886081886624e-05,
+      "loss": 0.8645,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557380216458767e-05,
+      "loss": 0.8789,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555871843232223e-05,
+      "loss": 0.8833,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554360962602784e-05,
+      "loss": 0.8271,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552847574966895e-05,
+      "loss": 0.8628,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551331680721662e-05,
+      "loss": 0.8359,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954981328026484e-05,
+      "loss": 0.8513,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548292373994854e-05,
+      "loss": 0.7964,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546768962310777e-05,
+      "loss": 0.7798,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545243045612346e-05,
+      "loss": 0.8513,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954371462429995e-05,
+      "loss": 0.8428,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954218369877464e-05,
+      "loss": 0.8091,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954065026943812e-05,
+      "loss": 0.7954,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539114336692746e-05,
+      "loss": 0.875,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953757590094155e-05,
+      "loss": 0.8276,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95360349625882e-05,
+      "loss": 0.8542,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534491522037033e-05,
+      "loss": 0.8457,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532945579693036e-05,
+      "loss": 0.8491,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531397135961855e-05,
+      "loss": 0.8037,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529846191249794e-05,
+      "loss": 0.8118,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952829274596381e-05,
+      "loss": 0.8184,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526736800511515e-05,
+      "loss": 0.7908,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525178355301184e-05,
+      "loss": 0.8256,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523617410741743e-05,
+      "loss": 0.7969,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522053967242774e-05,
+      "loss": 0.8774,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520488025214512e-05,
+      "loss": 0.7566,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518919585067855e-05,
+      "loss": 0.8054,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517348647214353e-05,
+      "loss": 0.856,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95157752120662e-05,
+      "loss": 0.8899,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514199280036273e-05,
+      "loss": 0.8511,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512620851538073e-05,
+      "loss": 0.8152,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511039926985776e-05,
+      "loss": 0.8022,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509456506794205e-05,
+      "loss": 0.8594,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507870591378842e-05,
+      "loss": 0.7891,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950628218115582e-05,
+      "loss": 0.7821,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950469127654193e-05,
+      "loss": 0.8513,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503097877954615e-05,
+      "loss": 0.8552,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501501985811973e-05,
+      "loss": 0.8291,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499903600532754e-05,
+      "loss": 0.8276,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498302722536368e-05,
+      "loss": 0.8245,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496699352242873e-05,
+      "loss": 0.8777,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495093490072982e-05,
+      "loss": 0.8564,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493485136448068e-05,
+      "loss": 0.8477,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491874291790156e-05,
+      "loss": 0.7957,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490260956521916e-05,
+      "loss": 0.8208,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488645131066678e-05,
+      "loss": 0.8733,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487026815848426e-05,
+      "loss": 0.804,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485406011291794e-05,
+      "loss": 0.8267,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483782717822075e-05,
+      "loss": 0.8025,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948215693586521e-05,
+      "loss": 0.7532,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480528665847796e-05,
+      "loss": 0.8481,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478897908197077e-05,
+      "loss": 0.8042,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477264663340956e-05,
+      "loss": 0.8206,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475628931707985e-05,
+      "loss": 0.8115,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473990713727376e-05,
+      "loss": 0.8757,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947235000982898e-05,
+      "loss": 0.8218,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947070682044331e-05,
+      "loss": 0.8247,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469061146001534e-05,
+      "loss": 0.8601,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946741298693546e-05,
+      "loss": 0.8652,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465762343677556e-05,
+      "loss": 0.8601,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464109216660945e-05,
+      "loss": 0.8391,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462453606319394e-05,
+      "loss": 0.7488,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946079551308733e-05,
+      "loss": 0.2564,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945913493739982e-05,
+      "loss": 0.7833,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945747187969259e-05,
+      "loss": 0.9097,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455806340402024e-05,
+      "loss": 0.795,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945413831996514e-05,
+      "loss": 0.853,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452467818819622e-05,
+      "loss": 0.7979,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450794837403792e-05,
+      "loss": 0.7858,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944911937615664e-05,
+      "loss": 0.8303,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944744143551779e-05,
+      "loss": 0.8281,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445761015927532e-05,
+      "loss": 0.7936,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444078117826785e-05,
+      "loss": 0.7972,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442392741657143e-05,
+      "loss": 0.8547,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944070488786083e-05,
+      "loss": 0.842,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439014556880737e-05,
+      "loss": 0.7399,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437321749160386e-05,
+      "loss": 0.8328,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435626465143967e-05,
+      "loss": 0.8589,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433928705276312e-05,
+      "loss": 0.8818,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432228470002902e-05,
+      "loss": 0.8591,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943052575976987e-05,
+      "loss": 0.281,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942882057502399e-05,
+      "loss": 0.8955,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94271129162127e-05,
+      "loss": 0.8064,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425402783784076e-05,
+      "loss": 0.7629,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423690178186846e-05,
+      "loss": 0.8301,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942197509987039e-05,
+      "loss": 0.8599,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420257549284732e-05,
+      "loss": 0.7556,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418537526880543e-05,
+      "loss": 0.8317,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416815033109154e-05,
+      "loss": 0.8575,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415090068422535e-05,
+      "loss": 0.8281,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413362633273303e-05,
+      "loss": 0.8467,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411632728114725e-05,
+      "loss": 0.8403,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409900353400725e-05,
+      "loss": 0.8708,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408165509585862e-05,
+      "loss": 0.824,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406428197125354e-05,
+      "loss": 0.7842,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404688416475053e-05,
+      "loss": 0.8938,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402946168091472e-05,
+      "loss": 0.7516,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401201452431765e-05,
+      "loss": 0.8043,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939945426995374e-05,
+      "loss": 0.8635,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939770462111584e-05,
+      "loss": 0.813,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395952506377164e-05,
+      "loss": 0.8536,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394197926197456e-05,
+      "loss": 0.8796,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939244088103711e-05,
+      "loss": 0.8374,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939068137135716e-05,
+      "loss": 0.8604,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388919397619297e-05,
+      "loss": 0.8025,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387154960285847e-05,
+      "loss": 0.8386,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385388059819787e-05,
+      "loss": 0.8389,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383618696684743e-05,
+      "loss": 0.8562,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938184687134499e-05,
+      "loss": 0.842,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380072584265432e-05,
+      "loss": 0.7354,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378295835911645e-05,
+      "loss": 0.8467,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937651662674983e-05,
+      "loss": 0.7955,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374734957246837e-05,
+      "loss": 0.8438,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937295082787017e-05,
+      "loss": 0.8533,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371164239087975e-05,
+      "loss": 0.8015,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369375191369042e-05,
+      "loss": 0.8213,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367583685182804e-05,
+      "loss": 0.7888,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936578972099934e-05,
+      "loss": 0.9062,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936399329928938e-05,
+      "loss": 0.9202,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936219442052429e-05,
+      "loss": 0.7996,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360393085176084e-05,
+      "loss": 0.8701,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935858929371743e-05,
+      "loss": 0.8485,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356783046621623e-05,
+      "loss": 0.8435,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354974344362613e-05,
+      "loss": 0.8708,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353163187414993e-05,
+      "loss": 0.8207,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351349576254003e-05,
+      "loss": 0.8696,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349533511355517e-05,
+      "loss": 0.8209,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347714993196064e-05,
+      "loss": 0.8376,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345894022252817e-05,
+      "loss": 0.7572,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344070599003575e-05,
+      "loss": 0.8094,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342244723926808e-05,
+      "loss": 0.7792,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93404163975016e-05,
+      "loss": 0.7705,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338585620207702e-05,
+      "loss": 0.2796,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336752392525496e-05,
+      "loss": 0.8807,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334916714936015e-05,
+      "loss": 0.791,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933307858792092e-05,
+      "loss": 0.795,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933123801196253e-05,
+      "loss": 0.8435,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329394987543802e-05,
+      "loss": 0.8284,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327549515148333e-05,
+      "loss": 0.8342,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325701595260366e-05,
+      "loss": 0.8459,
+      "step": 911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932385122836478e-05,
+      "loss": 0.9072,
+      "step": 912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93219984149471e-05,
+      "loss": 0.8843,
+      "step": 913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320143155493496e-05,
+      "loss": 0.8284,
+      "step": 914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318285450490777e-05,
+      "loss": 0.8483,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931642530042639e-05,
+      "loss": 0.8687,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314562705788434e-05,
+      "loss": 0.9065,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312697667065633e-05,
+      "loss": 0.8306,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310830184747366e-05,
+      "loss": 0.8496,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308960259323655e-05,
+      "loss": 0.2795,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307087891285148e-05,
+      "loss": 0.7997,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305213081123146e-05,
+      "loss": 0.876,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303335829329594e-05,
+      "loss": 0.79,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930145613639706e-05,
+      "loss": 0.8621,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299574002818776e-05,
+      "loss": 0.7639,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929768942908859e-05,
+      "loss": 0.8177,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295802415701017e-05,
+      "loss": 0.8574,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293912963151187e-05,
+      "loss": 0.8601,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292021071934885e-05,
+      "loss": 0.8168,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929012674254853e-05,
+      "loss": 0.8386,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288229975489182e-05,
+      "loss": 0.824,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286330771254548e-05,
+      "loss": 0.9285,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928442913034296e-05,
+      "loss": 0.8103,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282525053253397e-05,
+      "loss": 0.7424,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928061854048548e-05,
+      "loss": 0.8301,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927870959253947e-05,
+      "loss": 0.8325,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276798209916258e-05,
+      "loss": 0.8357,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927488439311738e-05,
+      "loss": 0.8621,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272968142645013e-05,
+      "loss": 0.8218,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271049459001963e-05,
+      "loss": 0.845,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269128342691693e-05,
+      "loss": 0.8094,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267204794218276e-05,
+      "loss": 0.7993,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265278814086455e-05,
+      "loss": 0.8796,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263350402801584e-05,
+      "loss": 0.8879,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261419560869673e-05,
+      "loss": 0.8364,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925948628879736e-05,
+      "loss": 0.8346,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257550587091926e-05,
+      "loss": 0.821,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925561245626129e-05,
+      "loss": 0.8026,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253671896813998e-05,
+      "loss": 0.8145,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251728909259248e-05,
+      "loss": 0.8276,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924978349410686e-05,
+      "loss": 0.75,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924783565186731e-05,
+      "loss": 0.824,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924588538305169e-05,
+      "loss": 0.8674,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924393268817175e-05,
+      "loss": 0.877,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241977567739852e-05,
+      "loss": 0.7225,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924002002226902e-05,
+      "loss": 0.8594,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238060052272895e-05,
+      "loss": 0.8374,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236097658265765e-05,
+      "loss": 0.8301,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234132840762542e-05,
+      "loss": 0.8572,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232165600278795e-05,
+      "loss": 0.8928,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230195937330705e-05,
+      "loss": 0.8306,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228223852435106e-05,
+      "loss": 0.7903,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922624934610946e-05,
+      "loss": 0.7822,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922427241887186e-05,
+      "loss": 0.7578,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222293071241046e-05,
+      "loss": 0.8235,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922031130373639e-05,
+      "loss": 0.8276,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218327116877888e-05,
+      "loss": 0.8989,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921634051118618e-05,
+      "loss": 0.7932,
+      "step": 968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214351487182542e-05,
+      "loss": 0.8274,
+      "step": 969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921236004538888e-05,
+      "loss": 0.8328,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210366186327735e-05,
+      "loss": 0.7749,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208369910522287e-05,
+      "loss": 0.8481,
+      "step": 972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920637121849635e-05,
+      "loss": 0.2694,
+      "step": 973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920437011077436e-05,
+      "loss": 0.8853,
+      "step": 974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92023665878814e-05,
+      "loss": 0.8875,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200360650343183e-05,
+      "loss": 0.8811,
+      "step": 976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919835229868605e-05,
+      "loss": 0.8141,
+      "step": 977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196341533436986e-05,
+      "loss": 0.8767,
+      "step": 978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.91943283551236e-05,
+      "loss": 0.8544,
+      "step": 979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192312764274144e-05,
+      "loss": 0.8467,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190294761417488e-05,
+      "loss": 0.8279,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188274347083147e-05,
+      "loss": 0.855,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918625152180127e-05,
+      "loss": 0.8733,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184226286102626e-05,
+      "loss": 0.803,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182198640518636e-05,
+      "loss": 0.8684,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918016858558133e-05,
+      "loss": 0.8386,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178136121823386e-05,
+      "loss": 0.8239,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917610124977811e-05,
+      "loss": 0.821,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174063969979446e-05,
+      "loss": 0.8142,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917202428296196e-05,
+      "loss": 0.765,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916998218926085e-05,
+      "loss": 0.8298,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916793768941195e-05,
+      "loss": 0.7885,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916589078395173e-05,
+      "loss": 0.8269,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163841473417278e-05,
+      "loss": 0.8279,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161789758346328e-05,
+      "loss": 0.8726,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159735639277234e-05,
+      "loss": 0.8601,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915767911674898e-05,
+      "loss": 0.8369,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155620191301193e-05,
+      "loss": 0.8701,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915355886347412e-05,
+      "loss": 0.778,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151495133808638e-05,
+      "loss": 0.822,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914942900284626e-05,
+      "loss": 0.8442,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147360471129124e-05,
+      "loss": 0.7925,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145289539200003e-05,
+      "loss": 0.8616,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143216207602297e-05,
+      "loss": 0.8156,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141140476880032e-05,
+      "loss": 0.8069,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913906234757787e-05,
+      "loss": 0.8123,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91369818202411e-05,
+      "loss": 0.7996,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134898895415635e-05,
+      "loss": 0.856,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913281357364803e-05,
+      "loss": 0.2719,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130725855485453e-05,
+      "loss": 0.8682,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128635741475716e-05,
+      "loss": 0.2458,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126543232167247e-05,
+      "loss": 0.7853,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124448328109106e-05,
+      "loss": 0.843,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122351029850987e-05,
+      "loss": 0.8652,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120251337943208e-05,
+      "loss": 0.7334,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118149252936716e-05,
+      "loss": 0.8728,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116044775383086e-05,
+      "loss": 0.7504,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113937905834518e-05,
+      "loss": 0.7955,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911182864484384e-05,
+      "loss": 0.7723,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910971699296451e-05,
+      "loss": 0.8579,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107602950750616e-05,
+      "loss": 0.8179,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105486518756872e-05,
+      "loss": 0.7491,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910336769753861e-05,
+      "loss": 0.7866,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101246487651798e-05,
+      "loss": 0.806,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909912288965303e-05,
+      "loss": 0.8391,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096996904099528e-05,
+      "loss": 0.8901,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094868531549127e-05,
+      "loss": 0.7966,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092737772560315e-05,
+      "loss": 0.8383,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909060462769218e-05,
+      "loss": 0.7922,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088469097504447e-05,
+      "loss": 0.8125,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086331182557466e-05,
+      "loss": 0.925,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084190883412215e-05,
+      "loss": 0.8313,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082048200630297e-05,
+      "loss": 0.8037,
+      "step": 1033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079903134773935e-05,
+      "loss": 0.7917,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077755686405982e-05,
+      "loss": 0.7551,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907560585608992e-05,
+      "loss": 0.8489,
+      "step": 1036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073453644389843e-05,
+      "loss": 0.7976,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071299051870484e-05,
+      "loss": 0.8341,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90691420790972e-05,
+      "loss": 0.8152,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066982726635955e-05,
+      "loss": 0.8533,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906482099505336e-05,
+      "loss": 0.8562,
+      "step": 1041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906265688491664e-05,
+      "loss": 0.8538,
+      "step": 1042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906049039679364e-05,
+      "loss": 0.8674,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905832153125283e-05,
+      "loss": 0.7903,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905615028886332e-05,
+      "loss": 0.8525,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053976670194818e-05,
+      "loss": 0.8335,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051800675817674e-05,
+      "loss": 0.8113,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049622306302857e-05,
+      "loss": 0.8154,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047441562221956e-05,
+      "loss": 0.7778,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045258444147185e-05,
+      "loss": 0.8149,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904307295265138e-05,
+      "loss": 0.7654,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040885088308e-05,
+      "loss": 0.8877,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038694851691132e-05,
+      "loss": 0.8704,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036502243375475e-05,
+      "loss": 0.8008,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034307263936365e-05,
+      "loss": 0.8325,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903210991394974e-05,
+      "loss": 0.8213,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902991019399218e-05,
+      "loss": 0.8235,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027708104640874e-05,
+      "loss": 0.7954,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902550364647364e-05,
+      "loss": 0.8149,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023296820068907e-05,
+      "loss": 0.7988,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021087626005742e-05,
+      "loss": 0.2889,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901887606486382e-05,
+      "loss": 0.8472,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901666213722344e-05,
+      "loss": 0.8433,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901444584366553e-05,
+      "loss": 0.8181,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012227184771623e-05,
+      "loss": 0.8154,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901000616112388e-05,
+      "loss": 0.824,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007782773305097e-05,
+      "loss": 0.8171,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005557021898667e-05,
+      "loss": 0.8013,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900332890748862e-05,
+      "loss": 0.7284,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001098430659595e-05,
+      "loss": 0.7986,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899886559199686e-05,
+      "loss": 0.8262,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899663039208629e-05,
+      "loss": 0.8358,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994392831514396e-05,
+      "loss": 0.877,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992152910868297e-05,
+      "loss": 0.8137,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898991063073574e-05,
+      "loss": 0.8516,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987665991705072e-05,
+      "loss": 0.8372,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985418994365288e-05,
+      "loss": 0.8936,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898316963930598e-05,
+      "loss": 0.8564,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980917927117364e-05,
+      "loss": 0.8643,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978663858390277e-05,
+      "loss": 0.9045,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976407433716176e-05,
+      "loss": 0.8401,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974148653687126e-05,
+      "loss": 0.8257,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971887518895827e-05,
+      "loss": 0.8345,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969624029935575e-05,
+      "loss": 0.8093,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967358187400306e-05,
+      "loss": 0.8801,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896508999188456e-05,
+      "loss": 0.8259,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962819443983496e-05,
+      "loss": 0.7942,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960546544292894e-05,
+      "loss": 0.8152,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895827129340915e-05,
+      "loss": 0.3242,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955993691929272e-05,
+      "loss": 0.7859,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953713740450893e-05,
+      "loss": 0.7849,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895143143957226e-05,
+      "loss": 0.7727,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949146789892227e-05,
+      "loss": 0.8046,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894685979201028e-05,
+      "loss": 0.7805,
+      "step": 1094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944570446526506e-05,
+      "loss": 0.8721,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942278754041625e-05,
+      "loss": 0.7585,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939984715156952e-05,
+      "loss": 0.7631,
+      "step": 1097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893768833047444e-05,
+      "loss": 0.9182,
+      "step": 1098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935389600596644e-05,
+      "loss": 0.8098,
+      "step": 1099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893308852612673e-05,
+      "loss": 0.2743,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893078510766849e-05,
+      "loss": 0.812,
+      "step": 1101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892847934582633e-05,
+      "loss": 0.8477,
+      "step": 1102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926171241205262e-05,
+      "loss": 0.8369,
+      "step": 1103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892386079441092e-05,
+      "loss": 0.8643,
+      "step": 1104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892154800604956e-05,
+      "loss": 0.8337,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919232876728033e-05,
+      "loss": 0.8279,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891691540705382e-05,
+      "loss": 0.8362,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891459559763501e-05,
+      "loss": 0.8809,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912273449080305e-05,
+      "loss": 0.7849,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909948961999023e-05,
+      "loss": 0.7946,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89076221370011e-05,
+      "loss": 0.8264,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905292974697077e-05,
+      "loss": 0.276,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890296147569811e-05,
+      "loss": 0.7903,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900627640615975e-05,
+      "loss": 0.8523,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898291470063052e-05,
+      "loss": 0.8311,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889595296465234e-05,
+      "loss": 0.8547,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893612124997453e-05,
+      "loss": 0.2618,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891268951712605e-05,
+      "loss": 0.8005,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888923445412635e-05,
+      "loss": 0.7957,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886575606712988e-05,
+      "loss": 0.7429,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884225436229726e-05,
+      "loss": 0.8521,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881872934579518e-05,
+      "loss": 0.7822,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879518102379643e-05,
+      "loss": 0.8075,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877160940247995e-05,
+      "loss": 0.833,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874801448803084e-05,
+      "loss": 0.825,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887243962866402e-05,
+      "loss": 0.8264,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870075480450538e-05,
+      "loss": 0.8567,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867709004782968e-05,
+      "loss": 0.8231,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865340202282266e-05,
+      "loss": 0.8271,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862969073569986e-05,
+      "loss": 0.2682,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860595619268302e-05,
+      "loss": 0.8694,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885821983999999e-05,
+      "loss": 0.8679,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885584173638844e-05,
+      "loss": 0.826,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853461309057655e-05,
+      "loss": 0.791,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885107855863225e-05,
+      "loss": 0.8621,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884869348573743e-05,
+      "loss": 0.8358,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846306090999036e-05,
+      "loss": 0.7762,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88439163750435e-05,
+      "loss": 0.804,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884152433849787e-05,
+      "loss": 0.8086,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839129981989805e-05,
+      "loss": 0.752,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836733306147566e-05,
+      "loss": 0.248,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834334311600026e-05,
+      "loss": 0.873,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883193299897667e-05,
+      "loss": 0.8081,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829529368907584e-05,
+      "loss": 0.8337,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827123422023467e-05,
+      "loss": 0.8369,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882471515895563e-05,
+      "loss": 0.8279,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822304580335983e-05,
+      "loss": 0.834,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881989168679704e-05,
+      "loss": 0.8306,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817476478971943e-05,
+      "loss": 0.8267,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815058957494416e-05,
+      "loss": 0.8513,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881263912299881e-05,
+      "loss": 0.8491,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881021697612007e-05,
+      "loss": 0.7838,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807792517493757e-05,
+      "loss": 0.8511,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805365747756033e-05,
+      "loss": 0.8354,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802936667543665e-05,
+      "loss": 0.854,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880050527749403e-05,
+      "loss": 0.2827,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798071578245113e-05,
+      "loss": 0.8293,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795635570435496e-05,
+      "loss": 0.7748,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879319725470438e-05,
+      "loss": 0.8074,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790756631691557e-05,
+      "loss": 0.7704,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788313702037437e-05,
+      "loss": 0.8101,
+      "step": 1161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878586846638303e-05,
+      "loss": 0.7896,
+      "step": 1162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783420925369948e-05,
+      "loss": 0.8706,
+      "step": 1163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780971079640412e-05,
+      "loss": 0.9021,
+      "step": 1164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778518929837247e-05,
+      "loss": 0.8375,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877606447660388e-05,
+      "loss": 0.7839,
+      "step": 1166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877360772058435e-05,
+      "loss": 0.7732,
+      "step": 1167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877114866242329e-05,
+      "loss": 0.8135,
+      "step": 1168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768687302765943e-05,
+      "loss": 0.7898,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766223642258158e-05,
+      "loss": 0.8423,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763757681546383e-05,
+      "loss": 0.8181,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761289421277667e-05,
+      "loss": 0.8138,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875881886209967e-05,
+      "loss": 0.845,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756346004660654e-05,
+      "loss": 0.8677,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753870849609477e-05,
+      "loss": 0.8655,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875139339759561e-05,
+      "loss": 0.8145,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748913649269116e-05,
+      "loss": 0.8274,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874643160528067e-05,
+      "loss": 0.8289,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874394726628154e-05,
+      "loss": 0.811,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874146063292361e-05,
+      "loss": 0.8251,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738971705859354e-05,
+      "loss": 0.8031,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736480485741846e-05,
+      "loss": 0.863,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733986973224772e-05,
+      "loss": 0.8301,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731491168962416e-05,
+      "loss": 0.7745,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728993073609657e-05,
+      "loss": 0.8545,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872649268782198e-05,
+      "loss": 0.8079,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872399001225548e-05,
+      "loss": 0.2948,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721485047566835e-05,
+      "loss": 0.7822,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718977794413338e-05,
+      "loss": 0.8325,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871646825345287e-05,
+      "loss": 0.8862,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713956425343927e-05,
+      "loss": 0.8169,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711442310745597e-05,
+      "loss": 0.8384,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708925910317562e-05,
+      "loss": 0.8704,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706407224720114e-05,
+      "loss": 0.8055,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870388625461414e-05,
+      "loss": 0.8242,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870136300066113e-05,
+      "loss": 0.8132,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698837463523167e-05,
+      "loss": 0.8047,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869630964386294e-05,
+      "loss": 0.9153,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693779542343734e-05,
+      "loss": 0.8057,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691247159629427e-05,
+      "loss": 0.778,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688712496384505e-05,
+      "loss": 0.7893,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686175553274046e-05,
+      "loss": 0.7932,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868363633096373e-05,
+      "loss": 0.8394,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681094830119834e-05,
+      "loss": 0.8416,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678551051409232e-05,
+      "loss": 0.8466,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676004995499398e-05,
+      "loss": 0.7764,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673456663058397e-05,
+      "loss": 0.8391,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86709060547549e-05,
+      "loss": 0.8059,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866835317125817e-05,
+      "loss": 0.7609,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866579801323807e-05,
+      "loss": 0.9006,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866324058136505e-05,
+      "loss": 0.8206,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660680876310175e-05,
+      "loss": 0.8182,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658118898745087e-05,
+      "loss": 0.8684,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655554649342043e-05,
+      "loss": 0.8313,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865298812877388e-05,
+      "loss": 0.8137,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865041933771404e-05,
+      "loss": 0.7554,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647848276836557e-05,
+      "loss": 0.8369,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645274946816062e-05,
+      "loss": 0.8357,
+      "step": 1218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642699348327782e-05,
+      "loss": 0.8412,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640121482047537e-05,
+      "loss": 0.8247,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637541348651742e-05,
+      "loss": 0.8352,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634958948817412e-05,
+      "loss": 0.7683,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632374283222153e-05,
+      "loss": 0.8723,
+      "step": 1223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629787352544163e-05,
+      "loss": 0.8491,
+      "step": 1224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627198157462237e-05,
+      "loss": 0.8662,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862460669865577e-05,
+      "loss": 0.844,
+      "step": 1226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862201297680474e-05,
+      "loss": 0.8364,
+      "step": 1227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619416992589722e-05,
+      "loss": 0.8274,
+      "step": 1228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616818746691886e-05,
+      "loss": 0.8254,
+      "step": 1229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614218239793e-05,
+      "loss": 0.8354,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611615472575423e-05,
+      "loss": 0.856,
+      "step": 1231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86090104457221e-05,
+      "loss": 0.8462,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8606403159916578e-05,
+      "loss": 0.8198,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603793615842992e-05,
+      "loss": 0.762,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601181814186072e-05,
+      "loss": 0.834,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598567755631135e-05,
+      "loss": 0.7887,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595951440864097e-05,
+      "loss": 0.8396,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593332870571462e-05,
+      "loss": 0.8359,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590712045440327e-05,
+      "loss": 0.7257,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588088966158383e-05,
+      "loss": 0.7811,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585463633413903e-05,
+      "loss": 0.8572,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582836047895772e-05,
+      "loss": 0.8115,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580206210293437e-05,
+      "loss": 0.7888,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577574121296963e-05,
+      "loss": 0.8145,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857493978159699e-05,
+      "loss": 0.7986,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572303191884753e-05,
+      "loss": 0.8003,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569664352852073e-05,
+      "loss": 0.8162,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567023265191372e-05,
+      "loss": 0.8162,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856437992959565e-05,
+      "loss": 0.8301,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856173434675851e-05,
+      "loss": 0.7917,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559086517374126e-05,
+      "loss": 0.8662,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855643644213728e-05,
+      "loss": 0.7871,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553784121743335e-05,
+      "loss": 0.8096,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551129556888244e-05,
+      "loss": 0.8118,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548472748268547e-05,
+      "loss": 0.8049,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545813696581373e-05,
+      "loss": 0.8657,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543152402524444e-05,
+      "loss": 0.7776,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540488866796066e-05,
+      "loss": 0.8152,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537823090095134e-05,
+      "loss": 0.86,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853515507312113e-05,
+      "loss": 0.8384,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532484816574136e-05,
+      "loss": 0.8025,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529812321154798e-05,
+      "loss": 0.8652,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852713758756437e-05,
+      "loss": 0.8782,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524460616504683e-05,
+      "loss": 0.8828,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521781408678158e-05,
+      "loss": 0.8176,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519099964787804e-05,
+      "loss": 0.7834,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516416285537217e-05,
+      "loss": 0.875,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513730371630577e-05,
+      "loss": 0.7535,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851104222377265e-05,
+      "loss": 0.8271,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508351842668793e-05,
+      "loss": 0.8159,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505659229024938e-05,
+      "loss": 0.8416,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502964383547622e-05,
+      "loss": 0.8408,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500267306943947e-05,
+      "loss": 0.8699,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497567999921612e-05,
+      "loss": 0.7968,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84948664631889e-05,
+      "loss": 0.8018,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849216269745468e-05,
+      "loss": 0.8674,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84894567034284e-05,
+      "loss": 0.7534,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486748481820094e-05,
+      "loss": 0.8733,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848403803334039e-05,
+      "loss": 0.8918,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848132535870049e-05,
+      "loss": 0.8074,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478610458612186e-05,
+      "loss": 0.78,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475893333787846e-05,
+      "loss": 0.7937,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473173984940425e-05,
+      "loss": 0.8098,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470452412783473e-05,
+      "loss": 0.8398,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467728618031108e-05,
+      "loss": 0.7891,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465002601398043e-05,
+      "loss": 0.8387,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846227436359956e-05,
+      "loss": 0.8303,
+      "step": 1287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459543905351534e-05,
+      "loss": 0.8145,
+      "step": 1288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845681122737043e-05,
+      "loss": 0.7854,
+      "step": 1289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845407633037327e-05,
+      "loss": 0.7936,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451339215077685e-05,
+      "loss": 0.7633,
+      "step": 1291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844859988220188e-05,
+      "loss": 0.8083,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445858332464634e-05,
+      "loss": 0.8745,
+      "step": 1293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443114566585314e-05,
+      "loss": 0.8779,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440368585283866e-05,
+      "loss": 0.8032,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843762038928082e-05,
+      "loss": 0.822,
+      "step": 1296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434869979297288e-05,
+      "loss": 0.865,
+      "step": 1297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843211735605496e-05,
+      "loss": 0.7686,
+      "step": 1298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842936252027611e-05,
+      "loss": 0.8486,
+      "step": 1299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842660547268358e-05,
+      "loss": 0.8528,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423846214000815e-05,
+      "loss": 0.8523,
+      "step": 1301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842108474495182e-05,
+      "loss": 0.8291,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418321066261192e-05,
+      "loss": 0.8,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415555178654103e-05,
+      "loss": 0.8752,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412787082856304e-05,
+      "loss": 0.8279,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410016779594124e-05,
+      "loss": 0.7611,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840724426959448e-05,
+      "loss": 0.8,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404469553584856e-05,
+      "loss": 0.76,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401692632293322e-05,
+      "loss": 0.7975,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839891350644853e-05,
+      "loss": 0.8301,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83961321767797e-05,
+      "loss": 0.8215,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393348644016636e-05,
+      "loss": 0.8386,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839056290888973e-05,
+      "loss": 0.7859,
+      "step": 1313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387774972129924e-05,
+      "loss": 0.7009,
+      "step": 1314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384984834468773e-05,
+      "loss": 0.8466,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382192496638385e-05,
+      "loss": 0.8004,
+      "step": 1316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837939795937145e-05,
+      "loss": 0.8237,
+      "step": 1317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837660122340124e-05,
+      "loss": 0.8569,
+      "step": 1318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373802289461606e-05,
+      "loss": 0.7886,
+      "step": 1319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371001158286965e-05,
+      "loss": 0.7328,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368197830612322e-05,
+      "loss": 0.8074,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365392307173246e-05,
+      "loss": 0.7986,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362584588705895e-05,
+      "loss": 0.8508,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359774675947e-05,
+      "loss": 0.843,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356962569633858e-05,
+      "loss": 0.8716,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835414827050435e-05,
+      "loss": 0.7798,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835133177929694e-05,
+      "loss": 0.7888,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348513096750645e-05,
+      "loss": 0.8569,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345692223605074e-05,
+      "loss": 0.8484,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342869160600413e-05,
+      "loss": 0.8167,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340043908477417e-05,
+      "loss": 0.8977,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337216467977404e-05,
+      "loss": 0.8037,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833438683984229e-05,
+      "loss": 0.8501,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331555024814544e-05,
+      "loss": 0.8262,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832872102363722e-05,
+      "loss": 0.8369,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325884837053943e-05,
+      "loss": 0.3071,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832304646580891e-05,
+      "loss": 0.8413,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320205910646892e-05,
+      "loss": 0.8528,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317363172313238e-05,
+      "loss": 0.8545,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314518251553862e-05,
+      "loss": 0.9014,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831167114911525e-05,
+      "loss": 0.8481,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830882186574447e-05,
+      "loss": 0.8479,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305970402189158e-05,
+      "loss": 0.8145,
+      "step": 1343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303116759197516e-05,
+      "loss": 0.3062,
+      "step": 1344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300260937518318e-05,
+      "loss": 0.8114,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297402937900928e-05,
+      "loss": 0.2878,
+      "step": 1346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294542761095262e-05,
+      "loss": 0.3135,
+      "step": 1347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291680407851807e-05,
+      "loss": 0.8228,
+      "step": 1348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828881587892163e-05,
+      "loss": 0.7957,
+      "step": 1349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285949175056375e-05,
+      "loss": 0.7733,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283080297008237e-05,
+      "loss": 0.7996,
+      "step": 1351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280209245529996e-05,
+      "loss": 0.8167,
+      "step": 1352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277336021374995e-05,
+      "loss": 0.8118,
+      "step": 1353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274460625297157e-05,
+      "loss": 0.8213,
+      "step": 1354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271583058050966e-05,
+      "loss": 0.7664,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268703320391478e-05,
+      "loss": 0.78,
+      "step": 1356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265821413074314e-05,
+      "loss": 0.8237,
+      "step": 1357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262937336855678e-05,
+      "loss": 0.8293,
+      "step": 1358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826005109249233e-05,
+      "loss": 0.828,
+      "step": 1359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257162680741597e-05,
+      "loss": 0.8623,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254272102361393e-05,
+      "loss": 0.826,
+      "step": 1361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251379358110174e-05,
+      "loss": 0.7983,
+      "step": 1362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248484448746993e-05,
+      "loss": 0.7717,
+      "step": 1363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245587375031445e-05,
+      "loss": 0.8245,
+      "step": 1364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242688137723713e-05,
+      "loss": 0.7922,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239786737584536e-05,
+      "loss": 0.295,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823688317537522e-05,
+      "loss": 0.7783,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233977451857646e-05,
+      "loss": 0.8396,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231069567794263e-05,
+      "loss": 0.7703,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228159523948073e-05,
+      "loss": 0.8125,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225247321082656e-05,
+      "loss": 0.8499,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222332959962162e-05,
+      "loss": 0.797,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219416441351295e-05,
+      "loss": 0.7998,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821649776601534e-05,
+      "loss": 0.8628,
+      "step": 1374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821357693472013e-05,
+      "loss": 0.7305,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821065394823208e-05,
+      "loss": 0.813,
+      "step": 1376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207728807318163e-05,
+      "loss": 0.8621,
+      "step": 1377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204801512745914e-05,
+      "loss": 0.8059,
+      "step": 1378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201872065283444e-05,
+      "loss": 0.8228,
+      "step": 1379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198940465699418e-05,
+      "loss": 0.8367,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819600671476307e-05,
+      "loss": 0.8066,
+      "step": 1381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81930708132442e-05,
+      "loss": 0.8285,
+      "step": 1382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819013276191317e-05,
+      "loss": 0.8304,
+      "step": 1383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187192561540908e-05,
+      "loss": 0.7861,
+      "step": 1384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184250212898903e-05,
+      "loss": 0.8408,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818130571675921e-05,
+      "loss": 0.785,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178359073894446e-05,
+      "loss": 0.8354,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175410285077798e-05,
+      "loss": 0.8484,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172459351083e-05,
+      "loss": 0.8164,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169506272684367e-05,
+      "loss": 0.8733,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816655105065677e-05,
+      "loss": 0.8375,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163593685775636e-05,
+      "loss": 0.2903,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816063417881696e-05,
+      "loss": 0.8545,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157672530557305e-05,
+      "loss": 0.7634,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154708741773782e-05,
+      "loss": 0.8016,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151742813244074e-05,
+      "loss": 0.8342,
+      "step": 1396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148774745746425e-05,
+      "loss": 0.8302,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145804540059634e-05,
+      "loss": 0.2975,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142832196963068e-05,
+      "loss": 0.7822,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813985771723665e-05,
+      "loss": 0.8413,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136881101660864e-05,
+      "loss": 0.7865,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133902351016764e-05,
+      "loss": 0.7836,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130921466085944e-05,
+      "loss": 0.769,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812793844765058e-05,
+      "loss": 0.8386,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81249532964934e-05,
+      "loss": 0.856,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121966013397677e-05,
+      "loss": 0.7911,
+      "step": 1406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118976599147263e-05,
+      "loss": 0.8411,
+      "step": 1407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811598505452656e-05,
+      "loss": 0.8181,
+      "step": 1408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811299138032054e-05,
+      "loss": 0.2856,
+      "step": 1409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109995577314713e-05,
+      "loss": 0.8362,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106997646295168e-05,
+      "loss": 0.8423,
+      "step": 1411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810399758804854e-05,
+      "loss": 0.825,
+      "step": 1412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100995403362027e-05,
+      "loss": 0.8345,
+      "step": 1413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097991093023385e-05,
+      "loss": 0.8704,
+      "step": 1414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094984657820923e-05,
+      "loss": 0.7961,
+      "step": 1415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091976098543515e-05,
+      "loss": 0.8196,
+      "step": 1416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088965415980588e-05,
+      "loss": 0.8315,
+      "step": 1417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808595261092213e-05,
+      "loss": 0.6676,
+      "step": 1418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082937684158676e-05,
+      "loss": 0.856,
+      "step": 1419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079920636481333e-05,
+      "loss": 0.8711,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076901468681747e-05,
+      "loss": 0.8425,
+      "step": 1421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8073880181552136e-05,
+      "loss": 0.7795,
+      "step": 1422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070856775885263e-05,
+      "loss": 0.835,
+      "step": 1423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806783125247445e-05,
+      "loss": 0.8774,
+      "step": 1424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064803612113586e-05,
+      "loss": 0.8121,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061773855597094e-05,
+      "loss": 0.9009,
+      "step": 1426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058741983719975e-05,
+      "loss": 0.8098,
+      "step": 1427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805570799727776e-05,
+      "loss": 0.275,
+      "step": 1428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805267189706656e-05,
+      "loss": 0.7628,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049633683883025e-05,
+      "loss": 0.8755,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046593358524366e-05,
+      "loss": 0.2444,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043550921788343e-05,
+      "loss": 0.8291,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040506374473274e-05,
+      "loss": 0.7705,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037459717378027e-05,
+      "loss": 0.8193,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034410951302034e-05,
+      "loss": 0.7815,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031360077045262e-05,
+      "loss": 0.8445,
+      "step": 1436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028307095408254e-05,
+      "loss": 0.7329,
+      "step": 1437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025252007192086e-05,
+      "loss": 0.7778,
+      "step": 1438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802219481319839e-05,
+      "loss": 0.8627,
+      "step": 1439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801913551422937e-05,
+      "loss": 0.7993,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801607411108776e-05,
+      "loss": 0.2667,
+      "step": 1441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801301060457685e-05,
+      "loss": 0.8303,
+      "step": 1442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009944995500485e-05,
+      "loss": 0.8682,
+      "step": 1443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800687728466307e-05,
+      "loss": 0.8549,
+      "step": 1444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003807472869547e-05,
+      "loss": 0.7942,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800073556092542e-05,
+      "loss": 0.8296,
+      "step": 1446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997661549636737e-05,
+      "loss": 0.8174,
+      "step": 1447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994585439810102e-05,
+      "loss": 0.3021,
+      "step": 1448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991507232252668e-05,
+      "loss": 0.7649,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988426927772134e-05,
+      "loss": 0.8695,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798534452717676e-05,
+      "loss": 0.2701,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982260031275343e-05,
+      "loss": 0.2782,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797917344087724e-05,
+      "loss": 0.9011,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976084756792353e-05,
+      "loss": 0.7368,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797299397983113e-05,
+      "loss": 0.7728,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969901110804576e-05,
+      "loss": 0.8477,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796680615052424e-05,
+      "loss": 0.8008,
+      "step": 1457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963709099802223e-05,
+      "loss": 0.7993,
+      "step": 1458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960609959451174e-05,
+      "loss": 0.7352,
+      "step": 1459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957508730284283e-05,
+      "loss": 0.8075,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795440541311529e-05,
+      "loss": 0.7874,
+      "step": 1461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951300008758502e-05,
+      "loss": 0.7791,
+      "step": 1462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948192518028742e-05,
+      "loss": 0.7891,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945082941741408e-05,
+      "loss": 0.8533,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941971280712427e-05,
+      "loss": 0.3018,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938857535758284e-05,
+      "loss": 0.8115,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935741707696008e-05,
+      "loss": 0.8379,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932623797343163e-05,
+      "loss": 0.7825,
+      "step": 1468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929503805517887e-05,
+      "loss": 0.833,
+      "step": 1469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926381733038833e-05,
+      "loss": 0.8165,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923257580725217e-05,
+      "loss": 0.8113,
+      "step": 1471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920131349396804e-05,
+      "loss": 0.8276,
+      "step": 1472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791700303987389e-05,
+      "loss": 0.8042,
+      "step": 1473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913872652977328e-05,
+      "loss": 0.8625,
+      "step": 1474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910740189528515e-05,
+      "loss": 0.8079,
+      "step": 1475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907605650349384e-05,
+      "loss": 0.8447,
+      "step": 1476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904469036262425e-05,
+      "loss": 0.8372,
+      "step": 1477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901330348090663e-05,
+      "loss": 0.8312,
+      "step": 1478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898189586657676e-05,
+      "loss": 0.86,
+      "step": 1479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895046752787575e-05,
+      "loss": 0.8433,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789190184730502e-05,
+      "loss": 0.7642,
+      "step": 1481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788875487103522e-05,
+      "loss": 0.8184,
+      "step": 1482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885605824803923e-05,
+      "loss": 0.8301,
+      "step": 1483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882454709437413e-05,
+      "loss": 0.8323,
+      "step": 1484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879301525762528e-05,
+      "loss": 0.7662,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876146274606647e-05,
+      "loss": 0.7522,
+      "step": 1486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872988956797684e-05,
+      "loss": 0.7972,
+      "step": 1487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78698295731641e-05,
+      "loss": 0.7825,
+      "step": 1488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866668124534903e-05,
+      "loss": 0.8208,
+      "step": 1489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863504611739633e-05,
+      "loss": 0.833,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860339035608376e-05,
+      "loss": 0.7791,
+      "step": 1491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857171396971765e-05,
+      "loss": 0.8195,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854001696660965e-05,
+      "loss": 0.7615,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785082993550769e-05,
+      "loss": 0.8484,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784765611434419e-05,
+      "loss": 0.8474,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844480234003254e-05,
+      "loss": 0.8069,
+      "step": 1496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784130229531821e-05,
+      "loss": 0.7993,
+      "step": 1497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783812229912294e-05,
+      "loss": 0.7295,
+      "step": 1498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834940246251848e-05,
+      "loss": 0.7683,
+      "step": 1499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783175613753989e-05,
+      "loss": 0.8207,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828569973822554e-05,
+      "loss": 0.8284,
+      "step": 1501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782538175593587e-05,
+      "loss": 0.7749,
+      "step": 1502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782219148471641e-05,
+      "loss": 0.7791,
+      "step": 1503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818999161001275e-05,
+      "loss": 0.7739,
+      "step": 1504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781580478562812e-05,
+      "loss": 0.7882,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812608359435122e-05,
+      "loss": 0.8032,
+      "step": 1506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809409883261015e-05,
+      "loss": 0.8389,
+      "step": 1507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806209357945048e-05,
+      "loss": 0.8588,
+      "step": 1508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803006784327022e-05,
+      "loss": 0.8425,
+      "step": 1509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799802163247278e-05,
+      "loss": 0.7493,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796595495546686e-05,
+      "loss": 0.7905,
+      "step": 1511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779338678206666e-05,
+      "loss": 0.7375,
+      "step": 1512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779017602364914e-05,
+      "loss": 0.8314,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778696322113661e-05,
+      "loss": 0.8025,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7783748375372092e-05,
+      "loss": 0.7957,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780531487199147e-05,
+      "loss": 0.8053,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777731255746186e-05,
+      "loss": 0.8491,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774091587004863e-05,
+      "loss": 0.776,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770868576673317e-05,
+      "loss": 0.7903,
+      "step": 1519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767643527312918e-05,
+      "loss": 0.7717,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764416439769907e-05,
+      "loss": 0.8181,
+      "step": 1521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761187314891047e-05,
+      "loss": 0.7106,
+      "step": 1522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775795615352364e-05,
+      "loss": 0.2512,
+      "step": 1523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754722956515525e-05,
+      "loss": 0.8821,
+      "step": 1524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775148772471507e-05,
+      "loss": 0.7983,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748250458971186e-05,
+      "loss": 0.8411,
+      "step": 1526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774501116013331e-05,
+      "loss": 0.8606,
+      "step": 1527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741769829051412e-05,
+      "loss": 0.8438,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738526466576e-05,
+      "loss": 0.8235,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735281073558114e-05,
+      "loss": 0.825,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732033650849318e-05,
+      "loss": 0.8677,
+      "step": 1531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728784199301725e-05,
+      "loss": 0.8469,
+      "step": 1532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725532719767967e-05,
+      "loss": 0.7264,
+      "step": 1533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772227921310121e-05,
+      "loss": 0.8125,
+      "step": 1534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771902368015516e-05,
+      "loss": 0.8523,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771576612178405e-05,
+      "loss": 0.8054,
+      "step": 1536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712506538842635e-05,
+      "loss": 0.8313,
+      "step": 1537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770924493218622e-05,
+      "loss": 0.7657,
+      "step": 1538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705981302670623e-05,
+      "loss": 0.8251,
+      "step": 1539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702715651152204e-05,
+      "loss": 0.7557,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769944797848785e-05,
+      "loss": 0.8105,
+      "step": 1541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769617828553498e-05,
+      "loss": 0.803,
+      "step": 1542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769290657315154e-05,
+      "loss": 0.8149,
+      "step": 1543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689632842196005e-05,
+      "loss": 0.7979,
+      "step": 1544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686357093527385e-05,
+      "loss": 0.798,
+      "step": 1545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768307932800522e-05,
+      "loss": 0.8779,
+      "step": 1546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679799546489568e-05,
+      "loss": 0.825,
+      "step": 1547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767651774984103e-05,
+      "loss": 0.7671,
+      "step": 1548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7673233938920727e-05,
+      "loss": 0.7922,
+      "step": 1549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766994811459031e-05,
+      "loss": 0.8479,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766666027771196e-05,
+      "loss": 0.8132,
+      "step": 1551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766337042914839e-05,
+      "loss": 0.887,
+      "step": 1552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766007856976283e-05,
+      "loss": 0.7854,
+      "step": 1553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765678470041905e-05,
+      "loss": 0.719,
+      "step": 1554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653488821981338e-05,
+      "loss": 0.7872,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650190935314507e-05,
+      "loss": 0.8005,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646891041283916e-05,
+      "loss": 0.8618,
+      "step": 1557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643589140755427e-05,
+      "loss": 0.7928,
+      "step": 1558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640285234595438e-05,
+      "loss": 0.7771,
+      "step": 1559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636979323670876e-05,
+      "loss": 0.7781,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633671408849198e-05,
+      "loss": 0.8826,
+      "step": 1561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630361490998372e-05,
+      "loss": 0.8376,
+      "step": 1562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762704957098691e-05,
+      "loss": 0.7963,
+      "step": 1563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623735649683826e-05,
+      "loss": 0.8047,
+      "step": 1564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620419727958685e-05,
+      "loss": 0.8157,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761710180668156e-05,
+      "loss": 0.8191,
+      "step": 1566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613781886723057e-05,
+      "loss": 0.8169,
+      "step": 1567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610459968954295e-05,
+      "loss": 0.7837,
+      "step": 1568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607136054246932e-05,
+      "loss": 0.7682,
+      "step": 1569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603810143473144e-05,
+      "loss": 0.797,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600482237505625e-05,
+      "loss": 0.7123,
+      "step": 1571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75971523372176e-05,
+      "loss": 0.8145,
+      "step": 1572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759382044348281e-05,
+      "loss": 0.2881,
+      "step": 1573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590486557175534e-05,
+      "loss": 0.8181,
+      "step": 1574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587150679170553e-05,
+      "loss": 0.8782,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758381281034319e-05,
+      "loss": 0.7892,
+      "step": 1576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580472951569273e-05,
+      "loss": 0.8,
+      "step": 1577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577131103725164e-05,
+      "loss": 0.8069,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573787267687747e-05,
+      "loss": 0.8647,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757044144433442e-05,
+      "loss": 0.8104,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567093634543112e-05,
+      "loss": 0.8003,
+      "step": 1581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756374383919226e-05,
+      "loss": 0.8202,
+      "step": 1582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756039205916084e-05,
+      "loss": 0.8348,
+      "step": 1583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755703829532833e-05,
+      "loss": 0.7949,
+      "step": 1584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755368254857474e-05,
+      "loss": 0.8264,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550324819780603e-05,
+      "loss": 0.2882,
+      "step": 1586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546965109826957e-05,
+      "loss": 0.7488,
+      "step": 1587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543603419595373e-05,
+      "loss": 0.833,
+      "step": 1588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540239749967947e-05,
+      "loss": 0.7826,
+      "step": 1589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536874101827272e-05,
+      "loss": 0.8213,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533506476056485e-05,
+      "loss": 0.8604,
+      "step": 1591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530136873539222e-05,
+      "loss": 0.8657,
+      "step": 1592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526765295159653e-05,
+      "loss": 0.7671,
+      "step": 1593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752339174180245e-05,
+      "loss": 0.8403,
+      "step": 1594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520016214352823e-05,
+      "loss": 0.7314,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751663871369649e-05,
+      "loss": 0.7653,
+      "step": 1596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751325924071968e-05,
+      "loss": 0.8418,
+      "step": 1597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750987779630915e-05,
+      "loss": 0.281,
+      "step": 1598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750649438135217e-05,
+      "loss": 0.7759,
+      "step": 1599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503108996736528e-05,
+      "loss": 0.7998,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499721643350528e-05,
+      "loss": 0.7847,
+      "step": 1601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749633232208299e-05,
+      "loss": 0.7839,
+      "step": 1602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492941033823254e-05,
+      "loss": 0.76,
+      "step": 1603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748954777946117e-05,
+      "loss": 0.7932,
+      "step": 1604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486152559887112e-05,
+      "loss": 0.7656,
+      "step": 1605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748275537599196e-05,
+      "loss": 0.8179,
+      "step": 1606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479356228667115e-05,
+      "loss": 0.7345,
+      "step": 1607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74759551188045e-05,
+      "loss": 0.8047,
+      "step": 1608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747255204729654e-05,
+      "loss": 0.7882,
+      "step": 1609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469147015036174e-05,
+      "loss": 0.8206,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465740022916873e-05,
+      "loss": 0.8245,
+      "step": 1611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462331071832604e-05,
+      "loss": 0.8025,
+      "step": 1612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745892016267786e-05,
+      "loss": 0.7299,
+      "step": 1613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745550729634764e-05,
+      "loss": 0.7405,
+      "step": 1614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745209247373746e-05,
+      "loss": 0.7976,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744867569574335e-05,
+      "loss": 0.8281,
+      "step": 1616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445256963261846e-05,
+      "loss": 0.8462,
+      "step": 1617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744183627719001e-05,
+      "loss": 0.781,
+      "step": 1618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743841363842541e-05,
+      "loss": 0.8369,
+      "step": 1619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434989047866118e-05,
+      "loss": 0.8032,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743156250641073e-05,
+      "loss": 0.8167,
+      "step": 1621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428134014958347e-05,
+      "loss": 0.7223,
+      "step": 1622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742470357440859e-05,
+      "loss": 0.7412,
+      "step": 1623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421271185661588e-05,
+      "loss": 0.832,
+      "step": 1624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741783684961797e-05,
+      "loss": 0.8185,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741440056717889e-05,
+      "loss": 0.8403,
+      "step": 1626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410962339246005e-05,
+      "loss": 0.8591,
+      "step": 1627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407522166721487e-05,
+      "loss": 0.8286,
+      "step": 1628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404080050508016e-05,
+      "loss": 0.8049,
+      "step": 1629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7400635991508782e-05,
+      "loss": 0.7828,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739718999062749e-05,
+      "loss": 0.7043,
+      "step": 1631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739374204876834e-05,
+      "loss": 0.8174,
+      "step": 1632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739029216683606e-05,
+      "loss": 0.801,
+      "step": 1633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386840345735878e-05,
+      "loss": 0.7881,
+      "step": 1634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7383386586373525e-05,
+      "loss": 0.8655,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379930889655247e-05,
+      "loss": 0.8203,
+      "step": 1636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376473256487805e-05,
+      "loss": 0.8284,
+      "step": 1637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373013687778454e-05,
+      "loss": 0.8114,
+      "step": 1638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369552184434962e-05,
+      "loss": 0.8014,
+      "step": 1639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366088747365616e-05,
+      "loss": 0.8469,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362623377479194e-05,
+      "loss": 0.7754,
+      "step": 1641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735915607568499e-05,
+      "loss": 0.8035,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73556868428928e-05,
+      "loss": 0.8,
+      "step": 1643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352215680012938e-05,
+      "loss": 0.801,
+      "step": 1644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348742587956208e-05,
+      "loss": 0.814,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734526756763393e-05,
+      "loss": 0.7996,
+      "step": 1646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734179061995793e-05,
+      "loss": 0.8381,
+      "step": 1647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338311745840536e-05,
+      "loss": 0.7947,
+      "step": 1648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334830946194584e-05,
+      "loss": 0.8665,
+      "step": 1649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331348221933416e-05,
+      "loss": 0.8174,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732786357397088e-05,
+      "loss": 0.7792,
+      "step": 1651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732437700322132e-05,
+      "loss": 0.8416,
+      "step": 1652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320888510599595e-05,
+      "loss": 0.7886,
+      "step": 1653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317398097021064e-05,
+      "loss": 0.8354,
+      "step": 1654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731390576340159e-05,
+      "loss": 0.7473,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310411510657542e-05,
+      "loss": 0.7515,
+      "step": 1656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730691533970579e-05,
+      "loss": 0.8718,
+      "step": 1657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303417251463707e-05,
+      "loss": 0.874,
+      "step": 1658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729991724684917e-05,
+      "loss": 0.8545,
+      "step": 1659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296415326780558e-05,
+      "loss": 0.7898,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292911492176756e-05,
+      "loss": 0.8169,
+      "step": 1661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728940574395715e-05,
+      "loss": 0.2984,
+      "step": 1662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285898083041624e-05,
+      "loss": 0.8499,
+      "step": 1663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7282388510350564e-05,
+      "loss": 0.7976,
+      "step": 1664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727887702680487e-05,
+      "loss": 0.8171,
+      "step": 1665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275363633325928e-05,
+      "loss": 0.8186,
+      "step": 1666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271848330835634e-05,
+      "loss": 0.8335,
+      "step": 1667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268331120256378e-05,
+      "loss": 0.7708,
+      "step": 1668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264812002511057e-05,
+      "loss": 0.2887,
+      "step": 1669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261290978523065e-05,
+      "loss": 0.8201,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257768049216302e-05,
+      "loss": 0.8252,
+      "step": 1671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254243215515156e-05,
+      "loss": 0.8113,
+      "step": 1672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250716478344528e-05,
+      "loss": 0.8267,
+      "step": 1673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247187838629805e-05,
+      "loss": 0.8971,
+      "step": 1674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243657297296894e-05,
+      "loss": 0.7581,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240124855272172e-05,
+      "loss": 0.8108,
+      "step": 1676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236590513482543e-05,
+      "loss": 0.7985,
+      "step": 1677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723305427285539e-05,
+      "loss": 0.8584,
+      "step": 1678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229516134318607e-05,
+      "loss": 0.8376,
+      "step": 1679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722597609880057e-05,
+      "loss": 0.8181,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722243416723017e-05,
+      "loss": 0.7507,
+      "step": 1681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721889034053679e-05,
+      "loss": 0.7861,
+      "step": 1682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215344619650308e-05,
+      "loss": 0.8118,
+      "step": 1683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7211797005501095e-05,
+      "loss": 0.7678,
+      "step": 1684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208247499020022e-05,
+      "loss": 0.7697,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720469610113847e-05,
+      "loss": 0.7843,
+      "step": 1686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201142812788295e-05,
+      "loss": 0.7446,
+      "step": 1687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197587634901854e-05,
+      "loss": 0.7644,
+      "step": 1688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194030568412014e-05,
+      "loss": 0.2905,
+      "step": 1689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190471614252123e-05,
+      "loss": 0.8586,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186910773356033e-05,
+      "loss": 0.2797,
+      "step": 1691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183348046658078e-05,
+      "loss": 0.8091,
+      "step": 1692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179783435093105e-05,
+      "loss": 0.7935,
+      "step": 1693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176216939596444e-05,
+      "loss": 0.8301,
+      "step": 1694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7172648561103918e-05,
+      "loss": 0.8119,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716907830055185e-05,
+      "loss": 0.8242,
+      "step": 1696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165506158877062e-05,
+      "loss": 0.7908,
+      "step": 1697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716193213701685e-05,
+      "loss": 0.8655,
+      "step": 1698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158356235909024e-05,
+      "loss": 0.8411,
+      "step": 1699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154778456491877e-05,
+      "loss": 0.8273,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151198799704196e-05,
+      "loss": 0.833,
+      "step": 1701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714761726648526e-05,
+      "loss": 0.8218,
+      "step": 1702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144033857774847e-05,
+      "loss": 0.8176,
+      "step": 1703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140448574513213e-05,
+      "loss": 0.829,
+      "step": 1704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136861417641123e-05,
+      "loss": 0.8037,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133272388099818e-05,
+      "loss": 0.8,
+      "step": 1706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129681486831042e-05,
+      "loss": 0.856,
+      "step": 1707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126088714777025e-05,
+      "loss": 0.8246,
+      "step": 1708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712249407288049e-05,
+      "loss": 0.8108,
+      "step": 1709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118897562084645e-05,
+      "loss": 0.7607,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71152991833332e-05,
+      "loss": 0.8789,
+      "step": 1711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111698937570343e-05,
+      "loss": 0.8066,
+      "step": 1712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108096825740754e-05,
+      "loss": 0.7905,
+      "step": 1713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710449284878961e-05,
+      "loss": 0.7925,
+      "step": 1714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710088700766257e-05,
+      "loss": 0.7352,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097279303305784e-05,
+      "loss": 0.2925,
+      "step": 1716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70936697366659e-05,
+      "loss": 0.8867,
+      "step": 1717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090058308690034e-05,
+      "loss": 0.7827,
+      "step": 1718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086445020325812e-05,
+      "loss": 0.7947,
+      "step": 1719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082829872521332e-05,
+      "loss": 0.76,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079212866225193e-05,
+      "loss": 0.7988,
+      "step": 1721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075594002386475e-05,
+      "loss": 0.7773,
+      "step": 1722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071973281954742e-05,
+      "loss": 0.8376,
+      "step": 1723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068350705880054e-05,
+      "loss": 0.8024,
+      "step": 1724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706472627511295e-05,
+      "loss": 0.8672,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061099990604458e-05,
+      "loss": 0.8381,
+      "step": 1726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057471853306095e-05,
+      "loss": 0.8033,
+      "step": 1727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705384186416986e-05,
+      "loss": 0.792,
+      "step": 1728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050210024148244e-05,
+      "loss": 0.749,
+      "step": 1729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046576334194215e-05,
+      "loss": 0.2806,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042940795261235e-05,
+      "loss": 0.2555,
+      "step": 1731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039303408303246e-05,
+      "loss": 0.803,
+      "step": 1732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7035664174274676e-05,
+      "loss": 0.7485,
+      "step": 1733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032023094130438e-05,
+      "loss": 0.8096,
+      "step": 1734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702838016882593e-05,
+      "loss": 0.8313,
+      "step": 1735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024735399317033e-05,
+      "loss": 0.8011,
+      "step": 1736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021088786560117e-05,
+      "loss": 0.8037,
+      "step": 1737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701744033151202e-05,
+      "loss": 0.8567,
+      "step": 1738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013790035130084e-05,
+      "loss": 0.785,
+      "step": 1739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010137898372122e-05,
+      "loss": 0.8115,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006483922196436e-05,
+      "loss": 0.2679,
+      "step": 1741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.70028281075618e-05,
+      "loss": 0.8113,
+      "step": 1742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699917045542748e-05,
+      "loss": 0.2806,
+      "step": 1743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995510966753225e-05,
+      "loss": 0.7448,
+      "step": 1744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991849642499264e-05,
+      "loss": 0.2895,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988186483626302e-05,
+      "loss": 0.7542,
+      "step": 1746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984521491095527e-05,
+      "loss": 0.7947,
+      "step": 1747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980854665868615e-05,
+      "loss": 0.8314,
+      "step": 1748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697718600890772e-05,
+      "loss": 0.7737,
+      "step": 1749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973515521175477e-05,
+      "loss": 0.7832,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696984320363499e-05,
+      "loss": 0.8386,
+      "step": 1751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966169057249865e-05,
+      "loss": 0.8474,
+      "step": 1752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962493082984167e-05,
+      "loss": 0.7734,
+      "step": 1753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958815281802453e-05,
+      "loss": 0.7646,
+      "step": 1754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955135654669758e-05,
+      "loss": 0.7876,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951454202551588e-05,
+      "loss": 0.7795,
+      "step": 1756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694777092641394e-05,
+      "loss": 0.2698,
+      "step": 1757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944085827223274e-05,
+      "loss": 0.799,
+      "step": 1758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694039890594655e-05,
+      "loss": 0.8235,
+      "step": 1759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693671016355119e-05,
+      "loss": 0.8699,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693301960100509e-05,
+      "loss": 0.7568,
+      "step": 1761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929327219276638e-05,
+      "loss": 0.7329,
+      "step": 1762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925633019334692e-05,
+      "loss": 0.7927,
+      "step": 1763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6921937002148586e-05,
+      "loss": 0.7732,
+      "step": 1764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918239168688133e-05,
+      "loss": 0.7837,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914539519923625e-05,
+      "loss": 0.8149,
+      "step": 1766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910838056825826e-05,
+      "loss": 0.8423,
+      "step": 1767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690713478036597e-05,
+      "loss": 0.7795,
+      "step": 1768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903429691515785e-05,
+      "loss": 0.7576,
+      "step": 1769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899722791247463e-05,
+      "loss": 0.7959,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896014080533664e-05,
+      "loss": 0.7383,
+      "step": 1771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892303560347537e-05,
+      "loss": 0.2855,
+      "step": 1772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68885912316627e-05,
+      "loss": 0.832,
+      "step": 1773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884877095453246e-05,
+      "loss": 0.8124,
+      "step": 1774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881161152693742e-05,
+      "loss": 0.8167,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877443404359224e-05,
+      "loss": 0.7668,
+      "step": 1776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687372385142521e-05,
+      "loss": 0.8157,
+      "step": 1777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687000249486769e-05,
+      "loss": 0.7881,
+      "step": 1778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866279335663125e-05,
+      "loss": 0.7905,
+      "step": 1779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862554374788445e-05,
+      "loss": 0.8359,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685882761322106e-05,
+      "loss": 0.8618,
+      "step": 1781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855099051938855e-05,
+      "loss": 0.8206,
+      "step": 1782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851368691920173e-05,
+      "loss": 0.8499,
+      "step": 1783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684763653414384e-05,
+      "loss": 0.8062,
+      "step": 1784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6843902579589157e-05,
+      "loss": 0.8044,
+      "step": 1785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840166829235886e-05,
+      "loss": 0.8606,
+      "step": 1786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836429284064265e-05,
+      "loss": 0.7639,
+      "step": 1787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832689945055007e-05,
+      "loss": 0.8708,
+      "step": 1788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682894881318929e-05,
+      "loss": 0.7876,
+      "step": 1789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6825205889448762e-05,
+      "loss": 0.7473,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821461174815547e-05,
+      "loss": 0.7527,
+      "step": 1791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817714670272236e-05,
+      "loss": 0.8096,
+      "step": 1792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813966376801884e-05,
+      "loss": 0.76,
+      "step": 1793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810216295388027e-05,
+      "loss": 0.7985,
+      "step": 1794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806464427014656e-05,
+      "loss": 0.8855,
+      "step": 1795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802710772666248e-05,
+      "loss": 0.8511,
+      "step": 1796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679895533332773e-05,
+      "loss": 0.8425,
+      "step": 1797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679519810998451e-05,
+      "loss": 0.8809,
+      "step": 1798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791439103622463e-05,
+      "loss": 0.8257,
+      "step": 1799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678767831522793e-05,
+      "loss": 0.7747,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783915745787717e-05,
+      "loss": 0.2887,
+      "step": 1801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6780151396289096e-05,
+      "loss": 0.8701,
+      "step": 1802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776385267719813e-05,
+      "loss": 0.8235,
+      "step": 1803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772617361068078e-05,
+      "loss": 0.7629,
+      "step": 1804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768847677322564e-05,
+      "loss": 0.8977,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765076217472417e-05,
+      "loss": 0.7927,
+      "step": 1806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761302982507245e-05,
+      "loss": 0.2759,
+      "step": 1807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675752797341712e-05,
+      "loss": 0.8499,
+      "step": 1808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675375119119258e-05,
+      "loss": 0.7957,
+      "step": 1809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749972636824632e-05,
+      "loss": 0.8472,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746192311304748e-05,
+      "loss": 0.8149,
+      "step": 1811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742410215624856e-05,
+      "loss": 0.7637,
+      "step": 1812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673862635077736e-05,
+      "loss": 0.2858,
+      "step": 1813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734840717755127e-05,
+      "loss": 0.7808,
+      "step": 1814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731053317551474e-05,
+      "loss": 0.7957,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67272641511602e-05,
+      "loss": 0.8179,
+      "step": 1816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723473219575554e-05,
+      "loss": 0.8062,
+      "step": 1817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719680523792256e-05,
+      "loss": 0.7971,
+      "step": 1818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671588606480549e-05,
+      "loss": 0.7531,
+      "step": 1819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6712089843610892e-05,
+      "loss": 0.7445,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708291861204577e-05,
+      "loss": 0.7451,
+      "step": 1821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67044921185831e-05,
+      "loss": 0.2611,
+      "step": 1822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700690616743497e-05,
+      "loss": 0.7964,
+      "step": 1823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696887356683264e-05,
+      "loss": 0.7529,
+      "step": 1824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693082339400344e-05,
+      "loss": 0.8748,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668927556589316e-05,
+      "loss": 0.7781,
+      "step": 1826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685467037160577e-05,
+      "loss": 0.79,
+      "step": 1827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681656754201934e-05,
+      "loss": 0.7531,
+      "step": 1828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677844718017027e-05,
+      "loss": 0.8394,
+      "step": 1829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674030929606112e-05,
+      "loss": 0.8169,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670215389969903e-05,
+      "loss": 0.7238,
+      "step": 1831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666398100109572e-05,
+      "loss": 0.8499,
+      "step": 1832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662579061026757e-05,
+      "loss": 0.8511,
+      "step": 1833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658758273723544e-05,
+      "loss": 0.8156,
+      "step": 1834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665493573920249e-05,
+      "loss": 0.8284,
+      "step": 1835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651111458466605e-05,
+      "loss": 0.8588,
+      "step": 1836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647285432519358e-05,
+      "loss": 0.8225,
+      "step": 1837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643457662364674e-05,
+      "loss": 0.8193,
+      "step": 1838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663962814900693e-05,
+      "loss": 0.8059,
+      "step": 1839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635796893450976e-05,
+      "loss": 0.8315,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631963896702107e-05,
+      "loss": 0.8262,
+      "step": 1841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628129159766077e-05,
+      "loss": 0.7793,
+      "step": 1842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624292683649105e-05,
+      "loss": 0.8357,
+      "step": 1843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620454469357845e-05,
+      "loss": 0.8469,
+      "step": 1844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616614517899435e-05,
+      "loss": 0.7915,
+      "step": 1845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612772830281446e-05,
+      "loss": 0.7745,
+      "step": 1846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608929407511917e-05,
+      "loss": 0.8026,
+      "step": 1847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605084250599337e-05,
+      "loss": 0.8018,
+      "step": 1848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6601237360552656e-05,
+      "loss": 0.8149,
+      "step": 1849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659738873838127e-05,
+      "loss": 0.765,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659353838509504e-05,
+      "loss": 0.845,
+      "step": 1851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658968630170427e-05,
+      "loss": 0.7888,
+      "step": 1852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585832489219726e-05,
+      "loss": 0.8169,
+      "step": 1853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581976948652624e-05,
+      "loss": 0.8005,
+      "step": 1854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657811968101463e-05,
+      "loss": 0.8699,
+      "step": 1855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574260687317877e-05,
+      "loss": 0.8687,
+      "step": 1856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6570399968574932e-05,
+      "loss": 0.766,
+      "step": 1857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566537525798835e-05,
+      "loss": 0.7827,
+      "step": 1858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656267336000306e-05,
+      "loss": 0.8047,
+      "step": 1859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6558807472201543e-05,
+      "loss": 0.8018,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554939863408662e-05,
+      "loss": 0.7471,
+      "step": 1861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551070534639267e-05,
+      "loss": 0.8298,
+      "step": 1862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547199486908643e-05,
+      "loss": 0.8192,
+      "step": 1863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654332672123252e-05,
+      "loss": 0.7729,
+      "step": 1864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65394522386271e-05,
+      "loss": 0.7939,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653557604010902e-05,
+      "loss": 0.8309,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531698126695367e-05,
+      "loss": 0.8315,
+      "step": 1867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527818499403684e-05,
+      "loss": 0.7699,
+      "step": 1868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6523937159251966e-05,
+      "loss": 0.8896,
+      "step": 1869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652005410725865e-05,
+      "loss": 0.8131,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516169344442622e-05,
+      "loss": 0.8564,
+      "step": 1871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512282871823225e-05,
+      "loss": 0.8455,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508394690420245e-05,
+      "loss": 0.7939,
+      "step": 1873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504504801253915e-05,
+      "loss": 0.8455,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500613205344918e-05,
+      "loss": 0.8008,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496719903714393e-05,
+      "loss": 0.7501,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492824897383907e-05,
+      "loss": 0.814,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488928187375496e-05,
+      "loss": 0.7731,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485029774711626e-05,
+      "loss": 0.7684,
+      "step": 1879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481129660415224e-05,
+      "loss": 0.2968,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477227845509646e-05,
+      "loss": 0.7266,
+      "step": 1881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6473324331018713e-05,
+      "loss": 0.7456,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469419117966686e-05,
+      "loss": 0.7664,
+      "step": 1883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646551220737826e-05,
+      "loss": 0.269,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461603600278592e-05,
+      "loss": 0.8501,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645769329769327e-05,
+      "loss": 0.7499,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645378130064834e-05,
+      "loss": 0.7919,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449867610170287e-05,
+      "loss": 0.8401,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445952227286035e-05,
+      "loss": 0.8538,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442035153022958e-05,
+      "loss": 0.7961,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438116388408874e-05,
+      "loss": 0.8022,
+      "step": 1891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434195934472042e-05,
+      "loss": 0.8118,
+      "step": 1892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643027379224116e-05,
+      "loss": 0.2668,
+      "step": 1893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642634996274539e-05,
+      "loss": 0.7668,
+      "step": 1894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422424447014305e-05,
+      "loss": 0.7522,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418497246077946e-05,
+      "loss": 0.7871,
+      "step": 1896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641456836096678e-05,
+      "loss": 0.8218,
+      "step": 1897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641063779271173e-05,
+      "loss": 0.7928,
+      "step": 1898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406705542344144e-05,
+      "loss": 0.873,
+      "step": 1899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6402771610895834e-05,
+      "loss": 0.2766,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398835999399025e-05,
+      "loss": 0.8042,
+      "step": 1901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394898708886406e-05,
+      "loss": 0.8564,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63909597403911e-05,
+      "loss": 0.7788,
+      "step": 1903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638701909494666e-05,
+      "loss": 0.8369,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383076773587097e-05,
+      "loss": 0.8447,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379132777346843e-05,
+      "loss": 0.7666,
+      "step": 1906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637518710726079e-05,
+      "loss": 0.8401,
+      "step": 1907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637123976436425e-05,
+      "loss": 0.7954,
+      "step": 1908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636729074969298e-05,
+      "loss": 0.8169,
+      "step": 1909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363340064283183e-05,
+      "loss": 0.8032,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635938770917149e-05,
+      "loss": 0.8921,
+      "step": 1911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635543368539498e-05,
+      "loss": 0.8606,
+      "step": 1912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351477993991163e-05,
+      "loss": 0.8193,
+      "step": 1913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634752063599799e-05,
+      "loss": 0.8035,
+      "step": 1914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343561612453844e-05,
+      "loss": 0.8119,
+      "step": 1915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633960092439755e-05,
+      "loss": 0.8325,
+      "step": 1916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335638572868373e-05,
+      "loss": 0.759,
+      "step": 1917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331674558906003e-05,
+      "loss": 0.8281,
+      "step": 1918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632770888355058e-05,
+      "loss": 0.8896,
+      "step": 1919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323741547842665e-05,
+      "loss": 0.7693,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319772552823272e-05,
+      "loss": 0.8464,
+      "step": 1921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315801899533835e-05,
+      "loss": 0.8132,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311829589016233e-05,
+      "loss": 0.7744,
+      "step": 1923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307855622312772e-05,
+      "loss": 0.7399,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63038800004662e-05,
+      "loss": 0.8405,
+      "step": 1925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.62999027245197e-05,
+      "loss": 0.8494,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295923795516878e-05,
+      "loss": 0.8396,
+      "step": 1927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629194321450178e-05,
+      "loss": 0.8621,
+      "step": 1928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287960982518894e-05,
+      "loss": 0.7639,
+      "step": 1929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283977100613127e-05,
+      "loss": 0.8452,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279991569829827e-05,
+      "loss": 0.8091,
+      "step": 1931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276004391214776e-05,
+      "loss": 0.7874,
+      "step": 1932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6272015565814184e-05,
+      "loss": 0.802,
+      "step": 1933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268025094674694e-05,
+      "loss": 0.8411,
+      "step": 1934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626403297884338e-05,
+      "loss": 0.8306,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626003921936775e-05,
+      "loss": 0.7716,
+      "step": 1936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256043817295747e-05,
+      "loss": 0.7195,
+      "step": 1937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252046773675732e-05,
+      "loss": 0.7683,
+      "step": 1938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624804808955651e-05,
+      "loss": 0.7997,
+      "step": 1939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244047765987314e-05,
+      "loss": 0.848,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62400458040178e-05,
+      "loss": 0.8206,
+      "step": 1941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236042204698058e-05,
+      "loss": 0.7931,
+      "step": 1942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232036969078613e-05,
+      "loss": 0.7101,
+      "step": 1943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622803009821041e-05,
+      "loss": 0.8064,
+      "step": 1944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622402159314483e-05,
+      "loss": 0.791,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622001145493368e-05,
+      "loss": 0.8057,
+      "step": 1946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215999684629195e-05,
+      "loss": 0.7478,
+      "step": 1947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211986283284042e-05,
+      "loss": 0.8167,
+      "step": 1948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207971251951307e-05,
+      "loss": 0.7914,
+      "step": 1949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203954591684518e-05,
+      "loss": 0.2838,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199936303537617e-05,
+      "loss": 0.7859,
+      "step": 1951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195916388564983e-05,
+      "loss": 0.8326,
+      "step": 1952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191894847821413e-05,
+      "loss": 0.8374,
+      "step": 1953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187871682362135e-05,
+      "loss": 0.8647,
+      "step": 1954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183846893242804e-05,
+      "loss": 0.7383,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179820481519503e-05,
+      "loss": 0.8096,
+      "step": 1956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175792448248735e-05,
+      "loss": 0.7563,
+      "step": 1957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617176279448743e-05,
+      "loss": 0.8459,
+      "step": 1958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167731521292952e-05,
+      "loss": 0.76,
+      "step": 1959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163698629723074e-05,
+      "loss": 0.765,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159664120836008e-05,
+      "loss": 0.7996,
+      "step": 1961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155627995690383e-05,
+      "loss": 0.7898,
+      "step": 1962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151590255345254e-05,
+      "loss": 0.824,
+      "step": 1963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6147550900860096e-05,
+      "loss": 0.7699,
+      "step": 1964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614350993329482e-05,
+      "loss": 0.8003,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613946735370974e-05,
+      "loss": 0.8103,
+      "step": 1966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135423163165614e-05,
+      "loss": 0.8159,
+      "step": 1967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131377362723608e-05,
+      "loss": 0.8063,
+      "step": 1968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6127329953445317e-05,
+      "loss": 0.7944,
+      "step": 1969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612328093639276e-05,
+      "loss": 0.825,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119230312628363e-05,
+      "loss": 0.9009,
+      "step": 1971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115178083214998e-05,
+      "loss": 0.8357,
+      "step": 1972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111124249215943e-05,
+      "loss": 0.8047,
+      "step": 1973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107068811694895e-05,
+      "loss": 0.2632,
+      "step": 1974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610301177171598e-05,
+      "loss": 0.7385,
+      "step": 1975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098953130343735e-05,
+      "loss": 0.8425,
+      "step": 1976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609489288864313e-05,
+      "loss": 0.8315,
+      "step": 1977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090831047679546e-05,
+      "loss": 0.8013,
+      "step": 1978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086767608518782e-05,
+      "loss": 0.7611,
+      "step": 1979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608270257222706e-05,
+      "loss": 0.8281,
+      "step": 1980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607863593987103e-05,
+      "loss": 0.8289,
+      "step": 1981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074567712517735e-05,
+      "loss": 0.8337,
+      "step": 1982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607049789123467e-05,
+      "loss": 0.827,
+      "step": 1983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066426477089726e-05,
+      "loss": 0.8403,
+      "step": 1984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606235347115121e-05,
+      "loss": 0.8342,
+      "step": 1985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605827887448787e-05,
+      "loss": 0.7635,
+      "step": 1986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054202688168837e-05,
+      "loss": 0.8533,
+      "step": 1987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050124913263686e-05,
+      "loss": 0.7781,
+      "step": 1988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60460455508424e-05,
+      "loss": 0.79,
+      "step": 1989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041964601975384e-05,
+      "loss": 0.7594,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037882067733445e-05,
+      "loss": 0.8362,
+      "step": 1991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033797949187825e-05,
+      "loss": 0.8076,
+      "step": 1992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602971224741016e-05,
+      "loss": 0.8245,
+      "step": 1993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602562496347253e-05,
+      "loss": 0.8169,
+      "step": 1994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021536098447398e-05,
+      "loss": 0.8708,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017445653407663e-05,
+      "loss": 0.7668,
+      "step": 1996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601335362942664e-05,
+      "loss": 0.8143,
+      "step": 1997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009260027578037e-05,
+      "loss": 0.7716,
+      "step": 1998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005164848936e-05,
+      "loss": 0.786,
+      "step": 1999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001068094575075e-05,
+      "loss": 0.7734,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996969765570233e-05,
+      "loss": 0.746,
+      "step": 2001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599286986299684e-05,
+      "loss": 0.2916,
+      "step": 2002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988768387930693e-05,
+      "loss": 0.8018,
+      "step": 2003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984665341447994e-05,
+      "loss": 0.8374,
+      "step": 2004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980560724625354e-05,
+      "loss": 0.8491,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59764545385398e-05,
+      "loss": 0.7524,
+      "step": 2006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972346784268774e-05,
+      "loss": 0.8303,
+      "step": 2007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596823746289012e-05,
+      "loss": 0.7954,
+      "step": 2008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5964126575482105e-05,
+      "loss": 0.8123,
+      "step": 2009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960014123123403e-05,
+      "loss": 0.7535,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955900106893085e-05,
+      "loss": 0.8394,
+      "step": 2011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951784527870656e-05,
+      "loss": 0.8281,
+      "step": 2012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947667387136008e-05,
+      "loss": 0.8037,
+      "step": 2013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943548685769467e-05,
+      "loss": 0.7004,
+      "step": 2014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939428424851745e-05,
+      "loss": 0.7889,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935306605463975e-05,
+      "loss": 0.8173,
+      "step": 2016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931183228687702e-05,
+      "loss": 0.7695,
+      "step": 2017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927058295604867e-05,
+      "loss": 0.8152,
+      "step": 2018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922931807297833e-05,
+      "loss": 0.7877,
+      "step": 2019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591880376484937e-05,
+      "loss": 0.8154,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914674169342646e-05,
+      "loss": 0.8208,
+      "step": 2021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591054302186124e-05,
+      "loss": 0.2886,
+      "step": 2022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906410323489145e-05,
+      "loss": 0.8313,
+      "step": 2023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5902276075310753e-05,
+      "loss": 0.7711,
+      "step": 2024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898140278410865e-05,
+      "loss": 0.772,
+      "step": 2025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589400293387469e-05,
+      "loss": 0.8049,
+      "step": 2026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889864042787842e-05,
+      "loss": 0.7847,
+      "step": 2027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885723606236343e-05,
+      "loss": 0.7754,
+      "step": 2028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881581625306615e-05,
+      "loss": 0.7618,
+      "step": 2029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587743810108549e-05,
+      "loss": 0.8545,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873293034660205e-05,
+      "loss": 0.8521,
+      "step": 2031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58691464271184e-05,
+      "loss": 0.7839,
+      "step": 2032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586499827954812e-05,
+      "loss": 0.7993,
+      "step": 2033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860848593037813e-05,
+      "loss": 0.2821,
+      "step": 2034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585669736867633e-05,
+      "loss": 0.7661,
+      "step": 2035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852544607552926e-05,
+      "loss": 0.8621,
+      "step": 2036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848390310757267e-05,
+      "loss": 0.8481,
+      "step": 2037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584423447937941e-05,
+      "loss": 0.7404,
+      "step": 2038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5840077114509823e-05,
+      "loss": 0.823,
+      "step": 2039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583591821723937e-05,
+      "loss": 0.873,
+      "step": 2040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5831757788659325e-05,
+      "loss": 0.781,
+      "step": 2041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827595829861354e-05,
+      "loss": 0.7665,
+      "step": 2042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823432341937534e-05,
+      "loss": 0.781,
+      "step": 2043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819267325980338e-05,
+      "loss": 0.8398,
+      "step": 2044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815100783082634e-05,
+      "loss": 0.7875,
+      "step": 2045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810932714337707e-05,
+      "loss": 0.7961,
+      "step": 2046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806763120839233e-05,
+      "loss": 0.7991,
+      "step": 2047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580259200368128e-05,
+      "loss": 0.7638,
+      "step": 2048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798419363958334e-05,
+      "loss": 0.8203,
+      "step": 2049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5794245202765263e-05,
+      "loss": 0.8198,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790069521197345e-05,
+      "loss": 0.8021,
+      "step": 2051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785892320350248e-05,
+      "loss": 0.834,
+      "step": 2052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781713601320054e-05,
+      "loss": 0.7671,
+      "step": 2053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777533365203225e-05,
+      "loss": 0.7896,
+      "step": 2054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577335161309663e-05,
+      "loss": 0.7886,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576916834609754e-05,
+      "loss": 0.8071,
+      "step": 2056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764983565303622e-05,
+      "loss": 0.7686,
+      "step": 2057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760797271812925e-05,
+      "loss": 0.855,
+      "step": 2058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575660946672392e-05,
+      "loss": 0.8228,
+      "step": 2059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752420151135455e-05,
+      "loss": 0.7839,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5748229326146782e-05,
+      "loss": 0.7925,
+      "step": 2061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744036992857545e-05,
+      "loss": 0.8635,
+      "step": 2062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5739843152367792e-05,
+      "loss": 0.7368,
+      "step": 2063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573564780577796e-05,
+      "loss": 0.7737,
+      "step": 2064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731450954188883e-05,
+      "loss": 0.8438,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727252598701786e-05,
+      "loss": 0.801,
+      "step": 2066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5723052740418293e-05,
+      "loss": 0.7815,
+      "step": 2067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571885138044043e-05,
+      "loss": 0.8007,
+      "step": 2068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714648519870593e-05,
+      "loss": 0.7908,
+      "step": 2069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57104441598116e-05,
+      "loss": 0.7539,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570623830136664e-05,
+      "loss": 0.269,
+      "step": 2071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702030945639314e-05,
+      "loss": 0.7966,
+      "step": 2072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697822093733602e-05,
+      "loss": 0.7556,
+      "step": 2073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693611746753877e-05,
+      "loss": 0.811,
+      "step": 2074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689399905804918e-05,
+      "loss": 0.7767,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568518657199188e-05,
+      "loss": 0.2728,
+      "step": 2076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5680971746420322e-05,
+      "loss": 0.7903,
+      "step": 2077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567675543019618e-05,
+      "loss": 0.8398,
+      "step": 2078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672537624425798e-05,
+      "loss": 0.7646,
+      "step": 2079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668318330215896e-05,
+      "loss": 0.7913,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664097548673594e-05,
+      "loss": 0.8105,
+      "step": 2081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659875280906406e-05,
+      "loss": 0.7745,
+      "step": 2082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565565152802222e-05,
+      "loss": 0.807,
+      "step": 2083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5651426291129322e-05,
+      "loss": 0.8652,
+      "step": 2084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647199571336397e-05,
+      "loss": 0.811,
+      "step": 2085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642971369752507e-05,
+      "loss": 0.759,
+      "step": 2086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638741687487107e-05,
+      "loss": 0.7483,
+      "step": 2087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563451052565004e-05,
+      "loss": 0.8544,
+      "step": 2088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563027788535153e-05,
+      "loss": 0.8002,
+      "step": 2089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626043767702213e-05,
+      "loss": 0.8054,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562180817381308e-05,
+      "loss": 0.7633,
+      "step": 2091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561757110479553e-05,
+      "loss": 0.8176,
+      "step": 2092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613332561761347e-05,
+      "loss": 0.8313,
+      "step": 2093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609092545822696e-05,
+      "loss": 0.7697,
+      "step": 2094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604851058092135e-05,
+      "loss": 0.7869,
+      "step": 2095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56006080996826e-05,
+      "loss": 0.7856,
+      "step": 2096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559636367170742e-05,
+      "loss": 0.8131,
+      "step": 2097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592117775280308e-05,
+      "loss": 0.8738,
+      "step": 2098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587870411515363e-05,
+      "loss": 0.776,
+      "step": 2099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558362158152706e-05,
+      "loss": 0.7195,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579371286430272e-05,
+      "loss": 0.8108,
+      "step": 2101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575119527340254e-05,
+      "loss": 0.7581,
+      "step": 2102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5570866305372637e-05,
+      "loss": 0.7998,
+      "step": 2103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556661162164344e-05,
+      "loss": 0.7991,
+      "step": 2104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5562355477269068e-05,
+      "loss": 0.8625,
+      "step": 2105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558097873366308e-05,
+      "loss": 0.7759,
+      "step": 2106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553838811052323e-05,
+      "loss": 0.7742,
+      "step": 2107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549578291444678e-05,
+      "loss": 0.2682,
+      "step": 2108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55453163156613e-05,
+      "loss": 0.8386,
+      "step": 2109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5541052884820505e-05,
+      "loss": 0.7147,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536788000040994e-05,
+      "loss": 0.7983,
+      "step": 2111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553252166244184e-05,
+      "loss": 0.8728,
+      "step": 2112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528253873142514e-05,
+      "loss": 0.8337,
+      "step": 2113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523984633262853e-05,
+      "loss": 0.7627,
+      "step": 2114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519713943923083e-05,
+      "loss": 0.7408,
+      "step": 2115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55154418062438e-05,
+      "loss": 0.7461,
+      "step": 2116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511168221345993e-05,
+      "loss": 0.7827,
+      "step": 2117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5506893190351024e-05,
+      "loss": 0.7773,
+      "step": 2118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502616714380635e-05,
+      "loss": 0.835,
+      "step": 2119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498338794556946e-05,
+      "loss": 0.8137,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549405943200246e-05,
+      "loss": 0.7686,
+      "step": 2121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548977862784005e-05,
+      "loss": 0.7622,
+      "step": 2122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548549638319298e-05,
+      "loss": 0.7935,
+      "step": 2123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548121269918488e-05,
+      "loss": 0.8093,
+      "step": 2124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5476927576939766e-05,
+      "loss": 0.7793,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472641017582026e-05,
+      "loss": 0.8037,
+      "step": 2126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546835302223643e-05,
+      "loss": 0.7537,
+      "step": 2127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5464063592028117e-05,
+      "loss": 0.7898,
+      "step": 2128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459772728082617e-05,
+      "loss": 0.8044,
+      "step": 2129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455480431525817e-05,
+      "loss": 0.7196,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545118670348399e-05,
+      "loss": 0.7977,
+      "step": 2131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446891545083792e-05,
+      "loss": 0.8164,
+      "step": 2132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544259495745224e-05,
+      "loss": 0.8076,
+      "step": 2133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438296941716733e-05,
+      "loss": 0.8306,
+      "step": 2134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433997499005046e-05,
+      "loss": 0.8323,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429696630445326e-05,
+      "loss": 0.8064,
+      "step": 2136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54253943371661e-05,
+      "loss": 0.7444,
+      "step": 2137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421090620296254e-05,
+      "loss": 0.7806,
+      "step": 2138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541678548096506e-05,
+      "loss": 0.7817,
+      "step": 2139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541247892030217e-05,
+      "loss": 0.8107,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408170939437588e-05,
+      "loss": 0.7589,
+      "step": 2141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403861539501708e-05,
+      "loss": 0.8599,
+      "step": 2142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399550721625285e-05,
+      "loss": 0.8198,
+      "step": 2143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539523848693946e-05,
+      "loss": 0.761,
+      "step": 2144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539092483657573e-05,
+      "loss": 0.7988,
+      "step": 2145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386609771665972e-05,
+      "loss": 0.801,
+      "step": 2146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382293293342435e-05,
+      "loss": 0.7701,
+      "step": 2147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377975402737738e-05,
+      "loss": 0.7651,
+      "step": 2148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373656100984865e-05,
+      "loss": 0.7378,
+      "step": 2149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369335389217173e-05,
+      "loss": 0.7889,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365013268568403e-05,
+      "loss": 0.7559,
+      "step": 2151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360689740172636e-05,
+      "loss": 0.7761,
+      "step": 2152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5356364805164355e-05,
+      "loss": 0.7917,
+      "step": 2153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352038464678387e-05,
+      "loss": 0.8048,
+      "step": 2154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347710719849943e-05,
+      "loss": 0.8162,
+      "step": 2155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534338157181459e-05,
+      "loss": 0.7596,
+      "step": 2156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339051021708278e-05,
+      "loss": 0.8306,
+      "step": 2157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5334719070667314e-05,
+      "loss": 0.7588,
+      "step": 2158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330385719828376e-05,
+      "loss": 0.7693,
+      "step": 2159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532605097032851e-05,
+      "loss": 0.8625,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321714823305123e-05,
+      "loss": 0.8379,
+      "step": 2161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317377279896e-05,
+      "loss": 0.7263,
+      "step": 2162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313038341239282e-05,
+      "loss": 0.8848,
+      "step": 2163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308698008473483e-05,
+      "loss": 0.8115,
+      "step": 2164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530435628273748e-05,
+      "loss": 0.8018,
+      "step": 2165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530001316517051e-05,
+      "loss": 0.7891,
+      "step": 2166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529566865691219e-05,
+      "loss": 0.7603,
+      "step": 2167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291322759102485e-05,
+      "loss": 0.8608,
+      "step": 2168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286975472881736e-05,
+      "loss": 0.8252,
+      "step": 2169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282626799390647e-05,
+      "loss": 0.7708,
+      "step": 2170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278276739770275e-05,
+      "loss": 0.8606,
+      "step": 2171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273925295162056e-05,
+      "loss": 0.8101,
+      "step": 2172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269572466707782e-05,
+      "loss": 0.757,
+      "step": 2173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265218255549606e-05,
+      "loss": 0.7562,
+      "step": 2174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260862662830055e-05,
+      "loss": 0.8533,
+      "step": 2175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256505689691996e-05,
+      "loss": 0.7944,
+      "step": 2176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252147337278682e-05,
+      "loss": 0.8293,
+      "step": 2177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247787606733718e-05,
+      "loss": 0.7964,
+      "step": 2178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243426499201068e-05,
+      "loss": 0.7891,
+      "step": 2179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239064015825065e-05,
+      "loss": 0.7505,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234700157750391e-05,
+      "loss": 0.7756,
+      "step": 2181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230334926122101e-05,
+      "loss": 0.746,
+      "step": 2182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225968322085605e-05,
+      "loss": 0.7449,
+      "step": 2183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221600346786671e-05,
+      "loss": 0.7937,
+      "step": 2184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217231001371431e-05,
+      "loss": 0.8416,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212860286986377e-05,
+      "loss": 0.7645,
+      "step": 2186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520848820477835e-05,
+      "loss": 0.8821,
+      "step": 2187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520411475589457e-05,
+      "loss": 0.8423,
+      "step": 2188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199739941482594e-05,
+      "loss": 0.7729,
+      "step": 2189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195363762690349e-05,
+      "loss": 0.7861,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190986220666122e-05,
+      "loss": 0.7772,
+      "step": 2191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186607316558546e-05,
+      "loss": 0.7759,
+      "step": 2192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182227051516622e-05,
+      "loss": 0.7728,
+      "step": 2193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517784542668971e-05,
+      "loss": 0.8119,
+      "step": 2194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173462443227515e-05,
+      "loss": 0.7908,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169078102280108e-05,
+      "loss": 0.8442,
+      "step": 2196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164692404997916e-05,
+      "loss": 0.8127,
+      "step": 2197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5160305352531714e-05,
+      "loss": 0.8452,
+      "step": 2198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155916946032646e-05,
+      "loss": 0.7698,
+      "step": 2199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151527186652193e-05,
+      "loss": 0.726,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147136075542208e-05,
+      "loss": 0.7629,
+      "step": 2201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142743613854891e-05,
+      "loss": 0.7883,
+      "step": 2202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138349802742798e-05,
+      "loss": 0.8049,
+      "step": 2203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133954643358837e-05,
+      "loss": 0.7917,
+      "step": 2204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512955813685627e-05,
+      "loss": 0.76,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125160284388716e-05,
+      "loss": 0.7771,
+      "step": 2206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120761087110145e-05,
+      "loss": 0.7942,
+      "step": 2207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116360546174878e-05,
+      "loss": 0.781,
+      "step": 2208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511195866273759e-05,
+      "loss": 0.7461,
+      "step": 2209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107555437953311e-05,
+      "loss": 0.8149,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103150872977419e-05,
+      "loss": 0.7629,
+      "step": 2211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098744968965644e-05,
+      "loss": 0.8177,
+      "step": 2212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094337727074073e-05,
+      "loss": 0.7703,
+      "step": 2213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508992914845913e-05,
+      "loss": 0.7697,
+      "step": 2214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085519234277612e-05,
+      "loss": 0.804,
+      "step": 2215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5081107985686645e-05,
+      "loss": 0.7703,
+      "step": 2216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076695403843717e-05,
+      "loss": 0.7991,
+      "step": 2217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072281489906665e-05,
+      "loss": 0.8118,
+      "step": 2218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067866245033664e-05,
+      "loss": 0.8245,
+      "step": 2219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063449670383258e-05,
+      "loss": 0.783,
+      "step": 2220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059031767114327e-05,
+      "loss": 0.8096,
+      "step": 2221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054612536386097e-05,
+      "loss": 0.8164,
+      "step": 2222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050191979358153e-05,
+      "loss": 0.8359,
+      "step": 2223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504577009719042e-05,
+      "loss": 0.7803,
+      "step": 2224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041346891043174e-05,
+      "loss": 0.7816,
+      "step": 2225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036922362077036e-05,
+      "loss": 0.8562,
+      "step": 2226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032496511452976e-05,
+      "loss": 0.7725,
+      "step": 2227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502806934033231e-05,
+      "loss": 0.7278,
+      "step": 2228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023640849876704e-05,
+      "loss": 0.801,
+      "step": 2229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019211041248164e-05,
+      "loss": 0.7812,
+      "step": 2230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014779915609043e-05,
+      "loss": 0.2822,
+      "step": 2231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010347474122045e-05,
+      "loss": 0.8096,
+      "step": 2232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005913717950215e-05,
+      "loss": 0.7881,
+      "step": 2233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001478648256949e-05,
+      "loss": 0.7444,
+      "step": 2234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997042266205974e-05,
+      "loss": 0.8107,
+      "step": 2235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992604572961372e-05,
+      "loss": 0.814,
+      "step": 2236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498816556968757e-05,
+      "loss": 0.7893,
+      "step": 2237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983725257549331e-05,
+      "loss": 0.8215,
+      "step": 2238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497928363771177e-05,
+      "loss": 0.8093,
+      "step": 2239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497484071134034e-05,
+      "loss": 0.8152,
+      "step": 2240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970396479600836e-05,
+      "loss": 0.745,
+      "step": 2241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49659509436594e-05,
+      "loss": 0.745,
+      "step": 2242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961504104682513e-05,
+      "loss": 0.7495,
+      "step": 2243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957055963837e-05,
+      "loss": 0.8032,
+      "step": 2244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952606522290024e-05,
+      "loss": 0.7993,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494815578120909e-05,
+      "loss": 0.289,
+      "step": 2246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943703741762048e-05,
+      "loss": 0.8262,
+      "step": 2247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939250405117088e-05,
+      "loss": 0.7854,
+      "step": 2248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934795772442734e-05,
+      "loss": 0.7679,
+      "step": 2249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930339844907857e-05,
+      "loss": 0.7285,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925882623681667e-05,
+      "loss": 0.291,
+      "step": 2251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921424109933709e-05,
+      "loss": 0.8059,
+      "step": 2252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916964304833872e-05,
+      "loss": 0.834,
+      "step": 2253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491250320955238e-05,
+      "loss": 0.8158,
+      "step": 2254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908040825259799e-05,
+      "loss": 0.8521,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903577153127034e-05,
+      "loss": 0.8193,
+      "step": 2256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489911219432532e-05,
+      "loss": 0.7454,
+      "step": 2257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894645950026236e-05,
+      "loss": 0.8047,
+      "step": 2258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890178421401703e-05,
+      "loss": 0.748,
+      "step": 2259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885709609623968e-05,
+      "loss": 0.8418,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881239515865624e-05,
+      "loss": 0.7634,
+      "step": 2261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487676814129959e-05,
+      "loss": 0.8022,
+      "step": 2262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872295487099137e-05,
+      "loss": 0.6886,
+      "step": 2263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867821554437856e-05,
+      "loss": 0.771,
+      "step": 2264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486334634448968e-05,
+      "loss": 0.7988,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858869858428877e-05,
+      "loss": 0.7615,
+      "step": 2266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854392097430052e-05,
+      "loss": 0.8198,
+      "step": 2267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849913062668141e-05,
+      "loss": 0.838,
+      "step": 2268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845432755318418e-05,
+      "loss": 0.8318,
+      "step": 2269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840951176556484e-05,
+      "loss": 0.8062,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836468327558283e-05,
+      "loss": 0.7998,
+      "step": 2271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831984209500083e-05,
+      "loss": 0.8489,
+      "step": 2272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827498823558492e-05,
+      "loss": 0.7983,
+      "step": 2273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823012170910449e-05,
+      "loss": 0.7887,
+      "step": 2274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818524252733221e-05,
+      "loss": 0.7853,
+      "step": 2275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814035070204413e-05,
+      "loss": 0.791,
+      "step": 2276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809544624501962e-05,
+      "loss": 0.774,
+      "step": 2277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805052916804127e-05,
+      "loss": 0.8074,
+      "step": 2278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480055994828951e-05,
+      "loss": 0.796,
+      "step": 2279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796065720137039e-05,
+      "loss": 0.7167,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791570233525968e-05,
+      "loss": 0.7903,
+      "step": 2281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787073489635886e-05,
+      "loss": 0.8108,
+      "step": 2282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782575489646717e-05,
+      "loss": 0.7971,
+      "step": 2283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778076234738706e-05,
+      "loss": 0.8127,
+      "step": 2284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773575726092426e-05,
+      "loss": 0.8132,
+      "step": 2285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769073964888786e-05,
+      "loss": 0.3132,
+      "step": 2286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764570952309024e-05,
+      "loss": 0.7891,
+      "step": 2287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760066689534704e-05,
+      "loss": 0.804,
+      "step": 2288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475556117774771e-05,
+      "loss": 0.812,
+      "step": 2289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751054418130267e-05,
+      "loss": 0.7025,
+      "step": 2290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474654641186492e-05,
+      "loss": 0.7869,
+      "step": 2291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474203716013454e-05,
+      "loss": 0.8208,
+      "step": 2292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737526664122329e-05,
+      "loss": 0.8098,
+      "step": 2293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733014925011818e-05,
+      "loss": 0.7515,
+      "step": 2294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728501943986853e-05,
+      "loss": 0.8418,
+      "step": 2295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723987722231619e-05,
+      "loss": 0.8198,
+      "step": 2296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719472260930614e-05,
+      "loss": 0.7844,
+      "step": 2297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714955561268674e-05,
+      "loss": 0.7998,
+      "step": 2298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471043762443095e-05,
+      "loss": 0.8508,
+      "step": 2299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705918451602923e-05,
+      "loss": 0.8145,
+      "step": 2300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701398043970394e-05,
+      "loss": 0.8269,
+      "step": 2301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696876402719495e-05,
+      "loss": 0.7217,
+      "step": 2302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692353529036672e-05,
+      "loss": 0.7998,
+      "step": 2303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687829424108703e-05,
+      "loss": 0.7145,
+      "step": 2304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683304089122686e-05,
+      "loss": 0.7898,
+      "step": 2305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467877752526604e-05,
+      "loss": 0.8289,
+      "step": 2306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467424973372651e-05,
+      "loss": 0.8425,
+      "step": 2307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4669720715692157e-05,
+      "loss": 0.7925,
+      "step": 2308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665190472351371e-05,
+      "loss": 0.7959,
+      "step": 2309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660659004892863e-05,
+      "loss": 0.8118,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656126314505659e-05,
+      "loss": 0.7935,
+      "step": 2311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651592402379109e-05,
+      "loss": 0.7886,
+      "step": 2312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647057269702885e-05,
+      "loss": 0.7771,
+      "step": 2313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642520917666976e-05,
+      "loss": 0.8108,
+      "step": 2314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46379833474617e-05,
+      "loss": 0.7896,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633444560277678e-05,
+      "loss": 0.7798,
+      "step": 2316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628904557305869e-05,
+      "loss": 0.8083,
+      "step": 2317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624363339737541e-05,
+      "loss": 0.7423,
+      "step": 2318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461982090876428e-05,
+      "loss": 0.7163,
+      "step": 2319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615277265577991e-05,
+      "loss": 0.7595,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46107324113709e-05,
+      "loss": 0.7301,
+      "step": 2321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460618634733555e-05,
+      "loss": 0.7719,
+      "step": 2322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46016390746648e-05,
+      "loss": 0.8333,
+      "step": 2323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597090594551827e-05,
+      "loss": 0.7616,
+      "step": 2324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4592540908190123e-05,
+      "loss": 0.7981,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587990016773501e-05,
+      "loss": 0.8147,
+      "step": 2326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583437921496084e-05,
+      "loss": 0.7976,
+      "step": 2327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578884623552318e-05,
+      "loss": 0.7903,
+      "step": 2328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457433012413696e-05,
+      "loss": 0.8306,
+      "step": 2329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456977442444508e-05,
+      "loss": 0.833,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565217525672068e-05,
+      "loss": 0.7999,
+      "step": 2331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560659429013627e-05,
+      "loss": 0.7756,
+      "step": 2332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556100135665773e-05,
+      "loss": 0.8359,
+      "step": 2333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455153964682484e-05,
+      "loss": 0.2994,
+      "step": 2334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546977963687467e-05,
+      "loss": 0.7006,
+      "step": 2335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542415087450617e-05,
+      "loss": 0.8091,
+      "step": 2336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453785101931156e-05,
+      "loss": 0.7396,
+      "step": 2337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533285760467877e-05,
+      "loss": 0.8671,
+      "step": 2338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528719312117467e-05,
+      "loss": 0.2639,
+      "step": 2339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524151675458538e-05,
+      "loss": 0.7778,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519582851689609e-05,
+      "loss": 0.759,
+      "step": 2341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515012842009515e-05,
+      "loss": 0.6786,
+      "step": 2342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510441647617395e-05,
+      "loss": 0.8093,
+      "step": 2343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505869269712705e-05,
+      "loss": 0.7349,
+      "step": 2344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450129570949521e-05,
+      "loss": 0.8418,
+      "step": 2345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496720968164979e-05,
+      "loss": 0.8699,
+      "step": 2346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492145046922406e-05,
+      "loss": 0.7656,
+      "step": 2347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487567946968174e-05,
+      "loss": 0.7942,
+      "step": 2348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482989669503293e-05,
+      "loss": 0.7646,
+      "step": 2349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478410215729077e-05,
+      "loss": 0.7754,
+      "step": 2350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473829586847141e-05,
+      "loss": 0.8198,
+      "step": 2351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446924778405942e-05,
+      "loss": 0.7103,
+      "step": 2352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464664808568148e-05,
+      "loss": 0.7961,
+      "step": 2353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460080661575869e-05,
+      "loss": 0.8198,
+      "step": 2354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455495344285437e-05,
+      "loss": 0.7651,
+      "step": 2355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450908857900012e-05,
+      "loss": 0.808,
+      "step": 2356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444632120362306e-05,
+      "loss": 0.8018,
+      "step": 2357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441732382658354e-05,
+      "loss": 0.77,
+      "step": 2358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437142396209969e-05,
+      "loss": 0.822,
+      "step": 2359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432551245482295e-05,
+      "loss": 0.8821,
+      "step": 2360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427958931680021e-05,
+      "loss": 0.7556,
+      "step": 2361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423365456008143e-05,
+      "loss": 0.8269,
+      "step": 2362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418770819671956e-05,
+      "loss": 0.769,
+      "step": 2363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414175023877071e-05,
+      "loss": 0.7528,
+      "step": 2364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409578069829393e-05,
+      "loss": 0.8379,
+      "step": 2365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404979958735137e-05,
+      "loss": 0.7698,
+      "step": 2366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400380691800818e-05,
+      "loss": 0.7988,
+      "step": 2367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395780270233255e-05,
+      "loss": 0.8174,
+      "step": 2368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4391178695239576e-05,
+      "loss": 0.7695,
+      "step": 2369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.43865759680272e-05,
+      "loss": 0.8159,
+      "step": 2370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381972089803857e-05,
+      "loss": 0.7806,
+      "step": 2371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437736706177758e-05,
+      "loss": 0.7582,
+      "step": 2372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372760885156694e-05,
+      "loss": 0.8491,
+      "step": 2373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368153561149836e-05,
+      "loss": 0.8136,
+      "step": 2374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436354509096594e-05,
+      "loss": 0.8826,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358935475814239e-05,
+      "loss": 0.8206,
+      "step": 2376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354324716904273e-05,
+      "loss": 0.8506,
+      "step": 2377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349712815445871e-05,
+      "loss": 0.8391,
+      "step": 2378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434509977264917e-05,
+      "loss": 0.7726,
+      "step": 2379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340485589724606e-05,
+      "loss": 0.8281,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433587026788291e-05,
+      "loss": 0.7803,
+      "step": 2381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433125380833512e-05,
+      "loss": 0.7683,
+      "step": 2382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326636212292557e-05,
+      "loss": 0.7908,
+      "step": 2383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322017480966859e-05,
+      "loss": 0.873,
+      "step": 2384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317397615569952e-05,
+      "loss": 0.7781,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312776617314057e-05,
+      "loss": 0.2964,
+      "step": 2386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308154487411698e-05,
+      "loss": 0.7747,
+      "step": 2387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303531227075697e-05,
+      "loss": 0.7693,
+      "step": 2388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298906837519164e-05,
+      "loss": 0.2816,
+      "step": 2389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294281319955513e-05,
+      "loss": 0.2776,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4289654675598455e-05,
+      "loss": 0.8315,
+      "step": 2391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285026905661985e-05,
+      "loss": 0.7788,
+      "step": 2392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428039801136041e-05,
+      "loss": 0.8158,
+      "step": 2393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275767993908321e-05,
+      "loss": 0.8044,
+      "step": 2394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271136854520608e-05,
+      "loss": 0.8279,
+      "step": 2395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266504594412451e-05,
+      "loss": 0.8105,
+      "step": 2396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426187121479933e-05,
+      "loss": 0.7814,
+      "step": 2397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257236716897016e-05,
+      "loss": 0.7812,
+      "step": 2398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252601101921568e-05,
+      "loss": 0.821,
+      "step": 2399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247964371089349e-05,
+      "loss": 0.8413,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243326525617008e-05,
+      "loss": 0.8147,
+      "step": 2401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238687566721484e-05,
+      "loss": 0.3008,
+      "step": 2402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234047495620014e-05,
+      "loss": 0.7664,
+      "step": 2403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229406313530127e-05,
+      "loss": 0.7178,
+      "step": 2404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224764021669637e-05,
+      "loss": 0.8384,
+      "step": 2405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220120621256655e-05,
+      "loss": 0.7914,
+      "step": 2406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215476113509583e-05,
+      "loss": 0.7478,
+      "step": 2407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210830499647109e-05,
+      "loss": 0.845,
+      "step": 2408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206183780888214e-05,
+      "loss": 0.8147,
+      "step": 2409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201535958452168e-05,
+      "loss": 0.7443,
+      "step": 2410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196887033558538e-05,
+      "loss": 0.752,
+      "step": 2411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192237007427168e-05,
+      "loss": 0.7867,
+      "step": 2412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187585881278195e-05,
+      "loss": 0.2852,
+      "step": 2413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182933656332049e-05,
+      "loss": 0.7457,
+      "step": 2414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178280333809448e-05,
+      "loss": 0.8164,
+      "step": 2415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173625914931392e-05,
+      "loss": 0.7828,
+      "step": 2416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168970400919175e-05,
+      "loss": 0.7698,
+      "step": 2417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164313792994373e-05,
+      "loss": 0.7805,
+      "step": 2418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159656092378852e-05,
+      "loss": 0.7715,
+      "step": 2419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154997300294768e-05,
+      "loss": 0.7765,
+      "step": 2420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150337417964557e-05,
+      "loss": 0.7737,
+      "step": 2421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414567644661094e-05,
+      "loss": 0.7683,
+      "step": 2422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141014387456937e-05,
+      "loss": 0.7488,
+      "step": 2423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136351241725836e-05,
+      "loss": 0.8464,
+      "step": 2424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131687010641222e-05,
+      "loss": 0.7861,
+      "step": 2425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127021695426961e-05,
+      "loss": 0.7164,
+      "step": 2426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122355297307202e-05,
+      "loss": 0.7944,
+      "step": 2427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117687817506378e-05,
+      "loss": 0.7883,
+      "step": 2428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113019257249212e-05,
+      "loss": 0.2718,
+      "step": 2429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41083496177607e-05,
+      "loss": 0.7837,
+      "step": 2430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103678900266135e-05,
+      "loss": 0.7898,
+      "step": 2431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4099007105991078e-05,
+      "loss": 0.7617,
+      "step": 2432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094334236161384e-05,
+      "loss": 0.8209,
+      "step": 2433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089660292003183e-05,
+      "loss": 0.7976,
+      "step": 2434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408498527474289e-05,
+      "loss": 0.7428,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080309185607203e-05,
+      "loss": 0.8245,
+      "step": 2436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075632025823097e-05,
+      "loss": 0.7411,
+      "step": 2437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070953796617832e-05,
+      "loss": 0.2932,
+      "step": 2438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066274499218949e-05,
+      "loss": 0.2749,
+      "step": 2439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061594134854262e-05,
+      "loss": 0.8311,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056912704751873e-05,
+      "loss": 0.2595,
+      "step": 2441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052230210140163e-05,
+      "loss": 0.8618,
+      "step": 2442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047546652247785e-05,
+      "loss": 0.7603,
+      "step": 2443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042862032303682e-05,
+      "loss": 0.7507,
+      "step": 2444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038176351537067e-05,
+      "loss": 0.7478,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033489611177433e-05,
+      "loss": 0.7493,
+      "step": 2446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4028801812454558e-05,
+      "loss": 0.7844,
+      "step": 2447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4024112956598483e-05,
+      "loss": 0.7793,
+      "step": 2448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019423044839542e-05,
+      "loss": 0.3007,
+      "step": 2449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014732078408341e-05,
+      "loss": 0.7568,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010040058535754e-05,
+      "loss": 0.7421,
+      "step": 2451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005346986452946e-05,
+      "loss": 0.7896,
+      "step": 2452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000652863391344e-05,
+      "loss": 0.8047,
+      "step": 2453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995957690582662e-05,
+      "loss": 0.853,
+      "step": 2454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991261469258885e-05,
+      "loss": 0.7428,
+      "step": 2455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398656420065227e-05,
+      "loss": 0.7869,
+      "step": 2456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981865885995353e-05,
+      "loss": 0.8262,
+      "step": 2457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977166526520946e-05,
+      "loss": 0.8007,
+      "step": 2458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972466123462129e-05,
+      "loss": 0.7437,
+      "step": 2459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396776467805226e-05,
+      "loss": 0.7656,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963062191524972e-05,
+      "loss": 0.792,
+      "step": 2461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958358665114167e-05,
+      "loss": 0.7606,
+      "step": 2462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953654100054021e-05,
+      "loss": 0.7778,
+      "step": 2463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948948497578982e-05,
+      "loss": 0.7445,
+      "step": 2464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944241858923777e-05,
+      "loss": 0.8077,
+      "step": 2465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939534185323398e-05,
+      "loss": 0.8845,
+      "step": 2466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934825478013107e-05,
+      "loss": 0.7754,
+      "step": 2467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930115738228445e-05,
+      "loss": 0.8374,
+      "step": 2468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925404967205211e-05,
+      "loss": 0.7659,
+      "step": 2469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392069316617949e-05,
+      "loss": 0.7976,
+      "step": 2470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915980336387631e-05,
+      "loss": 0.8196,
+      "step": 2471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911266479066245e-05,
+      "loss": 0.7617,
+      "step": 2472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906551595452225e-05,
+      "loss": 0.7808,
+      "step": 2473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901835686782727e-05,
+      "loss": 0.7711,
+      "step": 2474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897118754295176e-05,
+      "loss": 0.7917,
+      "step": 2475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892400799227266e-05,
+      "loss": 0.8154,
+      "step": 2476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887681822816964e-05,
+      "loss": 0.7461,
+      "step": 2477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882961826302493e-05,
+      "loss": 0.8086,
+      "step": 2478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878240810922359e-05,
+      "loss": 0.3102,
+      "step": 2479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873518777915322e-05,
+      "loss": 0.78,
+      "step": 2480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868795728520419e-05,
+      "loss": 0.7856,
+      "step": 2481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386407166397695e-05,
+      "loss": 0.7876,
+      "step": 2482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859346585524476e-05,
+      "loss": 0.2797,
+      "step": 2483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854620494402837e-05,
+      "loss": 0.7769,
+      "step": 2484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849893391852122e-05,
+      "loss": 0.7343,
+      "step": 2485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38451652791127e-05,
+      "loss": 0.77,
+      "step": 2486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840436157425197e-05,
+      "loss": 0.7781,
+      "step": 2487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835706028030504e-05,
+      "loss": 0.8381,
+      "step": 2488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383097489216978e-05,
+      "loss": 0.7227,
+      "step": 2489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826242751084451e-05,
+      "loss": 0.8223,
+      "step": 2490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821509606016194e-05,
+      "loss": 0.7653,
+      "step": 2491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816775458206962e-05,
+      "loss": 0.7263,
+      "step": 2492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812040308898969e-05,
+      "loss": 0.7849,
+      "step": 2493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807304159334682e-05,
+      "loss": 0.8354,
+      "step": 2494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802567010756844e-05,
+      "loss": 0.7639,
+      "step": 2495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797828864408456e-05,
+      "loss": 0.8101,
+      "step": 2496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793089721532773e-05,
+      "loss": 0.7871,
+      "step": 2497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3788349583373321e-05,
+      "loss": 0.7593,
+      "step": 2498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378360845117388e-05,
+      "loss": 0.7242,
+      "step": 2499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37788663261785e-05,
+      "loss": 0.8008,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774123209631483e-05,
+      "loss": 0.7756,
+      "step": 2501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769379102777392e-05,
+      "loss": 0.7228,
+      "step": 2502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764634006861055e-05,
+      "loss": 0.7812,
+      "step": 2503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759887923127553e-05,
+      "loss": 0.2591,
+      "step": 2504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755140852822234e-05,
+      "loss": 0.7568,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750392797190699e-05,
+      "loss": 0.8098,
+      "step": 2506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745643757478805e-05,
+      "loss": 0.8198,
+      "step": 2507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740893734932677e-05,
+      "loss": 0.7961,
+      "step": 2508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736142730798692e-05,
+      "loss": 0.833,
+      "step": 2509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3731390746323479e-05,
+      "loss": 0.7263,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726637782753934e-05,
+      "loss": 0.8035,
+      "step": 2511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721883841337206e-05,
+      "loss": 0.7656,
+      "step": 2512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37171289233207e-05,
+      "loss": 0.7996,
+      "step": 2513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712373029952078e-05,
+      "loss": 0.7231,
+      "step": 2514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707616162479255e-05,
+      "loss": 0.7656,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702858322150405e-05,
+      "loss": 0.8062,
+      "step": 2516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369809951021396e-05,
+      "loss": 0.782,
+      "step": 2517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693339727918602e-05,
+      "loss": 0.7523,
+      "step": 2518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688578976513268e-05,
+      "loss": 0.8186,
+      "step": 2519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368381725724715e-05,
+      "loss": 0.8306,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679054571369691e-05,
+      "loss": 0.7642,
+      "step": 2521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674290920130601e-05,
+      "loss": 0.7992,
+      "step": 2522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669526304779823e-05,
+      "loss": 0.7896,
+      "step": 2523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366476072656757e-05,
+      "loss": 0.7549,
+      "step": 2524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659994186744297e-05,
+      "loss": 0.8174,
+      "step": 2525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655226686560714e-05,
+      "loss": 0.8411,
+      "step": 2526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650458227267789e-05,
+      "loss": 0.824,
+      "step": 2527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645688810116735e-05,
+      "loss": 0.8655,
+      "step": 2528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640918436359017e-05,
+      "loss": 0.7793,
+      "step": 2529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363614710724635e-05,
+      "loss": 0.832,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631374824030707e-05,
+      "loss": 0.7396,
+      "step": 2531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626601587964304e-05,
+      "loss": 0.7653,
+      "step": 2532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621827400299608e-05,
+      "loss": 0.8088,
+      "step": 2533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617052262289337e-05,
+      "loss": 0.7883,
+      "step": 2534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612276175186457e-05,
+      "loss": 0.8071,
+      "step": 2535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360749914024419e-05,
+      "loss": 0.7852,
+      "step": 2536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602721158715997e-05,
+      "loss": 0.8101,
+      "step": 2537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597942231855592e-05,
+      "loss": 0.7598,
+      "step": 2538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593162360916936e-05,
+      "loss": 0.8223,
+      "step": 2539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588381547154237e-05,
+      "loss": 0.7109,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3583599791821958e-05,
+      "loss": 0.7832,
+      "step": 2541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578817096174795e-05,
+      "loss": 0.7664,
+      "step": 2542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574033461467698e-05,
+      "loss": 0.7395,
+      "step": 2543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569248888955872e-05,
+      "loss": 0.7495,
+      "step": 2544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564463379894752e-05,
+      "loss": 0.8301,
+      "step": 2545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559676935540029e-05,
+      "loss": 0.7927,
+      "step": 2546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554889557147638e-05,
+      "loss": 0.7524,
+      "step": 2547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550101245973759e-05,
+      "loss": 0.814,
+      "step": 2548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545312003274816e-05,
+      "loss": 0.7947,
+      "step": 2549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354052183030747e-05,
+      "loss": 0.2963,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535730728328642e-05,
+      "loss": 0.8213,
+      "step": 2551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530938698595485e-05,
+      "loss": 0.754,
+      "step": 2552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526145742365398e-05,
+      "loss": 0.7422,
+      "step": 2553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521351860896025e-05,
+      "loss": 0.7881,
+      "step": 2554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351655705544525e-05,
+      "loss": 0.8057,
+      "step": 2555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511761327271203e-05,
+      "loss": 0.8055,
+      "step": 2556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506964677632253e-05,
+      "loss": 0.7411,
+      "step": 2557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350216710778701e-05,
+      "loss": 0.7949,
+      "step": 2558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497368618994329e-05,
+      "loss": 0.8131,
+      "step": 2559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492569212513306e-05,
+      "loss": 0.8066,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487768889603274e-05,
+      "loss": 0.8555,
+      "step": 2561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3482967651523809e-05,
+      "loss": 0.7745,
+      "step": 2562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478165499534728e-05,
+      "loss": 0.7433,
+      "step": 2563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473362434896085e-05,
+      "loss": 0.7291,
+      "step": 2564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468558458868173e-05,
+      "loss": 0.7615,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463753572711533e-05,
+      "loss": 0.757,
+      "step": 2566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345894777768693e-05,
+      "loss": 0.7887,
+      "step": 2567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454141075055382e-05,
+      "loss": 0.7522,
+      "step": 2568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449333466078133e-05,
+      "loss": 0.7957,
+      "step": 2569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444524952016676e-05,
+      "loss": 0.7975,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439715534132733e-05,
+      "loss": 0.7845,
+      "step": 2571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434905213688265e-05,
+      "loss": 0.8503,
+      "step": 2572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430093991945476e-05,
+      "loss": 0.7942,
+      "step": 2573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425281870166792e-05,
+      "loss": 0.8621,
+      "step": 2574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420468849614896e-05,
+      "loss": 0.7731,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341565493155269e-05,
+      "loss": 0.7844,
+      "step": 2576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410840117243315e-05,
+      "loss": 0.7504,
+      "step": 2577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406024407950152e-05,
+      "loss": 0.7927,
+      "step": 2578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3401207804936813e-05,
+      "loss": 0.835,
+      "step": 2579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396390309467145e-05,
+      "loss": 0.7561,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339157192280523e-05,
+      "loss": 0.7775,
+      "step": 2581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386752646215383e-05,
+      "loss": 0.8572,
+      "step": 2582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381932480962154e-05,
+      "loss": 0.8192,
+      "step": 2583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377111428310327e-05,
+      "loss": 0.8127,
+      "step": 2584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372289489524912e-05,
+      "loss": 0.894,
+      "step": 2585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336746666587116e-05,
+      "loss": 0.7461,
+      "step": 2586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362642958614552e-05,
+      "loss": 0.762,
+      "step": 2587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3357818369020792e-05,
+      "loss": 0.7157,
+      "step": 2588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352992898355834e-05,
+      "loss": 0.8167,
+      "step": 2589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3348166547885846e-05,
+      "loss": 0.8176,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343339318877233e-05,
+      "loss": 0.7833,
+      "step": 2591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338511212596635e-05,
+      "loss": 0.8345,
+      "step": 2592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333368223031091e-05,
+      "loss": 0.8149,
+      "step": 2593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332885237328716e-05,
+      "loss": 0.8606,
+      "step": 2594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324021642792708e-05,
+      "loss": 0.7981,
+      "step": 2595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319190040095111e-05,
+      "loss": 0.7413,
+      "step": 2596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314357566462148e-05,
+      "loss": 0.8036,
+      "step": 2597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309524223161835e-05,
+      "loss": 0.7769,
+      "step": 2598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304690011462408e-05,
+      "loss": 0.7984,
+      "step": 2599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329985493263234e-05,
+      "loss": 0.7764,
+      "step": 2600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329501898794032e-05,
+      "loss": 0.8057,
+      "step": 2601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290182178655275e-05,
+      "loss": 0.7699,
+      "step": 2602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285344506046356e-05,
+      "loss": 0.7651,
+      "step": 2603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280505971382933e-05,
+      "loss": 0.2789,
+      "step": 2604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275666575934612e-05,
+      "loss": 0.7615,
+      "step": 2605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270826320971222e-05,
+      "loss": 0.2792,
+      "step": 2606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265985207762813e-05,
+      "loss": 0.8191,
+      "step": 2607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261143237579667e-05,
+      "loss": 0.8162,
+      "step": 2608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256300411692289e-05,
+      "loss": 0.7776,
+      "step": 2609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251456731371401e-05,
+      "loss": 0.7632,
+      "step": 2610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246612197887963e-05,
+      "loss": 0.2953,
+      "step": 2611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241766812513146e-05,
+      "loss": 0.7777,
+      "step": 2612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236920576518349e-05,
+      "loss": 0.8403,
+      "step": 2613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232073491175202e-05,
+      "loss": 0.7976,
+      "step": 2614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227225557755544e-05,
+      "loss": 0.7847,
+      "step": 2615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222376777531443e-05,
+      "loss": 0.7805,
+      "step": 2616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217527151775193e-05,
+      "loss": 0.8818,
+      "step": 2617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212676681759307e-05,
+      "loss": 0.7687,
+      "step": 2618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207825368756518e-05,
+      "loss": 0.7612,
+      "step": 2619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202973214039778e-05,
+      "loss": 0.7817,
+      "step": 2620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198120218882266e-05,
+      "loss": 0.8533,
+      "step": 2621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319326638455738e-05,
+      "loss": 0.8159,
+      "step": 2622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188411712338733e-05,
+      "loss": 0.8152,
+      "step": 2623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3183556203500164e-05,
+      "loss": 0.7798,
+      "step": 2624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178699859315727e-05,
+      "loss": 0.7007,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173842681059698e-05,
+      "loss": 0.8558,
+      "step": 2626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168984670006575e-05,
+      "loss": 0.8291,
+      "step": 2627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164125827431067e-05,
+      "loss": 0.8381,
+      "step": 2628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159266154608104e-05,
+      "loss": 0.733,
+      "step": 2629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154405652812839e-05,
+      "loss": 0.7834,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149544323320633e-05,
+      "loss": 0.8662,
+      "step": 2631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144682167407072e-05,
+      "loss": 0.8345,
+      "step": 2632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313981918634796e-05,
+      "loss": 0.8037,
+      "step": 2633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134955381419303e-05,
+      "loss": 0.7944,
+      "step": 2634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130090753897347e-05,
+      "loss": 0.8362,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312522530505853e-05,
+      "loss": 0.864,
+      "step": 2636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120359036179523e-05,
+      "loss": 0.761,
+      "step": 2637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115491948537207e-05,
+      "loss": 0.7401,
+      "step": 2638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110624043408668e-05,
+      "loss": 0.8438,
+      "step": 2639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105755322071221e-05,
+      "loss": 0.7874,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100885785802388e-05,
+      "loss": 0.7888,
+      "step": 2641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096015435879903e-05,
+      "loss": 0.7195,
+      "step": 2642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091144273581723e-05,
+      "loss": 0.7426,
+      "step": 2643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086272300186002e-05,
+      "loss": 0.8101,
+      "step": 2644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081399516971125e-05,
+      "loss": 0.7579,
+      "step": 2645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076525925215677e-05,
+      "loss": 0.808,
+      "step": 2646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307165152619846e-05,
+      "loss": 0.769,
+      "step": 2647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066776321198483e-05,
+      "loss": 0.7639,
+      "step": 2648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061900311494977e-05,
+      "loss": 0.267,
+      "step": 2649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057023498367371e-05,
+      "loss": 0.7527,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052145883095315e-05,
+      "loss": 0.7933,
+      "step": 2651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047267466958667e-05,
+      "loss": 0.7638,
+      "step": 2652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042388251237488e-05,
+      "loss": 0.8369,
+      "step": 2653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303750823721206e-05,
+      "loss": 0.6934,
+      "step": 2654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032627426162867e-05,
+      "loss": 0.7174,
+      "step": 2655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3027745819370603e-05,
+      "loss": 0.8516,
+      "step": 2656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022863418116179e-05,
+      "loss": 0.7891,
+      "step": 2657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017980223680695e-05,
+      "loss": 0.8472,
+      "step": 2658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013096237345482e-05,
+      "loss": 0.8252,
+      "step": 2659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008211460392063e-05,
+      "loss": 0.7833,
+      "step": 2660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003325894102178e-05,
+      "loss": 0.8159,
+      "step": 2661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998439539757772e-05,
+      "loss": 0.7693,
+      "step": 2662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993552398640985e-05,
+      "loss": 0.7942,
+      "step": 2663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988664472034183e-05,
+      "loss": 0.8357,
+      "step": 2664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983775761219923e-05,
+      "loss": 0.8271,
+      "step": 2665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978886267480974e-05,
+      "loss": 0.854,
+      "step": 2666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297399599210031e-05,
+      "loss": 0.7834,
+      "step": 2667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969104936361111e-05,
+      "loss": 0.7871,
+      "step": 2668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296421310154676e-05,
+      "loss": 0.8225,
+      "step": 2669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295932048894084e-05,
+      "loss": 0.7532,
+      "step": 2670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954427099827149e-05,
+      "loss": 0.7791,
+      "step": 2671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2949532935489682e-05,
+      "loss": 0.6802,
+      "step": 2672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944637997212637e-05,
+      "loss": 0.7979,
+      "step": 2673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939742286280415e-05,
+      "loss": 0.7981,
+      "step": 2674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934845803977619e-05,
+      "loss": 0.7734,
+      "step": 2675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929948551589062e-05,
+      "loss": 0.7419,
+      "step": 2676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292505053039975e-05,
+      "loss": 0.2803,
+      "step": 2677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2920151741694893e-05,
+      "loss": 0.8396,
+      "step": 2678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915252186759906e-05,
+      "loss": 0.8062,
+      "step": 2679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29103518668804e-05,
+      "loss": 0.7817,
+      "step": 2680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905450783342192e-05,
+      "loss": 0.8486,
+      "step": 2681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900548937431296e-05,
+      "loss": 0.844,
+      "step": 2682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895646330433924e-05,
+      "loss": 0.723,
+      "step": 2683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890742963636495e-05,
+      "loss": 0.7566,
+      "step": 2684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288583883832562e-05,
+      "loss": 0.7271,
+      "step": 2685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880933955788113e-05,
+      "loss": 0.77,
+      "step": 2686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876028317310982e-05,
+      "loss": 0.7742,
+      "step": 2687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287112192418144e-05,
+      "loss": 0.7653,
+      "step": 2688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2866214777686894e-05,
+      "loss": 0.7864,
+      "step": 2689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286130687911495e-05,
+      "loss": 0.7626,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285639822975341e-05,
+      "loss": 0.7566,
+      "step": 2691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851488830890276e-05,
+      "loss": 0.8074,
+      "step": 2692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846578683813742e-05,
+      "loss": 0.8044,
+      "step": 2693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28416677898122e-05,
+      "loss": 0.7603,
+      "step": 2694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283675615017424e-05,
+      "loss": 0.8008,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831843766188649e-05,
+      "loss": 0.8114,
+      "step": 2696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826930639144405e-05,
+      "loss": 0.7397,
+      "step": 2697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822016770330682e-05,
+      "loss": 0.761,
+      "step": 2698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817102161036852e-05,
+      "loss": 0.804,
+      "step": 2699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812186812552476e-05,
+      "loss": 0.8262,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807270726167312e-05,
+      "loss": 0.7686,
+      "step": 2701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802353903171309e-05,
+      "loss": 0.7881,
+      "step": 2702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797436344854618e-05,
+      "loss": 0.7854,
+      "step": 2703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792518052507572e-05,
+      "loss": 0.7786,
+      "step": 2704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787599027420704e-05,
+      "loss": 0.7415,
+      "step": 2705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782679270884733e-05,
+      "loss": 0.8127,
+      "step": 2706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777758784190573e-05,
+      "loss": 0.8025,
+      "step": 2707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772837568629333e-05,
+      "loss": 0.8259,
+      "step": 2708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767915625492306e-05,
+      "loss": 0.7817,
+      "step": 2709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762992956070985e-05,
+      "loss": 0.7773,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275806956165705e-05,
+      "loss": 0.7214,
+      "step": 2711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275314544354236e-05,
+      "loss": 0.8188,
+      "step": 2712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748220603018983e-05,
+      "loss": 0.8181,
+      "step": 2713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743295041379164e-05,
+      "loss": 0.6945,
+      "step": 2714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273836875991534e-05,
+      "loss": 0.7771,
+      "step": 2715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733441759920142e-05,
+      "loss": 0.8127,
+      "step": 2716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272851404268638e-05,
+      "loss": 0.7534,
+      "step": 2717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723585609507057e-05,
+      "loss": 0.7732,
+      "step": 2718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718656461675364e-05,
+      "loss": 0.7697,
+      "step": 2719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713726600484683e-05,
+      "loss": 0.7632,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708796027228582e-05,
+      "loss": 0.7328,
+      "step": 2721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703864743200805e-05,
+      "loss": 0.8547,
+      "step": 2722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2698932749695294e-05,
+      "loss": 0.7927,
+      "step": 2723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269400004800618e-05,
+      "loss": 0.7891,
+      "step": 2724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689066639427764e-05,
+      "loss": 0.8455,
+      "step": 2725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684132525254553e-05,
+      "loss": 0.8262,
+      "step": 2726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679197706781224e-05,
+      "loss": 0.8425,
+      "step": 2727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2674262185302641e-05,
+      "loss": 0.8237,
+      "step": 2728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669325962113856e-05,
+      "loss": 0.7786,
+      "step": 2729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266438903851011e-05,
+      "loss": 0.8008,
+      "step": 2730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659451415786812e-05,
+      "loss": 0.7207,
+      "step": 2731
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654513095239573e-05,
+      "loss": 0.7311,
+      "step": 2732
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649574078164172e-05,
+      "loss": 0.7616,
+      "step": 2733
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644634365856583e-05,
+      "loss": 0.8181,
+      "step": 2734
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639693959612953e-05,
+      "loss": 0.7797,
+      "step": 2735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634752860729616e-05,
+      "loss": 0.6941,
+      "step": 2736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629811070503085e-05,
+      "loss": 0.8319,
+      "step": 2737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624868590230056e-05,
+      "loss": 0.7625,
+      "step": 2738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619925421207409e-05,
+      "loss": 0.7761,
+      "step": 2739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614981564732199e-05,
+      "loss": 0.7996,
+      "step": 2740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261003702210166e-05,
+      "loss": 0.7666,
+      "step": 2741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605091794613218e-05,
+      "loss": 0.7396,
+      "step": 2742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600145883564465e-05,
+      "loss": 0.7499,
+      "step": 2743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595199290253181e-05,
+      "loss": 0.8127,
+      "step": 2744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259025201597732e-05,
+      "loss": 0.854,
+      "step": 2745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258530406203502e-05,
+      "loss": 0.7518,
+      "step": 2746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580355429724589e-05,
+      "loss": 0.8127,
+      "step": 2747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575406120344521e-05,
+      "loss": 0.7268,
+      "step": 2748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570456135193488e-05,
+      "loss": 0.802,
+      "step": 2749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565505475570332e-05,
+      "loss": 0.8311,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560554142774077e-05,
+      "loss": 0.7734,
+      "step": 2751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555602138103924e-05,
+      "loss": 0.7686,
+      "step": 2752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550649462859249e-05,
+      "loss": 0.837,
+      "step": 2753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545696118339601e-05,
+      "loss": 0.7722,
+      "step": 2754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254074210584471e-05,
+      "loss": 0.7778,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535787426674479e-05,
+      "loss": 0.7883,
+      "step": 2756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530832082128983e-05,
+      "loss": 0.7528,
+      "step": 2757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2525876073508478e-05,
+      "loss": 0.8242,
+      "step": 2758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520919402113392e-05,
+      "loss": 0.7522,
+      "step": 2759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515962069244321e-05,
+      "loss": 0.8979,
+      "step": 2760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251100407620204e-05,
+      "loss": 0.2977,
+      "step": 2761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25060454242875e-05,
+      "loss": 0.7344,
+      "step": 2762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501086114801814e-05,
+      "loss": 0.7217,
+      "step": 2763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2496126149046284e-05,
+      "loss": 0.755,
+      "step": 2764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491165528322368e-05,
+      "loss": 0.7668,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486204253931706e-05,
+      "loss": 0.7485,
+      "step": 2766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481242327176105e-05,
+      "loss": 0.7715,
+      "step": 2767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476279749357545e-05,
+      "loss": 0.7037,
+      "step": 2768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471316521778176e-05,
+      "loss": 0.7506,
+      "step": 2769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466352645740321e-05,
+      "loss": 0.8093,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461388122546466e-05,
+      "loss": 0.7769,
+      "step": 2771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456422953499278e-05,
+      "loss": 0.8003,
+      "step": 2772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451457139901586e-05,
+      "loss": 0.8171,
+      "step": 2773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446490683056384e-05,
+      "loss": 0.8037,
+      "step": 2774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2441523584266848e-05,
+      "loss": 0.7932,
+      "step": 2775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436555844836307e-05,
+      "loss": 0.7449,
+      "step": 2776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431587466068269e-05,
+      "loss": 0.7646,
+      "step": 2777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242661844926641e-05,
+      "loss": 0.754,
+      "step": 2778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421648795734565e-05,
+      "loss": 0.76,
+      "step": 2779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416678506776743e-05,
+      "loss": 0.7898,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411707583697118e-05,
+      "loss": 0.7577,
+      "step": 2781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240673602780003e-05,
+      "loss": 0.7655,
+      "step": 2782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401763840389982e-05,
+      "loss": 0.7935,
+      "step": 2783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396791022771652e-05,
+      "loss": 0.8026,
+      "step": 2784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391817576249869e-05,
+      "loss": 0.6927,
+      "step": 2785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386843502129646e-05,
+      "loss": 0.7916,
+      "step": 2786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238186880171614e-05,
+      "loss": 0.8097,
+      "step": 2787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237689347631469e-05,
+      "loss": 0.7534,
+      "step": 2788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371917527230785e-05,
+      "loss": 0.7776,
+      "step": 2789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366940955770087e-05,
+      "loss": 0.7893,
+      "step": 2790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361963763238419e-05,
+      "loss": 0.7417,
+      "step": 2791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356985950941761e-05,
+      "loss": 0.7377,
+      "step": 2792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352007520186266e-05,
+      "loss": 0.7817,
+      "step": 2793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347028472278247e-05,
+      "loss": 0.7424,
+      "step": 2794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342048808524167e-05,
+      "loss": 0.7476,
+      "step": 2795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337068530230665e-05,
+      "loss": 0.8474,
+      "step": 2796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332087638704538e-05,
+      "loss": 0.7483,
+      "step": 2797
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327106135252736e-05,
+      "loss": 0.2574,
+      "step": 2798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322124021182377e-05,
+      "loss": 0.66,
+      "step": 2799
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2317141297800743e-05,
+      "loss": 0.7693,
+      "step": 2800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312157966415263e-05,
+      "loss": 0.8142,
+      "step": 2801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307174028333537e-05,
+      "loss": 0.7283,
+      "step": 2802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302189484863318e-05,
+      "loss": 0.7747,
+      "step": 2803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297204337312524e-05,
+      "loss": 0.7738,
+      "step": 2804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292218586989226e-05,
+      "loss": 0.8575,
+      "step": 2805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287232235201651e-05,
+      "loss": 0.7301,
+      "step": 2806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282245283258193e-05,
+      "loss": 0.759,
+      "step": 2807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2277257732467392e-05,
+      "loss": 0.7341,
+      "step": 2808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2272269584137955e-05,
+      "loss": 0.2922,
+      "step": 2809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267280839578743e-05,
+      "loss": 0.719,
+      "step": 2810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262291500098769e-05,
+      "loss": 0.7987,
+      "step": 2811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257301567007207e-05,
+      "loss": 0.832,
+      "step": 2812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252311041613387e-05,
+      "loss": 0.7118,
+      "step": 2813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247319925226788e-05,
+      "loss": 0.2738,
+      "step": 2814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224232821915705e-05,
+      "loss": 0.798,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237335924713976e-05,
+      "loss": 0.2507,
+      "step": 2816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22323430432075e-05,
+      "loss": 0.7397,
+      "step": 2817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222734957594773e-05,
+      "loss": 0.7963,
+      "step": 2818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222355524244922e-05,
+      "loss": 0.7073,
+      "step": 2819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217360889409482e-05,
+      "loss": 0.7908,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212365672751977e-05,
+      "loss": 0.751,
+      "step": 2821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2207369875583117e-05,
+      "loss": 0.759,
+      "step": 2822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202373499213772e-05,
+      "loss": 0.7896,
+      "step": 2823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219737654495496e-05,
+      "loss": 0.7534,
+      "step": 2824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192379014117848e-05,
+      "loss": 0.7697,
+      "step": 2825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187380908013764e-05,
+      "loss": 0.7322,
+      "step": 2826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182382227954177e-05,
+      "loss": 0.7762,
+      "step": 2827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177382975250712e-05,
+      "loss": 0.7664,
+      "step": 2828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217238315121514e-05,
+      "loss": 0.6909,
+      "step": 2829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167382757159384e-05,
+      "loss": 0.6647,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162381794395523e-05,
+      "loss": 0.7249,
+      "step": 2831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157380264235777e-05,
+      "loss": 0.7616,
+      "step": 2832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152378167992511e-05,
+      "loss": 0.7783,
+      "step": 2833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147375506978252e-05,
+      "loss": 0.8159,
+      "step": 2834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142372282505669e-05,
+      "loss": 0.7559,
+      "step": 2835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213736849588757e-05,
+      "loss": 0.7974,
+      "step": 2836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132364148436924e-05,
+      "loss": 0.7817,
+      "step": 2837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127359241466838e-05,
+      "loss": 0.7559,
+      "step": 2838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122353776290572e-05,
+      "loss": 0.8066,
+      "step": 2839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117347754221533e-05,
+      "loss": 0.7878,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112341176573264e-05,
+      "loss": 0.8405,
+      "step": 2841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107334044659465e-05,
+      "loss": 0.7749,
+      "step": 2842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102326359793974e-05,
+      "loss": 0.7622,
+      "step": 2843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209731812329078e-05,
+      "loss": 0.8489,
+      "step": 2844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092309336464015e-05,
+      "loss": 0.7772,
+      "step": 2845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208730000062795e-05,
+      "loss": 0.7476,
+      "step": 2846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082290117097006e-05,
+      "loss": 0.8538,
+      "step": 2847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207727968718575e-05,
+      "loss": 0.78,
+      "step": 2848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207226871220888e-05,
+      "loss": 0.2738,
+      "step": 2849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067257193481254e-05,
+      "loss": 0.8083,
+      "step": 2850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062245132317862e-05,
+      "loss": 0.8081,
+      "step": 2851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057232530033836e-05,
+      "loss": 0.7679,
+      "step": 2852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052219387944452e-05,
+      "loss": 0.8501,
+      "step": 2853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047205707365133e-05,
+      "loss": 0.8215,
+      "step": 2854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042191489611434e-05,
+      "loss": 0.7349,
+      "step": 2855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037176735999057e-05,
+      "loss": 0.7327,
+      "step": 2856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032161447843839e-05,
+      "loss": 0.7227,
+      "step": 2857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202714562646177e-05,
+      "loss": 0.7948,
+      "step": 2858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022129273168965e-05,
+      "loss": 0.8123,
+      "step": 2859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2017112389281684e-05,
+      "loss": 0.8162,
+      "step": 2860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201209497611633e-05,
+      "loss": 0.745,
+      "step": 2861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007077034989441e-05,
+      "loss": 0.7666,
+      "step": 2862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002058567217692e-05,
+      "loss": 0.8274,
+      "step": 2863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997039574117904e-05,
+      "loss": 0.2709,
+      "step": 2864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992020057007023e-05,
+      "loss": 0.8247,
+      "step": 2865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987000017202145e-05,
+      "loss": 0.7455,
+      "step": 2866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981979456020498e-05,
+      "loss": 0.8374,
+      "step": 2867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976958374779445e-05,
+      "loss": 0.7866,
+      "step": 2868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971936774796487e-05,
+      "loss": 0.7683,
+      "step": 2869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966914657389266e-05,
+      "loss": 0.7094,
+      "step": 2870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961892023875548e-05,
+      "loss": 0.8115,
+      "step": 2871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956868875573246e-05,
+      "loss": 0.8342,
+      "step": 2872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951845213800404e-05,
+      "loss": 0.8264,
+      "step": 2873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946821039875194e-05,
+      "loss": 0.7822,
+      "step": 2874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194179635511594e-05,
+      "loss": 0.7629,
+      "step": 2875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193677116084108e-05,
+      "loss": 0.7908,
+      "step": 2876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931745458369202e-05,
+      "loss": 0.7656,
+      "step": 2877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192671924901901e-05,
+      "loss": 0.7421,
+      "step": 2878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921692534109357e-05,
+      "loss": 0.7771,
+      "step": 2879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916665314959227e-05,
+      "loss": 0.8499,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911637592887722e-05,
+      "loss": 0.792,
+      "step": 2881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906609369214088e-05,
+      "loss": 0.7544,
+      "step": 2882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901580645257709e-05,
+      "loss": 0.7788,
+      "step": 2883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896551422338083e-05,
+      "loss": 0.7991,
+      "step": 2884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891521701774849e-05,
+      "loss": 0.814,
+      "step": 2885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886491484887779e-05,
+      "loss": 0.7986,
+      "step": 2886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881460772996767e-05,
+      "loss": 0.7612,
+      "step": 2887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876429567421843e-05,
+      "loss": 0.7737,
+      "step": 2888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187139786948317e-05,
+      "loss": 0.7632,
+      "step": 2889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866365680501025e-05,
+      "loss": 0.8403,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861333001795832e-05,
+      "loss": 0.7468,
+      "step": 2891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856299834688127e-05,
+      "loss": 0.7831,
+      "step": 2892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851266180498596e-05,
+      "loss": 0.7888,
+      "step": 2893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846232040548027e-05,
+      "loss": 0.8066,
+      "step": 2894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841197416157351e-05,
+      "loss": 0.7935,
+      "step": 2895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836162308647628e-05,
+      "loss": 0.7568,
+      "step": 2896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831126719340034e-05,
+      "loss": 0.7587,
+      "step": 2897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826090649555879e-05,
+      "loss": 0.7495,
+      "step": 2898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821054100616598e-05,
+      "loss": 0.7675,
+      "step": 2899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816017073843748e-05,
+      "loss": 0.7407,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181097957055902e-05,
+      "loss": 0.7939,
+      "step": 2901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180594159208422e-05,
+      "loss": 0.7559,
+      "step": 2902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800903139741284e-05,
+      "loss": 0.8239,
+      "step": 2903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795864214852268e-05,
+      "loss": 0.7203,
+      "step": 2904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790824818739363e-05,
+      "loss": 0.731,
+      "step": 2905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178578495272487e-05,
+      "loss": 0.7467,
+      "step": 2906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780744618131224e-05,
+      "loss": 0.2726,
+      "step": 2907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775703816280972e-05,
+      "loss": 0.8215,
+      "step": 2908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770662548496791e-05,
+      "loss": 0.823,
+      "step": 2909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765620816101487e-05,
+      "loss": 0.7781,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760578620417968e-05,
+      "loss": 0.2856,
+      "step": 2911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755535962769288e-05,
+      "loss": 0.7605,
+      "step": 2912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750492844478601e-05,
+      "loss": 0.7617,
+      "step": 2913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745449266869194e-05,
+      "loss": 0.2694,
+      "step": 2914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1740405231264471e-05,
+      "loss": 0.719,
+      "step": 2915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735360738987956e-05,
+      "loss": 0.7793,
+      "step": 2916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730315791363292e-05,
+      "loss": 0.7439,
+      "step": 2917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725270389714247e-05,
+      "loss": 0.8064,
+      "step": 2918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720224535364697e-05,
+      "loss": 0.7495,
+      "step": 2919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171517822963865e-05,
+      "loss": 0.7358,
+      "step": 2920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710131473860224e-05,
+      "loss": 0.6931,
+      "step": 2921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705084269353655e-05,
+      "loss": 0.6951,
+      "step": 2922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700036617443305e-05,
+      "loss": 0.8062,
+      "step": 2923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694988519453635e-05,
+      "loss": 0.7852,
+      "step": 2924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689939976709251e-05,
+      "loss": 0.7737,
+      "step": 2925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684890990534847e-05,
+      "loss": 0.7554,
+      "step": 2926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679841562255253e-05,
+      "loss": 0.8074,
+      "step": 2927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674791693195409e-05,
+      "loss": 0.7822,
+      "step": 2928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669741384680364e-05,
+      "loss": 0.782,
+      "step": 2929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664690638035297e-05,
+      "loss": 0.733,
+      "step": 2930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659639454585487e-05,
+      "loss": 0.7742,
+      "step": 2931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654587835656334e-05,
+      "loss": 0.8264,
+      "step": 2932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649535782573355e-05,
+      "loss": 0.8051,
+      "step": 2933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644483296662182e-05,
+      "loss": 0.7856,
+      "step": 2934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1639430379248546e-05,
+      "loss": 0.7991,
+      "step": 2935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634377031658307e-05,
+      "loss": 0.7806,
+      "step": 2936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162932325521744e-05,
+      "loss": 0.8191,
+      "step": 2937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624269051252012e-05,
+      "loss": 0.7915,
+      "step": 2938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619214421088225e-05,
+      "loss": 0.8164,
+      "step": 2939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614159366052384e-05,
+      "loss": 0.7955,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609103887470898e-05,
+      "loss": 0.6995,
+      "step": 2941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1604047986670299e-05,
+      "loss": 0.783,
+      "step": 2942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598991664977222e-05,
+      "loss": 0.8047,
+      "step": 2943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593934923718415e-05,
+      "loss": 0.3001,
+      "step": 2944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158887776422074e-05,
+      "loss": 0.7197,
+      "step": 2945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158382018781116e-05,
+      "loss": 0.7778,
+      "step": 2946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157876219581676e-05,
+      "loss": 0.8037,
+      "step": 2947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573703789564718e-05,
+      "loss": 0.848,
+      "step": 2948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568644970382334e-05,
+      "loss": 0.6741,
+      "step": 2949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563585739597012e-05,
+      "loss": 0.8308,
+      "step": 2950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558526098536258e-05,
+      "loss": 0.2719,
+      "step": 2951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553466048527697e-05,
+      "loss": 0.7488,
+      "step": 2952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548405590899054e-05,
+      "loss": 0.7885,
+      "step": 2953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154334472697816e-05,
+      "loss": 0.7694,
+      "step": 2954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538283458092958e-05,
+      "loss": 0.6873,
+      "step": 2955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533221785571495e-05,
+      "loss": 0.8083,
+      "step": 2956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152815971074192e-05,
+      "loss": 0.7822,
+      "step": 2957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523097234932494e-05,
+      "loss": 0.7717,
+      "step": 2958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518034359471577e-05,
+      "loss": 0.7427,
+      "step": 2959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512971085687637e-05,
+      "loss": 0.7593,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150790741490925e-05,
+      "loss": 0.7209,
+      "step": 2961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502843348465084e-05,
+      "loss": 0.7976,
+      "step": 2962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149777888768393e-05,
+      "loss": 0.8047,
+      "step": 2963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492714033894665e-05,
+      "loss": 0.728,
+      "step": 2964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148764878842628e-05,
+      "loss": 0.8411,
+      "step": 2965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482583152607865e-05,
+      "loss": 0.7667,
+      "step": 2966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477517127768608e-05,
+      "loss": 0.7715,
+      "step": 2967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472450715237804e-05,
+      "loss": 0.7545,
+      "step": 2968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467383916344852e-05,
+      "loss": 0.7524,
+      "step": 2969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462316732419248e-05,
+      "loss": 0.7908,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457249164790587e-05,
+      "loss": 0.7428,
+      "step": 2971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452181214788573e-05,
+      "loss": 0.7529,
+      "step": 2972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447112883743003e-05,
+      "loss": 0.7554,
+      "step": 2973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442044172983778e-05,
+      "loss": 0.7979,
+      "step": 2974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436975083840896e-05,
+      "loss": 0.8408,
+      "step": 2975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431905617644452e-05,
+      "loss": 0.8379,
+      "step": 2976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142683577572465e-05,
+      "loss": 0.7866,
+      "step": 2977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421765559411779e-05,
+      "loss": 0.8241,
+      "step": 2978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416694970036237e-05,
+      "loss": 0.7644,
+      "step": 2979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411624008928518e-05,
+      "loss": 0.7747,
+      "step": 2980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1406552677419205e-05,
+      "loss": 0.7642,
+      "step": 2981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140148097683899e-05,
+      "loss": 0.8008,
+      "step": 2982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396408908518652e-05,
+      "loss": 0.769,
+      "step": 2983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391336473789076e-05,
+      "loss": 0.8077,
+      "step": 2984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138626367398124e-05,
+      "loss": 0.7527,
+      "step": 2985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381190510426206e-05,
+      "loss": 0.7646,
+      "step": 2986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376116984455147e-05,
+      "loss": 0.7378,
+      "step": 2987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137104309739933e-05,
+      "loss": 0.7844,
+      "step": 2988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365968850590106e-05,
+      "loss": 0.8325,
+      "step": 2989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136089424535893e-05,
+      "loss": 0.7292,
+      "step": 2990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355819283037346e-05,
+      "loss": 0.8672,
+      "step": 2991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135074396495699e-05,
+      "loss": 0.2789,
+      "step": 2992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345668292449604e-05,
+      "loss": 0.7915,
+      "step": 2993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340592266847004e-05,
+      "loss": 0.7389,
+      "step": 2994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335515889481112e-05,
+      "loss": 0.8242,
+      "step": 2995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330439161683944e-05,
+      "loss": 0.8242,
+      "step": 2996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325362084787593e-05,
+      "loss": 0.7858,
+      "step": 2997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320284660124259e-05,
+      "loss": 0.7662,
+      "step": 2998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315206889026226e-05,
+      "loss": 0.8184,
+      "step": 2999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310128772825872e-05,
+      "loss": 0.8405,
+      "step": 3000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305050312855666e-05,
+      "loss": 0.8369,
+      "step": 3001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1299971510448157e-05,
+      "loss": 0.7458,
+      "step": 3002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1294892366935998e-05,
+      "loss": 0.8364,
+      "step": 3003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289812883651928e-05,
+      "loss": 0.7798,
+      "step": 3004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284733061928765e-05,
+      "loss": 0.8198,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279652903099429e-05,
+      "loss": 0.7808,
+      "step": 3006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274572408496921e-05,
+      "loss": 0.7421,
+      "step": 3007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269491579454334e-05,
+      "loss": 0.7505,
+      "step": 3008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264410417304848e-05,
+      "loss": 0.8467,
+      "step": 3009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259328923381721e-05,
+      "loss": 0.7517,
+      "step": 3010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254247099018312e-05,
+      "loss": 0.751,
+      "step": 3011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249164945548063e-05,
+      "loss": 0.8386,
+      "step": 3012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244082464304497e-05,
+      "loss": 0.835,
+      "step": 3013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238999656621226e-05,
+      "loss": 0.7744,
+      "step": 3014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123391652383195e-05,
+      "loss": 0.8428,
+      "step": 3015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122883306727045e-05,
+      "loss": 0.7937,
+      "step": 3016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12237492882706e-05,
+      "loss": 0.7595,
+      "step": 3017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218665188166344e-05,
+      "loss": 0.782,
+      "step": 3018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213580768291725e-05,
+      "loss": 0.735,
+      "step": 3019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208496029980864e-05,
+      "loss": 0.7168,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203410974567963e-05,
+      "loss": 0.7493,
+      "step": 3021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119832560338731e-05,
+      "loss": 0.7982,
+      "step": 3022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193239917773278e-05,
+      "loss": 0.7974,
+      "step": 3023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188153919060316e-05,
+      "loss": 0.7816,
+      "step": 3024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183067608582962e-05,
+      "loss": 0.7551,
+      "step": 3025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177980987675837e-05,
+      "loss": 0.778,
+      "step": 3026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1172894057673631e-05,
+      "loss": 0.7205,
+      "step": 3027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167806819911132e-05,
+      "loss": 0.8477,
+      "step": 3028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162719275723191e-05,
+      "loss": 0.7993,
+      "step": 3029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157631426444754e-05,
+      "loss": 0.7747,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152543273410848e-05,
+      "loss": 0.744,
+      "step": 3031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147454817956563e-05,
+      "loss": 0.7405,
+      "step": 3032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142366061417082e-05,
+      "loss": 0.8015,
+      "step": 3033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113727700512767e-05,
+      "loss": 0.7517,
+      "step": 3034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132187650423653e-05,
+      "loss": 0.7816,
+      "step": 3035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127097998640461e-05,
+      "loss": 0.7676,
+      "step": 3036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122008051113578e-05,
+      "loss": 0.7773,
+      "step": 3037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116917809178576e-05,
+      "loss": 0.7842,
+      "step": 3038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111827274171107e-05,
+      "loss": 0.8066,
+      "step": 3039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106736447426896e-05,
+      "loss": 0.8159,
+      "step": 3040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101645330281741e-05,
+      "loss": 0.269,
+      "step": 3041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096553924071527e-05,
+      "loss": 0.8102,
+      "step": 3042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091462230132202e-05,
+      "loss": 0.792,
+      "step": 3043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086370249799803e-05,
+      "loss": 0.8843,
+      "step": 3044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081277984410423e-05,
+      "loss": 0.7671,
+      "step": 3045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076185435300253e-05,
+      "loss": 0.7721,
+      "step": 3046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071092603805544e-05,
+      "loss": 0.7617,
+      "step": 3047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106599949126262e-05,
+      "loss": 0.2535,
+      "step": 3048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060906099007885e-05,
+      "loss": 0.79,
+      "step": 3049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055812428377819e-05,
+      "loss": 0.7688,
+      "step": 3050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050718480708962e-05,
+      "loss": 0.781,
+      "step": 3051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104562425733794e-05,
+      "loss": 0.7969,
+      "step": 3052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040529759601449e-05,
+      "loss": 0.6968,
+      "step": 3053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035434988836247e-05,
+      "loss": 0.3073,
+      "step": 3054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030339946379178e-05,
+      "loss": 0.7793,
+      "step": 3055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1025244633567147e-05,
+      "loss": 0.8169,
+      "step": 3056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020149051737132e-05,
+      "loss": 0.7362,
+      "step": 3057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015053202226188e-05,
+      "loss": 0.6844,
+      "step": 3058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009957086371425e-05,
+      "loss": 0.7561,
+      "step": 3059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004860705510043e-05,
+      "loss": 0.314,
+      "step": 3060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.09997640609793e-05,
+      "loss": 0.8372,
+      "step": 3061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994667154116521e-05,
+      "loss": 0.7426,
+      "step": 3062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0989569986259105e-05,
+      "loss": 0.8018,
+      "step": 3063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984472558744516e-05,
+      "loss": 0.2659,
+      "step": 3064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979374872910293e-05,
+      "loss": 0.8219,
+      "step": 3065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0974276930094035e-05,
+      "loss": 0.7642,
+      "step": 3066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969178731633407e-05,
+      "loss": 0.8098,
+      "step": 3067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964080278866154e-05,
+      "loss": 0.7986,
+      "step": 3068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958981573130074e-05,
+      "loss": 0.8369,
+      "step": 3069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953882615763035e-05,
+      "loss": 0.8577,
+      "step": 3070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0948783408102977e-05,
+      "loss": 0.7869,
+      "step": 3071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943683951487898e-05,
+      "loss": 0.7371,
+      "step": 3072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938584247255865e-05,
+      "loss": 0.7737,
+      "step": 3073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933484296745012e-05,
+      "loss": 0.7876,
+      "step": 3074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928384101293533e-05,
+      "loss": 0.7751,
+      "step": 3075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923283662239687e-05,
+      "loss": 0.7629,
+      "step": 3076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918182980921801e-05,
+      "loss": 0.7772,
+      "step": 3077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913082058678261e-05,
+      "loss": 0.8237,
+      "step": 3078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090798089684752e-05,
+      "loss": 0.8508,
+      "step": 3079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902879496768092e-05,
+      "loss": 0.807,
+      "step": 3080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897777859778552e-05,
+      "loss": 0.761,
+      "step": 3081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892675987217539e-05,
+      "loss": 0.8291,
+      "step": 3082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088757388042375e-05,
+      "loss": 0.7366,
+      "step": 3083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882471540735952e-05,
+      "loss": 0.8257,
+      "step": 3084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087736896949297e-05,
+      "loss": 0.7778,
+      "step": 3085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872266168033679e-05,
+      "loss": 0.739,
+      "step": 3086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086716313769703e-05,
+      "loss": 0.7776,
+      "step": 3087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862059879822026e-05,
+      "loss": 0.7542,
+      "step": 3088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856956395747731e-05,
+      "loss": 0.8031,
+      "step": 3089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085185268681327e-05,
+      "loss": 0.6908,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084674875435782e-05,
+      "loss": 0.7798,
+      "step": 3091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841644599720624e-05,
+      "loss": 0.8101,
+      "step": 3092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836540224240988e-05,
+      "loss": 0.7673,
+      "step": 3093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831435629258261e-05,
+      "loss": 0.8198,
+      "step": 3094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826330816111859e-05,
+      "loss": 0.7556,
+      "step": 3095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821225786141255e-05,
+      "loss": 0.8401,
+      "step": 3096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816120540685982e-05,
+      "loss": 0.7887,
+      "step": 3097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081101508108562e-05,
+      "loss": 0.8173,
+      "step": 3098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805909408679813e-05,
+      "loss": 0.7228,
+      "step": 3099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800803524808256e-05,
+      "loss": 0.7825,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079569743081071e-05,
+      "loss": 0.7563,
+      "step": 3101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790591128026973e-05,
+      "loss": 0.2889,
+      "step": 3102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078548461779691e-05,
+      "loss": 0.769,
+      "step": 3103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780377901460447e-05,
+      "loss": 0.7363,
+      "step": 3104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775270980357542e-05,
+      "loss": 0.8147,
+      "step": 3105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077016385582823e-05,
+      "loss": 0.7542,
+      "step": 3106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765056529212588e-05,
+      "loss": 0.7419,
+      "step": 3107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759949001850744e-05,
+      "loss": 0.7778,
+      "step": 3108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754841275082887e-05,
+      "loss": 0.6985,
+      "step": 3109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0749733350249252e-05,
+      "loss": 0.8125,
+      "step": 3110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744625228690124e-05,
+      "loss": 0.2551,
+      "step": 3111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073951691174585e-05,
+      "loss": 0.8142,
+      "step": 3112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073440840075681e-05,
+      "loss": 0.736,
+      "step": 3113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072929969706346e-05,
+      "loss": 0.7988,
+      "step": 3114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724190802006284e-05,
+      "loss": 0.8071,
+      "step": 3115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719081716925827e-05,
+      "loss": 0.7289,
+      "step": 3116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713972443162686e-05,
+      "loss": 0.7759,
+      "step": 3117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708862982057495e-05,
+      "loss": 0.7051,
+      "step": 3118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703753334950953e-05,
+      "loss": 0.7761,
+      "step": 3119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06986435031838e-05,
+      "loss": 0.2863,
+      "step": 3120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069353348809682e-05,
+      "loss": 0.8127,
+      "step": 3121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688423291030856e-05,
+      "loss": 0.746,
+      "step": 3122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683312913326787e-05,
+      "loss": 0.7732,
+      "step": 3123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678202356325549e-05,
+      "loss": 0.7068,
+      "step": 3124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673091621368123e-05,
+      "loss": 0.7766,
+      "step": 3125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066798070979553e-05,
+      "loss": 0.6818,
+      "step": 3126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0662869622948844e-05,
+      "loss": 0.8032,
+      "step": 3127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657758362169184e-05,
+      "loss": 0.7722,
+      "step": 3128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652646928797715e-05,
+      "loss": 0.8362,
+      "step": 3129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0647535324175643e-05,
+      "loss": 0.7468,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642423549644222e-05,
+      "loss": 0.7939,
+      "step": 3131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637311606544756e-05,
+      "loss": 0.7909,
+      "step": 3132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632199496218586e-05,
+      "loss": 0.8242,
+      "step": 3133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062708722000709e-05,
+      "loss": 0.7742,
+      "step": 3134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621974779251709e-05,
+      "loss": 0.7639,
+      "step": 3135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616862175293917e-05,
+      "loss": 0.793,
+      "step": 3136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061174940947522e-05,
+      "loss": 0.7637,
+      "step": 3137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606636483137184e-05,
+      "loss": 0.7039,
+      "step": 3138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601523397621414e-05,
+      "loss": 0.8406,
+      "step": 3139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0596410154269543e-05,
+      "loss": 0.2897,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059129675442326e-05,
+      "loss": 0.7422,
+      "step": 3141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586183199424289e-05,
+      "loss": 0.3002,
+      "step": 3142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05810694906144e-05,
+      "loss": 0.8127,
+      "step": 3143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575955629335395e-05,
+      "loss": 0.8315,
+      "step": 3144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057084161692912e-05,
+      "loss": 0.7639,
+      "step": 3145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056572745473746e-05,
+      "loss": 0.7693,
+      "step": 3146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560613144102347e-05,
+      "loss": 0.7726,
+      "step": 3147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555498686365737e-05,
+      "loss": 0.251,
+      "step": 3148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055038408286964e-05,
+      "loss": 0.7847,
+      "step": 3149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054526933495609e-05,
+      "loss": 0.7354,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540154443967168e-05,
+      "loss": 0.8525,
+      "step": 3151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535039411244994e-05,
+      "loss": 0.7312,
+      "step": 3152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529924238131717e-05,
+      "loss": 0.7797,
+      "step": 3153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524808925969528e-05,
+      "loss": 0.7762,
+      "step": 3154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519693476100658e-05,
+      "loss": 0.7122,
+      "step": 3155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514577889867367e-05,
+      "loss": 0.726,
+      "step": 3156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509462168611953e-05,
+      "loss": 0.735,
+      "step": 3157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504346313676754e-05,
+      "loss": 0.7961,
+      "step": 3158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499230326404132e-05,
+      "loss": 0.7595,
+      "step": 3159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494114208136502e-05,
+      "loss": 0.7866,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488997960216292e-05,
+      "loss": 0.8306,
+      "step": 3161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048388158398598e-05,
+      "loss": 0.8105,
+      "step": 3162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478765080788074e-05,
+      "loss": 0.8271,
+      "step": 3163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473648451965106e-05,
+      "loss": 0.7671,
+      "step": 3164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468531698859656e-05,
+      "loss": 0.2656,
+      "step": 3165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463414822814324e-05,
+      "loss": 0.7847,
+      "step": 3166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045829782517175e-05,
+      "loss": 0.7236,
+      "step": 3167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453180707274604e-05,
+      "loss": 0.7421,
+      "step": 3168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448063470465583e-05,
+      "loss": 0.7832,
+      "step": 3169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442946116087421e-05,
+      "loss": 0.6962,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043782864548288e-05,
+      "loss": 0.7844,
+      "step": 3171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432711059994759e-05,
+      "loss": 0.7922,
+      "step": 3172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427593360965871e-05,
+      "loss": 0.8093,
+      "step": 3173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422475549739078e-05,
+      "loss": 0.7922,
+      "step": 3174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417357627657259e-05,
+      "loss": 0.7732,
+      "step": 3175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412239596063323e-05,
+      "loss": 0.7634,
+      "step": 3176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407121456300216e-05,
+      "loss": 0.7306,
+      "step": 3177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402003209710905e-05,
+      "loss": 0.7375,
+      "step": 3178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396884857638388e-05,
+      "loss": 0.7991,
+      "step": 3179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391766401425685e-05,
+      "loss": 0.7297,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386647842415852e-05,
+      "loss": 0.832,
+      "step": 3181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381529181951969e-05,
+      "loss": 0.7783,
+      "step": 3182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376410421377136e-05,
+      "loss": 0.2859,
+      "step": 3183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371291562034493e-05,
+      "loss": 0.7886,
+      "step": 3184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366172605267188e-05,
+      "loss": 0.7703,
+      "step": 3185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361053552418411e-05,
+      "loss": 0.7593,
+      "step": 3186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355934404831369e-05,
+      "loss": 0.7448,
+      "step": 3187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350815163849295e-05,
+      "loss": 0.7502,
+      "step": 3188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345695830815446e-05,
+      "loss": 0.7927,
+      "step": 3189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340576407073106e-05,
+      "loss": 0.7764,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335456893965578e-05,
+      "loss": 0.7891,
+      "step": 3191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330337292836193e-05,
+      "loss": 0.772,
+      "step": 3192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0325217605028305e-05,
+      "loss": 0.7708,
+      "step": 3193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320097831885282e-05,
+      "loss": 0.7957,
+      "step": 3194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031497797475053e-05,
+      "loss": 0.8159,
+      "step": 3195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309858034967466e-05,
+      "loss": 0.8013,
+      "step": 3196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304738013879528e-05,
+      "loss": 0.7623,
+      "step": 3197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299617912830182e-05,
+      "loss": 0.7354,
+      "step": 3198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294497733162911e-05,
+      "loss": 0.8186,
+      "step": 3199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289377476221219e-05,
+      "loss": 0.7346,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284257143348629e-05,
+      "loss": 0.7526,
+      "step": 3201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279136735888684e-05,
+      "loss": 0.7124,
+      "step": 3202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274016255184953e-05,
+      "loss": 0.7844,
+      "step": 3203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268895702581016e-05,
+      "loss": 0.764,
+      "step": 3204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263775079420472e-05,
+      "loss": 0.7734,
+      "step": 3205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025865438704695e-05,
+      "loss": 0.7139,
+      "step": 3206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253533626804079e-05,
+      "loss": 0.7764,
+      "step": 3207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024841280003552e-05,
+      "loss": 0.7767,
+      "step": 3208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243291908084952e-05,
+      "loss": 0.7517,
+      "step": 3209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238170952296055e-05,
+      "loss": 0.2709,
+      "step": 3210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233049934012544e-05,
+      "loss": 0.8523,
+      "step": 3211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0227928854578141e-05,
+      "loss": 0.8186,
+      "step": 3212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0222807715336589e-05,
+      "loss": 0.7649,
+      "step": 3213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0217686517631645e-05,
+      "loss": 0.2636,
+      "step": 3214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021256526280707e-05,
+      "loss": 0.821,
+      "step": 3215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207443952206664e-05,
+      "loss": 0.8215,
+      "step": 3216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202322587174219e-05,
+      "loss": 0.7767,
+      "step": 3217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019720116905355e-05,
+      "loss": 0.8093,
+      "step": 3218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019207969918849e-05,
+      "loss": 0.8198,
+      "step": 3219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018695817892288e-05,
+      "loss": 0.7964,
+      "step": 3220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181836609600577e-05,
+      "loss": 0.7616,
+      "step": 3221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176714992565451e-05,
+      "loss": 0.7369,
+      "step": 3222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171593329161378e-05,
+      "loss": 0.7729,
+      "step": 3223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166471620732257e-05,
+      "loss": 0.8013,
+      "step": 3224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016134986862199e-05,
+      "loss": 0.6947,
+      "step": 3225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156228074174495e-05,
+      "loss": 0.7981,
+      "step": 3226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151106238733697e-05,
+      "loss": 0.7595,
+      "step": 3227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014598436364354e-05,
+      "loss": 0.7144,
+      "step": 3228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014086245024797e-05,
+      "loss": 0.8334,
+      "step": 3229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135740499890946e-05,
+      "loss": 0.2872,
+      "step": 3230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130618513916433e-05,
+      "loss": 0.7311,
+      "step": 3231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0125496493668417e-05,
+      "loss": 0.7842,
+      "step": 3232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120374440490881e-05,
+      "loss": 0.7712,
+      "step": 3233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011525235572782e-05,
+      "loss": 0.8071,
+      "step": 3234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110130240723242e-05,
+      "loss": 0.7659,
+      "step": 3235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105008096821152e-05,
+      "loss": 0.7769,
+      "step": 3236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099885925365574e-05,
+      "loss": 0.8062,
+      "step": 3237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094763727700536e-05,
+      "loss": 0.7188,
+      "step": 3238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089641505170066e-05,
+      "loss": 0.8057,
+      "step": 3239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008451925911821e-05,
+      "loss": 0.7426,
+      "step": 3240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079396990889014e-05,
+      "loss": 0.7538,
+      "step": 3241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074274701826528e-05,
+      "loss": 0.8237,
+      "step": 3242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069152393274808e-05,
+      "loss": 0.8047,
+      "step": 3243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006403006657792e-05,
+      "loss": 0.7249,
+      "step": 3244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058907723079932e-05,
+      "loss": 0.7096,
+      "step": 3245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053785364124913e-05,
+      "loss": 0.749,
+      "step": 3246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004866299105694e-05,
+      "loss": 0.8047,
+      "step": 3247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004354060522009e-05,
+      "loss": 0.764,
+      "step": 3248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038418207958453e-05,
+      "loss": 0.7449,
+      "step": 3249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003329580061611e-05,
+      "loss": 0.744,
+      "step": 3250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002817338453715e-05,
+      "loss": 0.7148,
+      "step": 3251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023050961065663e-05,
+      "loss": 0.749,
+      "step": 3252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0017928531545747e-05,
+      "loss": 0.7505,
+      "step": 3253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001280609732149e-05,
+      "loss": 0.7722,
+      "step": 3254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000768365973699e-05,
+      "loss": 0.6855,
+      "step": 3255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002561220136346e-05,
+      "loss": 0.7704,
+      "step": 3256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997438779863655e-06,
+      "loss": 0.749,
+      "step": 3257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992316340263011e-06,
+      "loss": 0.7413,
+      "step": 3258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987193902678511e-06,
+      "loss": 0.8025,
+      "step": 3259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982071468454258e-06,
+      "loss": 0.7595,
+      "step": 3260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976949038934339e-06,
+      "loss": 0.7585,
+      "step": 3261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971826615462853e-06,
+      "loss": 0.8041,
+      "step": 3262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966704199383893e-06,
+      "loss": 0.7954,
+      "step": 3263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96158179204155e-06,
+      "loss": 0.7793,
+      "step": 3264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956459394779913e-06,
+      "loss": 0.7589,
+      "step": 3265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951337008943066e-06,
+      "loss": 0.7821,
+      "step": 3266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94621463587509e-06,
+      "loss": 0.7834,
+      "step": 3267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941092276920073e-06,
+      "loss": 0.7356,
+      "step": 3268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.935969933422084e-06,
+      "loss": 0.7874,
+      "step": 3269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930847606725194e-06,
+      "loss": 0.7461,
+      "step": 3270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925725298173477e-06,
+      "loss": 0.7942,
+      "step": 3271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92060300911099e-06,
+      "loss": 0.7025,
+      "step": 3272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915480740881793e-06,
+      "loss": 0.7515,
+      "step": 3273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910358494829939e-06,
+      "loss": 0.7686,
+      "step": 3274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905236272299468e-06,
+      "loss": 0.6604,
+      "step": 3275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900114074634426e-06,
+      "loss": 0.7937,
+      "step": 3276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89499190317885e-06,
+      "loss": 0.8057,
+      "step": 3277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889869759276763e-06,
+      "loss": 0.8203,
+      "step": 3278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88474764427218e-06,
+      "loss": 0.7671,
+      "step": 3279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879625559509119e-06,
+      "loss": 0.7723,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.874503506331584e-06,
+      "loss": 0.7617,
+      "step": 3281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869381486083567e-06,
+      "loss": 0.717,
+      "step": 3282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864259500109056e-06,
+      "loss": 0.7898,
+      "step": 3283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859137549752033e-06,
+      "loss": 0.7607,
+      "step": 3284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854015636356462e-06,
+      "loss": 0.7615,
+      "step": 3285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848893761266307e-06,
+      "loss": 0.7805,
+      "step": 3286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84377192582551e-06,
+      "loss": 0.6989,
+      "step": 3287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838650131378013e-06,
+      "loss": 0.8096,
+      "step": 3288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833528379267747e-06,
+      "loss": 0.2966,
+      "step": 3289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828406670838624e-06,
+      "loss": 0.7727,
+      "step": 3290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823285007434552e-06,
+      "loss": 0.7795,
+      "step": 3291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818163390399424e-06,
+      "loss": 0.7294,
+      "step": 3292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813041821077121e-06,
+      "loss": 0.733,
+      "step": 3293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807920300811511e-06,
+      "loss": 0.74,
+      "step": 3294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802798830946453e-06,
+      "loss": 0.7251,
+      "step": 3295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797677412825786e-06,
+      "loss": 0.6844,
+      "step": 3296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792556047793339e-06,
+      "loss": 0.7762,
+      "step": 3297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787434737192932e-06,
+      "loss": 0.7737,
+      "step": 3298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78231348236836e-06,
+      "loss": 0.8004,
+      "step": 3299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777192284663415e-06,
+      "loss": 0.7773,
+      "step": 3300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77207114542186e-06,
+      "loss": 0.813,
+      "step": 3301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766950065987458e-06,
+      "loss": 0.8513,
+      "step": 3302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761829047703948e-06,
+      "loss": 0.7275,
+      "step": 3303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756708091915053e-06,
+      "loss": 0.777,
+      "step": 3304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751587199964481e-06,
+      "loss": 0.7302,
+      "step": 3305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746466373195926e-06,
+      "loss": 0.7224,
+      "step": 3306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741345612953055e-06,
+      "loss": 0.7898,
+      "step": 3307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73622492057953e-06,
+      "loss": 0.7781,
+      "step": 3308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731104297418989e-06,
+      "loss": 0.7279,
+      "step": 3309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725983744815049e-06,
+      "loss": 0.845,
+      "step": 3310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720863264111316e-06,
+      "loss": 0.7712,
+      "step": 3311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715742856651373e-06,
+      "loss": 0.7419,
+      "step": 3312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710622523778783e-06,
+      "loss": 0.748,
+      "step": 3313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70550226683709e-06,
+      "loss": 0.7396,
+      "step": 3314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700382087169818e-06,
+      "loss": 0.8025,
+      "step": 3315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695261986120474e-06,
+      "loss": 0.7971,
+      "step": 3316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690141965032536e-06,
+      "loss": 0.7335,
+      "step": 3317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68502202524947e-06,
+      "loss": 0.7731,
+      "step": 3318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.67990216811472e-06,
+      "loss": 0.7714,
+      "step": 3319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.674782394971699e-06,
+      "loss": 0.748,
+      "step": 3320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66966270716381e-06,
+      "loss": 0.8055,
+      "step": 3321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664543106034425e-06,
+      "loss": 0.7771,
+      "step": 3322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659423592926896e-06,
+      "loss": 0.7986,
+      "step": 3323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654304169184558e-06,
+      "loss": 0.7677,
+      "step": 3324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649184836150709e-06,
+      "loss": 0.7555,
+      "step": 3325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644065595168633e-06,
+      "loss": 0.7737,
+      "step": 3326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638946447581592e-06,
+      "loss": 0.7716,
+      "step": 3327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633827394732815e-06,
+      "loss": 0.3019,
+      "step": 3328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62870843796551e-06,
+      "loss": 0.8042,
+      "step": 3329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623589578622865e-06,
+      "loss": 0.8306,
+      "step": 3330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618470818048033e-06,
+      "loss": 0.791,
+      "step": 3331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61335215758415e-06,
+      "loss": 0.7379,
+      "step": 3332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608233598574319e-06,
+      "loss": 0.7852,
+      "step": 3333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603115142361616e-06,
+      "loss": 0.7876,
+      "step": 3334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5979967902891e-06,
+      "loss": 0.77,
+      "step": 3335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592878543699787e-06,
+      "loss": 0.7224,
+      "step": 3336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587760403936679e-06,
+      "loss": 0.3079,
+      "step": 3337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582642372342748e-06,
+      "loss": 0.8208,
+      "step": 3338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577524450260927e-06,
+      "loss": 0.7296,
+      "step": 3339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572406639034132e-06,
+      "loss": 0.7678,
+      "step": 3340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567288940005248e-06,
+      "loss": 0.7224,
+      "step": 3341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562171354517123e-06,
+      "loss": 0.7213,
+      "step": 3342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557053883912582e-06,
+      "loss": 0.7507,
+      "step": 3343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551936529534422e-06,
+      "loss": 0.7218,
+      "step": 3344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546819292725401e-06,
+      "loss": 0.7473,
+      "step": 3345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54170217482825e-06,
+      "loss": 0.8132,
+      "step": 3346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536585177185676e-06,
+      "loss": 0.8181,
+      "step": 3347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531468301140346e-06,
+      "loss": 0.8286,
+      "step": 3348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526351548034894e-06,
+      "loss": 0.7468,
+      "step": 3349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521234919211927e-06,
+      "loss": 0.7534,
+      "step": 3350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516118416014021e-06,
+      "loss": 0.6831,
+      "step": 3351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511002039783709e-06,
+      "loss": 0.7363,
+      "step": 3352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5058857918635e-06,
+      "loss": 0.726,
+      "step": 3353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50076967359587e-06,
+      "loss": 0.7545,
+      "step": 3354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49565368632325e-06,
+      "loss": 0.7668,
+      "step": 3355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490537831388048e-06,
+      "loss": 0.7175,
+      "step": 3356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485422110132636e-06,
+      "loss": 0.7986,
+      "step": 3357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480306523899344e-06,
+      "loss": 0.7788,
+      "step": 3358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475191074030474e-06,
+      "loss": 0.7991,
+      "step": 3359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470075761868287e-06,
+      "loss": 0.781,
+      "step": 3360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46496058875501e-06,
+      "loss": 0.7419,
+      "step": 3361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459845556032835e-06,
+      "loss": 0.8148,
+      "step": 3362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454730665043914e-06,
+      "loss": 0.7941,
+      "step": 3363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449615917130364e-06,
+      "loss": 0.7124,
+      "step": 3364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444501313634266e-06,
+      "loss": 0.8096,
+      "step": 3365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439386855897655e-06,
+      "loss": 0.8052,
+      "step": 3366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434272545262542e-06,
+      "loss": 0.7759,
+      "step": 3367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.429158383070884e-06,
+      "loss": 0.7524,
+      "step": 3368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42404437066461e-06,
+      "loss": 0.7734,
+      "step": 3369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418930509385603e-06,
+      "loss": 0.7266,
+      "step": 3370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413816800575713e-06,
+      "loss": 0.2704,
+      "step": 3371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408703245576742e-06,
+      "loss": 0.7517,
+      "step": 3372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403589845730462e-06,
+      "loss": 0.7767,
+      "step": 3373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398476602378591e-06,
+      "loss": 0.7355,
+      "step": 3374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393363516862817e-06,
+      "loss": 0.8308,
+      "step": 3375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388250590524785e-06,
+      "loss": 0.7256,
+      "step": 3376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383137824706088e-06,
+      "loss": 0.7295,
+      "step": 3377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378025220748293e-06,
+      "loss": 0.7333,
+      "step": 3378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372912779992914e-06,
+      "loss": 0.741,
+      "step": 3379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367800503781421e-06,
+      "loss": 0.7715,
+      "step": 3380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362688393455249e-06,
+      "loss": 0.8257,
+      "step": 3381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357576450355776e-06,
+      "loss": 0.7578,
+      "step": 3382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35246467582436e-06,
+      "loss": 0.2656,
+      "step": 3383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347353071202287e-06,
+      "loss": 0.8022,
+      "step": 3384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342241637830816e-06,
+      "loss": 0.8027,
+      "step": 3385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337130377051158e-06,
+      "loss": 0.7141,
+      "step": 3386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332019290204472e-06,
+      "loss": 0.752,
+      "step": 3387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326908378631878e-06,
+      "loss": 0.8132,
+      "step": 3388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321797643674451e-06,
+      "loss": 0.8252,
+      "step": 3389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316687086673213e-06,
+      "loss": 0.7871,
+      "step": 3390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311576708969149e-06,
+      "loss": 0.7991,
+      "step": 3391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306466511903181e-06,
+      "loss": 0.7361,
+      "step": 3392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301356496816203e-06,
+      "loss": 0.7494,
+      "step": 3393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296246665049049e-06,
+      "loss": 0.78,
+      "step": 3394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291137017942507e-06,
+      "loss": 0.7002,
+      "step": 3395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286027556837318e-06,
+      "loss": 0.7455,
+      "step": 3396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280918283074176e-06,
+      "loss": 0.7161,
+      "step": 3397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275809197993719e-06,
+      "loss": 0.2701,
+      "step": 3398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270700302936542e-06,
+      "loss": 0.8281,
+      "step": 3399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265591599243191e-06,
+      "loss": 0.7771,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260483088254156e-06,
+      "loss": 0.2913,
+      "step": 3401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255374771309879e-06,
+      "loss": 0.7917,
+      "step": 3402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.250266649750751e-06,
+      "loss": 0.7275,
+      "step": 3403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245158724917115e-06,
+      "loss": 0.7277,
+      "step": 3404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240050998149258e-06,
+      "loss": 0.7703,
+      "step": 3405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234943470787415e-06,
+      "loss": 0.7515,
+      "step": 3406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229836144171771e-06,
+      "loss": 0.2587,
+      "step": 3407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224729019642461e-06,
+      "loss": 0.7866,
+      "step": 3408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219622098539558e-06,
+      "loss": 0.7605,
+      "step": 3409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214515382203091e-06,
+      "loss": 0.8262,
+      "step": 3410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209408871973033e-06,
+      "loss": 0.7539,
+      "step": 3411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204302569189296e-06,
+      "loss": 0.7073,
+      "step": 3412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199196475191745e-06,
+      "loss": 0.7822,
+      "step": 3413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194090591320192e-06,
+      "loss": 0.7598,
+      "step": 3414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188984918914383e-06,
+      "loss": 0.8237,
+      "step": 3415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183879459314023e-06,
+      "loss": 0.6594,
+      "step": 3416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178774213858743e-06,
+      "loss": 0.7664,
+      "step": 3417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173669183888142e-06,
+      "loss": 0.2566,
+      "step": 3418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16856437074174e-06,
+      "loss": 0.7771,
+      "step": 3419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163459775759013e-06,
+      "loss": 0.8088,
+      "step": 3420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158355400279375e-06,
+      "loss": 0.7754,
+      "step": 3421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15325124564218e-06,
+      "loss": 0.7844,
+      "step": 3422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148147313186734e-06,
+      "loss": 0.7654,
+      "step": 3423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14304360425227e-06,
+      "loss": 0.7581,
+      "step": 3424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137940120177975e-06,
+      "loss": 0.8232,
+      "step": 3425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132836862302973e-06,
+      "loss": 0.7611,
+      "step": 3426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127733831966323e-06,
+      "loss": 0.7271,
+      "step": 3427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122631030507035e-06,
+      "loss": 0.7703,
+      "step": 3428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117528459264051e-06,
+      "loss": 0.7399,
+      "step": 3429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112426119576253e-06,
+      "loss": 0.7903,
+      "step": 3430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107324012782466e-06,
+      "loss": 0.7844,
+      "step": 3431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102222140221454e-06,
+      "loss": 0.2966,
+      "step": 3432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097120503231911e-06,
+      "loss": 0.7512,
+      "step": 3433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092019103152481e-06,
+      "loss": 0.8167,
+      "step": 3434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086917941321742e-06,
+      "loss": 0.7207,
+      "step": 3435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0818170190782e-06,
+      "loss": 0.7861,
+      "step": 3436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076716337760316e-06,
+      "loss": 0.8411,
+      "step": 3437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07161589870647e-06,
+      "loss": 0.7878,
+      "step": 3438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06651570325499e-06,
+      "loss": 0.7852,
+      "step": 3439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061415752744138e-06,
+      "loss": 0.8164,
+      "step": 3440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056316048512105e-06,
+      "loss": 0.7391,
+      "step": 3441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051216591897026e-06,
+      "loss": 0.7229,
+      "step": 3442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046117384236969e-06,
+      "loss": 0.7754,
+      "step": 3443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04101842686993e-06,
+      "loss": 0.7976,
+      "step": 3444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03591972113385e-06,
+      "loss": 0.771,
+      "step": 3445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.030821268366596e-06,
+      "loss": 0.7839,
+      "step": 3446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02572306990597e-06,
+      "loss": 0.7368,
+      "step": 3447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020625127089713e-06,
+      "loss": 0.7667,
+      "step": 3448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015527441255487e-06,
+      "loss": 0.834,
+      "step": 3449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.0104300137409e-06,
+      "loss": 0.7412,
+      "step": 3450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005332845883484e-06,
+      "loss": 0.6985,
+      "step": 3451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000235939020705e-06,
+      "loss": 0.7369,
+      "step": 3452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995139294489957e-06,
+      "loss": 0.7671,
+      "step": 3453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990042913628573e-06,
+      "loss": 0.7942,
+      "step": 3454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984946797773814e-06,
+      "loss": 0.7783,
+      "step": 3455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97985094826287e-06,
+      "loss": 0.7573,
+      "step": 3456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974755366432854e-06,
+      "loss": 0.7903,
+      "step": 3457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969660053620825e-06,
+      "loss": 0.6625,
+      "step": 3458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964565011163753e-06,
+      "loss": 0.736,
+      "step": 3459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959470240398553e-06,
+      "loss": 0.7385,
+      "step": 3460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954375742662061e-06,
+      "loss": 0.7896,
+      "step": 3461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94928151929104e-06,
+      "loss": 0.8291,
+      "step": 3462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944187571622184e-06,
+      "loss": 0.7015,
+      "step": 3463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939093900992117e-06,
+      "loss": 0.7812,
+      "step": 3464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934000508737385e-06,
+      "loss": 0.8115,
+      "step": 3465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92890739619446e-06,
+      "loss": 0.2729,
+      "step": 3466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92381456469975e-06,
+      "loss": 0.7798,
+      "step": 3467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918722015589579e-06,
+      "loss": 0.7759,
+      "step": 3468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913629750200202e-06,
+      "loss": 0.7896,
+      "step": 3469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9085377698678e-06,
+      "loss": 0.7224,
+      "step": 3470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903446075928477e-06,
+      "loss": 0.7196,
+      "step": 3471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89835466971826e-06,
+      "loss": 0.7139,
+      "step": 3472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893263552573108e-06,
+      "loss": 0.7836,
+      "step": 3473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888172725828895e-06,
+      "loss": 0.6976,
+      "step": 3474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883082190821428e-06,
+      "loss": 0.7288,
+      "step": 3475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877991948886427e-06,
+      "loss": 0.7623,
+      "step": 3476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872902001359542e-06,
+      "loss": 0.7686,
+      "step": 3477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867812349576348e-06,
+      "loss": 0.7253,
+      "step": 3478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862722994872336e-06,
+      "loss": 0.7743,
+      "step": 3479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85763393858292e-06,
+      "loss": 0.8022,
+      "step": 3480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852545182043442e-06,
+      "loss": 0.7438,
+      "step": 3481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847456726589158e-06,
+      "loss": 0.7751,
+      "step": 3482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84236857355525e-06,
+      "loss": 0.7545,
+      "step": 3483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837280724276814e-06,
+      "loss": 0.7383,
+      "step": 3484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832193180088875e-06,
+      "loss": 0.7993,
+      "step": 3485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.827105942326374e-06,
+      "loss": 0.7737,
+      "step": 3486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822019012324168e-06,
+      "loss": 0.6934,
+      "step": 3487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81693239141704e-06,
+      "loss": 0.8389,
+      "step": 3488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811846080939685e-06,
+      "loss": 0.7444,
+      "step": 3489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806760082226724e-06,
+      "loss": 0.7478,
+      "step": 3490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801674396612691e-06,
+      "loss": 0.7458,
+      "step": 3491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796589025432037e-06,
+      "loss": 0.6641,
+      "step": 3492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791503970019138e-06,
+      "loss": 0.7568,
+      "step": 3493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786419231708275e-06,
+      "loss": 0.7478,
+      "step": 3494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781334811833654e-06,
+      "loss": 0.7952,
+      "step": 3495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776250711729403e-06,
+      "loss": 0.7517,
+      "step": 3496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77116693272955e-06,
+      "loss": 0.7407,
+      "step": 3497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766083476168051e-06,
+      "loss": 0.7532,
+      "step": 3498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761000343378776e-06,
+      "loss": 0.7493,
+      "step": 3499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755917535695506e-06,
+      "loss": 0.7437,
+      "step": 3500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75083505445194e-06,
+      "loss": 0.8102,
+      "step": 3501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74575290098169e-06,
+      "loss": 0.7056,
+      "step": 3502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740671076618282e-06,
+      "loss": 0.7605,
+      "step": 3503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735589582695157e-06,
+      "loss": 0.7434,
+      "step": 3504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730508420545667e-06,
+      "loss": 0.8101,
+      "step": 3505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72542759150308e-06,
+      "loss": 0.7842,
+      "step": 3506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720347096900575e-06,
+      "loss": 0.3044,
+      "step": 3507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715266938071237e-06,
+      "loss": 0.7932,
+      "step": 3508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710187116348076e-06,
+      "loss": 0.7483,
+      "step": 3509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705107633064003e-06,
+      "loss": 0.79,
+      "step": 3510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700028489551846e-06,
+      "loss": 0.75,
+      "step": 3511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694949687144338e-06,
+      "loss": 0.7937,
+      "step": 3512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68987122717413e-06,
+      "loss": 0.7434,
+      "step": 3513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684793110973775e-06,
+      "loss": 0.7537,
+      "step": 3514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679715339875743e-06,
+      "loss": 0.7655,
+      "step": 3515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67463791521241e-06,
+      "loss": 0.7509,
+      "step": 3516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669560838316063e-06,
+      "loss": 0.8235,
+      "step": 3517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664484110518893e-06,
+      "loss": 0.7318,
+      "step": 3518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659407733153001e-06,
+      "loss": 0.7424,
+      "step": 3519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654331707550401e-06,
+      "loss": 0.7681,
+      "step": 3520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649256035043015e-06,
+      "loss": 0.7825,
+      "step": 3521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64418071696266e-06,
+      "loss": 0.8096,
+      "step": 3522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639105754641076e-06,
+      "loss": 0.7627,
+      "step": 3523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634031149409894e-06,
+      "loss": 0.2664,
+      "step": 3524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62895690260067e-06,
+      "loss": 0.7834,
+      "step": 3525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623883015544851e-06,
+      "loss": 0.7805,
+      "step": 3526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618809489573796e-06,
+      "loss": 0.824,
+      "step": 3527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613736326018765e-06,
+      "loss": 0.8011,
+      "step": 3528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608663526210923e-06,
+      "loss": 0.8518,
+      "step": 3529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603591091481348e-06,
+      "loss": 0.7806,
+      "step": 3530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598519023161012e-06,
+      "loss": 0.717,
+      "step": 3531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593447322580797e-06,
+      "loss": 0.7556,
+      "step": 3532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588375991071485e-06,
+      "loss": 0.78,
+      "step": 3533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583305029963765e-06,
+      "loss": 0.8325,
+      "step": 3534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578234440588223e-06,
+      "loss": 0.7109,
+      "step": 3535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573164224275352e-06,
+      "loss": 0.7708,
+      "step": 3536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56809438235555e-06,
+      "loss": 0.7673,
+      "step": 3537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563024916159108e-06,
+      "loss": 0.6827,
+      "step": 3538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557955827016226e-06,
+      "loss": 0.7384,
+      "step": 3539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552887116257e-06,
+      "loss": 0.7726,
+      "step": 3540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54781878521143e-06,
+      "loss": 0.7578,
+      "step": 3541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542750835209416e-06,
+      "loss": 0.7952,
+      "step": 3542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537683267580757e-06,
+      "loss": 0.7546,
+      "step": 3543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532616083655151e-06,
+      "loss": 0.2668,
+      "step": 3544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5275492847622e-06,
+      "loss": 0.7433,
+      "step": 3545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522482872231396e-06,
+      "loss": 0.2594,
+      "step": 3546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517416847392138e-06,
+      "loss": 0.783,
+      "step": 3547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512351211573722e-06,
+      "loss": 0.7584,
+      "step": 3548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507285966105337e-06,
+      "loss": 0.8184,
+      "step": 3549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502221112316073e-06,
+      "loss": 0.7782,
+      "step": 3550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49715665153492e-06,
+      "loss": 0.7634,
+      "step": 3551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.492092585090757e-06,
+      "loss": 0.7772,
+      "step": 3552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48702891431237e-06,
+      "loss": 0.7146,
+      "step": 3553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48196564052843e-06,
+      "loss": 0.672,
+      "step": 3554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476902765067511e-06,
+      "loss": 0.7485,
+      "step": 3555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471840289258084e-06,
+      "loss": 0.7363,
+      "step": 3556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46677821442851e-06,
+      "loss": 0.7964,
+      "step": 3557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461716541907044e-06,
+      "loss": 0.725,
+      "step": 3558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456655273021843e-06,
+      "loss": 0.6836,
+      "step": 3559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451594409100946e-06,
+      "loss": 0.7404,
+      "step": 3560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446533951472304e-06,
+      "loss": 0.7804,
+      "step": 3561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441473901463743e-06,
+      "loss": 0.8102,
+      "step": 3562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436414260402993e-06,
+      "loss": 0.7419,
+      "step": 3563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431355029617668e-06,
+      "loss": 0.7854,
+      "step": 3564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426296210435281e-06,
+      "loss": 0.7542,
+      "step": 3565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421237804183242e-06,
+      "loss": 0.812,
+      "step": 3566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41617981218884e-06,
+      "loss": 0.7734,
+      "step": 3567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411122235779261e-06,
+      "loss": 0.7314,
+      "step": 3568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406065076281587e-06,
+      "loss": 0.2776,
+      "step": 3569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401008335022781e-06,
+      "loss": 0.7441,
+      "step": 3570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395952013329704e-06,
+      "loss": 0.7666,
+      "step": 3571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390896112529105e-06,
+      "loss": 0.7303,
+      "step": 3572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38584063394762e-06,
+      "loss": 0.746,
+      "step": 3573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380785578911776e-06,
+      "loss": 0.7717,
+      "step": 3574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37573094874799e-06,
+      "loss": 0.8345,
+      "step": 3575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370676744782562e-06,
+      "loss": 0.7649,
+      "step": 3576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365622968341694e-06,
+      "loss": 0.7866,
+      "step": 3577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.360569620751456e-06,
+      "loss": 0.757,
+      "step": 3578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355516703337821e-06,
+      "loss": 0.7443,
+      "step": 3579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350464217426646e-06,
+      "loss": 0.7991,
+      "step": 3580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345412164343668e-06,
+      "loss": 0.8406,
+      "step": 3581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340360545414516e-06,
+      "loss": 0.7324,
+      "step": 3582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335309361964708e-06,
+      "loss": 0.7079,
+      "step": 3583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330258615319638e-06,
+      "loss": 0.8052,
+      "step": 3584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325208306804595e-06,
+      "loss": 0.8037,
+      "step": 3585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32015843774475e-06,
+      "loss": 0.7576,
+      "step": 3586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315109009465157e-06,
+      "loss": 0.783,
+      "step": 3587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310060023290756e-06,
+      "loss": 0.7178,
+      "step": 3588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305011480546368e-06,
+      "loss": 0.8264,
+      "step": 3589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299963382556702e-06,
+      "loss": 0.813,
+      "step": 3590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29491573064635e-06,
+      "loss": 0.2668,
+      "step": 3591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289868526139781e-06,
+      "loss": 0.7767,
+      "step": 3592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284821770361353e-06,
+      "loss": 0.7466,
+      "step": 3593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279775464635307e-06,
+      "loss": 0.7332,
+      "step": 3594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27472961028576e-06,
+      "loss": 0.7556,
+      "step": 3595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26968420863671e-06,
+      "loss": 0.7661,
+      "step": 3596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264639261012044e-06,
+      "loss": 0.8103,
+      "step": 3597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259594768735532e-06,
+      "loss": 0.782,
+      "step": 3598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254550733130808e-06,
+      "loss": 0.7764,
+      "step": 3599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249507155521399e-06,
+      "loss": 0.7788,
+      "step": 3600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244464037230715e-06,
+      "loss": 0.7399,
+      "step": 3601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23942137958203e-06,
+      "loss": 0.2621,
+      "step": 3602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234379183898515e-06,
+      "loss": 0.7693,
+      "step": 3603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22933745150321e-06,
+      "loss": 0.7791,
+      "step": 3604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224296183719031e-06,
+      "loss": 0.8064,
+      "step": 3605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21925538186878e-06,
+      "loss": 0.7711,
+      "step": 3606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21421504727513e-06,
+      "loss": 0.8154,
+      "step": 3607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209175181260639e-06,
+      "loss": 0.728,
+      "step": 3608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204135785147733e-06,
+      "loss": 0.2704,
+      "step": 3609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19909686025872e-06,
+      "loss": 0.7279,
+      "step": 3610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194058407915782e-06,
+      "loss": 0.7867,
+      "step": 3611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189020429440983e-06,
+      "loss": 0.7485,
+      "step": 3612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183982926156254e-06,
+      "loss": 0.2697,
+      "step": 3613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178945899383406e-06,
+      "loss": 0.7451,
+      "step": 3614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173909350444125e-06,
+      "loss": 0.6829,
+      "step": 3615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16887328065997e-06,
+      "loss": 0.777,
+      "step": 3616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163837691352375e-06,
+      "loss": 0.734,
+      "step": 3617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158802583842652e-06,
+      "loss": 0.7959,
+      "step": 3618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153767959451978e-06,
+      "loss": 0.7812,
+      "step": 3619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14873381950141e-06,
+      "loss": 0.7667,
+      "step": 3620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143700165311876e-06,
+      "loss": 0.752,
+      "step": 3621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138666998204174e-06,
+      "loss": 0.735,
+      "step": 3622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13363431949898e-06,
+      "loss": 0.8159,
+      "step": 3623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128602130516836e-06,
+      "loss": 0.7476,
+      "step": 3624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123570432578158e-06,
+      "loss": 0.6918,
+      "step": 3625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118539227003238e-06,
+      "loss": 0.7433,
+      "step": 3626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113508515112224e-06,
+      "loss": 0.8357,
+      "step": 3627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108478298225153e-06,
+      "loss": 0.6952,
+      "step": 3628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103448577661922e-06,
+      "loss": 0.7434,
+      "step": 3629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098419354742295e-06,
+      "loss": 0.7405,
+      "step": 3630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09339063078591e-06,
+      "loss": 0.7048,
+      "step": 3631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08836240711228e-06,
+      "loss": 0.7616,
+      "step": 3632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083334685040778e-06,
+      "loss": 0.707,
+      "step": 3633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078307465890643e-06,
+      "loss": 0.8135,
+      "step": 3634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073280750980991e-06,
+      "loss": 0.6786,
+      "step": 3635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068254541630803e-06,
+      "loss": 0.7566,
+      "step": 3636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063228839158921e-06,
+      "loss": 0.7585,
+      "step": 3637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05820364488406e-06,
+      "loss": 0.7271,
+      "step": 3638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053178960124806e-06,
+      "loss": 0.6958,
+      "step": 3639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0481547861996e-06,
+      "loss": 0.7546,
+      "step": 3640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043131124426755e-06,
+      "loss": 0.7162,
+      "step": 3641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038107976124456e-06,
+      "loss": 0.7196,
+      "step": 3642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033085342610738e-06,
+      "loss": 0.7407,
+      "step": 3643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028063225203515e-06,
+      "loss": 0.7822,
+      "step": 3644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023041625220558e-06,
+      "loss": 0.7605,
+      "step": 3645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018020543979504e-06,
+      "loss": 0.8093,
+      "step": 3646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012999982797858e-06,
+      "loss": 0.8318,
+      "step": 3647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007979942992979e-06,
+      "loss": 0.7401,
+      "step": 3648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0029604258821e-06,
+      "loss": 0.7463,
+      "step": 3649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99794143278231e-06,
+      "loss": 0.8101,
+      "step": 3650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992922965010562e-06,
+      "loss": 0.725,
+      "step": 3651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987905023883672e-06,
+      "loss": 0.7375,
+      "step": 3652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98288761071832e-06,
+      "loss": 0.6976,
+      "step": 3653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977870726831039e-06,
+      "loss": 0.7488,
+      "step": 3654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972854373538235e-06,
+      "loss": 0.7666,
+      "step": 3655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967838552156163e-06,
+      "loss": 0.8237,
+      "step": 3656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962823264000948e-06,
+      "loss": 0.7378,
+      "step": 3657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957808510388571e-06,
+      "loss": 0.7705,
+      "step": 3658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952794292634872e-06,
+      "loss": 0.7104,
+      "step": 3659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947780612055551e-06,
+      "loss": 0.8103,
+      "step": 3660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942767469966169e-06,
+      "loss": 0.7104,
+      "step": 3661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937754867682143e-06,
+      "loss": 0.7693,
+      "step": 3662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93274280651875e-06,
+      "loss": 0.7587,
+      "step": 3663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927731287791123e-06,
+      "loss": 0.8043,
+      "step": 3664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922720312814256e-06,
+      "loss": 0.749,
+      "step": 3665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917709882902997e-06,
+      "loss": 0.7969,
+      "step": 3666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912699999372052e-06,
+      "loss": 0.7252,
+      "step": 3667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907690663535987e-06,
+      "loss": 0.8076,
+      "step": 3668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90268187670922e-06,
+      "loss": 0.782,
+      "step": 3669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897673640206025e-06,
+      "loss": 0.7267,
+      "step": 3670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892665955340537e-06,
+      "loss": 0.8037,
+      "step": 3671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887658823426738e-06,
+      "loss": 0.7507,
+      "step": 3672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882652245778468e-06,
+      "loss": 0.7546,
+      "step": 3673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87764622370943e-06,
+      "loss": 0.7422,
+      "step": 3674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872640758533164e-06,
+      "loss": 0.7468,
+      "step": 3675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867635851563077e-06,
+      "loss": 0.7324,
+      "step": 3676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862631504112434e-06,
+      "loss": 0.261,
+      "step": 3677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857627717494335e-06,
+      "loss": 0.7279,
+      "step": 3678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85262449302175e-06,
+      "loss": 0.7861,
+      "step": 3679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84762183200749e-06,
+      "loss": 0.7571,
+      "step": 3680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842619735764226e-06,
+      "loss": 0.8462,
+      "step": 3681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83761820560448e-06,
+      "loss": 0.2918,
+      "step": 3682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832617242840617e-06,
+      "loss": 0.7749,
+      "step": 3683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827616848784862e-06,
+      "loss": 0.8389,
+      "step": 3684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822617024749293e-06,
+      "loss": 0.8103,
+      "step": 3685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817617772045826e-06,
+      "loss": 0.7748,
+      "step": 3686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812619091986237e-06,
+      "loss": 0.7701,
+      "step": 3687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807620985882153e-06,
+      "loss": 0.7058,
+      "step": 3688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802623455045044e-06,
+      "loss": 0.7424,
+      "step": 3689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797626500786232e-06,
+      "loss": 0.7197,
+      "step": 3690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792630124416886e-06,
+      "loss": 0.7866,
+      "step": 3691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787634327248026e-06,
+      "loss": 0.7615,
+      "step": 3692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782639110590521e-06,
+      "loss": 0.7502,
+      "step": 3693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777644475755083e-06,
+      "loss": 0.8293,
+      "step": 3694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772650424052274e-06,
+      "loss": 0.8149,
+      "step": 3695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767656956792506e-06,
+      "loss": 0.752,
+      "step": 3696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76266407528603e-06,
+      "loss": 0.7747,
+      "step": 3697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757671780842951e-06,
+      "loss": 0.6949,
+      "step": 3698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.752680074773217e-06,
+      "loss": 0.8315,
+      "step": 3699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747688958386618e-06,
+      "loss": 0.7239,
+      "step": 3700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742698432992796e-06,
+      "loss": 0.7587,
+      "step": 3701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737708499901236e-06,
+      "loss": 0.801,
+      "step": 3702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732719160421259e-06,
+      "loss": 0.7791,
+      "step": 3703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727730415862046e-06,
+      "loss": 0.731,
+      "step": 3704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722742267532608e-06,
+      "loss": 0.7108,
+      "step": 3705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717754716741809e-06,
+      "loss": 0.7644,
+      "step": 3706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71276776479835e-06,
+      "loss": 0.7209,
+      "step": 3707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707781413010775e-06,
+      "loss": 0.772,
+      "step": 3708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702795662687476e-06,
+      "loss": 0.7661,
+      "step": 3709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697810515136682e-06,
+      "loss": 0.7332,
+      "step": 3710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692825971666463e-06,
+      "loss": 0.7234,
+      "step": 3711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687842033584739e-06,
+      "loss": 0.729,
+      "step": 3712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68285870219926e-06,
+      "loss": 0.7402,
+      "step": 3713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677875978817626e-06,
+      "loss": 0.7874,
+      "step": 3714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672893864747269e-06,
+      "loss": 0.7034,
+      "step": 3715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667912361295467e-06,
+      "loss": 0.7073,
+      "step": 3716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662931469769338e-06,
+      "loss": 0.7734,
+      "step": 3717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657951191475836e-06,
+      "loss": 0.7527,
+      "step": 3718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652971527721757e-06,
+      "loss": 0.8153,
+      "step": 3719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647992479813736e-06,
+      "loss": 0.7737,
+      "step": 3720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643014049058242e-06,
+      "loss": 0.7625,
+      "step": 3721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638036236761584e-06,
+      "loss": 0.7673,
+      "step": 3722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633059044229917e-06,
+      "loss": 0.8066,
+      "step": 3723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628082472769218e-06,
+      "loss": 0.7666,
+      "step": 3724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623106523685315e-06,
+      "loss": 0.7485,
+      "step": 3725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618131198283863e-06,
+      "loss": 0.7537,
+      "step": 3726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613156497870356e-06,
+      "loss": 0.78,
+      "step": 3727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608182423750132e-06,
+      "loss": 0.8455,
+      "step": 3728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603208977228353e-06,
+      "loss": 0.7777,
+      "step": 3729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59823615961002e-06,
+      "loss": 0.822,
+      "step": 3730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593263972199976e-06,
+      "loss": 0.7905,
+      "step": 3731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588292416302886e-06,
+      "loss": 0.7085,
+      "step": 3732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58332149322326e-06,
+      "loss": 0.6968,
+      "step": 3733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5783512042654395e-06,
+      "loss": 0.74,
+      "step": 3734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573381550733594e-06,
+      "loss": 0.7939,
+      "step": 3735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568412533931733e-06,
+      "loss": 0.7629,
+      "step": 3736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563444155163698e-06,
+      "loss": 0.7549,
+      "step": 3737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5584764157331555e-06,
+      "loss": 0.7466,
+      "step": 3738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553509316943617e-06,
+      "loss": 0.7288,
+      "step": 3739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548542860098416e-06,
+      "loss": 0.75,
+      "step": 3740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543577046500723e-06,
+      "loss": 0.7347,
+      "step": 3741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5386118774535335e-06,
+      "loss": 0.7998,
+      "step": 3742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53364735425968e-06,
+      "loss": 0.6964,
+      "step": 3743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528683478221825e-06,
+      "loss": 0.6852,
+      "step": 3744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523720250642457e-06,
+      "loss": 0.719,
+      "step": 3745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5187576728238985e-06,
+      "loss": 0.7137,
+      "step": 3746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5137957460682975e-06,
+      "loss": 0.2897,
+      "step": 3747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508834471677634e-06,
+      "loss": 0.7461,
+      "step": 3748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50387385095372e-06,
+      "loss": 0.2502,
+      "step": 3749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498913885198187e-06,
+      "loss": 0.7546,
+      "step": 3750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493954575712503e-06,
+      "loss": 0.7695,
+      "step": 3751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.488995923797964e-06,
+      "loss": 0.7888,
+      "step": 3752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484037930755682e-06,
+      "loss": 0.7012,
+      "step": 3753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47908059788661e-06,
+      "loss": 0.802,
+      "step": 3754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474123926491523e-06,
+      "loss": 0.749,
+      "step": 3755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469167917871018e-06,
+      "loss": 0.7965,
+      "step": 3756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464212573325524e-06,
+      "loss": 0.7751,
+      "step": 3757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459257894155294e-06,
+      "loss": 0.7891,
+      "step": 3758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454303881660403e-06,
+      "loss": 0.7876,
+      "step": 3759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449350537140756e-06,
+      "loss": 0.8051,
+      "step": 3760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444397861896079e-06,
+      "loss": 0.702,
+      "step": 3761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.439445857225924e-06,
+      "loss": 0.7664,
+      "step": 3762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434494524429672e-06,
+      "loss": 0.7734,
+      "step": 3763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429543864806515e-06,
+      "loss": 0.6748,
+      "step": 3764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424593879655479e-06,
+      "loss": 0.7206,
+      "step": 3765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419644570275415e-06,
+      "loss": 0.7678,
+      "step": 3766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414695937964986e-06,
+      "loss": 0.7493,
+      "step": 3767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4097479840226835e-06,
+      "loss": 0.7158,
+      "step": 3768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4048007097468245e-06,
+      "loss": 0.71,
+      "step": 3769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399854116435539e-06,
+      "loss": 0.7445,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394908205386788e-06,
+      "loss": 0.7728,
+      "step": 3771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389962977898344e-06,
+      "loss": 0.7388,
+      "step": 3772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385018435267808e-06,
+      "loss": 0.7512,
+      "step": 3773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380074578792593e-06,
+      "loss": 0.8032,
+      "step": 3774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3751314097699434e-06,
+      "loss": 0.7385,
+      "step": 3775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370188929496916e-06,
+      "loss": 0.8098,
+      "step": 3776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.365247139270385e-06,
+      "loss": 0.7388,
+      "step": 3777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360306040387046e-06,
+      "loss": 0.7551,
+      "step": 3778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3553656341434175e-06,
+      "loss": 0.7872,
+      "step": 3779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350425921835826e-06,
+      "loss": 0.7242,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345486904760431e-06,
+      "loss": 0.7069,
+      "step": 3781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340548584213189e-06,
+      "loss": 0.8085,
+      "step": 3782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335610961489894e-06,
+      "loss": 0.7496,
+      "step": 3783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330674037886146e-06,
+      "loss": 0.6609,
+      "step": 3784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325737814697362e-06,
+      "loss": 0.6497,
+      "step": 3785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320802293218779e-06,
+      "loss": 0.7463,
+      "step": 3786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315867474745451e-06,
+      "loss": 0.7631,
+      "step": 3787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3109333605722365e-06,
+      "loss": 0.7432,
+      "step": 3788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305999951993823e-06,
+      "loss": 0.6896,
+      "step": 3789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.301067250304709e-06,
+      "loss": 0.7598,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2961352567992e-06,
+      "loss": 0.7776,
+      "step": 3791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291203972771421e-06,
+      "loss": 0.7238,
+      "step": 3792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286273399515318e-06,
+      "loss": 0.7139,
+      "step": 3793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281343538324637e-06,
+      "loss": 0.8132,
+      "step": 3794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276414390492947e-06,
+      "loss": 0.7891,
+      "step": 3795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2714859573136245e-06,
+      "loss": 0.8767,
+      "step": 3796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266558240079861e-06,
+      "loss": 0.7449,
+      "step": 3797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2616312400846625e-06,
+      "loss": 0.7386,
+      "step": 3798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2567049586208395e-06,
+      "loss": 0.8064,
+      "step": 3799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25177939698102e-06,
+      "loss": 0.74,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246854556457644e-06,
+      "loss": 0.7739,
+      "step": 3801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241930438342957e-06,
+      "loss": 0.7798,
+      "step": 3802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237007043929018e-06,
+      "loss": 0.7593,
+      "step": 3803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232084374507697e-06,
+      "loss": 0.7732,
+      "step": 3804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2271624313706724e-06,
+      "loss": 0.7566,
+      "step": 3805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222241215809433e-06,
+      "loss": 0.745,
+      "step": 3806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217320729115273e-06,
+      "loss": 0.6995,
+      "step": 3807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2124009725793006e-06,
+      "loss": 0.8047,
+      "step": 3808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207481947492432e-06,
+      "loss": 0.7208,
+      "step": 3809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202563655145382e-06,
+      "loss": 0.7456,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1976460968286894e-06,
+      "loss": 0.7502,
+      "step": 3811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1927292738326905e-06,
+      "loss": 0.7794,
+      "step": 3812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187813187447525e-06,
+      "loss": 0.8159,
+      "step": 3813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18289783896315e-06,
+      "loss": 0.7927,
+      "step": 3814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1779832296693176e-06,
+      "loss": 0.7524,
+      "step": 3815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173069360855597e-06,
+      "loss": 0.6841,
+      "step": 3816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168156233811352e-06,
+      "loss": 0.8035,
+      "step": 3817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163243849825761e-06,
+      "loss": 0.2808,
+      "step": 3818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158332210187804e-06,
+      "loss": 0.7703,
+      "step": 3819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153421316186262e-06,
+      "loss": 0.7544,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148511169109727e-06,
+      "loss": 0.7317,
+      "step": 3821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143601770246593e-06,
+      "loss": 0.7579,
+      "step": 3822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1386931208850516e-06,
+      "loss": 0.2737,
+      "step": 3823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.133785222313108e-06,
+      "loss": 0.7646,
+      "step": 3824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128878075818564e-06,
+      "loss": 0.6818,
+      "step": 3825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123971682689022e-06,
+      "loss": 0.7733,
+      "step": 3826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11906604421189e-06,
+      "loss": 0.6937,
+      "step": 3827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114161161674382e-06,
+      "loss": 0.8066,
+      "step": 3828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109257036363506e-06,
+      "loss": 0.7371,
+      "step": 3829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104353669566078e-06,
+      "loss": 0.7358,
+      "step": 3830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099451062568707e-06,
+      "loss": 0.8064,
+      "step": 3831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0945492166578086e-06,
+      "loss": 0.7432,
+      "step": 3832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0896481331196035e-06,
+      "loss": 0.2588,
+      "step": 3833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084747813240098e-06,
+      "loss": 0.7644,
+      "step": 3834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079848258305111e-06,
+      "loss": 0.7411,
+      "step": 3835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074949469600256e-06,
+      "loss": 0.2617,
+      "step": 3836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070051448410942e-06,
+      "loss": 0.7607,
+      "step": 3837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065154196022384e-06,
+      "loss": 0.762,
+      "step": 3838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060257713719592e-06,
+      "loss": 0.7267,
+      "step": 3839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055362002787368e-06,
+      "loss": 0.7153,
+      "step": 3840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050467064510323e-06,
+      "loss": 0.8291,
+      "step": 3841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045572900172854e-06,
+      "loss": 0.79,
+      "step": 3842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040679511059163e-06,
+      "loss": 0.7495,
+      "step": 3843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0357868984532475e-06,
+      "loss": 0.7672,
+      "step": 3844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030895063638889e-06,
+      "loss": 0.8098,
+      "step": 3845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026004007899691e-06,
+      "loss": 0.8058,
+      "step": 3846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021113732519027e-06,
+      "loss": 0.7942,
+      "step": 3847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016224238780078e-06,
+      "loss": 0.7817,
+      "step": 3848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011335527965819e-06,
+      "loss": 0.7952,
+      "step": 3849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006447601359015e-06,
+      "loss": 0.7451,
+      "step": 3850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001560460242231e-06,
+      "loss": 0.7598,
+      "step": 3851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996674105897822e-06,
+      "loss": 0.724,
+      "step": 3852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991788539607936e-06,
+      "loss": 0.7512,
+      "step": 3853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98690376265452e-06,
+      "loss": 0.7506,
+      "step": 3854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982019776319307e-06,
+      "loss": 0.7761,
+      "step": 3855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977136581883826e-06,
+      "loss": 0.7371,
+      "step": 3856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972254180629398e-06,
+      "loss": 0.7346,
+      "step": 3857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967372573837135e-06,
+      "loss": 0.7212,
+      "step": 3858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962491762787942e-06,
+      "loss": 0.752,
+      "step": 3859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957611748762514e-06,
+      "loss": 0.7047,
+      "step": 3860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952732533041337e-06,
+      "loss": 0.7694,
+      "step": 3861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947854116904688e-06,
+      "loss": 0.6705,
+      "step": 3862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942976501632631e-06,
+      "loss": 0.718,
+      "step": 3863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938099688505026e-06,
+      "loss": 0.7825,
+      "step": 3864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93322367880152e-06,
+      "loss": 0.7362,
+      "step": 3865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928348473801544e-06,
+      "loss": 0.8142,
+      "step": 3866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923474074784324e-06,
+      "loss": 0.8228,
+      "step": 3867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918600483028879e-06,
+      "loss": 0.7639,
+      "step": 3868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9137276998140005e-06,
+      "loss": 0.8062,
+      "step": 3869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908855726418281e-06,
+      "loss": 0.696,
+      "step": 3870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903984564120101e-06,
+      "loss": 0.8054,
+      "step": 3871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899114214197616e-06,
+      "loss": 0.7583,
+      "step": 3872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894244677928783e-06,
+      "loss": 0.8022,
+      "step": 3873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889375956591337e-06,
+      "loss": 0.7217,
+      "step": 3874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884508051462798e-06,
+      "loss": 0.783,
+      "step": 3875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87964096382048e-06,
+      "loss": 0.6875,
+      "step": 3876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874774694941473e-06,
+      "loss": 0.8081,
+      "step": 3877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869909246102657e-06,
+      "loss": 0.7018,
+      "step": 3878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8650446185807006e-06,
+      "loss": 0.7719,
+      "step": 3879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8601808136520465e-06,
+      "loss": 0.7944,
+      "step": 3880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8553178325929296e-06,
+      "loss": 0.7834,
+      "step": 3881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8504556766793675e-06,
+      "loss": 0.7305,
+      "step": 3882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.845594347187161e-06,
+      "loss": 0.7756,
+      "step": 3883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.840733845391896e-06,
+      "loss": 0.7269,
+      "step": 3884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835874172568933e-06,
+      "loss": 0.6962,
+      "step": 3885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831015329993427e-06,
+      "loss": 0.7432,
+      "step": 3886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826157318940303e-06,
+      "loss": 0.7567,
+      "step": 3887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821300140684275e-06,
+      "loss": 0.7844,
+      "step": 3888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816443796499841e-06,
+      "loss": 0.767,
+      "step": 3889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811588287661271e-06,
+      "loss": 0.7134,
+      "step": 3890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.806733615442623e-06,
+      "loss": 0.7207,
+      "step": 3891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801879781117737e-06,
+      "loss": 0.8027,
+      "step": 3892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797026785960226e-06,
+      "loss": 0.6913,
+      "step": 3893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7921746312434855e-06,
+      "loss": 0.7058,
+      "step": 3894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787323318240696e-06,
+      "loss": 0.722,
+      "step": 3895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782472848224808e-06,
+      "loss": 0.7545,
+      "step": 3896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7776232224685616e-06,
+      "loss": 0.7408,
+      "step": 3897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7727744422444605e-06,
+      "loss": 0.7058,
+      "step": 3898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767926508824802e-06,
+      "loss": 0.7578,
+      "step": 3899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7630794234816535e-06,
+      "loss": 0.7888,
+      "step": 3900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758233187486858e-06,
+      "loss": 0.7522,
+      "step": 3901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753387802112039e-06,
+      "loss": 0.8054,
+      "step": 3902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7485432686286015e-06,
+      "loss": 0.7361,
+      "step": 3903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743699588307715e-06,
+      "loss": 0.8114,
+      "step": 3904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738856762420334e-06,
+      "loss": 0.3065,
+      "step": 3905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73401479223719e-06,
+      "loss": 0.7725,
+      "step": 3906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729173679028782e-06,
+      "loss": 0.7524,
+      "step": 3907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72433342406539e-06,
+      "loss": 0.7644,
+      "step": 3908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719494028617072e-06,
+      "loss": 0.7344,
+      "step": 3909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714655493953649e-06,
+      "loss": 0.6941,
+      "step": 3910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7098178213447296e-06,
+      "loss": 0.865,
+      "step": 3911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704981012059684e-06,
+      "loss": 0.7164,
+      "step": 3912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7001450673676645e-06,
+      "loss": 0.8013,
+      "step": 3913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695309988537596e-06,
+      "loss": 0.749,
+      "step": 3914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69047577683817e-06,
+      "loss": 0.7461,
+      "step": 3915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685642433537856e-06,
+      "loss": 0.8167,
+      "step": 3916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68080995990489e-06,
+      "loss": 0.7542,
+      "step": 3917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675978357207293e-06,
+      "loss": 0.6755,
+      "step": 3918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671147626712842e-06,
+      "loss": 0.7169,
+      "step": 3919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666317769689091e-06,
+      "loss": 0.7482,
+      "step": 3920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6614887874033695e-06,
+      "loss": 0.7759,
+      "step": 3921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656660681122768e-06,
+      "loss": 0.7642,
+      "step": 3922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.651833452114156e-06,
+      "loss": 0.7327,
+      "step": 3923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647007101644167e-06,
+      "loss": 0.7278,
+      "step": 3924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6421816309792075e-06,
+      "loss": 0.7427,
+      "step": 3925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6373570413854505e-06,
+      "loss": 0.7244,
+      "step": 3926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6325333341288435e-06,
+      "loss": 0.803,
+      "step": 3927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627710510475091e-06,
+      "loss": 0.7454,
+      "step": 3928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622888571689676e-06,
+      "loss": 0.7156,
+      "step": 3929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618067519037848e-06,
+      "loss": 0.729,
+      "step": 3930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613247353784619e-06,
+      "loss": 0.7966,
+      "step": 3931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.608428077194774e-06,
+      "loss": 0.7469,
+      "step": 3932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603609690532858e-06,
+      "loss": 0.7649,
+      "step": 3933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59879219506319e-06,
+      "loss": 0.7509,
+      "step": 3934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593975592049852e-06,
+      "loss": 0.7311,
+      "step": 3935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589159882756688e-06,
+      "loss": 0.7507,
+      "step": 3936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584345068447313e-06,
+      "loss": 0.691,
+      "step": 3937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579531150385107e-06,
+      "loss": 0.7305,
+      "step": 3938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5747181298332085e-06,
+      "loss": 0.7443,
+      "step": 3939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569906008054527e-06,
+      "loss": 0.7551,
+      "step": 3940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565094786311739e-06,
+      "loss": 0.7786,
+      "step": 3941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560284465867271e-06,
+      "loss": 0.8184,
+      "step": 3942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5554750479833264e-06,
+      "loss": 0.7611,
+      "step": 3943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55066653392187e-06,
+      "loss": 0.7653,
+      "step": 3944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545858924944624e-06,
+      "loss": 0.7742,
+      "step": 3945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541052222313076e-06,
+      "loss": 0.7185,
+      "step": 3946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536246427288473e-06,
+      "loss": 0.7535,
+      "step": 3947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53144154113183e-06,
+      "loss": 0.6493,
+      "step": 3948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526637565103922e-06,
+      "loss": 0.2531,
+      "step": 3949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521834500465277e-06,
+      "loss": 0.781,
+      "step": 3950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517032348476195e-06,
+      "loss": 0.7532,
+      "step": 3951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.512231110396726e-06,
+      "loss": 0.7461,
+      "step": 3952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507430787486695e-06,
+      "loss": 0.7898,
+      "step": 3953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5026313810056705e-06,
+      "loss": 0.8176,
+      "step": 3954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49783289221299e-06,
+      "loss": 0.6653,
+      "step": 3955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493035322367749e-06,
+      "loss": 0.7322,
+      "step": 3956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488238672728799e-06,
+      "loss": 0.7522,
+      "step": 3957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483442944554749e-06,
+      "loss": 0.7362,
+      "step": 3958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478648139103978e-06,
+      "loss": 0.7415,
+      "step": 3959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473854257634604e-06,
+      "loss": 0.7634,
+      "step": 3960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469061301404518e-06,
+      "loss": 0.7367,
+      "step": 3961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464269271671362e-06,
+      "loss": 0.7645,
+      "step": 3962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.459478169692532e-06,
+      "loss": 0.8184,
+      "step": 3963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454687996725188e-06,
+      "loss": 0.7302,
+      "step": 3964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449898754026245e-06,
+      "loss": 0.8086,
+      "step": 3965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445110442852362e-06,
+      "loss": 0.8057,
+      "step": 3966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440323064459974e-06,
+      "loss": 0.7844,
+      "step": 3967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435536620105251e-06,
+      "loss": 0.7535,
+      "step": 3968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430751111044131e-06,
+      "loss": 0.7611,
+      "step": 3969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425966538532305e-06,
+      "loss": 0.7463,
+      "step": 3970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421182903825209e-06,
+      "loss": 0.74,
+      "step": 3971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4164002081780455e-06,
+      "loss": 0.761,
+      "step": 3972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411618452845765e-06,
+      "loss": 0.7529,
+      "step": 3973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406837639083067e-06,
+      "loss": 0.79,
+      "step": 3974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40205776814441e-06,
+      "loss": 0.7761,
+      "step": 3975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397278841284007e-06,
+      "loss": 0.7898,
+      "step": 3976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392500859755813e-06,
+      "loss": 0.7356,
+      "step": 3977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387723824813545e-06,
+      "loss": 0.793,
+      "step": 3978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382947737710668e-06,
+      "loss": 0.8062,
+      "step": 3979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3781725997003976e-06,
+      "loss": 0.7563,
+      "step": 3980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373398412035703e-06,
+      "loss": 0.7441,
+      "step": 3981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368625175969298e-06,
+      "loss": 0.7804,
+      "step": 3982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363852892753653e-06,
+      "loss": 0.7595,
+      "step": 3983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35908156364099e-06,
+      "loss": 0.2783,
+      "step": 3984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354311189883269e-06,
+      "loss": 0.772,
+      "step": 3985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349541772732213e-06,
+      "loss": 0.8018,
+      "step": 3986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34477331343929e-06,
+      "loss": 0.72,
+      "step": 3987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340005813255705e-06,
+      "loss": 0.7,
+      "step": 3988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335239273432432e-06,
+      "loss": 0.8918,
+      "step": 3989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330473695220176e-06,
+      "loss": 0.7412,
+      "step": 3990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325709079869402e-06,
+      "loss": 0.7405,
+      "step": 3991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320945428630308e-06,
+      "loss": 0.7207,
+      "step": 3992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316182742752852e-06,
+      "loss": 0.7858,
+      "step": 3993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311421023486735e-06,
+      "loss": 0.2682,
+      "step": 3994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3066602720814e-06,
+      "loss": 0.8342,
+      "step": 3995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30190048978604e-06,
+      "loss": 0.7891,
+      "step": 3996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297141677849597e-06,
+      "loss": 0.7151,
+      "step": 3997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292383837520748e-06,
+      "loss": 0.7815,
+      "step": 3998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287626970047925e-06,
+      "loss": 0.2682,
+      "step": 3999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2828710766793045e-06,
+      "loss": 0.2567,
+      "step": 4000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278116158662797e-06,
+      "loss": 0.7976,
+      "step": 4001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273362217246069e-06,
+      "loss": 0.792,
+      "step": 4002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268609253676524e-06,
+      "loss": 0.7603,
+      "step": 4003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263857269201312e-06,
+      "loss": 0.8213,
+      "step": 4004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259106265067325e-06,
+      "loss": 0.6918,
+      "step": 4005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254356242521196e-06,
+      "loss": 0.7452,
+      "step": 4006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249607202809304e-06,
+      "loss": 0.7217,
+      "step": 4007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24485914717777e-06,
+      "loss": 0.7045,
+      "step": 4008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240112076872449e-06,
+      "loss": 0.7861,
+      "step": 4009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235365993138948e-06,
+      "loss": 0.2985,
+      "step": 4010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230620897222612e-06,
+      "loss": 0.7657,
+      "step": 4011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225876790368521e-06,
+      "loss": 0.2943,
+      "step": 4012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221133673821504e-06,
+      "loss": 0.6754,
+      "step": 4013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216391548826122e-06,
+      "loss": 0.7416,
+      "step": 4014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.211650416626682e-06,
+      "loss": 0.2405,
+      "step": 4015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206910278467231e-06,
+      "loss": 0.7456,
+      "step": 4016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202171135591548e-06,
+      "loss": 0.8022,
+      "step": 4017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197432989243157e-06,
+      "loss": 0.7732,
+      "step": 4018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192695840665322e-06,
+      "loss": 0.8176,
+      "step": 4019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187959691101036e-06,
+      "loss": 0.7291,
+      "step": 4020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18322454179304e-06,
+      "loss": 0.6948,
+      "step": 4021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178490393983811e-06,
+      "loss": 0.7107,
+      "step": 4022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173757248915555e-06,
+      "loss": 0.6936,
+      "step": 4023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16902510783022e-06,
+      "loss": 0.657,
+      "step": 4024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164293971969497e-06,
+      "loss": 0.7048,
+      "step": 4025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159563842574807e-06,
+      "loss": 0.7375,
+      "step": 4026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154834720887301e-06,
+      "loss": 0.7671,
+      "step": 4027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1501066081478784e-06,
+      "loss": 0.8044,
+      "step": 4028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145379505597167e-06,
+      "loss": 0.7683,
+      "step": 4029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140653414475524e-06,
+      "loss": 0.7733,
+      "step": 4030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1359283360230514e-06,
+      "loss": 0.7281,
+      "step": 4031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131204271479583e-06,
+      "loss": 0.7354,
+      "step": 4032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126481222084679e-06,
+      "loss": 0.7493,
+      "step": 4033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121759189077643e-06,
+      "loss": 0.7571,
+      "step": 4034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117038173697509e-06,
+      "loss": 0.7659,
+      "step": 4035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1123181771830405e-06,
+      "loss": 0.7368,
+      "step": 4036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1075992007727366e-06,
+      "loss": 0.739,
+      "step": 4037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102881245704827e-06,
+      "loss": 0.7036,
+      "step": 4038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098164313217275e-06,
+      "loss": 0.7783,
+      "step": 4039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0934484045477785e-06,
+      "loss": 0.7319,
+      "step": 4040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0887335209337576e-06,
+      "loss": 0.6525,
+      "step": 4041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084019663612373e-06,
+      "loss": 0.7672,
+      "step": 4042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079306833820512e-06,
+      "loss": 0.7043,
+      "step": 4043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07459503279479e-06,
+      "loss": 0.76,
+      "step": 4044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0698842617715594e-06,
+      "loss": 0.7878,
+      "step": 4045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065174521986895e-06,
+      "loss": 0.7729,
+      "step": 4046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060465814676604e-06,
+      "loss": 0.7358,
+      "step": 4047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055758141076225e-06,
+      "loss": 0.7424,
+      "step": 4048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0510515024210194e-06,
+      "loss": 0.7632,
+      "step": 4049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046345899945982e-06,
+      "loss": 0.752,
+      "step": 4050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041641334885839e-06,
+      "loss": 0.7581,
+      "step": 4051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036937808475032e-06,
+      "loss": 0.7603,
+      "step": 4052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032235321947743e-06,
+      "loss": 0.7588,
+      "step": 4053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027533876537875e-06,
+      "loss": 0.7714,
+      "step": 4054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022833473479057e-06,
+      "loss": 0.7886,
+      "step": 4055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018134114004649e-06,
+      "loss": 0.7629,
+      "step": 4056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013435799347733e-06,
+      "loss": 0.7249,
+      "step": 4057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008738530741119e-06,
+      "loss": 0.734,
+      "step": 4058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004042309417337e-06,
+      "loss": 0.693,
+      "step": 4059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999347136608655e-06,
+      "loss": 0.7177,
+      "step": 4060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994653013547056e-06,
+      "loss": 0.7712,
+      "step": 4061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.989959941464245e-06,
+      "loss": 0.271,
+      "step": 4062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985267921591659e-06,
+      "loss": 0.7355,
+      "step": 4063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980576955160457e-06,
+      "loss": 0.7295,
+      "step": 4064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975887043401517e-06,
+      "loss": 0.7549,
+      "step": 4065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971198187545444e-06,
+      "loss": 0.2791,
+      "step": 4066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9665103888225685e-06,
+      "loss": 0.808,
+      "step": 4067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961823648462935e-06,
+      "loss": 0.8262,
+      "step": 4068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9571379676963215e-06,
+      "loss": 0.708,
+      "step": 4069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952453347752218e-06,
+      "loss": 0.7537,
+      "step": 4070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9477697898598405e-06,
+      "loss": 0.6815,
+      "step": 4071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.943087295248131e-06,
+      "loss": 0.6809,
+      "step": 4072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938405865145742e-06,
+      "loss": 0.2675,
+      "step": 4073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933725500781056e-06,
+      "loss": 0.7739,
+      "step": 4074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929046203382171e-06,
+      "loss": 0.7986,
+      "step": 4075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924367974176906e-06,
+      "loss": 0.7499,
+      "step": 4076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919690814392799e-06,
+      "loss": 0.8145,
+      "step": 4077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9150147252571135e-06,
+      "loss": 0.7451,
+      "step": 4078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.91033970799682e-06,
+      "loss": 0.8064,
+      "step": 4079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905665763838617e-06,
+      "loss": 0.802,
+      "step": 4080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900992894008923e-06,
+      "loss": 0.749,
+      "step": 4081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896321099733867e-06,
+      "loss": 0.7434,
+      "step": 4082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8916503822393025e-06,
+      "loss": 0.7479,
+      "step": 4083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886980742750792e-06,
+      "loss": 0.7407,
+      "step": 4084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8823121824936245e-06,
+      "loss": 0.7677,
+      "step": 4085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877644702692804e-06,
+      "loss": 0.7515,
+      "step": 4086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872978304573043e-06,
+      "loss": 0.7396,
+      "step": 4087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868312989358781e-06,
+      "loss": 0.7874,
+      "step": 4088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863648758274168e-06,
+      "loss": 0.7629,
+      "step": 4089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858985612543067e-06,
+      "loss": 0.7303,
+      "step": 4090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854323553389061e-06,
+      "loss": 0.7102,
+      "step": 4091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849662582035449e-06,
+      "loss": 0.739,
+      "step": 4092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845002699705236e-06,
+      "loss": 0.7175,
+      "step": 4093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840343907621151e-06,
+      "loss": 0.811,
+      "step": 4094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.835686207005627e-06,
+      "loss": 0.7456,
+      "step": 4095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8310295990808265e-06,
+      "loss": 0.7368,
+      "step": 4096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82637408506861e-06,
+      "loss": 0.7495,
+      "step": 4097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8217196661905516e-06,
+      "loss": 0.8086,
+      "step": 4098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817066343667952e-06,
+      "loss": 0.7139,
+      "step": 4099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.812414118721808e-06,
+      "loss": 0.7332,
+      "step": 4100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807762992572834e-06,
+      "loss": 0.7812,
+      "step": 4101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8031129664414645e-06,
+      "loss": 0.7762,
+      "step": 4102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798464041547833e-06,
+      "loss": 0.7521,
+      "step": 4103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793816219111787e-06,
+      "loss": 0.7751,
+      "step": 4104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789169500352894e-06,
+      "loss": 0.7729,
+      "step": 4105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784523886490422e-06,
+      "loss": 0.7942,
+      "step": 4106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779879378743346e-06,
+      "loss": 0.7148,
+      "step": 4107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775235978330367e-06,
+      "loss": 0.7542,
+      "step": 4108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7705936864698784e-06,
+      "loss": 0.761,
+      "step": 4109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765952504379988e-06,
+      "loss": 0.7844,
+      "step": 4110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7613124332785196e-06,
+      "loss": 0.7659,
+      "step": 4111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756673474382997e-06,
+      "loss": 0.7227,
+      "step": 4112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752035628910653e-06,
+      "loss": 0.7356,
+      "step": 4113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747398898078434e-06,
+      "loss": 0.7472,
+      "step": 4114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742763283102989e-06,
+      "loss": 0.7336,
+      "step": 4115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73812878520067e-06,
+      "loss": 0.783,
+      "step": 4116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73349540558755e-06,
+      "loss": 0.7417,
+      "step": 4117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.728863145479395e-06,
+      "loss": 0.7812,
+      "step": 4118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724232006091679e-06,
+      "loss": 0.3052,
+      "step": 4119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7196019886395915e-06,
+      "loss": 0.3257,
+      "step": 4120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714973094338019e-06,
+      "loss": 0.7225,
+      "step": 4121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710345324401549e-06,
+      "loss": 0.7732,
+      "step": 4122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70571868004449e-06,
+      "loss": 0.7476,
+      "step": 4123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701093162480842e-06,
+      "loss": 0.7626,
+      "step": 4124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6964687729243105e-06,
+      "loss": 0.7732,
+      "step": 4125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6918455125883036e-06,
+      "loss": 0.7617,
+      "step": 4126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687223382685947e-06,
+      "loss": 0.7222,
+      "step": 4127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682602384430054e-06,
+      "loss": 0.7708,
+      "step": 4128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677982519033143e-06,
+      "loss": 0.7434,
+      "step": 4129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673363787707447e-06,
+      "loss": 0.715,
+      "step": 4130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6687461916648844e-06,
+      "loss": 0.7766,
+      "step": 4131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664129732117092e-06,
+      "loss": 0.8145,
+      "step": 4132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659514410275395e-06,
+      "loss": 0.8049,
+      "step": 4133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.654900227350832e-06,
+      "loss": 0.8086,
+      "step": 4134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650287184554133e-06,
+      "loss": 0.7778,
+      "step": 4135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6456752830957285e-06,
+      "loss": 0.7754,
+      "step": 4136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641064524185761e-06,
+      "loss": 0.7542,
+      "step": 4137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6364549090340625e-06,
+      "loss": 0.7798,
+      "step": 4138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6318464388501646e-06,
+      "loss": 0.7976,
+      "step": 4139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627239114843309e-06,
+      "loss": 0.8057,
+      "step": 4140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622632938222426e-06,
+      "loss": 0.7593,
+      "step": 4141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.618027910196144e-06,
+      "loss": 0.7612,
+      "step": 4142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613424031972803e-06,
+      "loss": 0.7544,
+      "step": 4143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60882130476043e-06,
+      "loss": 0.7196,
+      "step": 4144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604219729766746e-06,
+      "loss": 0.6986,
+      "step": 4145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599619308199185e-06,
+      "loss": 0.7079,
+      "step": 4146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595020041264867e-06,
+      "loss": 0.8105,
+      "step": 4147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590421930170609e-06,
+      "loss": 0.8047,
+      "step": 4148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585824976122931e-06,
+      "loss": 0.6443,
+      "step": 4149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581229180328047e-06,
+      "loss": 0.7599,
+      "step": 4150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576634543991859e-06,
+      "loss": 0.7837,
+      "step": 4151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57204106831998e-06,
+      "loss": 0.2733,
+      "step": 4152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567448754517707e-06,
+      "loss": 0.2683,
+      "step": 4153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56285760379003e-06,
+      "loss": 0.7698,
+      "step": 4154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55826761734165e-06,
+      "loss": 0.7582,
+      "step": 4155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5536787963769445e-06,
+      "loss": 0.7183,
+      "step": 4156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549091142099994e-06,
+      "loss": 0.7061,
+      "step": 4157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544504655714567e-06,
+      "loss": 0.7878,
+      "step": 4158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539919338424135e-06,
+      "loss": 0.7556,
+      "step": 4159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5353351914318585e-06,
+      "loss": 0.741,
+      "step": 4160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530752215940583e-06,
+      "loss": 0.7551,
+      "step": 4161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526170413152863e-06,
+      "loss": 0.8271,
+      "step": 4162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5215897842709286e-06,
+      "loss": 0.7612,
+      "step": 4163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5170103304967084e-06,
+      "loss": 0.7427,
+      "step": 4164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51243205303183e-06,
+      "loss": 0.76,
+      "step": 4165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507854953077598e-06,
+      "loss": 0.783,
+      "step": 4166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503279031835023e-06,
+      "loss": 0.7583,
+      "step": 4167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4987042905047915e-06,
+      "loss": 0.6908,
+      "step": 4168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494130730287297e-06,
+      "loss": 0.7623,
+      "step": 4169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489558352382607e-06,
+      "loss": 0.7494,
+      "step": 4170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4849871579904855e-06,
+      "loss": 0.722,
+      "step": 4171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480417148310392e-06,
+      "loss": 0.7418,
+      "step": 4172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4758483245414655e-06,
+      "loss": 0.7307,
+      "step": 4173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471280687882534e-06,
+      "loss": 0.698,
+      "step": 4174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466714239532126e-06,
+      "loss": 0.8125,
+      "step": 4175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4621489806884455e-06,
+      "loss": 0.8108,
+      "step": 4176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457584912549385e-06,
+      "loss": 0.7573,
+      "step": 4177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453022036312535e-06,
+      "loss": 0.7471,
+      "step": 4178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448460353175165e-06,
+      "loss": 0.7429,
+      "step": 4179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443899864334228e-06,
+      "loss": 0.7577,
+      "step": 4180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439340570986376e-06,
+      "loss": 0.7006,
+      "step": 4181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434782474327935e-06,
+      "loss": 0.7739,
+      "step": 4182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430225575554921e-06,
+      "loss": 0.7714,
+      "step": 4183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425669875863043e-06,
+      "loss": 0.7476,
+      "step": 4184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421115376447685e-06,
+      "loss": 0.2551,
+      "step": 4185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416562078503915e-06,
+      "loss": 0.8094,
+      "step": 4186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412009983226502e-06,
+      "loss": 0.8257,
+      "step": 4187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4074590918098815e-06,
+      "loss": 0.7162,
+      "step": 4188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402909405448176e-06,
+      "loss": 0.7671,
+      "step": 4189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398360925335202e-06,
+      "loss": 0.7512,
+      "step": 4190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393813652664455e-06,
+      "loss": 0.7671,
+      "step": 4191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389267588629106e-06,
+      "loss": 0.7614,
+      "step": 4192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384722734422012e-06,
+      "loss": 0.7917,
+      "step": 4193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3801790912357244e-06,
+      "loss": 0.7605,
+      "step": 4194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375636660262464e-06,
+      "loss": 0.7578,
+      "step": 4195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371095442694132e-06,
+      "loss": 0.752,
+      "step": 4196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366555439722325e-06,
+      "loss": 0.7985,
+      "step": 4197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3620166525383076e-06,
+      "loss": 0.774,
+      "step": 4198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357479082333027e-06,
+      "loss": 0.7507,
+      "step": 4199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352942730297121e-06,
+      "loss": 0.8284,
+      "step": 4200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348407597620898e-06,
+      "loss": 0.7444,
+      "step": 4201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343873685494345e-06,
+      "loss": 0.7235,
+      "step": 4202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3393409951071366e-06,
+      "loss": 0.6538,
+      "step": 4203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334809527648628e-06,
+      "loss": 0.7592,
+      "step": 4204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330279284307844e-06,
+      "loss": 0.7864,
+      "step": 4205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3257502662734905e-06,
+      "loss": 0.7419,
+      "step": 4206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32122247473396e-06,
+      "loss": 0.7485,
+      "step": 4207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316695910877316e-06,
+      "loss": 0.7561,
+      "step": 4208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312170575891297e-06,
+      "loss": 0.7155,
+      "step": 4209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307646470963331e-06,
+      "loss": 0.7408,
+      "step": 4210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303123597280509e-06,
+      "loss": 0.7378,
+      "step": 4211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298601956029607e-06,
+      "loss": 0.2863,
+      "step": 4212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29408154839708e-06,
+      "loss": 0.8049,
+      "step": 4213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289562375569054e-06,
+      "loss": 0.7856,
+      "step": 4214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285044438731328e-06,
+      "loss": 0.7803,
+      "step": 4215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280527739069388e-06,
+      "loss": 0.7435,
+      "step": 4216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276012277768386e-06,
+      "loss": 0.8243,
+      "step": 4217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271498056013148e-06,
+      "loss": 0.7439,
+      "step": 4218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2669850749881855e-06,
+      "loss": 0.7579,
+      "step": 4219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.262473335877674e-06,
+      "loss": 0.3066,
+      "step": 4220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257962839865462e-06,
+      "loss": 0.7662,
+      "step": 4221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253453588135084e-06,
+      "loss": 0.7811,
+      "step": 4222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2489455818697376e-06,
+      "loss": 0.7332,
+      "step": 4223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244438822252291e-06,
+      "loss": 0.7694,
+      "step": 4224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239933310465299e-06,
+      "loss": 0.6915,
+      "step": 4225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235429047690977e-06,
+      "loss": 0.7407,
+      "step": 4226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230926035111217e-06,
+      "loss": 0.7522,
+      "step": 4227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226424273907578e-06,
+      "loss": 0.7256,
+      "step": 4228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221923765261301e-06,
+      "loss": 0.7566,
+      "step": 4229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2174245103532885e-06,
+      "loss": 0.752,
+      "step": 4230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212926510364116e-06,
+      "loss": 0.7192,
+      "step": 4231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208429766474038e-06,
+      "loss": 0.7133,
+      "step": 4232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203934279862969e-06,
+      "loss": 0.7524,
+      "step": 4233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199440051710494e-06,
+      "loss": 0.7194,
+      "step": 4234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194947083195877e-06,
+      "loss": 0.7356,
+      "step": 4235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1904553754980445e-06,
+      "loss": 0.7708,
+      "step": 4236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1859649297955885e-06,
+      "loss": 0.78,
+      "step": 4237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181475747266779e-06,
+      "loss": 0.8118,
+      "step": 4238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176987829089553e-06,
+      "loss": 0.6874,
+      "step": 4239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.17250117644151e-06,
+      "loss": 0.7786,
+      "step": 4240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168015790499917e-06,
+      "loss": 0.7538,
+      "step": 4241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1635316724417195e-06,
+      "loss": 0.7737,
+      "step": 4242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159048823443517e-06,
+      "loss": 0.2549,
+      "step": 4243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154567244681583e-06,
+      "loss": 0.6851,
+      "step": 4244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1500869373318594e-06,
+      "loss": 0.7502,
+      "step": 4245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145607902569951e-06,
+      "loss": 0.7924,
+      "step": 4246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.141130141571124e-06,
+      "loss": 0.7332,
+      "step": 4247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136653655510324e-06,
+      "loss": 0.7386,
+      "step": 4248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132178445562148e-06,
+      "loss": 0.7507,
+      "step": 4249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127704512900864e-06,
+      "loss": 0.7705,
+      "step": 4250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12323185870041e-06,
+      "loss": 0.7391,
+      "step": 4251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1187604841343795e-06,
+      "loss": 0.7576,
+      "step": 4252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114290390376031e-06,
+      "loss": 0.7808,
+      "step": 4253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109821578598297e-06,
+      "loss": 0.8108,
+      "step": 4254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105354049973765e-06,
+      "loss": 0.7781,
+      "step": 4255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1008878056746815e-06,
+      "loss": 0.707,
+      "step": 4256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096422846872968e-06,
+      "loss": 0.7886,
+      "step": 4257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091959174740202e-06,
+      "loss": 0.7424,
+      "step": 4258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087496790447619e-06,
+      "loss": 0.6976,
+      "step": 4259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08303569516613e-06,
+      "loss": 0.6735,
+      "step": 4260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078575890066295e-06,
+      "loss": 0.7593,
+      "step": 4261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074117376318339e-06,
+      "loss": 0.7432,
+      "step": 4262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069660155092146e-06,
+      "loss": 0.7891,
+      "step": 4263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065204227557271e-06,
+      "loss": 0.8036,
+      "step": 4264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060749594882919e-06,
+      "loss": 0.6884,
+      "step": 4265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056296258237955e-06,
+      "loss": 0.7344,
+      "step": 4266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051844218790916e-06,
+      "loss": 0.77,
+      "step": 4267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047393477709983e-06,
+      "loss": 0.7218,
+      "step": 4268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042944036163004e-06,
+      "loss": 0.7211,
+      "step": 4269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03849589531749e-06,
+      "loss": 0.6909,
+      "step": 4270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034049056340604e-06,
+      "loss": 0.7727,
+      "step": 4271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029603520399167e-06,
+      "loss": 0.6931,
+      "step": 4272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02515928865966e-06,
+      "loss": 0.705,
+      "step": 4273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.020716362288232e-06,
+      "loss": 0.2735,
+      "step": 4274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0162747424506705e-06,
+      "loss": 0.74,
+      "step": 4275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011834430312431e-06,
+      "loss": 0.7235,
+      "step": 4276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007395427038629e-06,
+      "loss": 0.7202,
+      "step": 4277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002957733794029e-06,
+      "loss": 0.7617,
+      "step": 4278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998521351743052e-06,
+      "loss": 0.2433,
+      "step": 4279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994086282049784e-06,
+      "loss": 0.7603,
+      "step": 4280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989652525877957e-06,
+      "loss": 0.7397,
+      "step": 4281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985220084390958e-06,
+      "loss": 0.7626,
+      "step": 4282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98078895875184e-06,
+      "loss": 0.7155,
+      "step": 4283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9763591501233015e-06,
+      "loss": 0.7278,
+      "step": 4284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971930659667691e-06,
+      "loss": 0.7081,
+      "step": 4285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967503488547028e-06,
+      "loss": 0.7266,
+      "step": 4286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9630776379229695e-06,
+      "loss": 0.8151,
+      "step": 4287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958653108956828e-06,
+      "loss": 0.7341,
+      "step": 4288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954229902809582e-06,
+      "loss": 0.6959,
+      "step": 4289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94980802064185e-06,
+      "loss": 0.7622,
+      "step": 4290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945387463613903e-06,
+      "loss": 0.2571,
+      "step": 4291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940968232885676e-06,
+      "loss": 0.7795,
+      "step": 4292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.936550329616744e-06,
+      "loss": 0.6711,
+      "step": 4293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932133754966335e-06,
+      "loss": 0.7618,
+      "step": 4294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927718510093339e-06,
+      "loss": 0.8032,
+      "step": 4295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923304596156285e-06,
+      "loss": 0.272,
+      "step": 4296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918892014313359e-06,
+      "loss": 0.7759,
+      "step": 4297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914480765722389e-06,
+      "loss": 0.7666,
+      "step": 4298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910070851540871e-06,
+      "loss": 0.7708,
+      "step": 4299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905662272925933e-06,
+      "loss": 0.7186,
+      "step": 4300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901255031034357e-06,
+      "loss": 0.7292,
+      "step": 4301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896849127022585e-06,
+      "loss": 0.8213,
+      "step": 4302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892444562046694e-06,
+      "loss": 0.7634,
+      "step": 4303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8880413372624125e-06,
+      "loss": 0.7437,
+      "step": 4304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883639453825126e-06,
+      "loss": 0.7434,
+      "step": 4305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879238912889861e-06,
+      "loss": 0.7646,
+      "step": 4306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8748397156112855e-06,
+      "loss": 0.2754,
+      "step": 4307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870441863143733e-06,
+      "loss": 0.7854,
+      "step": 4308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866045356641164e-06,
+      "loss": 0.7482,
+      "step": 4309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861650197257205e-06,
+      "loss": 0.6702,
+      "step": 4310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857256386145109e-06,
+      "loss": 0.7358,
+      "step": 4311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852863924457792e-06,
+      "loss": 0.712,
+      "step": 4312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84847281334781e-06,
+      "loss": 0.7502,
+      "step": 4313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844083053967357e-06,
+      "loss": 0.6903,
+      "step": 4314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839694647468286e-06,
+      "loss": 0.7578,
+      "step": 4315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835307595002088e-06,
+      "loss": 0.7336,
+      "step": 4316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8309218977198915e-06,
+      "loss": 0.7483,
+      "step": 4317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826537556772487e-06,
+      "loss": 0.7554,
+      "step": 4318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822154573310294e-06,
+      "loss": 0.8135,
+      "step": 4319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8177729484833775e-06,
+      "loss": 0.7351,
+      "step": 4320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813392683441457e-06,
+      "loss": 0.7788,
+      "step": 4321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8090137793338844e-06,
+      "loss": 0.7791,
+      "step": 4322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804636237309651e-06,
+      "loss": 0.718,
+      "step": 4323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8002600585174086e-06,
+      "loss": 0.7351,
+      "step": 4324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.795885244105434e-06,
+      "loss": 0.7678,
+      "step": 4325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7915117952216485e-06,
+      "loss": 0.7854,
+      "step": 4326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787139713013625e-06,
+      "loss": 0.6838,
+      "step": 4327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782768998628571e-06,
+      "loss": 0.2844,
+      "step": 4328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778399653213334e-06,
+      "loss": 0.7502,
+      "step": 4329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774031677914397e-06,
+      "loss": 0.6764,
+      "step": 4330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7696650738779015e-06,
+      "loss": 0.7207,
+      "step": 4331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.765299842249613e-06,
+      "loss": 0.7042,
+      "step": 4332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760935984174938e-06,
+      "loss": 0.6877,
+      "step": 4333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7565735007989345e-06,
+      "loss": 0.726,
+      "step": 4334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752212393266286e-06,
+      "loss": 0.7162,
+      "step": 4335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747852662721321e-06,
+      "loss": 0.6908,
+      "step": 4336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743494310308009e-06,
+      "loss": 0.6891,
+      "step": 4337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739137337169954e-06,
+      "loss": 0.7231,
+      "step": 4338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.734781744450396e-06,
+      "loss": 0.777,
+      "step": 4339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730427533292223e-06,
+      "loss": 0.7557,
+      "step": 4340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7260747048379495e-06,
+      "loss": 0.7472,
+      "step": 4341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721723260229729e-06,
+      "loss": 0.7473,
+      "step": 4342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71737320060936e-06,
+      "loss": 0.7686,
+      "step": 4343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713024527118269e-06,
+      "loss": 0.74,
+      "step": 4344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708677240897518e-06,
+      "loss": 0.7646,
+      "step": 4345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704331343087811e-06,
+      "loss": 0.7898,
+      "step": 4346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699986834829491e-06,
+      "loss": 0.7441,
+      "step": 4347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695643717262525e-06,
+      "loss": 0.7457,
+      "step": 4348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691301991526518e-06,
+      "loss": 0.2852,
+      "step": 4349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68696165876072e-06,
+      "loss": 0.7644,
+      "step": 4350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682622720104004e-06,
+      "loss": 0.8013,
+      "step": 4351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678285176694878e-06,
+      "loss": 0.7454,
+      "step": 4352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673949029671494e-06,
+      "loss": 0.7908,
+      "step": 4353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669614280171628e-06,
+      "loss": 0.7451,
+      "step": 4354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665280929332686e-06,
+      "loss": 0.7627,
+      "step": 4355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660948978291724e-06,
+      "loss": 0.7546,
+      "step": 4356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656618428185413e-06,
+      "loss": 0.8154,
+      "step": 4357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65228928015006e-06,
+      "loss": 0.7749,
+      "step": 4358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647961535321616e-06,
+      "loss": 0.7233,
+      "step": 4359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.643635194835649e-06,
+      "loss": 0.7629,
+      "step": 4360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639310259827363e-06,
+      "loss": 0.7384,
+      "step": 4361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634986731431601e-06,
+      "loss": 0.7303,
+      "step": 4362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6306646107828276e-06,
+      "loss": 0.7261,
+      "step": 4363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626343899015141e-06,
+      "loss": 0.7308,
+      "step": 4364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622024597262266e-06,
+      "loss": 0.7223,
+      "step": 4365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617706706657569e-06,
+      "loss": 0.7539,
+      "step": 4366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613390228334033e-06,
+      "loss": 0.7308,
+      "step": 4367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609075163424274e-06,
+      "loss": 0.7567,
+      "step": 4368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604761513060545e-06,
+      "loss": 0.731,
+      "step": 4369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600449278374719e-06,
+      "loss": 0.2355,
+      "step": 4370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596138460498295e-06,
+      "loss": 0.7651,
+      "step": 4371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591829060562416e-06,
+      "loss": 0.7856,
+      "step": 4372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587521079697835e-06,
+      "loss": 0.7323,
+      "step": 4373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583214519034939e-06,
+      "loss": 0.7445,
+      "step": 4374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5789093797037494e-06,
+      "loss": 0.7074,
+      "step": 4375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574605662833906e-06,
+      "loss": 0.7883,
+      "step": 4376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570303369554675e-06,
+      "loss": 0.6697,
+      "step": 4377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566002500994957e-06,
+      "loss": 0.7349,
+      "step": 4378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561703058283272e-06,
+      "loss": 0.8115,
+      "step": 4379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5574050425477636e-06,
+      "loss": 0.7693,
+      "step": 4380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5531084549162085e-06,
+      "loss": 0.7429,
+      "step": 4381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54881329651601e-06,
+      "loss": 0.7517,
+      "step": 4382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544519568474187e-06,
+      "loss": 0.6608,
+      "step": 4383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540227271917385e-06,
+      "loss": 0.7708,
+      "step": 4384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535936407971882e-06,
+      "loss": 0.8215,
+      "step": 4385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.531646977763572e-06,
+      "loss": 0.7808,
+      "step": 4386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527358982417973e-06,
+      "loss": 0.7485,
+      "step": 4387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5230724230602354e-06,
+      "loss": 0.7976,
+      "step": 4388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518787300815123e-06,
+      "loss": 0.6785,
+      "step": 4389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514503616807022e-06,
+      "loss": 0.709,
+      "step": 4390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510221372159952e-06,
+      "loss": 0.8008,
+      "step": 4391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505940567997545e-06,
+      "loss": 0.7606,
+      "step": 4392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501661205443055e-06,
+      "loss": 0.7703,
+      "step": 4393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497383285619368e-06,
+      "loss": 0.7759,
+      "step": 4394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4931068096489795e-06,
+      "loss": 0.7368,
+      "step": 4395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488831778654007e-06,
+      "loss": 0.715,
+      "step": 4396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484558193756202e-06,
+      "loss": 0.7637,
+      "step": 4397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480286056076922e-06,
+      "loss": 0.6957,
+      "step": 4398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476015366737151e-06,
+      "loss": 0.6799,
+      "step": 4399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471746126857487e-06,
+      "loss": 0.2615,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467478337558162e-06,
+      "loss": 0.7727,
+      "step": 4401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463211999959013e-06,
+      "loss": 0.788,
+      "step": 4402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458947115179497e-06,
+      "loss": 0.728,
+      "step": 4403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4546836843387035e-06,
+      "loss": 0.7576,
+      "step": 4404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450421708555326e-06,
+      "loss": 0.7788,
+      "step": 4405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446161188947677e-06,
+      "loss": 0.7007,
+      "step": 4406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441902126633697e-06,
+      "loss": 0.71,
+      "step": 4407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437644522730939e-06,
+      "loss": 0.6554,
+      "step": 4408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433388378356563e-06,
+      "loss": 0.7974,
+      "step": 4409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429133694627368e-06,
+      "loss": 0.7346,
+      "step": 4410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424880472659751e-06,
+      "loss": 0.772,
+      "step": 4411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420628713569728e-06,
+      "loss": 0.74,
+      "step": 4412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416378418472943e-06,
+      "loss": 0.7793,
+      "step": 4413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412129588484644e-06,
+      "loss": 0.7959,
+      "step": 4414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407882224719693e-06,
+      "loss": 0.7615,
+      "step": 4415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403636328292579e-06,
+      "loss": 0.2618,
+      "step": 4416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3993919003174e-06,
+      "loss": 0.8049,
+      "step": 4417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395148941907868e-06,
+      "loss": 0.6918,
+      "step": 4418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3909074541773025e-06,
+      "loss": 0.7185,
+      "step": 4419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386667438238653e-06,
+      "loss": 0.719,
+      "step": 4420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382428895204471e-06,
+      "loss": 0.8416,
+      "step": 4421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37819182618692e-06,
+      "loss": 0.7271,
+      "step": 4422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373956232297789e-06,
+      "loss": 0.764,
+      "step": 4423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369722114648469e-06,
+      "loss": 0.7749,
+      "step": 4424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365489474349962e-06,
+      "loss": 0.8019,
+      "step": 4425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361258312512895e-06,
+      "loss": 0.8137,
+      "step": 4426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357028630247495e-06,
+      "loss": 0.6631,
+      "step": 4427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352800428663602e-06,
+      "loss": 0.7406,
+      "step": 4428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348573708870679e-06,
+      "loss": 0.7947,
+      "step": 4429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344348471977786e-06,
+      "loss": 0.7556,
+      "step": 4430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340124719093597e-06,
+      "loss": 0.7576,
+      "step": 4431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335902451326406e-06,
+      "loss": 0.7805,
+      "step": 4432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331681669784107e-06,
+      "loss": 0.7588,
+      "step": 4433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327462375574207e-06,
+      "loss": 0.7469,
+      "step": 4434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3232445698038205e-06,
+      "loss": 0.6936,
+      "step": 4435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319028253579681e-06,
+      "loss": 0.761,
+      "step": 4436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314813428008121e-06,
+      "loss": 0.726,
+      "step": 4437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310600094195082e-06,
+      "loss": 0.7043,
+      "step": 4438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306388253246123e-06,
+      "loss": 0.7146,
+      "step": 4439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302177906266403e-06,
+      "loss": 0.7014,
+      "step": 4440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297969054360688e-06,
+      "loss": 0.78,
+      "step": 4441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.293761698633362e-06,
+      "loss": 0.7554,
+      "step": 4442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289555840188407e-06,
+      "loss": 0.7639,
+      "step": 4443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2853514801294095e-06,
+      "loss": 0.7418,
+      "step": 4444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281148619559576e-06,
+      "loss": 0.7693,
+      "step": 4445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276947259581709e-06,
+      "loss": 0.7734,
+      "step": 4446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272747401298216e-06,
+      "loss": 0.7729,
+      "step": 4447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268549045811121e-06,
+      "loss": 0.7209,
+      "step": 4448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264352194222043e-06,
+      "loss": 0.7146,
+      "step": 4449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260156847632209e-06,
+      "loss": 0.7267,
+      "step": 4450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255963007142457e-06,
+      "loss": 0.7527,
+      "step": 4451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251770673853221e-06,
+      "loss": 0.7612,
+      "step": 4452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247579848864547e-06,
+      "loss": 0.7485,
+      "step": 4453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2433905332760796e-06,
+      "loss": 0.7141,
+      "step": 4454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239202728187073e-06,
+      "loss": 0.7407,
+      "step": 4455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235016434696382e-06,
+      "loss": 0.7894,
+      "step": 4456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2308316539024574e-06,
+      "loss": 0.7488,
+      "step": 4457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22664838690337e-06,
+      "loss": 0.7291,
+      "step": 4458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222466634796779e-06,
+      "loss": 0.7749,
+      "step": 4459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218286398679948e-06,
+      "loss": 0.8003,
+      "step": 4460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214107679649753e-06,
+      "loss": 0.2648,
+      "step": 4461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209930478802661e-06,
+      "loss": 0.8059,
+      "step": 4462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205754797234738e-06,
+      "loss": 0.7542,
+      "step": 4463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2015806360416674e-06,
+      "loss": 0.7584,
+      "step": 4464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197407996318721e-06,
+      "loss": 0.7671,
+      "step": 4465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193236879160768e-06,
+      "loss": 0.7555,
+      "step": 4466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189067285662293e-06,
+      "loss": 0.7432,
+      "step": 4467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184899216917369e-06,
+      "loss": 0.7201,
+      "step": 4468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18073267401967e-06,
+      "loss": 0.7078,
+      "step": 4469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176567658062469e-06,
+      "loss": 0.804,
+      "step": 4470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172404170138649e-06,
+      "loss": 0.7756,
+      "step": 4471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.168242211340679e-06,
+      "loss": 0.7922,
+      "step": 4472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16408178276063e-06,
+      "loss": 0.6997,
+      "step": 4473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159922885490179e-06,
+      "loss": 0.2355,
+      "step": 4474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155765520620593e-06,
+      "loss": 0.8042,
+      "step": 4475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1516096892427335e-06,
+      "loss": 0.7144,
+      "step": 4476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147455392447075e-06,
+      "loss": 0.6985,
+      "step": 4477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1433026313236755e-06,
+      "loss": 0.8147,
+      "step": 4478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139151406962191e-06,
+      "loss": 0.739,
+      "step": 4479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135001720451883e-06,
+      "loss": 0.7871,
+      "step": 4480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130853572881603e-06,
+      "loss": 0.717,
+      "step": 4481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126706965339796e-06,
+      "loss": 0.8174,
+      "step": 4482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122561898914513e-06,
+      "loss": 0.8138,
+      "step": 4483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118418374693391e-06,
+      "loss": 0.2769,
+      "step": 4484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114276393763661e-06,
+      "loss": 0.7307,
+      "step": 4485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110135957212161e-06,
+      "loss": 0.7816,
+      "step": 4486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105997066125312e-06,
+      "loss": 0.7183,
+      "step": 4487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101859721589139e-06,
+      "loss": 0.7566,
+      "step": 4488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097723924689248e-06,
+      "loss": 0.7891,
+      "step": 4489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093589676510856e-06,
+      "loss": 0.7151,
+      "step": 4490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089456978138762e-06,
+      "loss": 0.7128,
+      "step": 4491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085325830657354e-06,
+      "loss": 0.7917,
+      "step": 4492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0811962351506296e-06,
+      "loss": 0.7534,
+      "step": 4493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077068192702166e-06,
+      "loss": 0.6937,
+      "step": 4494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072941704395133e-06,
+      "loss": 0.7092,
+      "step": 4495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068816771312303e-06,
+      "loss": 0.6539,
+      "step": 4496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064693394536029e-06,
+      "loss": 0.7368,
+      "step": 4497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0605715751482575e-06,
+      "loss": 0.6907,
+      "step": 4498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056451314230537e-06,
+      "loss": 0.7664,
+      "step": 4499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052332612863994e-06,
+      "loss": 0.8052,
+      "step": 4500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048215472129347e-06,
+      "loss": 0.6799,
+      "step": 4501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044099893106916e-06,
+      "loss": 0.7383,
+      "step": 4502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039985876876603e-06,
+      "loss": 0.7347,
+      "step": 4503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035873424517898e-06,
+      "loss": 0.7596,
+      "step": 4504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03176253710988e-06,
+      "loss": 0.7534,
+      "step": 4505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02765321573123e-06,
+      "loss": 0.7366,
+      "step": 4506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023545461460204e-06,
+      "loss": 0.7932,
+      "step": 4507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019439275374649e-06,
+      "loss": 0.6797,
+      "step": 4508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01533465855201e-06,
+      "loss": 0.7952,
+      "step": 4509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011231612069311e-06,
+      "loss": 0.7581,
+      "step": 4510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0071301370031625e-06,
+      "loss": 0.7317,
+      "step": 4511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0030302344297725e-06,
+      "loss": 0.7207,
+      "step": 4512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9989319054249284e-06,
+      "loss": 0.2406,
+      "step": 4513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994835151064004e-06,
+      "loss": 0.71,
+      "step": 4514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9907399724219684e-06,
+      "loss": 0.771,
+      "step": 4515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98664637057337e-06,
+      "loss": 0.7162,
+      "step": 4516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982554346592338e-06,
+      "loss": 0.6453,
+      "step": 4517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978463901552606e-06,
+      "loss": 0.8171,
+      "step": 4518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9743750365274755e-06,
+      "loss": 0.7271,
+      "step": 4519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9702877525898385e-06,
+      "loss": 0.7615,
+      "step": 4520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.966202050812179e-06,
+      "loss": 0.7756,
+      "step": 4521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962117932266558e-06,
+      "loss": 0.7527,
+      "step": 4522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958035398024619e-06,
+      "loss": 0.7301,
+      "step": 4523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953954449157599e-06,
+      "loss": 0.7351,
+      "step": 4524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949875086736315e-06,
+      "loss": 0.7175,
+      "step": 4525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945797311831168e-06,
+      "loss": 0.7776,
+      "step": 4526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941721125512135e-06,
+      "loss": 0.8247,
+      "step": 4527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9376465288487884e-06,
+      "loss": 0.7194,
+      "step": 4528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933573522910278e-06,
+      "loss": 0.7128,
+      "step": 4529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929502108765329e-06,
+      "loss": 0.7637,
+      "step": 4530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925432287482264e-06,
+      "loss": 0.7361,
+      "step": 4531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921364060128976e-06,
+      "loss": 0.2497,
+      "step": 4532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9172974277729394e-06,
+      "loss": 0.8107,
+      "step": 4533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913232391481222e-06,
+      "loss": 0.7139,
+      "step": 4534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909168952320459e-06,
+      "loss": 0.7294,
+      "step": 4535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905107111356875e-06,
+      "loss": 0.7251,
+      "step": 4536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901046869656268e-06,
+      "loss": 0.7616,
+      "step": 4537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8969882282840264e-06,
+      "loss": 0.7537,
+      "step": 4538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89293118830511e-06,
+      "loss": 0.7107,
+      "step": 4539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88887575078406e-06,
+      "loss": 0.7146,
+      "step": 4540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884821916785005e-06,
+      "loss": 0.7485,
+      "step": 4541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880769687371641e-06,
+      "loss": 0.71,
+      "step": 4542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876719063607245e-06,
+      "loss": 0.7004,
+      "step": 4543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8726700465546865e-06,
+      "loss": 0.7739,
+      "step": 4544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8686226372763966e-06,
+      "loss": 0.7474,
+      "step": 4545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864576836834388e-06,
+      "loss": 0.7644,
+      "step": 4546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860532646290263e-06,
+      "loss": 0.7573,
+      "step": 4547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856490066705187e-06,
+      "loss": 0.7815,
+      "step": 4548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8524490991399045e-06,
+      "loss": 0.7917,
+      "step": 4549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84840974465475e-06,
+      "loss": 0.7322,
+      "step": 4550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844372004309621e-06,
+      "loss": 0.7329,
+      "step": 4551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840335879163993e-06,
+      "loss": 0.7371,
+      "step": 4552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836301370276928e-06,
+      "loss": 0.8198,
+      "step": 4553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832268478707053e-06,
+      "loss": 0.7482,
+      "step": 4554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82823720551257e-06,
+      "loss": 0.6265,
+      "step": 4555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824207551751269e-06,
+      "loss": 0.8003,
+      "step": 4556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820179518480501e-06,
+      "loss": 0.7396,
+      "step": 4557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816153106757197e-06,
+      "loss": 0.7002,
+      "step": 4558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8121283176378654e-06,
+      "loss": 0.731,
+      "step": 4559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8081051521785893e-06,
+      "loss": 0.7667,
+      "step": 4560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80408361143502e-06,
+      "loss": 0.673,
+      "step": 4561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8000636964623816e-06,
+      "loss": 0.7581,
+      "step": 4562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796045408315482e-06,
+      "loss": 0.7083,
+      "step": 4563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7920287480486938e-06,
+      "loss": 0.7622,
+      "step": 4564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.788013716715959e-06,
+      "loss": 0.6909,
+      "step": 4565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784000315370807e-06,
+      "loss": 0.7246,
+      "step": 4566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779988545066323e-06,
+      "loss": 0.73,
+      "step": 4567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775978406855171e-06,
+      "loss": 0.7246,
+      "step": 4568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771969901789592e-06,
+      "loss": 0.7677,
+      "step": 4569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767963030921391e-06,
+      "loss": 0.7126,
+      "step": 4570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763957795301946e-06,
+      "loss": 0.7629,
+      "step": 4571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759954195982203e-06,
+      "loss": 0.7605,
+      "step": 4572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7559522340126884e-06,
+      "loss": 0.7755,
+      "step": 4573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751951910443492e-06,
+      "loss": 0.7572,
+      "step": 4574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747953226324269e-06,
+      "loss": 0.7041,
+      "step": 4575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743956182704257e-06,
+      "loss": 0.7766,
+      "step": 4576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7399607806322523e-06,
+      "loss": 0.8135,
+      "step": 4577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735967021156621e-06,
+      "loss": 0.2595,
+      "step": 4578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7319749053253097e-06,
+      "loss": 0.8125,
+      "step": 4579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7279844341858206e-06,
+      "loss": 0.7471,
+      "step": 4580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723995608785226e-06,
+      "loss": 0.7526,
+      "step": 4581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7200084301701755e-06,
+      "loss": 0.2528,
+      "step": 4582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7160228993868787e-06,
+      "loss": 0.7267,
+      "step": 4583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7120390174811094e-06,
+      "loss": 0.7465,
+      "step": 4584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708056785498223e-06,
+      "loss": 0.8135,
+      "step": 4585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7040762044831293e-06,
+      "loss": 0.7771,
+      "step": 4586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.700097275480303e-06,
+      "loss": 0.7809,
+      "step": 4587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6961199995338013e-06,
+      "loss": 0.2638,
+      "step": 4588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6921443776872314e-06,
+      "loss": 0.7419,
+      "step": 4589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.68817041098377e-06,
+      "loss": 0.751,
+      "step": 4590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6841981004661677e-06,
+      "loss": 0.7244,
+      "step": 4591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680227447176733e-06,
+      "loss": 0.7217,
+      "step": 4592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6762584521573363e-06,
+      "loss": 0.6752,
+      "step": 4593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6722911164494212e-06,
+      "loss": 0.7534,
+      "step": 4594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6683254410939973e-06,
+      "loss": 0.7866,
+      "step": 4595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6643614271316297e-06,
+      "loss": 0.2709,
+      "step": 4596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6603990756024487e-06,
+      "loss": 0.7712,
+      "step": 4597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6564383875461574e-06,
+      "loss": 0.7424,
+      "step": 4598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.652479364002013e-06,
+      "loss": 0.7356,
+      "step": 4599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6485220060088357e-06,
+      "loss": 0.7579,
+      "step": 4600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.64456631460502e-06,
+      "loss": 0.7729,
+      "step": 4601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.640612290828511e-06,
+      "loss": 0.734,
+      "step": 4602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6366599357168185e-06,
+      "loss": 0.7122,
+      "step": 4603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6327092503070226e-06,
+      "loss": 0.6844,
+      "step": 4604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6287602356357556e-06,
+      "loss": 0.6797,
+      "step": 4605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6248128927392147e-06,
+      "loss": 0.7458,
+      "step": 4606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6208672226531562e-06,
+      "loss": 0.7725,
+      "step": 4607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6169232264129063e-06,
+      "loss": 0.6917,
+      "step": 4608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.612980905053343e-06,
+      "loss": 0.72,
+      "step": 4609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6090402596089025e-06,
+      "loss": 0.7393,
+      "step": 4610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6051012911135953e-06,
+      "loss": 0.7017,
+      "step": 4611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6011640006009786e-06,
+      "loss": 0.7322,
+      "step": 4612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5972283891041694e-06,
+      "loss": 0.8037,
+      "step": 4613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593294457655856e-06,
+      "loss": 0.8252,
+      "step": 4614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589362207288274e-06,
+      "loss": 0.2537,
+      "step": 4615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58543163903322e-06,
+      "loss": 0.6841,
+      "step": 4616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581502753922057e-06,
+      "loss": 0.7169,
+      "step": 4617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5775755529856983e-06,
+      "loss": 0.7178,
+      "step": 4618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5736500372546123e-06,
+      "loss": 0.7078,
+      "step": 4619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5697262077588392e-06,
+      "loss": 0.7986,
+      "step": 4620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5658040655279637e-06,
+      "loss": 0.7202,
+      "step": 4621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.561883611591129e-06,
+      "loss": 0.7368,
+      "step": 4622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5579648469770455e-06,
+      "loss": 0.6763,
+      "step": 4623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55404777271397e-06,
+      "loss": 0.7705,
+      "step": 4624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5501323898297155e-06,
+      "loss": 0.7858,
+      "step": 4625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546218699351661e-06,
+      "loss": 0.7568,
+      "step": 4626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542306702306734e-06,
+      "loss": 0.7563,
+      "step": 4627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538396399721412e-06,
+      "loss": 0.7964,
+      "step": 4628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5344877926217438e-06,
+      "loss": 0.2583,
+      "step": 4629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5305808820333164e-06,
+      "loss": 0.7682,
+      "step": 4630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5266756689812875e-06,
+      "loss": 0.6318,
+      "step": 4631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5227721544903537e-06,
+      "loss": 0.7502,
+      "step": 4632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.51887033958478e-06,
+      "loss": 0.7761,
+      "step": 4633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5149702252883754e-06,
+      "loss": 0.276,
+      "step": 4634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5110718126245046e-06,
+      "loss": 0.8281,
+      "step": 4635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.507175102616094e-06,
+      "loss": 0.6782,
+      "step": 4636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503280096285611e-06,
+      "loss": 0.8022,
+      "step": 4637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4993867946550797e-06,
+      "loss": 0.7861,
+      "step": 4638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.495495198746086e-06,
+      "loss": 0.7083,
+      "step": 4639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4916053095797587e-06,
+      "loss": 0.7443,
+      "step": 4640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4877171281767787e-06,
+      "loss": 0.7357,
+      "step": 4641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4838306555573785e-06,
+      "loss": 0.7717,
+      "step": 4642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4799458927413533e-06,
+      "loss": 0.7169,
+      "step": 4643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4760628407480368e-06,
+      "loss": 0.7502,
+      "step": 4644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4721815005963156e-06,
+      "loss": 0.7134,
+      "step": 4645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.468301873304636e-06,
+      "loss": 0.7041,
+      "step": 4646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4644239598909846e-06,
+      "loss": 0.7515,
+      "step": 4647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4605477613729e-06,
+      "loss": 0.7654,
+      "step": 4648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4566732787674806e-06,
+      "loss": 0.7756,
+      "step": 4649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452800513091362e-06,
+      "loss": 0.6868,
+      "step": 4650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448929465360733e-06,
+      "loss": 0.7583,
+      "step": 4651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4450601365913383e-06,
+      "loss": 0.688,
+      "step": 4652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4411925277984635e-06,
+      "loss": 0.7092,
+      "step": 4653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437326639996943e-06,
+      "loss": 0.7378,
+      "step": 4654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433462474201168e-06,
+      "loss": 0.7137,
+      "step": 4655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42960003142507e-06,
+      "loss": 0.7566,
+      "step": 4656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4257393126821258e-06,
+      "loss": 0.8253,
+      "step": 4657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4218803189853735e-06,
+      "loss": 0.6802,
+      "step": 4658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4180230513473835e-06,
+      "loss": 0.7505,
+      "step": 4659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4141675107802787e-06,
+      "loss": 0.7355,
+      "step": 4660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.410313698295734e-06,
+      "loss": 0.7596,
+      "step": 4661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4064616149049647e-06,
+      "loss": 0.7886,
+      "step": 4662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40261126161873e-06,
+      "loss": 0.7352,
+      "step": 4663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3987626394473474e-06,
+      "loss": 0.7528,
+      "step": 4664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3949157494006637e-06,
+      "loss": 0.7622,
+      "step": 4665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.391070592488086e-06,
+      "loss": 0.751,
+      "step": 4666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3872271697185556e-06,
+      "loss": 0.7485,
+      "step": 4667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3833854821005684e-06,
+      "loss": 0.2695,
+      "step": 4668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3795455306421566e-06,
+      "loss": 0.7754,
+      "step": 4669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.375707316350898e-06,
+      "loss": 0.6984,
+      "step": 4670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3718708402339217e-06,
+      "loss": 0.7859,
+      "step": 4671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368036103297894e-06,
+      "loss": 0.6621,
+      "step": 4672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3642031065490234e-06,
+      "loss": 0.7791,
+      "step": 4673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3603718509930705e-06,
+      "loss": 0.71,
+      "step": 4674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3565423376353303e-06,
+      "loss": 0.6853,
+      "step": 4675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352714567480645e-06,
+      "loss": 0.6924,
+      "step": 4676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3488885415333937e-06,
+      "loss": 0.6927,
+      "step": 4677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3450642607975115e-06,
+      "loss": 0.7695,
+      "step": 4678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3412417262764595e-06,
+      "loss": 0.7516,
+      "step": 4679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337420938973248e-06,
+      "loss": 0.7148,
+      "step": 4680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.333601899890432e-06,
+      "loss": 0.7664,
+      "step": 4681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3297846100301033e-06,
+      "loss": 0.2847,
+      "step": 4682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325969070393891e-06,
+      "loss": 0.7207,
+      "step": 4683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3221552819829763e-06,
+      "loss": 0.7325,
+      "step": 4684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3183432457980703e-06,
+      "loss": 0.7242,
+      "step": 4685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3145329628394263e-06,
+      "loss": 0.7339,
+      "step": 4686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310724434106846e-06,
+      "loss": 0.7766,
+      "step": 4687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306917660599659e-06,
+      "loss": 0.7655,
+      "step": 4688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3031126433167395e-06,
+      "loss": 0.7308,
+      "step": 4689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.299309383256505e-06,
+      "loss": 0.7493,
+      "step": 4690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295507881416905e-06,
+      "loss": 0.6868,
+      "step": 4691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2917081387954273e-06,
+      "loss": 0.7871,
+      "step": 4692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.287910156389109e-06,
+      "loss": 0.7223,
+      "step": 4693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2841139351945138e-06,
+      "loss": 0.8044,
+      "step": 4694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280319476207744e-06,
+      "loss": 0.2794,
+      "step": 4695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2765267804244495e-06,
+      "loss": 0.7295,
+      "step": 4696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2727358488398053e-06,
+      "loss": 0.728,
+      "step": 4697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2689466824485284e-06,
+      "loss": 0.7441,
+      "step": 4698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2651592822448785e-06,
+      "loss": 0.2424,
+      "step": 4699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.261373649222642e-06,
+      "loss": 0.6942,
+      "step": 4700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257589784375146e-06,
+      "loss": 0.7997,
+      "step": 4701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253807688695253e-06,
+      "loss": 0.7245,
+      "step": 4702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250027363175369e-06,
+      "loss": 0.7227,
+      "step": 4703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246248808807422e-06,
+      "loss": 0.7114,
+      "step": 4704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2424720265828813e-06,
+      "loss": 0.7062,
+      "step": 4705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2386970174927556e-06,
+      "loss": 0.2799,
+      "step": 4706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2349237825275836e-06,
+      "loss": 0.6989,
+      "step": 4707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2311523226774342e-06,
+      "loss": 0.7139,
+      "step": 4708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227382638931924e-06,
+      "loss": 0.7068,
+      "step": 4709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223614732280189e-06,
+      "loss": 0.7283,
+      "step": 4710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2198486037109087e-06,
+      "loss": 0.7346,
+      "step": 4711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2160842542122862e-06,
+      "loss": 0.2643,
+      "step": 4712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2123216847720727e-06,
+      "loss": 0.7762,
+      "step": 4713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2085608963775384e-06,
+      "loss": 0.7585,
+      "step": 4714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2048018900154898e-06,
+      "loss": 0.771,
+      "step": 4715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2010446666722726e-06,
+      "loss": 0.2685,
+      "step": 4716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1972892273337562e-06,
+      "loss": 0.7651,
+      "step": 4717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1935355729853436e-06,
+      "loss": 0.7251,
+      "step": 4718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1897837046119763e-06,
+      "loss": 0.2549,
+      "step": 4719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1860336231981182e-06,
+      "loss": 0.7761,
+      "step": 4720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1822853297277657e-06,
+      "loss": 0.7759,
+      "step": 4721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.178538825184454e-06,
+      "loss": 0.7734,
+      "step": 4722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1747941105512402e-06,
+      "loss": 0.7654,
+      "step": 4723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171051186810712e-06,
+      "loss": 0.7723,
+      "step": 4724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1673100549449953e-06,
+      "loss": 0.7417,
+      "step": 4725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1635707159357375e-06,
+      "loss": 0.7231,
+      "step": 4726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159833170764116e-06,
+      "loss": 0.7399,
+      "step": 4727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1560974204108462e-06,
+      "loss": 0.7383,
+      "step": 4728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152363465856163e-06,
+      "loss": 0.7412,
+      "step": 4729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.14863130807983e-06,
+      "loss": 0.7759,
+      "step": 4730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1449009480611492e-06,
+      "loss": 0.2672,
+      "step": 4731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141172386778942e-06,
+      "loss": 0.7405,
+      "step": 4732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137445625211557e-06,
+      "loss": 0.7085,
+      "step": 4733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.133720664336879e-06,
+      "loss": 0.7585,
+      "step": 4734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.129997505132314e-06,
+      "loss": 0.7505,
+      "step": 4735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.126276148574794e-06,
+      "loss": 0.7605,
+      "step": 4736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1225565956407755e-06,
+      "loss": 0.7961,
+      "step": 4737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11883884730626e-06,
+      "loss": 0.734,
+      "step": 4738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1151229045467544e-06,
+      "loss": 0.834,
+      "step": 4739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1114087683372983e-06,
+      "loss": 0.7271,
+      "step": 4740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107696439652462e-06,
+      "loss": 0.7817,
+      "step": 4741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103985919466338e-06,
+      "loss": 0.7629,
+      "step": 4742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.100277208752541e-06,
+      "loss": 0.7463,
+      "step": 4743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096570308484215e-06,
+      "loss": 0.7371,
+      "step": 4744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0928652196340315e-06,
+      "loss": 0.7048,
+      "step": 4745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0891619431741806e-06,
+      "loss": 0.707,
+      "step": 4746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0854604800763767e-06,
+      "loss": 0.2643,
+      "step": 4747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0817608313118684e-06,
+      "loss": 0.678,
+      "step": 4748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0780629978514186e-06,
+      "loss": 0.7726,
+      "step": 4749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0743669806653107e-06,
+      "loss": 0.8059,
+      "step": 4750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070672780723365e-06,
+      "loss": 0.7468,
+      "step": 4751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066980398994914e-06,
+      "loss": 0.6688,
+      "step": 4752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063289836448814e-06,
+      "loss": 0.7313,
+      "step": 4753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0596010940534514e-06,
+      "loss": 0.2798,
+      "step": 4754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055914172776727e-06,
+      "loss": 0.7192,
+      "step": 4755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0522290735860626e-06,
+      "loss": 0.7015,
+      "step": 4756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0485457974484147e-06,
+      "loss": 0.7581,
+      "step": 4757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0448643453302463e-06,
+      "loss": 0.712,
+      "step": 4758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041184718197547e-06,
+      "loss": 0.7012,
+      "step": 4759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0375069170158346e-06,
+      "loss": 0.7362,
+      "step": 4760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.033830942750139e-06,
+      "loss": 0.7323,
+      "step": 4761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0301567963650104e-06,
+      "loss": 0.7563,
+      "step": 4762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0264844788245273e-06,
+      "loss": 0.6984,
+      "step": 4763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0228139910922827e-06,
+      "loss": 0.7001,
+      "step": 4764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0191453341313858e-06,
+      "loss": 0.7098,
+      "step": 4765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0154785089044768e-06,
+      "loss": 0.7074,
+      "step": 4766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011813516373705e-06,
+      "loss": 0.7357,
+      "step": 4767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008150357500742e-06,
+      "loss": 0.764,
+      "step": 4768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.004489033246776e-06,
+      "loss": 0.7368,
+      "step": 4769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000829544572522e-06,
+      "loss": 0.7556,
+      "step": 4770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9971718924382053e-06,
+      "loss": 0.7693,
+      "step": 4771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9935160778035653e-06,
+      "loss": 0.7722,
+      "step": 4772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989862101627877e-06,
+      "loss": 0.8312,
+      "step": 4773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9862099648699173e-06,
+      "loss": 0.7771,
+      "step": 4774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982559668487979e-06,
+      "loss": 0.7207,
+      "step": 4775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978911213439887e-06,
+      "loss": 0.7856,
+      "step": 4776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9752646006829678e-06,
+      "loss": 0.2761,
+      "step": 4777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9716198311740717e-06,
+      "loss": 0.7096,
+      "step": 4778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.967976905869563e-06,
+      "loss": 0.7189,
+      "step": 4779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.964335825725326e-06,
+      "loss": 0.7343,
+      "step": 4780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9606965916967578e-06,
+      "loss": 0.2632,
+      "step": 4781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.957059204738766e-06,
+      "loss": 0.7313,
+      "step": 4782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953423665805787e-06,
+      "loss": 0.7025,
+      "step": 4783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.94978997585176e-06,
+      "loss": 0.7157,
+      "step": 4784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9461581358301407e-06,
+      "loss": 0.7014,
+      "step": 4785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942528146693907e-06,
+      "loss": 0.7223,
+      "step": 4786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9389000093955456e-06,
+      "loss": 0.7677,
+      "step": 4787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935273724887051e-06,
+      "loss": 0.7045,
+      "step": 4788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9316492941199482e-06,
+      "loss": 0.7732,
+      "step": 4789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9280267180452594e-06,
+      "loss": 0.7954,
+      "step": 4790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9244059976135254e-06,
+      "loss": 0.2349,
+      "step": 4791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.920787133774807e-06,
+      "loss": 0.7052,
+      "step": 4792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91717012747867e-06,
+      "loss": 0.7738,
+      "step": 4793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9135549796741914e-06,
+      "loss": 0.7463,
+      "step": 4794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.90994169130997e-06,
+      "loss": 0.7499,
+      "step": 4795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.906330263334106e-06,
+      "loss": 0.7683,
+      "step": 4796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9027206966942166e-06,
+      "loss": 0.7651,
+      "step": 4797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899112992337434e-06,
+      "loss": 0.7177,
+      "step": 4798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8955071512103948e-06,
+      "loss": 0.718,
+      "step": 4799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8919031742592485e-06,
+      "loss": 0.7532,
+      "step": 4800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8883010624296615e-06,
+      "loss": 0.7432,
+      "step": 4801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8847008166668035e-06,
+      "loss": 0.702,
+      "step": 4802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8811024379153573e-06,
+      "loss": 0.6831,
+      "step": 4803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8775059271195125e-06,
+      "loss": 0.7333,
+      "step": 4804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873911285222978e-06,
+      "loss": 0.6924,
+      "step": 4805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8703185131689627e-06,
+      "loss": 0.6831,
+      "step": 4806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8667276119001863e-06,
+      "loss": 0.6992,
+      "step": 4807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.863138582358881e-06,
+      "loss": 0.8013,
+      "step": 4808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8595514254867894e-06,
+      "loss": 0.7261,
+      "step": 4809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855966142225155e-06,
+      "loss": 0.7073,
+      "step": 4810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85238273351474e-06,
+      "loss": 0.6492,
+      "step": 4811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8488012002958066e-06,
+      "loss": 0.7122,
+      "step": 4812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8452215435081255e-06,
+      "loss": 0.6808,
+      "step": 4813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8416437640909755e-06,
+      "loss": 0.687,
+      "step": 4814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8380678629831504e-06,
+      "loss": 0.7327,
+      "step": 4815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.834493841122942e-06,
+      "loss": 0.7214,
+      "step": 4816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830921699448148e-06,
+      "loss": 0.7343,
+      "step": 4817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8273514388960843e-06,
+      "loss": 0.7142,
+      "step": 4818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823783060403561e-06,
+      "loss": 0.7142,
+      "step": 4819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820216564906897e-06,
+      "loss": 0.7291,
+      "step": 4820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8166519533419232e-06,
+      "loss": 0.7555,
+      "step": 4821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8130892266439724e-06,
+      "loss": 0.7717,
+      "step": 4822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809528385747877e-06,
+      "loss": 0.7615,
+      "step": 4823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8059694315879883e-06,
+      "loss": 0.769,
+      "step": 4824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802412365098148e-06,
+      "loss": 0.7271,
+      "step": 4825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7988571872117098e-06,
+      "loss": 0.6637,
+      "step": 4826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7953038988615334e-06,
+      "loss": 0.7039,
+      "step": 4827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7917525009799783e-06,
+      "loss": 0.7219,
+      "step": 4828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7882029944989074e-06,
+      "loss": 0.7384,
+      "step": 4829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7846553803496956e-06,
+      "loss": 0.6929,
+      "step": 4830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7811096594632113e-06,
+      "loss": 0.7308,
+      "step": 4831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7775658327698285e-06,
+      "loss": 0.6923,
+      "step": 4832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774023901199432e-06,
+      "loss": 0.7432,
+      "step": 4833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.770483865681398e-06,
+      "loss": 0.7793,
+      "step": 4834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76694572714461e-06,
+      "loss": 0.7456,
+      "step": 4835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7634094865174588e-06,
+      "loss": 0.7319,
+      "step": 4836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7598751447278292e-06,
+      "loss": 0.6735,
+      "step": 4837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756342702703112e-06,
+      "loss": 0.7402,
+      "step": 4838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7528121613701954e-06,
+      "loss": 0.7487,
+      "step": 4839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.749283521655478e-06,
+      "loss": 0.736,
+      "step": 4840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7457567844848497e-06,
+      "loss": 0.7627,
+      "step": 4841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7422319507837027e-06,
+      "loss": 0.7377,
+      "step": 4842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7387090214769386e-06,
+      "loss": 0.7944,
+      "step": 4843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735187997488945e-06,
+      "loss": 0.7758,
+      "step": 4844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731668879743622e-06,
+      "loss": 0.7103,
+      "step": 4845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.728151669164367e-06,
+      "loss": 0.7253,
+      "step": 4846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7246363666740728e-06,
+      "loss": 0.7075,
+      "step": 4847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7211229731951317e-06,
+      "loss": 0.6571,
+      "step": 4848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7176114896494342e-06,
+      "loss": 0.7854,
+      "step": 4849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7141019169583794e-06,
+      "loss": 0.7463,
+      "step": 4850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.710594256042854e-06,
+      "loss": 0.697,
+      "step": 4851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7070885078232455e-06,
+      "loss": 0.7097,
+      "step": 4852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7035846732194447e-06,
+      "loss": 0.7223,
+      "step": 4853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7000827531508343e-06,
+      "loss": 0.7822,
+      "step": 4854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696582748536296e-06,
+      "loss": 0.6788,
+      "step": 4855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.693084660294213e-06,
+      "loss": 0.7488,
+      "step": 4856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6895884893424606e-06,
+      "loss": 0.7397,
+      "step": 4857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68609423659841e-06,
+      "loss": 0.7837,
+      "step": 4858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6826019029789387e-06,
+      "loss": 0.7706,
+      "step": 4859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.679111489400409e-06,
+      "loss": 0.77,
+      "step": 4860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.675622996778682e-06,
+      "loss": 0.7198,
+      "step": 4861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672136426029124e-06,
+      "loss": 0.7465,
+      "step": 4862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6686517780665867e-06,
+      "loss": 0.7552,
+      "step": 4863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6651690538054163e-06,
+      "loss": 0.7046,
+      "step": 4864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661688254159467e-06,
+      "loss": 0.722,
+      "step": 4865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6582093800420738e-06,
+      "loss": 0.6656,
+      "step": 4866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6547324323660716e-06,
+      "loss": 0.714,
+      "step": 4867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.651257412043795e-06,
+      "loss": 0.7264,
+      "step": 4868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6477843199870647e-06,
+      "loss": 0.7395,
+      "step": 4869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.644313157107198e-06,
+      "loss": 0.7437,
+      "step": 4870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640843924315012e-06,
+      "loss": 0.6404,
+      "step": 4871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6373766225208084e-06,
+      "loss": 0.7991,
+      "step": 4872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633911252634388e-06,
+      "loss": 0.7637,
+      "step": 4873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.630447815565038e-06,
+      "loss": 0.7557,
+      "step": 4874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626986312221551e-06,
+      "loss": 0.7339,
+      "step": 4875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6235267435122013e-06,
+      "loss": 0.6721,
+      "step": 4876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620069110344755e-06,
+      "loss": 0.7866,
+      "step": 4877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61661341362648e-06,
+      "loss": 0.7001,
+      "step": 4878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6131596542641246e-06,
+      "loss": 0.7358,
+      "step": 4879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6097078331639383e-06,
+      "loss": 0.7832,
+      "step": 4880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606257951231659e-06,
+      "loss": 0.7417,
+      "step": 4881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602810009372513e-06,
+      "loss": 0.7412,
+      "step": 4882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5993640084912197e-06,
+      "loss": 0.7449,
+      "step": 4883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595919949491985e-06,
+      "loss": 0.6863,
+      "step": 4884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.592477833278516e-06,
+      "loss": 0.6914,
+      "step": 4885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.589037660753999e-06,
+      "loss": 0.7035,
+      "step": 4886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5855994328211132e-06,
+      "loss": 0.7516,
+      "step": 4887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.582163150382032e-06,
+      "loss": 0.7825,
+      "step": 4888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578728814338416e-06,
+      "loss": 0.8035,
+      "step": 4889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575296425591407e-06,
+      "loss": 0.7634,
+      "step": 4890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5718659850416516e-06,
+      "loss": 0.693,
+      "step": 4891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5684374935892743e-06,
+      "loss": 0.7581,
+      "step": 4892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565010952133885e-06,
+      "loss": 0.7551,
+      "step": 4893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.561586361574595e-06,
+      "loss": 0.691,
+      "step": 4894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558163722809992e-06,
+      "loss": 0.7839,
+      "step": 4895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554743036738154e-06,
+      "loss": 0.7174,
+      "step": 4896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5513243042566536e-06,
+      "loss": 0.6781,
+      "step": 4897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547907526262543e-06,
+      "loss": 0.7866,
+      "step": 4898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5444927036523613e-06,
+      "loss": 0.7784,
+      "step": 4899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541079837322141e-06,
+      "loss": 0.8,
+      "step": 4900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5376689281673985e-06,
+      "loss": 0.7328,
+      "step": 4901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5342599770831287e-06,
+      "loss": 0.6759,
+      "step": 4902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5308529849638277e-06,
+      "loss": 0.7861,
+      "step": 4903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527447952703467e-06,
+      "loss": 0.2872,
+      "step": 4904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524044881195502e-06,
+      "loss": 0.7449,
+      "step": 4905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5206437713328858e-06,
+      "loss": 0.7698,
+      "step": 4906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5172446240080437e-06,
+      "loss": 0.6992,
+      "step": 4907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5138474401128944e-06,
+      "loss": 0.7069,
+      "step": 4908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5104522205388326e-06,
+      "loss": 0.709,
+      "step": 4909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507058966176751e-06,
+      "loss": 0.7794,
+      "step": 4910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503667677917014e-06,
+      "loss": 0.7438,
+      "step": 4911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500278356649476e-06,
+      "loss": 0.7312,
+      "step": 4912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4968910032634762e-06,
+      "loss": 0.7261,
+      "step": 4913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4935056186478356e-06,
+      "loss": 0.7705,
+      "step": 4914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4901222036908535e-06,
+      "loss": 0.8014,
+      "step": 4915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4867407592803217e-06,
+      "loss": 0.6792,
+      "step": 4916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.483361286303513e-06,
+      "loss": 0.7614,
+      "step": 4917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479983785647179e-06,
+      "loss": 0.7371,
+      "step": 4918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4766082581975503e-06,
+      "loss": 0.7418,
+      "step": 4919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.473234704840353e-06,
+      "loss": 0.7191,
+      "step": 4920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469863126460782e-06,
+      "loss": 0.7549,
+      "step": 4921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4664935239435184e-06,
+      "loss": 0.7573,
+      "step": 4922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.463125898172729e-06,
+      "loss": 0.6836,
+      "step": 4923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4597602500320574e-06,
+      "loss": 0.7478,
+      "step": 4924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4563965804046253e-06,
+      "loss": 0.78,
+      "step": 4925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453034890173045e-06,
+      "loss": 0.7383,
+      "step": 4926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.449675180219402e-06,
+      "loss": 0.731,
+      "step": 4927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4463174514252597e-06,
+      "loss": 0.7427,
+      "step": 4928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4429617046716723e-06,
+      "loss": 0.7682,
+      "step": 4929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439607940839164e-06,
+      "loss": 0.7709,
+      "step": 4930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4362561608077383e-06,
+      "loss": 0.6942,
+      "step": 4931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4329063654568906e-06,
+      "loss": 0.7122,
+      "step": 4932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429558555665581e-06,
+      "loss": 0.6859,
+      "step": 4933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4262127323122542e-06,
+      "loss": 0.7185,
+      "step": 4934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422868896274837e-06,
+      "loss": 0.7937,
+      "step": 4935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419527048430731e-06,
+      "loss": 0.7053,
+      "step": 4936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4161871896568133e-06,
+      "loss": 0.7073,
+      "step": 4937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412849320829449e-06,
+      "loss": 0.7251,
+      "step": 4938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4095134428244703e-06,
+      "loss": 0.7474,
+      "step": 4939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4061795565171897e-06,
+      "loss": 0.7715,
+      "step": 4940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402847662782404e-06,
+      "loss": 0.7117,
+      "step": 4941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3995177624943787e-06,
+      "loss": 0.7137,
+      "step": 4942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3961898565268605e-06,
+      "loss": 0.7588,
+      "step": 4943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392863945753069e-06,
+      "loss": 0.7153,
+      "step": 4944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3895400310457074e-06,
+      "loss": 0.7964,
+      "step": 4945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.386218113276949e-06,
+      "loss": 0.6774,
+      "step": 4946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3828981933184414e-06,
+      "loss": 0.7151,
+      "step": 4947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3795802720413175e-06,
+      "loss": 0.8054,
+      "step": 4948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3762643503161776e-06,
+      "loss": 0.7776,
+      "step": 4949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3729504290130956e-06,
+      "loss": 0.7759,
+      "step": 4950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.369638509001627e-06,
+      "loss": 0.7332,
+      "step": 4951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.366328591150804e-06,
+      "loss": 0.6887,
+      "step": 4952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3630206763291254e-06,
+      "loss": 0.2437,
+      "step": 4953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.359714765404564e-06,
+      "loss": 0.7064,
+      "step": 4954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3564108592445766e-06,
+      "loss": 0.7783,
+      "step": 4955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.353108958716088e-06,
+      "loss": 0.7036,
+      "step": 4956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349809064685491e-06,
+      "loss": 0.2412,
+      "step": 4957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3465111780186656e-06,
+      "loss": 0.7229,
+      "step": 4958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.343215299580952e-06,
+      "loss": 0.7437,
+      "step": 4959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3399214302371686e-06,
+      "loss": 0.708,
+      "step": 4960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3366295708516117e-06,
+      "loss": 0.7394,
+      "step": 4961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3333397222880417e-06,
+      "loss": 0.7373,
+      "step": 4962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330051885409691e-06,
+      "loss": 0.7336,
+      "step": 4963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3267660610792763e-06,
+      "loss": 0.7587,
+      "step": 4964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323482250158975e-06,
+      "loss": 0.7952,
+      "step": 4965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320200453510434e-06,
+      "loss": 0.7596,
+      "step": 4966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3169206719947834e-06,
+      "loss": 0.7756,
+      "step": 4967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3136429064726175e-06,
+      "loss": 0.8035,
+      "step": 4968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3103671578039966e-06,
+      "loss": 0.2842,
+      "step": 4969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3070934268484636e-06,
+      "loss": 0.7781,
+      "step": 4970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3038217144650233e-06,
+      "loss": 0.7446,
+      "step": 4971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3005520215121504e-06,
+      "loss": 0.7888,
+      "step": 4972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297284348847797e-06,
+      "loss": 0.6813,
+      "step": 4973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2940186973293797e-06,
+      "loss": 0.7551,
+      "step": 4974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2907550678137847e-06,
+      "loss": 0.709,
+      "step": 4975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287493461157365e-06,
+      "loss": 0.7404,
+      "step": 4976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2842338782159544e-06,
+      "loss": 0.2621,
+      "step": 4977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280976319844842e-06,
+      "loss": 0.7382,
+      "step": 4978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2777207868987905e-06,
+      "loss": 0.7181,
+      "step": 4979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2744672802320378e-06,
+      "loss": 0.7336,
+      "step": 4980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2712158006982797e-06,
+      "loss": 0.2723,
+      "step": 4981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267966349150684e-06,
+      "loss": 0.7515,
+      "step": 4982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.264718926441891e-06,
+      "loss": 0.7515,
+      "step": 4983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2614735334240034e-06,
+      "loss": 0.7439,
+      "step": 4984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25823017094859e-06,
+      "loss": 0.751,
+      "step": 4985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25498883986669e-06,
+      "loss": 0.7627,
+      "step": 4986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2517495410288135e-06,
+      "loss": 0.7559,
+      "step": 4987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2485122752849308e-06,
+      "loss": 0.2406,
+      "step": 4988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245277043484477e-06,
+      "loss": 0.8101,
+      "step": 4989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2420438464763628e-06,
+      "loss": 0.7644,
+      "step": 4990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2388126851089575e-06,
+      "loss": 0.7864,
+      "step": 4991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2355835602300947e-06,
+      "loss": 0.7766,
+      "step": 4992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2323564726870827e-06,
+      "loss": 0.7468,
+      "step": 4993
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2291314233266868e-06,
+      "loss": 0.7939,
+      "step": 4994
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225908412995139e-06,
+      "loss": 0.6489,
+      "step": 4995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2226874425381428e-06,
+      "loss": 0.7771,
+      "step": 4996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2194685128008576e-06,
+      "loss": 0.7625,
+      "step": 4997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2162516246279086e-06,
+      "loss": 0.7518,
+      "step": 4998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2130367788633943e-06,
+      "loss": 0.7258,
+      "step": 4999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2098239763508666e-06,
+      "loss": 0.6947,
+      "step": 5000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2066132179333445e-06,
+      "loss": 0.6808,
+      "step": 5001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203404504453315e-06,
+      "loss": 0.6215,
+      "step": 5002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2001978367527234e-06,
+      "loss": 0.2674,
+      "step": 5003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196993215672977e-06,
+      "loss": 0.78,
+      "step": 5004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1937906420549536e-06,
+      "loss": 0.7072,
+      "step": 5005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1905901167389888e-06,
+      "loss": 0.7036,
+      "step": 5006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1873916405648754e-06,
+      "loss": 0.7715,
+      "step": 5007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1841952143718813e-06,
+      "loss": 0.7056,
+      "step": 5008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181000838998727e-06,
+      "loss": 0.7737,
+      "step": 5009
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1778085152835957e-06,
+      "loss": 0.7478,
+      "step": 5010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1746182440641327e-06,
+      "loss": 0.7524,
+      "step": 5011
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17143002617745e-06,
+      "loss": 0.8088,
+      "step": 5012
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.168243862460114e-06,
+      "loss": 0.7051,
+      "step": 5013
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165059753748153e-06,
+      "loss": 0.6946,
+      "step": 5014
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161877700877064e-06,
+      "loss": 0.7622,
+      "step": 5015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1586977046817937e-06,
+      "loss": 0.7786,
+      "step": 5016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1555197659967506e-06,
+      "loss": 0.2679,
+      "step": 5017
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1523438856558152e-06,
+      "loss": 0.2707,
+      "step": 5018
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.149170064492314e-06,
+      "loss": 0.7466,
+      "step": 5019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1459983033390365e-06,
+      "loss": 0.7374,
+      "step": 5020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1428286030282385e-06,
+      "loss": 0.7062,
+      "step": 5021
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1396609643916256e-06,
+      "loss": 0.7576,
+      "step": 5022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136495388260371e-06,
+      "loss": 0.7812,
+      "step": 5023
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133331875465099e-06,
+      "loss": 0.7424,
+      "step": 5024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1301704268359013e-06,
+      "loss": 0.759,
+      "step": 5025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1270110432023194e-06,
+      "loss": 0.7371,
+      "step": 5026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1238537253933543e-06,
+      "loss": 0.6975,
+      "step": 5027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1206984742374712e-06,
+      "loss": 0.7803,
+      "step": 5028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1175452905625894e-06,
+      "loss": 0.697,
+      "step": 5029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1143941751960786e-06,
+      "loss": 0.76,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1112451289647794e-06,
+      "loss": 0.7157,
+      "step": 5031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1080981526949806e-06,
+      "loss": 0.743,
+      "step": 5032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1049532472124266e-06,
+      "loss": 0.7812,
+      "step": 5033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1018104133423266e-06,
+      "loss": 0.7355,
+      "step": 5034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0986696519093386e-06,
+      "loss": 0.6725,
+      "step": 5035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0955309637375754e-06,
+      "loss": 0.7377,
+      "step": 5036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092394349650618e-06,
+      "loss": 0.7373,
+      "step": 5037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08925981047149e-06,
+      "loss": 0.7079,
+      "step": 5038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.086127347022673e-06,
+      "loss": 0.6996,
+      "step": 5039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0829969601261124e-06,
+      "loss": 0.701,
+      "step": 5040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0798686506032007e-06,
+      "loss": 0.2828,
+      "step": 5041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.076742419274783e-06,
+      "loss": 0.7262,
+      "step": 5042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0736182669611705e-06,
+      "loss": 0.7194,
+      "step": 5043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070496194482118e-06,
+      "loss": 0.7496,
+      "step": 5044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067376202656838e-06,
+      "loss": 0.7544,
+      "step": 5045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0642582923039965e-06,
+      "loss": 0.7932,
+      "step": 5046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061142464241719e-06,
+      "loss": 0.771,
+      "step": 5047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058028719287576e-06,
+      "loss": 0.7727,
+      "step": 5048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0549170582585943e-06,
+      "loss": 0.7581,
+      "step": 5049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0518074819712596e-06,
+      "loss": 0.7576,
+      "step": 5050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0486999912415027e-06,
+      "loss": 0.7334,
+      "step": 5051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0455945868847094e-06,
+      "loss": 0.7134,
+      "step": 5052
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0424912697157216e-06,
+      "loss": 0.7267,
+      "step": 5053
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.039390040548831e-06,
+      "loss": 0.6711,
+      "step": 5054
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036290900197776e-06,
+      "loss": 0.6901,
+      "step": 5055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0331938494757597e-06,
+      "loss": 0.7339,
+      "step": 5056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0300988891954265e-06,
+      "loss": 0.7085,
+      "step": 5057
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027006020168872e-06,
+      "loss": 0.7115,
+      "step": 5058
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023915243207649e-06,
+      "loss": 0.7642,
+      "step": 5059
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0208265591227614e-06,
+      "loss": 0.7452,
+      "step": 5060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017739968724659e-06,
+      "loss": 0.6903,
+      "step": 5061
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0146554728232413e-06,
+      "loss": 0.7981,
+      "step": 5062
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011573072227867e-06,
+      "loss": 0.7537,
+      "step": 5063
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0084927677473364e-06,
+      "loss": 0.7181,
+      "step": 5064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0054145601899e-06,
+      "loss": 0.6931,
+      "step": 5065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002338450363266e-06,
+      "loss": 0.7205,
+      "step": 5066
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999264439074584e-06,
+      "loss": 0.7747,
+      "step": 5067
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.996192527130455e-06,
+      "loss": 0.7749,
+      "step": 5068
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9931227153369336e-06,
+      "loss": 0.726,
+      "step": 5069
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990055004499517e-06,
+      "loss": 0.671,
+      "step": 5070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9869893954231535e-06,
+      "loss": 0.7168,
+      "step": 5071
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9839258889122435e-06,
+      "loss": 0.7029,
+      "step": 5072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9808644857706316e-06,
+      "loss": 0.7043,
+      "step": 5073
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977805186801607e-06,
+      "loss": 0.7229,
+      "step": 5074
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9747479928079173e-06,
+      "loss": 0.7607,
+      "step": 5075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9716929045917497e-06,
+      "loss": 0.7526,
+      "step": 5076
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968639922954738e-06,
+      "loss": 0.7065,
+      "step": 5077
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.96558904869797e-06,
+      "loss": 0.7668,
+      "step": 5078
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962540282621975e-06,
+      "loss": 0.7339,
+      "step": 5079
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9594936255267318e-06,
+      "loss": 0.7303,
+      "step": 5080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956449078211661e-06,
+      "loss": 0.7952,
+      "step": 5081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9534066414756383e-06,
+      "loss": 0.7231,
+      "step": 5082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950366316116977e-06,
+      "loss": 0.668,
+      "step": 5083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9473281029334414e-06,
+      "loss": 0.7283,
+      "step": 5084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944292002722241e-06,
+      "loss": 0.7042,
+      "step": 5085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941258016280031e-06,
+      "loss": 0.7422,
+      "step": 5086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938226144402906e-06,
+      "loss": 0.7207,
+      "step": 5087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9351963878864164e-06,
+      "loss": 0.7751,
+      "step": 5088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9321687475255515e-06,
+      "loss": 0.7878,
+      "step": 5089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929143224114741e-06,
+      "loss": 0.7543,
+      "step": 5090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9261198184478693e-06,
+      "loss": 0.7319,
+      "step": 5091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9230985313182584e-06,
+      "loss": 0.7194,
+      "step": 5092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920079363518671e-06,
+      "loss": 0.7719,
+      "step": 5093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9170623158413235e-06,
+      "loss": 0.7211,
+      "step": 5094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914047389077872e-06,
+      "loss": 0.7681,
+      "step": 5095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110345840194124e-06,
+      "loss": 0.6887,
+      "step": 5096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908023901456485e-06,
+      "loss": 0.7352,
+      "step": 5097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9050153421790796e-06,
+      "loss": 0.7076,
+      "step": 5098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9020089069766202e-06,
+      "loss": 0.7118,
+      "step": 5099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8990045966379745e-06,
+      "loss": 0.7307,
+      "step": 5100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.896002411951463e-06,
+      "loss": 0.7585,
+      "step": 5101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8930023537048348e-06,
+      "loss": 0.7195,
+      "step": 5102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8900044226852876e-06,
+      "loss": 0.8015,
+      "step": 5103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8870086196794635e-06,
+      "loss": 0.6605,
+      "step": 5104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8840149454734414e-06,
+      "loss": 0.7621,
+      "step": 5105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8810234008527395e-06,
+      "loss": 0.7357,
+      "step": 5106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8780339866023278e-06,
+      "loss": 0.7734,
+      "step": 5107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8750467035066068e-06,
+      "loss": 0.7427,
+      "step": 5108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.872061552349419e-06,
+      "loss": 0.681,
+      "step": 5109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.869078533914056e-06,
+      "loss": 0.7644,
+      "step": 5110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8660976489832394e-06,
+      "loss": 0.7188,
+      "step": 5111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8631188983391356e-06,
+      "loss": 0.7426,
+      "step": 5112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8601422827633531e-06,
+      "loss": 0.7251,
+      "step": 5113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8571678030369367e-06,
+      "loss": 0.688,
+      "step": 5114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.854195459940371e-06,
+      "loss": 0.6798,
+      "step": 5115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.851225254253579e-06,
+      "loss": 0.7307,
+      "step": 5116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8482571867559295e-06,
+      "loss": 0.719,
+      "step": 5117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8452912582262239e-06,
+      "loss": 0.6766,
+      "step": 5118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8423274694426996e-06,
+      "loss": 0.7456,
+      "step": 5119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8393658211830423e-06,
+      "loss": 0.7911,
+      "step": 5120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8364063142243692e-06,
+      "loss": 0.7177,
+      "step": 5121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8334489493432329e-06,
+      "loss": 0.7596,
+      "step": 5122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8304937273156342e-06,
+      "loss": 0.7341,
+      "step": 5123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8275406489170033e-06,
+      "loss": 0.7554,
+      "step": 5124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8245897149222059e-06,
+      "loss": 0.7856,
+      "step": 5125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8216409261055556e-06,
+      "loss": 0.7095,
+      "step": 5126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8186942832407939e-06,
+      "loss": 0.7871,
+      "step": 5127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8157497871010999e-06,
+      "loss": 0.7502,
+      "step": 5128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8128074384590922e-06,
+      "loss": 0.7461,
+      "step": 5129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809867238086831e-06,
+      "loss": 0.8071,
+      "step": 5130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8069291867558014e-06,
+      "loss": 0.2576,
+      "step": 5131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8039932852369303e-06,
+      "loss": 0.7201,
+      "step": 5132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8010595343005832e-06,
+      "loss": 0.7219,
+      "step": 5133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7981279347165582e-06,
+      "loss": 0.724,
+      "step": 5134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.795198487254085e-06,
+      "loss": 0.7109,
+      "step": 5135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792271192681838e-06,
+      "loss": 0.739,
+      "step": 5136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7893460517679218e-06,
+      "loss": 0.7942,
+      "step": 5137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7864230652798697e-06,
+      "loss": 0.7825,
+      "step": 5138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783502233984662e-06,
+      "loss": 0.7773,
+      "step": 5139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7805835586487042e-06,
+      "loss": 0.6833,
+      "step": 5140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7776670400378383e-06,
+      "loss": 0.7396,
+      "step": 5141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774752678917344e-06,
+      "loss": 0.7168,
+      "step": 5142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7718404760519303e-06,
+      "loss": 0.7373,
+      "step": 5143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7689304322057398e-06,
+      "loss": 0.8059,
+      "step": 5144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7660225481423542e-06,
+      "loss": 0.749,
+      "step": 5145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763116824624782e-06,
+      "loss": 0.7318,
+      "step": 5146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7602132624154666e-06,
+      "loss": 0.6882,
+      "step": 5147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7573118622762885e-06,
+      "loss": 0.7041,
+      "step": 5148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7544126249685567e-06,
+      "loss": 0.7374,
+      "step": 5149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7515155512530113e-06,
+      "loss": 0.7169,
+      "step": 5150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7486206418898266e-06,
+      "loss": 0.7633,
+      "step": 5151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7457278976386128e-06,
+      "loss": 0.7466,
+      "step": 5152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7428373192584058e-06,
+      "loss": 0.7949,
+      "step": 5153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7399489075076748e-06,
+      "loss": 0.7771,
+      "step": 5154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7370626631443256e-06,
+      "loss": 0.689,
+      "step": 5155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.734178586925689e-06,
+      "loss": 0.2413,
+      "step": 5156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7312966796085262e-06,
+      "loss": 0.6744,
+      "step": 5157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728416941949037e-06,
+      "loss": 0.7433,
+      "step": 5158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7255393747028458e-06,
+      "loss": 0.7584,
+      "step": 5159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7226639786250066e-06,
+      "loss": 0.7822,
+      "step": 5160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7197907544700088e-06,
+      "loss": 0.7529,
+      "step": 5161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7169197029917684e-06,
+      "loss": 0.7172,
+      "step": 5162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7140508249436283e-06,
+      "loss": 0.7017,
+      "step": 5163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7111841210783697e-06,
+      "loss": 0.7214,
+      "step": 5164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083195921481943e-06,
+      "loss": 0.7566,
+      "step": 5165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054572389047418e-06,
+      "loss": 0.7245,
+      "step": 5166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025970620990717e-06,
+      "loss": 0.7397,
+      "step": 5167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6997390624816801e-06,
+      "loss": 0.7632,
+      "step": 5168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6968832408024894e-06,
+      "loss": 0.6512,
+      "step": 5169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6940295978108446e-06,
+      "loss": 0.7317,
+      "step": 5170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.691178134255531e-06,
+      "loss": 0.7184,
+      "step": 5171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6883288508847518e-06,
+      "loss": 0.6951,
+      "step": 5172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6854817484461406e-06,
+      "loss": 0.7102,
+      "step": 5173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6826368276867634e-06,
+      "loss": 0.7245,
+      "step": 5174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6797940893531084e-06,
+      "loss": 0.8093,
+      "step": 5175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6769535341910902e-06,
+      "loss": 0.7411,
+      "step": 5176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6741151629460584e-06,
+      "loss": 0.2554,
+      "step": 5177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6712789763627823e-06,
+      "loss": 0.7117,
+      "step": 5178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668444975185457e-06,
+      "loss": 0.7404,
+      "step": 5179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.665613160157712e-06,
+      "loss": 0.2419,
+      "step": 5180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6627835320225971e-06,
+      "loss": 0.7402,
+      "step": 5181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6599560915225876e-06,
+      "loss": 0.7357,
+      "step": 5182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6571308393995867e-06,
+      "loss": 0.7773,
+      "step": 5183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6543077763949266e-06,
+      "loss": 0.7661,
+      "step": 5184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6514869032493608e-06,
+      "loss": 0.7609,
+      "step": 5185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6486682207030646e-06,
+      "loss": 0.739,
+      "step": 5186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6458517294956512e-06,
+      "loss": 0.7029,
+      "step": 5187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6430374303661467e-06,
+      "loss": 0.684,
+      "step": 5188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640225324053003e-06,
+      "loss": 0.7239,
+      "step": 5189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6374154112941064e-06,
+      "loss": 0.775,
+      "step": 5190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6346076928267575e-06,
+      "loss": 0.7067,
+      "step": 5191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6318021693876818e-06,
+      "loss": 0.7002,
+      "step": 5192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6289988417130375e-06,
+      "loss": 0.7013,
+      "step": 5193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6261977105383974e-06,
+      "loss": 0.7151,
+      "step": 5194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6233987765987602e-06,
+      "loss": 0.8069,
+      "step": 5195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6206020406285529e-06,
+      "loss": 0.7412,
+      "step": 5196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6178075033616204e-06,
+      "loss": 0.71,
+      "step": 5197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6150151655312284e-06,
+      "loss": 0.769,
+      "step": 5198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.612225027870077e-06,
+      "loss": 0.8035,
+      "step": 5199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6094370911102741e-06,
+      "loss": 0.7322,
+      "step": 5200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.606651355983364e-06,
+      "loss": 0.7388,
+      "step": 5201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6038678232203008e-06,
+      "loss": 0.7426,
+      "step": 5202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6010864935514725e-06,
+      "loss": 0.8074,
+      "step": 5203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5983073677066795e-06,
+      "loss": 0.7593,
+      "step": 5204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5955304464151456e-06,
+      "loss": 0.7181,
+      "step": 5205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592755730405523e-06,
+      "loss": 0.7195,
+      "step": 5206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5899832204058774e-06,
+      "loss": 0.6958,
+      "step": 5207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5872129171436978e-06,
+      "loss": 0.7344,
+      "step": 5208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5844448213458985e-06,
+      "loss": 0.7393,
+      "step": 5209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5816789337388094e-06,
+      "loss": 0.2536,
+      "step": 5210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.57891525504818e-06,
+      "loss": 0.7699,
+      "step": 5211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5761537859991872e-06,
+      "loss": 0.6956,
+      "step": 5212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5733945273164219e-06,
+      "loss": 0.6927,
+      "step": 5213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5706374797238944e-06,
+      "loss": 0.7769,
+      "step": 5214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.567882643945041e-06,
+      "loss": 0.712,
+      "step": 5215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565130020702713e-06,
+      "loss": 0.7952,
+      "step": 5216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5623796107191824e-06,
+      "loss": 0.7001,
+      "step": 5217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5596314147161362e-06,
+      "loss": 0.7438,
+      "step": 5218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5568854334146898e-06,
+      "loss": 0.7271,
+      "step": 5219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5541416675353694e-06,
+      "loss": 0.7839,
+      "step": 5220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5514001177981208e-06,
+      "loss": 0.7703,
+      "step": 5221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5486607849223146e-06,
+      "loss": 0.7275,
+      "step": 5222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5459236696267333e-06,
+      "loss": 0.7729,
+      "step": 5223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5431887726295757e-06,
+      "loss": 0.7648,
+      "step": 5224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.540456094648467e-06,
+      "loss": 0.6981,
+      "step": 5225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5377256364004445e-06,
+      "loss": 0.7343,
+      "step": 5226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5349973986019607e-06,
+      "loss": 0.6548,
+      "step": 5227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5322713819688928e-06,
+      "loss": 0.7734,
+      "step": 5228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.52954758721653e-06,
+      "loss": 0.7498,
+      "step": 5229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526826015059576e-06,
+      "loss": 0.7437,
+      "step": 5230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5241066662121595e-06,
+      "loss": 0.7246,
+      "step": 5231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.52138954138782e-06,
+      "loss": 0.708,
+      "step": 5232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5186746412995113e-06,
+      "loss": 0.7336,
+      "step": 5233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5159619666596115e-06,
+      "loss": 0.7528,
+      "step": 5234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513251518179908e-06,
+      "loss": 0.73,
+      "step": 5235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5105432965716038e-06,
+      "loss": 0.7639,
+      "step": 5236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507837302545322e-06,
+      "loss": 0.6781,
+      "step": 5237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5051335368111009e-06,
+      "loss": 0.7236,
+      "step": 5238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5024320000783898e-06,
+      "loss": 0.7356,
+      "step": 5239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499732693056055e-06,
+      "loss": 0.6558,
+      "step": 5240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4970356164523813e-06,
+      "loss": 0.6843,
+      "step": 5241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4943407709750634e-06,
+      "loss": 0.7375,
+      "step": 5242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4916481573312103e-06,
+      "loss": 0.7336,
+      "step": 5243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4889577762273521e-06,
+      "loss": 0.7522,
+      "step": 5244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4862696283694256e-06,
+      "loss": 0.7593,
+      "step": 5245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4835837144627829e-06,
+      "loss": 0.7566,
+      "step": 5246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.480900035212196e-06,
+      "loss": 0.7267,
+      "step": 5247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4782185913218438e-06,
+      "loss": 0.7465,
+      "step": 5248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.475539383495318e-06,
+      "loss": 0.7347,
+      "step": 5249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4728624124356317e-06,
+      "loss": 0.6782,
+      "step": 5250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.470187678845204e-06,
+      "loss": 0.7361,
+      "step": 5251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4675151834258683e-06,
+      "loss": 0.731,
+      "step": 5252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.464844926878869e-06,
+      "loss": 0.7733,
+      "step": 5253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4621769099048687e-06,
+      "loss": 0.7573,
+      "step": 5254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.459511133203939e-06,
+      "loss": 0.7544,
+      "step": 5255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4568475974755592e-06,
+      "loss": 0.7141,
+      "step": 5256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454186303418631e-06,
+      "loss": 0.7257,
+      "step": 5257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4515272517314583e-06,
+      "loss": 0.772,
+      "step": 5258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4488704431117584e-06,
+      "loss": 0.7729,
+      "step": 5259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.446215878256667e-06,
+      "loss": 0.6777,
+      "step": 5260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435635578627216e-06,
+      "loss": 0.7532,
+      "step": 5261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4409134826258752e-06,
+      "loss": 0.7368,
+      "step": 5262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4382656532414951e-06,
+      "loss": 0.7554,
+      "step": 5263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4356200704043522e-06,
+      "loss": 0.7239,
+      "step": 5264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.432976734808631e-06,
+      "loss": 0.7336,
+      "step": 5265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4303356471479302e-06,
+      "loss": 0.7174,
+      "step": 5266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4276968081152532e-06,
+      "loss": 0.6625,
+      "step": 5267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4250602184030126e-06,
+      "loss": 0.7471,
+      "step": 5268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4224258787030388e-06,
+      "loss": 0.7175,
+      "step": 5269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.419793789706564e-06,
+      "loss": 0.7273,
+      "step": 5270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4171639521042302e-06,
+      "loss": 0.6974,
+      "step": 5271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4145363665860945e-06,
+      "loss": 0.7615,
+      "step": 5272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4119110338416186e-06,
+      "loss": 0.8186,
+      "step": 5273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4092879545596748e-06,
+      "loss": 0.7894,
+      "step": 5274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4066671294285384e-06,
+      "loss": 0.7542,
+      "step": 5275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4040485591359044e-06,
+      "loss": 0.7029,
+      "step": 5276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4014322443688677e-06,
+      "loss": 0.7483,
+      "step": 5277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3988181858139293e-06,
+      "loss": 0.7698,
+      "step": 5278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3962063841570083e-06,
+      "loss": 0.667,
+      "step": 5279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393596840083422e-06,
+      "loss": 0.7827,
+      "step": 5280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3909895542778996e-06,
+      "loss": 0.7332,
+      "step": 5281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3883845274245778e-06,
+      "loss": 0.7164,
+      "step": 5282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.385781760207e-06,
+      "loss": 0.7556,
+      "step": 5283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3831812533081146e-06,
+      "loss": 0.7574,
+      "step": 5284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3805830074102823e-06,
+      "loss": 0.76,
+      "step": 5285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3779870231952653e-06,
+      "loss": 0.7788,
+      "step": 5286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3753933013442333e-06,
+      "loss": 0.6309,
+      "step": 5287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.372801842537762e-06,
+      "loss": 0.7908,
+      "step": 5288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3702126474558387e-06,
+      "loss": 0.6812,
+      "step": 5289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3676257167778496e-06,
+      "loss": 0.7786,
+      "step": 5290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3650410511825885e-06,
+      "loss": 0.7373,
+      "step": 5291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3624586513482596e-06,
+      "loss": 0.7664,
+      "step": 5292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3598785179524676e-06,
+      "loss": 0.79,
+      "step": 5293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3573006516722208e-06,
+      "loss": 0.7488,
+      "step": 5294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3547250531839407e-06,
+      "loss": 0.7068,
+      "step": 5295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3521517231634463e-06,
+      "loss": 0.7268,
+      "step": 5296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3495806622859608e-06,
+      "loss": 0.7201,
+      "step": 5297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3470118712261215e-06,
+      "loss": 0.6788,
+      "step": 5298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344445350657959e-06,
+      "loss": 0.7441,
+      "step": 5299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3418811012549126e-06,
+      "loss": 0.7568,
+      "step": 5300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3393191236898283e-06,
+      "loss": 0.7271,
+      "step": 5301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336759418634953e-06,
+      "loss": 0.7026,
+      "step": 5302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3342019867619348e-06,
+      "loss": 0.6881,
+      "step": 5303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3316468287418326e-06,
+      "loss": 0.7035,
+      "step": 5304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3290939452451035e-06,
+      "loss": 0.7271,
+      "step": 5305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3265433369416035e-06,
+      "loss": 0.7471,
+      "step": 5306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3239950045006022e-06,
+      "loss": 0.6958,
+      "step": 5307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3214489485907679e-06,
+      "loss": 0.6613,
+      "step": 5308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.318905169880167e-06,
+      "loss": 0.7371,
+      "step": 5309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3163636690362691e-06,
+      "loss": 0.7944,
+      "step": 5310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3138244467259553e-06,
+      "loss": 0.7771,
+      "step": 5311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3112875036154981e-06,
+      "loss": 0.7101,
+      "step": 5312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.308752840370574e-06,
+      "loss": 0.7324,
+      "step": 5313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3062204576562687e-06,
+      "loss": 0.7607,
+      "step": 5314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3036903561370606e-06,
+      "loss": 0.7328,
+      "step": 5315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301162536476832e-06,
+      "loss": 0.709,
+      "step": 5316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2986369993388704e-06,
+      "loss": 0.7408,
+      "step": 5317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2961137453858607e-06,
+      "loss": 0.7062,
+      "step": 5318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2935927752798871e-06,
+      "loss": 0.7952,
+      "step": 5319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2910740896824404e-06,
+      "loss": 0.8059,
+      "step": 5320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2885576892544061e-06,
+      "loss": 0.7184,
+      "step": 5321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.286043574656074e-06,
+      "loss": 0.7321,
+      "step": 5322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2835317465471297e-06,
+      "loss": 0.7861,
+      "step": 5323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281022205586665e-06,
+      "loss": 0.7725,
+      "step": 5324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2785149524331665e-06,
+      "loss": 0.7834,
+      "step": 5325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2760099877445208e-06,
+      "loss": 0.6595,
+      "step": 5326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2735073121780194e-06,
+      "loss": 0.7422,
+      "step": 5327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710069263903478e-06,
+      "loss": 0.6265,
+      "step": 5328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2685088310375881e-06,
+      "loss": 0.792,
+      "step": 5329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2660130267752313e-06,
+      "loss": 0.6632,
+      "step": 5330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263519514258159e-06,
+      "loss": 0.7185,
+      "step": 5331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2610282941406505e-06,
+      "loss": 0.7766,
+      "step": 5332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2585393670763923e-06,
+      "loss": 0.7478,
+      "step": 5333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2560527337184613e-06,
+      "loss": 0.7769,
+      "step": 5334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2535683947193333e-06,
+      "loss": 0.7107,
+      "step": 5335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2510863507308867e-06,
+      "loss": 0.6724,
+      "step": 5336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.248606602404394e-06,
+      "loss": 0.2686,
+      "step": 5337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2461291503905248e-06,
+      "loss": 0.78,
+      "step": 5338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2436539953393489e-06,
+      "loss": 0.7683,
+      "step": 5339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2411811379003324e-06,
+      "loss": 0.7385,
+      "step": 5340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2387105787223352e-06,
+      "loss": 0.7483,
+      "step": 5341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.236242318453621e-06,
+      "loss": 0.7559,
+      "step": 5342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2337763577418427e-06,
+      "loss": 0.6989,
+      "step": 5343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2313126972340572e-06,
+      "loss": 0.7812,
+      "step": 5344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2288513375767098e-06,
+      "loss": 0.7426,
+      "step": 5345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2263922794156512e-06,
+      "loss": 0.7151,
+      "step": 5346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2239355233961214e-06,
+      "loss": 0.7563,
+      "step": 5347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2214810701627544e-06,
+      "loss": 0.7859,
+      "step": 5348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2190289203595907e-06,
+      "loss": 0.7767,
+      "step": 5349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2165790746300553e-06,
+      "loss": 0.6555,
+      "step": 5350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2141315336169713e-06,
+      "loss": 0.7556,
+      "step": 5351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2116862979625633e-06,
+      "loss": 0.77,
+      "step": 5352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2092433683084447e-06,
+      "loss": 0.7891,
+      "step": 5353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.206802745295622e-06,
+      "loss": 0.7268,
+      "step": 5354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2043644295645052e-06,
+      "loss": 0.7698,
+      "step": 5355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201928421754891e-06,
+      "loss": 0.6805,
+      "step": 5356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1994947225059728e-06,
+      "loss": 0.6888,
+      "step": 5357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1970633324563374e-06,
+      "loss": 0.7076,
+      "step": 5358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1946342522439702e-06,
+      "loss": 0.7477,
+      "step": 5359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1922074825062458e-06,
+      "loss": 0.6533,
+      "step": 5360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1897830238799302e-06,
+      "loss": 0.7214,
+      "step": 5361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1873608770011923e-06,
+      "loss": 0.7236,
+      "step": 5362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1849410425055873e-06,
+      "loss": 0.7563,
+      "step": 5363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1825235210280605e-06,
+      "loss": 0.7542,
+      "step": 5364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180108313202961e-06,
+      "loss": 0.7284,
+      "step": 5365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1776954196640234e-06,
+      "loss": 0.7679,
+      "step": 5366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1752848410443718e-06,
+      "loss": 0.6964,
+      "step": 5367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1728765779765327e-06,
+      "loss": 0.7852,
+      "step": 5368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1704706310924186e-06,
+      "loss": 0.7046,
+      "step": 5369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1680670010233319e-06,
+      "loss": 0.6559,
+      "step": 5370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1656656883999761e-06,
+      "loss": 0.7271,
+      "step": 5371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1632666938524384e-06,
+      "loss": 0.7246,
+      "step": 5372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1608700180101973e-06,
+      "loss": 0.6743,
+      "step": 5373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158475661502132e-06,
+      "loss": 0.7744,
+      "step": 5374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1560836249565032e-06,
+      "loss": 0.7374,
+      "step": 5375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.153693909000967e-06,
+      "loss": 0.7795,
+      "step": 5376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1513065142625724e-06,
+      "loss": 0.2542,
+      "step": 5377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1489214413677564e-06,
+      "loss": 0.7109,
+      "step": 5378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1465386909423448e-06,
+      "loss": 0.7163,
+      "step": 5379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14415826361156e-06,
+      "loss": 0.7085,
+      "step": 5380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1417801600000134e-06,
+      "loss": 0.7003,
+      "step": 5381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1394043807317023e-06,
+      "loss": 0.2871,
+      "step": 5382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.137030926430015e-06,
+      "loss": 0.7035,
+      "step": 5383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1346597977177364e-06,
+      "loss": 0.7563,
+      "step": 5384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1322909952170336e-06,
+      "loss": 0.7278,
+      "step": 5385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1299245195494634e-06,
+      "loss": 0.803,
+      "step": 5386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1275603713359806e-06,
+      "loss": 0.6838,
+      "step": 5387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1251985511969199e-06,
+      "loss": 0.6812,
+      "step": 5388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1228390597520077e-06,
+      "loss": 0.7659,
+      "step": 5389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1204818976203612e-06,
+      "loss": 0.748,
+      "step": 5390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1181270654204868e-06,
+      "loss": 0.7343,
+      "step": 5391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1157745637702766e-06,
+      "loss": 0.7561,
+      "step": 5392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1134243932870125e-06,
+      "loss": 0.7269,
+      "step": 5393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1110765545873669e-06,
+      "loss": 0.7529,
+      "step": 5394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1087310482873981e-06,
+      "loss": 0.751,
+      "step": 5395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1063878750025492e-06,
+      "loss": 0.792,
+      "step": 5396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1040470353476606e-06,
+      "loss": 0.741,
+      "step": 5397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10170852993695e-06,
+      "loss": 0.2667,
+      "step": 5398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0993723593840278e-06,
+      "loss": 0.7988,
+      "step": 5399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0970385243018934e-06,
+      "loss": 0.7544,
+      "step": 5400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094707025302928e-06,
+      "loss": 0.6921,
+      "step": 5401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0923778629989024e-06,
+      "loss": 0.7474,
+      "step": 5402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0900510380009787e-06,
+      "loss": 0.7748,
+      "step": 5403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0877265509196998e-06,
+      "loss": 0.7629,
+      "step": 5404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0854044023649934e-06,
+      "loss": 0.7201,
+      "step": 5405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083084592946183e-06,
+      "loss": 0.6571,
+      "step": 5406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0807671232719697e-06,
+      "loss": 0.7397,
+      "step": 5407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0784519939504423e-06,
+      "loss": 0.6898,
+      "step": 5408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0761392055890796e-06,
+      "loss": 0.7029,
+      "step": 5409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0738287587947415e-06,
+      "loss": 0.7751,
+      "step": 5410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0715206541736733e-06,
+      "loss": 0.7434,
+      "step": 5411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0692148923315127e-06,
+      "loss": 0.7876,
+      "step": 5412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0669114738732733e-06,
+      "loss": 0.7275,
+      "step": 5413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0646103994033597e-06,
+      "loss": 0.6941,
+      "step": 5414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0623116695255586e-06,
+      "loss": 0.2659,
+      "step": 5415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0600152848430455e-06,
+      "loss": 0.6879,
+      "step": 5416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0577212459583775e-06,
+      "loss": 0.7205,
+      "step": 5417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0554295534734938e-06,
+      "loss": 0.7515,
+      "step": 5418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0531402079897235e-06,
+      "loss": 0.7317,
+      "step": 5419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.050853210107775e-06,
+      "loss": 0.7369,
+      "step": 5420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0485685604277428e-06,
+      "loss": 0.6782,
+      "step": 5421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0462862595491074e-06,
+      "loss": 0.761,
+      "step": 5422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0440063080707286e-06,
+      "loss": 0.7267,
+      "step": 5423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0417287065908521e-06,
+      "loss": 0.724,
+      "step": 5424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0394534557071056e-06,
+      "loss": 0.2295,
+      "step": 5425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0371805560165048e-06,
+      "loss": 0.6726,
+      "step": 5426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0349100081154418e-06,
+      "loss": 0.7715,
+      "step": 5427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.032641812599694e-06,
+      "loss": 0.7135,
+      "step": 5428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0303759700644255e-06,
+      "loss": 0.7659,
+      "step": 5429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0281124811041777e-06,
+      "loss": 0.7535,
+      "step": 5430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.025851346312874e-06,
+      "loss": 0.765,
+      "step": 5431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0235925662838265e-06,
+      "loss": 0.7196,
+      "step": 5432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0213361416097246e-06,
+      "loss": 0.7366,
+      "step": 5433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0190820728826368e-06,
+      "loss": 0.7573,
+      "step": 5434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0168303606940221e-06,
+      "loss": 0.2592,
+      "step": 5435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.014581005634715e-06,
+      "loss": 0.2453,
+      "step": 5436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0123340082949286e-06,
+      "loss": 0.8015,
+      "step": 5437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0100893692642666e-06,
+      "loss": 0.7747,
+      "step": 5438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007847089131706e-06,
+      "loss": 0.7354,
+      "step": 5439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0056071684856073e-06,
+      "loss": 0.7147,
+      "step": 5440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0033696079137135e-06,
+      "loss": 0.683,
+      "step": 5441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.001134408003147e-06,
+      "loss": 0.7019,
+      "step": 5442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.98901569340407e-07,
+      "loss": 0.7382,
+      "step": 5443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.96671092511383e-07,
+      "loss": 0.7542,
+      "step": 5444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.944429781013344e-07,
+      "loss": 0.713,
+      "step": 5445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.922172266949048e-07,
+      "loss": 0.7362,
+      "step": 5446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89993838876121e-07,
+      "loss": 0.7704,
+      "step": 5447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.877728152283839e-07,
+      "loss": 0.7419,
+      "step": 5448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.855541563344773e-07,
+      "loss": 0.7258,
+      "step": 5449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.8333786277656e-07,
+      "loss": 0.7366,
+      "step": 5450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.811239351361823e-07,
+      "loss": 0.7212,
+      "step": 5451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.789123739942608e-07,
+      "loss": 0.7764,
+      "step": 5452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.767031799310932e-07,
+      "loss": 0.7148,
+      "step": 5453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74496353526364e-07,
+      "loss": 0.7493,
+      "step": 5454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.722918953591288e-07,
+      "loss": 0.7184,
+      "step": 5455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700898060078212e-07,
+      "loss": 0.7454,
+      "step": 5456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.678900860502605e-07,
+      "loss": 0.7428,
+      "step": 5457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65692736063638e-07,
+      "loss": 0.7317,
+      "step": 5458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.63497756624525e-07,
+      "loss": 0.7109,
+      "step": 5459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.613051483088687e-07,
+      "loss": 0.7435,
+      "step": 5460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.591149116920008e-07,
+      "loss": 0.7681,
+      "step": 5461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.569270473486236e-07,
+      "loss": 0.6294,
+      "step": 5462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.547415558528184e-07,
+      "loss": 0.7214,
+      "step": 5463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.525584377780472e-07,
+      "loss": 0.7305,
+      "step": 5464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.503776936971454e-07,
+      "loss": 0.8018,
+      "step": 5465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.481993241823261e-07,
+      "loss": 0.7561,
+      "step": 5466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.460233298051835e-07,
+      "loss": 0.7584,
+      "step": 5467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.43849711136684e-07,
+      "loss": 0.7791,
+      "step": 5468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.41678468747168e-07,
+      "loss": 0.6738,
+      "step": 5469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.395096032063633e-07,
+      "loss": 0.7537,
+      "step": 5470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.373431150833634e-07,
+      "loss": 0.6605,
+      "step": 5471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.351790049466392e-07,
+      "loss": 0.7086,
+      "step": 5472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.330172733640452e-07,
+      "loss": 0.686,
+      "step": 5473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.308579209028045e-07,
+      "loss": 0.6847,
+      "step": 5474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.287009481295162e-07,
+      "loss": 0.6698,
+      "step": 5475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.265463556101595e-07,
+      "loss": 0.7194,
+      "step": 5476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.243941439100857e-07,
+      "loss": 0.6736,
+      "step": 5477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.222443135940207e-07,
+      "loss": 0.6803,
+      "step": 5478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.200968652260689e-07,
+      "loss": 0.7679,
+      "step": 5479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.179517993697085e-07,
+      "loss": 0.7638,
+      "step": 5480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.158091165877869e-07,
+      "loss": 0.2684,
+      "step": 5481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.136688174425368e-07,
+      "loss": 0.771,
+      "step": 5482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.115309024955576e-07,
+      "loss": 0.6974,
+      "step": 5483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.093953723078253e-07,
+      "loss": 0.7925,
+      "step": 5484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.072622274396869e-07,
+      "loss": 0.7399,
+      "step": 5485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.051314684508705e-07,
+      "loss": 0.6991,
+      "step": 5486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.030030959004754e-07,
+      "loss": 0.708,
+      "step": 5487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.008771103469694e-07,
+      "loss": 0.2468,
+      "step": 5488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.987535123482028e-07,
+      "loss": 0.698,
+      "step": 5489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.966323024613932e-07,
+      "loss": 0.7175,
+      "step": 5490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.9451348124313e-07,
+      "loss": 0.733,
+      "step": 5491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.923970492493827e-07,
+      "loss": 0.7749,
+      "step": 5492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.902830070354906e-07,
+      "loss": 0.7063,
+      "step": 5493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.881713551561632e-07,
+      "loss": 0.2455,
+      "step": 5494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.860620941654852e-07,
+      "loss": 0.2621,
+      "step": 5495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.83955224616917e-07,
+      "loss": 0.7991,
+      "step": 5496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.818507470632854e-07,
+      "loss": 0.7313,
+      "step": 5497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.797486620567919e-07,
+      "loss": 0.7634,
+      "step": 5498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.776489701490143e-07,
+      "loss": 0.7253,
+      "step": 5499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.755516718908975e-07,
+      "loss": 0.7874,
+      "step": 5500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.734567678327566e-07,
+      "loss": 0.7871,
+      "step": 5501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.713642585242876e-07,
+      "loss": 0.7576,
+      "step": 5502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.692741445145491e-07,
+      "loss": 0.7498,
+      "step": 5503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.671864263519725e-07,
+      "loss": 0.7968,
+      "step": 5504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.65101104584366e-07,
+      "loss": 0.7429,
+      "step": 5505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.630181797589044e-07,
+      "loss": 0.7705,
+      "step": 5506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.609376524221324e-07,
+      "loss": 0.7631,
+      "step": 5507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.588595231199703e-07,
+      "loss": 0.672,
+      "step": 5508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.567837923977073e-07,
+      "loss": 0.6724,
+      "step": 5509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.547104607999978e-07,
+      "loss": 0.7009,
+      "step": 5510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.526395288708778e-07,
+      "loss": 0.8386,
+      "step": 5511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.505709971537435e-07,
+      "loss": 0.7164,
+      "step": 5512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.48504866191363e-07,
+      "loss": 0.2622,
+      "step": 5513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.464411365258829e-07,
+      "loss": 0.7255,
+      "step": 5514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.443798086988087e-07,
+      "loss": 0.6799,
+      "step": 5515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.423208832510199e-07,
+      "loss": 0.7827,
+      "step": 5516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.402643607227701e-07,
+      "loss": 0.6833,
+      "step": 5517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.382102416536741e-07,
+      "loss": 0.2526,
+      "step": 5518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.361585265827232e-07,
+      "loss": 0.7478,
+      "step": 5519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.341092160482723e-07,
+      "loss": 0.7367,
+      "step": 5520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.320623105880499e-07,
+      "loss": 0.6802,
+      "step": 5521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.300178107391521e-07,
+      "loss": 0.7628,
+      "step": 5522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.27975717038042e-07,
+      "loss": 0.6715,
+      "step": 5523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.259360300205544e-07,
+      "loss": 0.7604,
+      "step": 5524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.238987502218887e-07,
+      "loss": 0.6748,
+      "step": 5525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.218638781766142e-07,
+      "loss": 0.7048,
+      "step": 5526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198314144186725e-07,
+      "loss": 0.6941,
+      "step": 5527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.178013594813683e-07,
+      "loss": 0.7573,
+      "step": 5528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.157737138973743e-07,
+      "loss": 0.7751,
+      "step": 5529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.137484781987315e-07,
+      "loss": 0.6979,
+      "step": 5530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.117256529168537e-07,
+      "loss": 0.6931,
+      "step": 5531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.097052385825155e-07,
+      "loss": 0.761,
+      "step": 5532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.076872357258591e-07,
+      "loss": 0.7732,
+      "step": 5533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.056716448764001e-07,
+      "loss": 0.7332,
+      "step": 5534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.036584665630165e-07,
+      "loss": 0.7062,
+      "step": 5535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.016477013139512e-07,
+      "loss": 0.7013,
+      "step": 5536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.996393496568211e-07,
+      "loss": 0.6801,
+      "step": 5537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.976334121186025e-07,
+      "loss": 0.7413,
+      "step": 5538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.956298892256408e-07,
+      "loss": 0.7196,
+      "step": 5539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.936287815036525e-07,
+      "loss": 0.7551,
+      "step": 5540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.916300894777118e-07,
+      "loss": 0.2614,
+      "step": 5541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.896338136722648e-07,
+      "loss": 0.7111,
+      "step": 5542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876399546111224e-07,
+      "loss": 0.6886,
+      "step": 5543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.856485128174618e-07,
+      "loss": 0.69,
+      "step": 5544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.836594888138227e-07,
+      "loss": 0.7454,
+      "step": 5545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.816728831221165e-07,
+      "loss": 0.7808,
+      "step": 5546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.796886962636141e-07,
+      "loss": 0.7111,
+      "step": 5547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.777069287589534e-07,
+      "loss": 0.7605,
+      "step": 5548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.757275811281417e-07,
+      "loss": 0.7917,
+      "step": 5549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.737506538905449e-07,
+      "loss": 0.7402,
+      "step": 5550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.717761475648966e-07,
+      "loss": 0.7192,
+      "step": 5551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.698040626692971e-07,
+      "loss": 0.726,
+      "step": 5552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.67834399721209e-07,
+      "loss": 0.7032,
+      "step": 5553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.6586715923746e-07,
+      "loss": 0.6974,
+      "step": 5554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.639023417342395e-07,
+      "loss": 0.7649,
+      "step": 5555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.619399477271072e-07,
+      "loss": 0.2539,
+      "step": 5556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.599799777309803e-07,
+      "loss": 0.7312,
+      "step": 5557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.580224322601459e-07,
+      "loss": 0.7388,
+      "step": 5558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.560673118282513e-07,
+      "loss": 0.7483,
+      "step": 5559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.541146169483094e-07,
+      "loss": 0.6818,
+      "step": 5560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.521643481326935e-07,
+      "loss": 0.7084,
+      "step": 5561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.50216505893141e-07,
+      "loss": 0.7634,
+      "step": 5562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.482710907407564e-07,
+      "loss": 0.7754,
+      "step": 5563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463281031860059e-07,
+      "loss": 0.6785,
+      "step": 5564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.443875437387138e-07,
+      "loss": 0.7552,
+      "step": 5565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.424494129080751e-07,
+      "loss": 0.6807,
+      "step": 5566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.405137112026417e-07,
+      "loss": 0.7175,
+      "step": 5567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.385804391303286e-07,
+      "loss": 0.7502,
+      "step": 5568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.366495971984178e-07,
+      "loss": 0.7192,
+      "step": 5569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.347211859135483e-07,
+      "loss": 0.7045,
+      "step": 5570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.327952057817234e-07,
+      "loss": 0.7322,
+      "step": 5571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.308716573083107e-07,
+      "loss": 0.7131,
+      "step": 5572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.289505409980369e-07,
+      "loss": 0.7854,
+      "step": 5573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.270318573549884e-07,
+      "loss": 0.6832,
+      "step": 5574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.251156068826215e-07,
+      "loss": 0.7229,
+      "step": 5575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.23201790083744e-07,
+      "loss": 0.6873,
+      "step": 5576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.212904074605309e-07,
+      "loss": 0.7437,
+      "step": 5577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.193814595145199e-07,
+      "loss": 0.2412,
+      "step": 5578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.174749467466057e-07,
+      "loss": 0.7603,
+      "step": 5579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.155708696570441e-07,
+      "loss": 0.6802,
+      "step": 5580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.13669228745456e-07,
+      "loss": 0.7662,
+      "step": 5581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.117700245108194e-07,
+      "loss": 0.7294,
+      "step": 5582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098732574514722e-07,
+      "loss": 0.7239,
+      "step": 5583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.079789280651184e-07,
+      "loss": 0.6902,
+      "step": 5584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.060870368488159e-07,
+      "loss": 0.7595,
+      "step": 5585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.041975842989857e-07,
+      "loss": 0.7969,
+      "step": 5586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.023105709114097e-07,
+      "loss": 0.7351,
+      "step": 5587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.004259971812277e-07,
+      "loss": 0.7731,
+      "step": 5588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.985438636029418e-07,
+      "loss": 0.6903,
+      "step": 5589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.966641706704103e-07,
+      "loss": 0.6967,
+      "step": 5590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.947869188768553e-07,
+      "loss": 0.2542,
+      "step": 5591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.92912108714856e-07,
+      "loss": 0.736,
+      "step": 5592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.910397406763469e-07,
+      "loss": 0.7338,
+      "step": 5593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.891698152526338e-07,
+      "loss": 0.7551,
+      "step": 5594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.873023329343697e-07,
+      "loss": 0.7378,
+      "step": 5595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.854372942115705e-07,
+      "loss": 0.728,
+      "step": 5596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.835746995736103e-07,
+      "loss": 0.7292,
+      "step": 5597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817145495092248e-07,
+      "loss": 0.6969,
+      "step": 5598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.798568445065057e-07,
+      "loss": 0.7615,
+      "step": 5599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.78001585052901e-07,
+      "loss": 0.6416,
+      "step": 5600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.761487716352222e-07,
+      "loss": 0.7383,
+      "step": 5601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.742984047396361e-07,
+      "loss": 0.7513,
+      "step": 5602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.724504848516666e-07,
+      "loss": 0.7418,
+      "step": 5603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.70605012456198e-07,
+      "loss": 0.7509,
+      "step": 5604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.687619880374708e-07,
+      "loss": 0.7659,
+      "step": 5605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.669214120790812e-07,
+      "loss": 0.7639,
+      "step": 5606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.650832850639888e-07,
+      "loss": 0.7783,
+      "step": 5607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.632476074745054e-07,
+      "loss": 0.7173,
+      "step": 5608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.614143797922989e-07,
+      "loss": 0.7412,
+      "step": 5609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.595836024984026e-07,
+      "loss": 0.7473,
+      "step": 5610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.577552760731976e-07,
+      "loss": 0.7373,
+      "step": 5611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.559294009964257e-07,
+      "loss": 0.7085,
+      "step": 5612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.54105977747187e-07,
+      "loss": 0.7634,
+      "step": 5613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.522850068039366e-07,
+      "loss": 0.7644,
+      "step": 5614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.504664886444844e-07,
+      "loss": 0.7903,
+      "step": 5615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.486504237460012e-07,
+      "loss": 0.7363,
+      "step": 5616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.46836812585011e-07,
+      "loss": 0.7383,
+      "step": 5617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.450256556373913e-07,
+      "loss": 0.6885,
+      "step": 5618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.432169533783816e-07,
+      "loss": 0.2595,
+      "step": 5619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.414107062825747e-07,
+      "loss": 0.6859,
+      "step": 5620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.396069148239181e-07,
+      "loss": 0.6964,
+      "step": 5621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.378055794757132e-07,
+      "loss": 0.7925,
+      "step": 5622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.360067007106241e-07,
+      "loss": 0.7146,
+      "step": 5623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.342102790006632e-07,
+      "loss": 0.749,
+      "step": 5624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324163148171991e-07,
+      "loss": 0.6607,
+      "step": 5625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.306248086309608e-07,
+      "loss": 0.7253,
+      "step": 5626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.288357609120266e-07,
+      "loss": 0.7383,
+      "step": 5627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.270491721298288e-07,
+      "loss": 0.7795,
+      "step": 5628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.252650427531648e-07,
+      "loss": 0.7654,
+      "step": 5629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.234833732501743e-07,
+      "loss": 0.7598,
+      "step": 5630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.217041640883581e-07,
+      "loss": 0.7742,
+      "step": 5631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.199274157345669e-07,
+      "loss": 0.7567,
+      "step": 5632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.181531286550135e-07,
+      "loss": 0.7485,
+      "step": 5633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.163813033152577e-07,
+      "loss": 0.7495,
+      "step": 5634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.146119401802142e-07,
+      "loss": 0.2557,
+      "step": 5635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.128450397141561e-07,
+      "loss": 0.7377,
+      "step": 5636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.110806023807058e-07,
+      "loss": 0.2459,
+      "step": 5637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.093186286428398e-07,
+      "loss": 0.7646,
+      "step": 5638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.075591189628927e-07,
+      "loss": 0.673,
+      "step": 5639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.058020738025472e-07,
+      "loss": 0.7917,
+      "step": 5640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.040474936228391e-07,
+      "loss": 0.7488,
+      "step": 5641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022953788841635e-07,
+      "loss": 0.7219,
+      "step": 5642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.005457300462635e-07,
+      "loss": 0.7458,
+      "step": 5643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.987985475682345e-07,
+      "loss": 0.6587,
+      "step": 5644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970538319085295e-07,
+      "loss": 0.7109,
+      "step": 5645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.95311583524949e-07,
+      "loss": 0.7654,
+      "step": 5646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.935718028746485e-07,
+      "loss": 0.7634,
+      "step": 5647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.918344904141371e-07,
+      "loss": 0.7391,
+      "step": 5648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.900996465992759e-07,
+      "loss": 0.7625,
+      "step": 5649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.88367271885274e-07,
+      "loss": 0.7152,
+      "step": 5650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.866373667267e-07,
+      "loss": 0.7661,
+      "step": 5651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.849099315774687e-07,
+      "loss": 0.2623,
+      "step": 5652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.831849668908462e-07,
+      "loss": 0.72,
+      "step": 5653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.814624731194574e-07,
+      "loss": 0.7511,
+      "step": 5654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.797424507152715e-07,
+      "loss": 0.748,
+      "step": 5655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.780249001296134e-07,
+      "loss": 0.7188,
+      "step": 5656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.763098218131546e-07,
+      "loss": 0.8259,
+      "step": 5657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.745972162159263e-07,
+      "loss": 0.7778,
+      "step": 5658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.728870837873024e-07,
+      "loss": 0.7202,
+      "step": 5659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.711794249760105e-07,
+      "loss": 0.6831,
+      "step": 5660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.694742402301334e-07,
+      "loss": 0.7112,
+      "step": 5661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.677715299970998e-07,
+      "loss": 0.7172,
+      "step": 5662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.660712947236891e-07,
+      "loss": 0.7751,
+      "step": 5663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.643735348560331e-07,
+      "loss": 0.7366,
+      "step": 5664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.626782508396156e-07,
+      "loss": 0.7495,
+      "step": 5665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.609854431192685e-07,
+      "loss": 0.2531,
+      "step": 5666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.592951121391709e-07,
+      "loss": 0.7891,
+      "step": 5667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.576072583428604e-07,
+      "loss": 0.7661,
+      "step": 5668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.559218821732171e-07,
+      "loss": 0.7524,
+      "step": 5669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.542389840724716e-07,
+      "loss": 0.6819,
+      "step": 5670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.525585644822096e-07,
+      "loss": 0.7485,
+      "step": 5671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.508806238433629e-07,
+      "loss": 0.6794,
+      "step": 5672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.492051625962092e-07,
+      "loss": 0.7828,
+      "step": 5673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.475321811803824e-07,
+      "loss": 0.7104,
+      "step": 5674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.458616800348637e-07,
+      "loss": 0.6912,
+      "step": 5675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.441936595979791e-07,
+      "loss": 0.2411,
+      "step": 5676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.425281203074096e-07,
+      "loss": 0.8167,
+      "step": 5677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.408650626001821e-07,
+      "loss": 0.7993,
+      "step": 5678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.392044869126722e-07,
+      "loss": 0.7341,
+      "step": 5679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.375463936806058e-07,
+      "loss": 0.7444,
+      "step": 5680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.35890783339057e-07,
+      "loss": 0.7986,
+      "step": 5681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.342376563224449e-07,
+      "loss": 0.7003,
+      "step": 5682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.325870130645427e-07,
+      "loss": 0.7235,
+      "step": 5683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.309388539984706e-07,
+      "loss": 0.7163,
+      "step": 5684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.292931795566903e-07,
+      "loss": 0.7534,
+      "step": 5685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.27649990171023e-07,
+      "loss": 0.8059,
+      "step": 5686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.260092862726273e-07,
+      "loss": 0.7274,
+      "step": 5687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.243710682920155e-07,
+      "loss": 0.7434,
+      "step": 5688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227353366590471e-07,
+      "loss": 0.7358,
+      "step": 5689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.211020918029264e-07,
+      "loss": 0.7383,
+      "step": 5690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.194713341522073e-07,
+      "loss": 0.7247,
+      "step": 5691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.178430641347909e-07,
+      "loss": 0.767,
+      "step": 5692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.162172821779255e-07,
+      "loss": 0.7908,
+      "step": 5693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.145939887082074e-07,
+      "loss": 0.6923,
+      "step": 5694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.129731841515762e-07,
+      "loss": 0.7073,
+      "step": 5695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.113548689333247e-07,
+      "loss": 0.7039,
+      "step": 5696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.09739043478088e-07,
+      "loss": 0.793,
+      "step": 5697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.081257082098457e-07,
+      "loss": 0.7526,
+      "step": 5698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.065148635519323e-07,
+      "loss": 0.7297,
+      "step": 5699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.049065099270189e-07,
+      "loss": 0.7264,
+      "step": 5700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.033006477571311e-07,
+      "loss": 0.7311,
+      "step": 5701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.016972774636364e-07,
+      "loss": 0.7166,
+      "step": 5702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.000963994672492e-07,
+      "loss": 0.7068,
+      "step": 5703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984980141880313e-07,
+      "loss": 0.7219,
+      "step": 5704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.969021220453863e-07,
+      "loss": 0.6909,
+      "step": 5705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9530872345807e-07,
+      "loss": 0.2883,
+      "step": 5706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.937178188441804e-07,
+      "loss": 0.6829,
+      "step": 5707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.921294086211581e-07,
+      "loss": 0.7319,
+      "step": 5708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.90543493205795e-07,
+      "loss": 0.7518,
+      "step": 5709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.889600730142263e-07,
+      "loss": 0.7153,
+      "step": 5710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.873791484619284e-07,
+      "loss": 0.6879,
+      "step": 5711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.858007199637293e-07,
+      "loss": 0.7607,
+      "step": 5712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.842247879337992e-07,
+      "loss": 0.6108,
+      "step": 5713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.826513527856502e-07,
+      "loss": 0.7703,
+      "step": 5714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.810804149321469e-07,
+      "loss": 0.791,
+      "step": 5715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795119747854893e-07,
+      "loss": 0.7764,
+      "step": 5716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.779460327572283e-07,
+      "loss": 0.705,
+      "step": 5717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7638258925825855e-07,
+      "loss": 0.7407,
+      "step": 5718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.748216446988174e-07,
+      "loss": 0.792,
+      "step": 5719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.732631994884862e-07,
+      "loss": 0.7212,
+      "step": 5720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.717072540361933e-07,
+      "loss": 0.6744,
+      "step": 5721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701538087502089e-07,
+      "loss": 0.7053,
+      "step": 5722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.686028640381457e-07,
+      "loss": 0.7458,
+      "step": 5723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.670544203069649e-07,
+      "loss": 0.7839,
+      "step": 5724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6550847796296817e-07,
+      "loss": 0.7209,
+      "step": 5725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.63965037411801e-07,
+      "loss": 0.7822,
+      "step": 5726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6242409905845165e-07,
+      "loss": 0.2578,
+      "step": 5727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6088566330725446e-07,
+      "loss": 0.7693,
+      "step": 5728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.593497305618866e-07,
+      "loss": 0.7539,
+      "step": 5729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.578163012253645e-07,
+      "loss": 0.7428,
+      "step": 5730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5628537570005295e-07,
+      "loss": 0.708,
+      "step": 5731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5475695438765844e-07,
+      "loss": 0.7209,
+      "step": 5732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.532310376892257e-07,
+      "loss": 0.6691,
+      "step": 5733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5170762600515004e-07,
+      "loss": 0.7728,
+      "step": 5734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.501867197351617e-07,
+      "loss": 0.6826,
+      "step": 5735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4866831927834143e-07,
+      "loss": 0.7168,
+      "step": 5736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4715242503310495e-07,
+      "loss": 0.7834,
+      "step": 5737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.456390373972175e-07,
+      "loss": 0.6992,
+      "step": 5738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.441281567677791e-07,
+      "loss": 0.7253,
+      "step": 5739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4261978354123604e-07,
+      "loss": 0.2724,
+      "step": 5740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4111391811337836e-07,
+      "loss": 0.7604,
+      "step": 5741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.396105608793344e-07,
+      "loss": 0.7236,
+      "step": 5742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3810971223357533e-07,
+      "loss": 0.7881,
+      "step": 5743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.366113725699184e-07,
+      "loss": 0.262,
+      "step": 5744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3511554228151566e-07,
+      "loss": 0.6748,
+      "step": 5745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.336222217608632e-07,
+      "loss": 0.6824,
+      "step": 5746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3213141139980207e-07,
+      "loss": 0.7771,
+      "step": 5747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.306431115895104e-07,
+      "loss": 0.7451,
+      "step": 5748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2915732272050813e-07,
+      "loss": 0.6738,
+      "step": 5749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2767404518266e-07,
+      "loss": 0.7405,
+      "step": 5750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.261932793651669e-07,
+      "loss": 0.7135,
+      "step": 5751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.247150256565713e-07,
+      "loss": 0.781,
+      "step": 5752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2323928444476304e-07,
+      "loss": 0.6992,
+      "step": 5753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2176605611696343e-07,
+      "loss": 0.7422,
+      "step": 5754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.202953410597388e-07,
+      "loss": 0.6934,
+      "step": 5755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1882713965899825e-07,
+      "loss": 0.7734,
+      "step": 5756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1736145229998693e-07,
+      "loss": 0.7643,
+      "step": 5757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1589827936729277e-07,
+      "loss": 0.702,
+      "step": 5758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.144376212448442e-07,
+      "loss": 0.7385,
+      "step": 5759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1297947831590914e-07,
+      "loss": 0.6954,
+      "step": 5760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.115238509630948e-07,
+      "loss": 0.7339,
+      "step": 5761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100707395683479e-07,
+      "loss": 0.7113,
+      "step": 5762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.086201445129578e-07,
+      "loss": 0.7246,
+      "step": 5763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0717206617755244e-07,
+      "loss": 0.7859,
+      "step": 5764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0572650494209555e-07,
+      "loss": 0.6306,
+      "step": 5765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0428346118589813e-07,
+      "loss": 0.7147,
+      "step": 5766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0284293528760287e-07,
+      "loss": 0.7739,
+      "step": 5767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0140492762519524e-07,
+      "loss": 0.7671,
+      "step": 5768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.999694385760011e-07,
+      "loss": 0.7213,
+      "step": 5769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.985364685166837e-07,
+      "loss": 0.2684,
+      "step": 5770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9710601782324444e-07,
+      "loss": 0.7568,
+      "step": 5771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9567808687102526e-07,
+      "loss": 0.772,
+      "step": 5772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9425267603470987e-07,
+      "loss": 0.7107,
+      "step": 5773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9282978568831344e-07,
+      "loss": 0.7089,
+      "step": 5774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9140941620519403e-07,
+      "loss": 0.7473,
+      "step": 5775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8999156795805126e-07,
+      "loss": 0.683,
+      "step": 5776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8857624131891645e-07,
+      "loss": 0.7319,
+      "step": 5777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.871634366591637e-07,
+      "loss": 0.7656,
+      "step": 5778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.857531543495063e-07,
+      "loss": 0.7483,
+      "step": 5779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.843453947599918e-07,
+      "loss": 0.6912,
+      "step": 5780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.829401582600067e-07,
+      "loss": 0.7668,
+      "step": 5781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.815374452182796e-07,
+      "loss": 0.729,
+      "step": 5782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8013725600287267e-07,
+      "loss": 0.71,
+      "step": 5783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.787395909811853e-07,
+      "loss": 0.7822,
+      "step": 5784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.773444505199597e-07,
+      "loss": 0.7292,
+      "step": 5785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.759518349852698e-07,
+      "loss": 0.7363,
+      "step": 5786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7456174474252874e-07,
+      "loss": 0.7375,
+      "step": 5787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.731741801564903e-07,
+      "loss": 0.745,
+      "step": 5788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7178914159124226e-07,
+      "loss": 0.6783,
+      "step": 5789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.704066294102093e-07,
+      "loss": 0.7494,
+      "step": 5790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.690266439761547e-07,
+      "loss": 0.7537,
+      "step": 5791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.676491856511799e-07,
+      "loss": 0.6575,
+      "step": 5792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6627425479671796e-07,
+      "loss": 0.762,
+      "step": 5793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.649018517735459e-07,
+      "loss": 0.7269,
+      "step": 5794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.635319769417733e-07,
+      "loss": 0.7488,
+      "step": 5795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.621646306608462e-07,
+      "loss": 0.6604,
+      "step": 5796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6079981328954626e-07,
+      "loss": 0.7371,
+      "step": 5797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.594375251859972e-07,
+      "loss": 0.7263,
+      "step": 5798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.580777667076529e-07,
+      "loss": 0.6425,
+      "step": 5799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5672053821130484e-07,
+      "loss": 0.741,
+      "step": 5800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.553658400530846e-07,
+      "loss": 0.6727,
+      "step": 5801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5401367258845466e-07,
+      "loss": 0.7111,
+      "step": 5802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5266403617221446e-07,
+      "loss": 0.7172,
+      "step": 5803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.51316931158503e-07,
+      "loss": 0.7434,
+      "step": 5804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4997235790079077e-07,
+      "loss": 0.7418,
+      "step": 5805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4863031675188563e-07,
+      "loss": 0.7505,
+      "step": 5806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4729080806393145e-07,
+      "loss": 0.717,
+      "step": 5807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4595383218840927e-07,
+      "loss": 0.7312,
+      "step": 5808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.446193894761307e-07,
+      "loss": 0.7826,
+      "step": 5809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4328748027724454e-07,
+      "loss": 0.7,
+      "step": 5810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4195810494123903e-07,
+      "loss": 0.6489,
+      "step": 5811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.406312638169329e-07,
+      "loss": 0.7291,
+      "step": 5812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.393069572524799e-07,
+      "loss": 0.6731,
+      "step": 5813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3798518559537197e-07,
+      "loss": 0.6926,
+      "step": 5814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.36665949192434e-07,
+      "loss": 0.6768,
+      "step": 5815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.353492483898224e-07,
+      "loss": 0.7397,
+      "step": 5816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.340350835330353e-07,
+      "loss": 0.7744,
+      "step": 5817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.327234549669001e-07,
+      "loss": 0.2661,
+      "step": 5818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3141436303557816e-07,
+      "loss": 0.7789,
+      "step": 5819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.301078080825715e-07,
+      "loss": 0.6951,
+      "step": 5820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.288037904507102e-07,
+      "loss": 0.7109,
+      "step": 5821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.275023104821584e-07,
+      "loss": 0.2396,
+      "step": 5822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2620336851842073e-07,
+      "loss": 0.706,
+      "step": 5823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.249069649003289e-07,
+      "loss": 0.7646,
+      "step": 5824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.236130999680509e-07,
+      "loss": 0.7725,
+      "step": 5825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.223217740610918e-07,
+      "loss": 0.7284,
+      "step": 5826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2103298751828606e-07,
+      "loss": 0.7368,
+      "step": 5827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.197467406778032e-07,
+      "loss": 0.7371,
+      "step": 5828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.184630338771455e-07,
+      "loss": 0.6941,
+      "step": 5829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.171818674531524e-07,
+      "loss": 0.7042,
+      "step": 5830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1590324174199273e-07,
+      "loss": 0.7483,
+      "step": 5831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1462715707917033e-07,
+      "loss": 0.7766,
+      "step": 5832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.133536137995219e-07,
+      "loss": 0.6742,
+      "step": 5833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.120826122372189e-07,
+      "loss": 0.6802,
+      "step": 5834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1081415272576243e-07,
+      "loss": 0.754,
+      "step": 5835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.095482355979895e-07,
+      "loss": 0.7681,
+      "step": 5836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0828486118607e-07,
+      "loss": 0.7715,
+      "step": 5837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.070240298215033e-07,
+      "loss": 0.8052,
+      "step": 5838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0576574183512676e-07,
+      "loss": 0.7566,
+      "step": 5839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0450999755710533e-07,
+      "loss": 0.7495,
+      "step": 5840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0325679731693866e-07,
+      "loss": 0.7698,
+      "step": 5841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.020061414434605e-07,
+      "loss": 0.2684,
+      "step": 5842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.007580302648361e-07,
+      "loss": 0.7291,
+      "step": 5843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9951246410856004e-07,
+      "loss": 0.6764,
+      "step": 5844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.98269443301461e-07,
+      "loss": 0.6332,
+      "step": 5845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.970289681697025e-07,
+      "loss": 0.6855,
+      "step": 5846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9579103903877636e-07,
+      "loss": 0.2487,
+      "step": 5847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.945556562335061e-07,
+      "loss": 0.7593,
+      "step": 5848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9332282007805244e-07,
+      "loss": 0.6685,
+      "step": 5849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.920925308959022e-07,
+      "loss": 0.7568,
+      "step": 5850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.908647890098737e-07,
+      "loss": 0.7422,
+      "step": 5851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896395947421227e-07,
+      "loss": 0.7004,
+      "step": 5852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.884169484141308e-07,
+      "loss": 0.7593,
+      "step": 5853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8719685034671154e-07,
+      "loss": 0.7263,
+      "step": 5854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8597930086001423e-07,
+      "loss": 0.7084,
+      "step": 5855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8476430027351567e-07,
+      "loss": 0.7207,
+      "step": 5856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.835518489060229e-07,
+      "loss": 0.7781,
+      "step": 5857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.823419470756772e-07,
+      "loss": 0.2656,
+      "step": 5858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8113459509995e-07,
+      "loss": 0.6792,
+      "step": 5859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7992979329564127e-07,
+      "loss": 0.6901,
+      "step": 5860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7872754197888705e-07,
+      "loss": 0.7329,
+      "step": 5861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.775278414651472e-07,
+      "loss": 0.7012,
+      "step": 5862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.763306920692177e-07,
+      "loss": 0.7528,
+      "step": 5863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.751360941052228e-07,
+      "loss": 0.7129,
+      "step": 5864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7394404788661844e-07,
+      "loss": 0.6632,
+      "step": 5865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.727545537261889e-07,
+      "loss": 0.6843,
+      "step": 5866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.715676119360511e-07,
+      "loss": 0.7333,
+      "step": 5867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.703832228276515e-07,
+      "loss": 0.7098,
+      "step": 5868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.692013867117682e-07,
+      "loss": 0.6887,
+      "step": 5869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6802210389850316e-07,
+      "loss": 0.7301,
+      "step": 5870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.668453746972988e-07,
+      "loss": 0.812,
+      "step": 5871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.656711994169181e-07,
+      "loss": 0.7311,
+      "step": 5872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6449957836545805e-07,
+      "loss": 0.7412,
+      "step": 5873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6333051185034486e-07,
+      "loss": 0.2645,
+      "step": 5874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6216400017833656e-07,
+      "loss": 0.7365,
+      "step": 5875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.61000043655516e-07,
+      "loss": 0.7384,
+      "step": 5876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5983864258730006e-07,
+      "loss": 0.7697,
+      "step": 5877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5867979727843163e-07,
+      "loss": 0.7383,
+      "step": 5878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5752350803298633e-07,
+      "loss": 0.7869,
+      "step": 5879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5636977515436703e-07,
+      "loss": 0.7969,
+      "step": 5880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.55218598945306e-07,
+      "loss": 0.7527,
+      "step": 5881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5406997970786604e-07,
+      "loss": 0.7339,
+      "step": 5882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5292391774343614e-07,
+      "loss": 0.7379,
+      "step": 5883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.517804133527368e-07,
+      "loss": 0.7976,
+      "step": 5884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5063946683581696e-07,
+      "loss": 0.6927,
+      "step": 5885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.495010784920537e-07,
+      "loss": 0.7063,
+      "step": 5886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4836524862015376e-07,
+      "loss": 0.6892,
+      "step": 5887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4723197751815085e-07,
+      "loss": 0.7488,
+      "step": 5888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.461012654834083e-07,
+      "loss": 0.6982,
+      "step": 5889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.449731128126209e-07,
+      "loss": 0.782,
+      "step": 5890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.438475198018064e-07,
+      "loss": 0.6963,
+      "step": 5891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.42724486746313e-07,
+      "loss": 0.7072,
+      "step": 5892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4160401394082045e-07,
+      "loss": 0.7446,
+      "step": 5893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.404861016793325e-07,
+      "loss": 0.6877,
+      "step": 5894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.393707502551823e-07,
+      "loss": 0.7551,
+      "step": 5895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.382579599610324e-07,
+      "loss": 0.7974,
+      "step": 5896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3714773108887147e-07,
+      "loss": 0.7231,
+      "step": 5897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3604006393001643e-07,
+      "loss": 0.6631,
+      "step": 5898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3493495877511264e-07,
+      "loss": 0.7322,
+      "step": 5899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3383241591413365e-07,
+      "loss": 0.7158,
+      "step": 5900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3273243563638026e-07,
+      "loss": 0.7302,
+      "step": 5901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3163501823047828e-07,
+      "loss": 0.666,
+      "step": 5902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.305401639843874e-07,
+      "loss": 0.7086,
+      "step": 5903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2944787318538774e-07,
+      "loss": 0.7507,
+      "step": 5904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.28358146120089e-07,
+      "loss": 0.7568,
+      "step": 5905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.272709830744324e-07,
+      "loss": 0.7634,
+      "step": 5906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.261863843336809e-07,
+      "loss": 0.8306,
+      "step": 5907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2510435018242573e-07,
+      "loss": 0.7313,
+      "step": 5908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2402488090458752e-07,
+      "loss": 0.6882,
+      "step": 5909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2294797678341305e-07,
+      "loss": 0.7219,
+      "step": 5910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.21873638101473e-07,
+      "loss": 0.719,
+      "step": 5911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.208018651406696e-07,
+      "loss": 0.7554,
+      "step": 5912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.197326581822279e-07,
+      "loss": 0.6344,
+      "step": 5913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1866601750670346e-07,
+      "loss": 0.749,
+      "step": 5914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1760194339397466e-07,
+      "loss": 0.8044,
+      "step": 5915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1654043612324927e-07,
+      "loss": 0.7297,
+      "step": 5916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.154814959730589e-07,
+      "loss": 0.7119,
+      "step": 5917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1442512322126352e-07,
+      "loss": 0.752,
+      "step": 5918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1337131814504918e-07,
+      "loss": 0.7529,
+      "step": 5919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1232008102092917e-07,
+      "loss": 0.7046,
+      "step": 5920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1127141212473944e-07,
+      "loss": 0.7197,
+      "step": 5921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1022531173164552e-07,
+      "loss": 0.793,
+      "step": 5922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0918178011613888e-07,
+      "loss": 0.7433,
+      "step": 5923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0814081755203385e-07,
+      "loss": 0.7045,
+      "step": 5924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0710242431247307e-07,
+      "loss": 0.7837,
+      "step": 5925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0606660066992634e-07,
+      "loss": 0.78,
+      "step": 5926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0503334689618515e-07,
+      "loss": 0.7045,
+      "step": 5927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.040026632623715e-07,
+      "loss": 0.7031,
+      "step": 5928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0297455003892906e-07,
+      "loss": 0.6859,
+      "step": 5929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0194900749562873e-07,
+      "loss": 0.7037,
+      "step": 5930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0092603590156633e-07,
+      "loss": 0.7097,
+      "step": 5931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9990563552516383e-07,
+      "loss": 0.7588,
+      "step": 5932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9888780663416818e-07,
+      "loss": 0.687,
+      "step": 5933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9787254949565127e-07,
+      "loss": 0.6654,
+      "step": 5934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685986437601112e-07,
+      "loss": 0.7959,
+      "step": 5935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9584975154096852e-07,
+      "loss": 0.7471,
+      "step": 5936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9484221125557257e-07,
+      "loss": 0.7312,
+      "step": 5937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9383724378419512e-07,
+      "loss": 0.7468,
+      "step": 5938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.928348493905341e-07,
+      "loss": 0.7396,
+      "step": 5939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9183502833761025e-07,
+      "loss": 0.7942,
+      "step": 5940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.908377808877715e-07,
+      "loss": 0.7644,
+      "step": 5941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8984310730269072e-07,
+      "loss": 0.7761,
+      "step": 5942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8885100784336252e-07,
+      "loss": 0.718,
+      "step": 5943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8786148277010864e-07,
+      "loss": 0.2489,
+      "step": 5944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.868745323425736e-07,
+      "loss": 0.7452,
+      "step": 5945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.85890156819728e-07,
+      "loss": 0.7334,
+      "step": 5946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.849083564598664e-07,
+      "loss": 0.6775,
+      "step": 5947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.83929131520606e-07,
+      "loss": 0.7365,
+      "step": 5948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8295248225889018e-07,
+      "loss": 0.7344,
+      "step": 5949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8197840893098507e-07,
+      "loss": 0.6821,
+      "step": 5950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8100691179248397e-07,
+      "loss": 0.7327,
+      "step": 5951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8003799109829966e-07,
+      "loss": 0.7046,
+      "step": 5952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.79071647102671e-07,
+      "loss": 0.2609,
+      "step": 5953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.781078800591618e-07,
+      "loss": 0.7378,
+      "step": 5954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7714669022065868e-07,
+      "loss": 0.7444,
+      "step": 5955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7618807783937098e-07,
+      "loss": 0.7037,
+      "step": 5956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7523204316683416e-07,
+      "loss": 0.7664,
+      "step": 5957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.742785864539054e-07,
+      "loss": 0.7686,
+      "step": 5958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7332770795076558e-07,
+      "loss": 0.2302,
+      "step": 5959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7237940790691964e-07,
+      "loss": 0.6843,
+      "step": 5960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.714336865711963e-07,
+      "loss": 0.2805,
+      "step": 5961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7049054419174483e-07,
+      "loss": 0.6842,
+      "step": 5962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.695499810160439e-07,
+      "loss": 0.7433,
+      "step": 5963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.686119972908884e-07,
+      "loss": 0.2388,
+      "step": 5964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.676765932624014e-07,
+      "loss": 0.7507,
+      "step": 5965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6674376917602653e-07,
+      "loss": 0.7786,
+      "step": 5966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6581352527653138e-07,
+      "loss": 0.6746,
+      "step": 5967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6488586180800626e-07,
+      "loss": 0.707,
+      "step": 5968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6396077901386308e-07,
+      "loss": 0.7224,
+      "step": 5969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6303827713683995e-07,
+      "loss": 0.7825,
+      "step": 5970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6211835641899543e-07,
+      "loss": 0.7451,
+      "step": 5971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6120101710171087e-07,
+      "loss": 0.7571,
+      "step": 5972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6028625942569043e-07,
+      "loss": 0.7445,
+      "step": 5973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5937408363096098e-07,
+      "loss": 0.7644,
+      "step": 5974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5846448995687103e-07,
+      "loss": 0.6632,
+      "step": 5975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5755747864209415e-07,
+      "loss": 0.6788,
+      "step": 5976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5665304992462438e-07,
+      "loss": 0.7028,
+      "step": 5977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5575120404177635e-07,
+      "loss": 0.7681,
+      "step": 5978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5485194123019188e-07,
+      "loss": 0.7705,
+      "step": 5979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5395526172582997e-07,
+      "loss": 0.7313,
+      "step": 5980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5306116576397468e-07,
+      "loss": 0.782,
+      "step": 5981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5216965357923163e-07,
+      "loss": 0.6654,
+      "step": 5982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5128072540552708e-07,
+      "loss": 0.7103,
+      "step": 5983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5039438147611107e-07,
+      "loss": 0.7301,
+      "step": 5984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4951062202355538e-07,
+      "loss": 0.6868,
+      "step": 5985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4862944727975336e-07,
+      "loss": 0.7003,
+      "step": 5986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.47750857475919e-07,
+      "loss": 0.6818,
+      "step": 5987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4687485284258784e-07,
+      "loss": 0.7122,
+      "step": 5988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4600143360962049e-07,
+      "loss": 0.6973,
+      "step": 5989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4513060000619696e-07,
+      "loss": 0.7266,
+      "step": 5990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4426235226081664e-07,
+      "loss": 0.7256,
+      "step": 5991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4339669060130401e-07,
+      "loss": 0.7258,
+      "step": 5992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4253361525480292e-07,
+      "loss": 0.7686,
+      "step": 5993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4167312644777776e-07,
+      "loss": 0.7247,
+      "step": 5994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4081522440601902e-07,
+      "loss": 0.7424,
+      "step": 5995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.399599093546322e-07,
+      "loss": 0.7795,
+      "step": 5996
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.391071815180467e-07,
+      "loss": 0.7413,
+      "step": 5997
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.382570411200157e-07,
+      "loss": 0.7253,
+      "step": 5998
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3740948838360858e-07,
+      "loss": 0.7245,
+      "step": 5999
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365645235312174e-07,
+      "loss": 0.6907,
+      "step": 6000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3572214678455818e-07,
+      "loss": 0.2525,
+      "step": 6001
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3488235836466524e-07,
+      "loss": 0.2565,
+      "step": 6002
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3404515849189225e-07,
+      "loss": 0.7195,
+      "step": 6003
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3321054738591577e-07,
+      "loss": 0.7123,
+      "step": 6004
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3237852526573392e-07,
+      "loss": 0.7798,
+      "step": 6005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3154909234966317e-07,
+      "loss": 0.2494,
+      "step": 6006
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3072224885534168e-07,
+      "loss": 0.6812,
+      "step": 6007
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.298979949997292e-07,
+      "loss": 0.6759,
+      "step": 6008
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2907633099910389e-07,
+      "loss": 0.7132,
+      "step": 6009
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2825725706906544e-07,
+      "loss": 0.7023,
+      "step": 6010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.274407734245353e-07,
+      "loss": 0.7725,
+      "step": 6011
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2662688027975211e-07,
+      "loss": 0.7219,
+      "step": 6012
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2581557784827726e-07,
+      "loss": 0.2759,
+      "step": 6013
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2500686634299153e-07,
+      "loss": 0.7405,
+      "step": 6014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2420074597609633e-07,
+      "loss": 0.7041,
+      "step": 6015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2339721695911133e-07,
+      "loss": 0.7576,
+      "step": 6016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.225962795028801e-07,
+      "loss": 0.6931,
+      "step": 6017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2179793381756122e-07,
+      "loss": 0.786,
+      "step": 6018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.210021801126371e-07,
+      "loss": 0.7944,
+      "step": 6019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2020901859690849e-07,
+      "loss": 0.7445,
+      "step": 6020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1941844947849667e-07,
+      "loss": 0.7161,
+      "step": 6021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1863047296484131e-07,
+      "loss": 0.7356,
+      "step": 6022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1784508926270256e-07,
+      "loss": 0.7734,
+      "step": 6023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.170622985781622e-07,
+      "loss": 0.7235,
+      "step": 6024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1628210111661819e-07,
+      "loss": 0.7593,
+      "step": 6025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1550449708278899e-07,
+      "loss": 0.7406,
+      "step": 6026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1472948668071359e-07,
+      "loss": 0.7756,
+      "step": 6027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1395707011375157e-07,
+      "loss": 0.7114,
+      "step": 6028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1318724758457855e-07,
+      "loss": 0.6534,
+      "step": 6029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1242001929519297e-07,
+      "loss": 0.6698,
+      "step": 6030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1165538544690935e-07,
+      "loss": 0.7471,
+      "step": 6031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1089334624036385e-07,
+      "loss": 0.6884,
+      "step": 6032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1013390187550987e-07,
+      "loss": 0.6846,
+      "step": 6033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0937705255162246e-07,
+      "loss": 0.6796,
+      "step": 6034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0862279846729385e-07,
+      "loss": 0.6973,
+      "step": 6035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0787113982043462e-07,
+      "loss": 0.7168,
+      "step": 6036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0712207680827813e-07,
+      "loss": 0.7849,
+      "step": 6037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0637560962737047e-07,
+      "loss": 0.2692,
+      "step": 6038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0563173847358277e-07,
+      "loss": 0.7279,
+      "step": 6039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0489046354210108e-07,
+      "loss": 0.7076,
+      "step": 6040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0415178502743207e-07,
+      "loss": 0.7073,
+      "step": 6041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0341570312339843e-07,
+      "loss": 0.6932,
+      "step": 6042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0268221802314682e-07,
+      "loss": 0.761,
+      "step": 6043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0195132991913881e-07,
+      "loss": 0.7578,
+      "step": 6044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0122303900315212e-07,
+      "loss": 0.7388,
+      "step": 6045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0049734546628942e-07,
+      "loss": 0.6914,
+      "step": 6046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.977424949896508e-08,
+      "loss": 0.771,
+      "step": 6047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.905375129091732e-08,
+      "loss": 0.749,
+      "step": 6048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.833585103120048e-08,
+      "loss": 0.6876,
+      "step": 6049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.762054890818607e-08,
+      "loss": 0.756,
+      "step": 6050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.690784510956508e-08,
+      "loss": 0.762,
+      "step": 6051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.619773982234904e-08,
+      "loss": 0.2567,
+      "step": 6052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.549023323286222e-08,
+      "loss": 0.2291,
+      "step": 6053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.478532552675057e-08,
+      "loss": 0.6909,
+      "step": 6054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.408301688897948e-08,
+      "loss": 0.7759,
+      "step": 6055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.338330750382818e-08,
+      "loss": 0.7281,
+      "step": 6056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.268619755489761e-08,
+      "loss": 0.7094,
+      "step": 6057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.199168722510476e-08,
+      "loss": 0.7529,
+      "step": 6058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.12997766966861e-08,
+      "loss": 0.714,
+      "step": 6059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.061046615119306e-08,
+      "loss": 0.7031,
+      "step": 6060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.992375576949763e-08,
+      "loss": 0.7283,
+      "step": 6061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.923964573178789e-08,
+      "loss": 0.6707,
+      "step": 6062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.855813621757025e-08,
+      "loss": 0.8069,
+      "step": 6063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.787922740566834e-08,
+      "loss": 0.7188,
+      "step": 6064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.72029194742241e-08,
+      "loss": 0.6843,
+      "step": 6065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.65292126006978e-08,
+      "loss": 0.7069,
+      "step": 6066
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.585810696186359e-08,
+      "loss": 0.7275,
+      "step": 6067
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.518960273381726e-08,
+      "loss": 0.726,
+      "step": 6068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.452370009197075e-08,
+      "loss": 0.7395,
+      "step": 6069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.386039921105094e-08,
+      "loss": 0.6897,
+      "step": 6070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.319970026510526e-08,
+      "loss": 0.7268,
+      "step": 6071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.254160342749618e-08,
+      "loss": 0.6462,
+      "step": 6072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.188610887090664e-08,
+      "loss": 0.2667,
+      "step": 6073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.123321676733132e-08,
+      "loss": 0.6719,
+      "step": 6074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.058292728808759e-08,
+      "loss": 0.7195,
+      "step": 6075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.993524060380564e-08,
+      "loss": 0.7059,
+      "step": 6076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.92901568844362e-08,
+      "loss": 0.8174,
+      "step": 6077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.864767629924386e-08,
+      "loss": 0.7263,
+      "step": 6078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.800779901681377e-08,
+      "loss": 0.7297,
+      "step": 6079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.737052520504273e-08,
+      "loss": 0.7327,
+      "step": 6080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.673585503114922e-08,
+      "loss": 0.6993,
+      "step": 6081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.610378866166778e-08,
+      "loss": 0.7404,
+      "step": 6082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.547432626244799e-08,
+      "loss": 0.2503,
+      "step": 6083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.484746799865549e-08,
+      "loss": 0.6684,
+      "step": 6084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.42232140347765e-08,
+      "loss": 0.7506,
+      "step": 6085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.360156453460999e-08,
+      "loss": 0.7708,
+      "step": 6086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.298251966127435e-08,
+      "loss": 0.7795,
+      "step": 6087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.236607957720188e-08,
+      "loss": 0.6969,
+      "step": 6088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.175224444414209e-08,
+      "loss": 0.7483,
+      "step": 6089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.11410144231639e-08,
+      "loss": 0.731,
+      "step": 6090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.053238967464903e-08,
+      "loss": 0.7278,
+      "step": 6091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.992637035829641e-08,
+      "loss": 0.7263,
+      "step": 6092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.932295663312217e-08,
+      "loss": 0.7522,
+      "step": 6093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.872214865745963e-08,
+      "loss": 0.7498,
+      "step": 6094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.812394658895605e-08,
+      "loss": 0.7522,
+      "step": 6095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.752835058457474e-08,
+      "loss": 0.7769,
+      "step": 6096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.693536080059959e-08,
+      "loss": 0.7289,
+      "step": 6097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.634497739262613e-08,
+      "loss": 0.2416,
+      "step": 6098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.5757200515566e-08,
+      "loss": 0.6541,
+      "step": 6099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.517203032365027e-08,
+      "loss": 0.7284,
+      "step": 6100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.458946697042278e-08,
+      "loss": 0.751,
+      "step": 6101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.40095106087446e-08,
+      "loss": 0.731,
+      "step": 6102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.3432161390794e-08,
+      "loss": 0.7095,
+      "step": 6103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.285741946806312e-08,
+      "loss": 0.7061,
+      "step": 6104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.228528499136133e-08,
+      "loss": 0.6591,
+      "step": 6105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.171575811081187e-08,
+      "loss": 0.7615,
+      "step": 6106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.114883897585633e-08,
+      "loss": 0.7444,
+      "step": 6107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.058452773525014e-08,
+      "loss": 0.7448,
+      "step": 6108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.002282453706598e-08,
+      "loss": 0.7391,
+      "step": 6109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9463729528690396e-08,
+      "loss": 0.7251,
+      "step": 6110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.890724285682825e-08,
+      "loss": 0.7552,
+      "step": 6111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.835336466749608e-08,
+      "loss": 0.7122,
+      "step": 6112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.780209510602874e-08,
+      "loss": 0.7544,
+      "step": 6113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.72534343170772e-08,
+      "loss": 0.7773,
+      "step": 6114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.670738244460627e-08,
+      "loss": 0.6936,
+      "step": 6115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.616393963189581e-08,
+      "loss": 0.7144,
+      "step": 6116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.562310602154286e-08,
+      "loss": 0.7079,
+      "step": 6117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.508488175545723e-08,
+      "loss": 0.7037,
+      "step": 6118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.454926697486818e-08,
+      "loss": 0.6659,
+      "step": 6119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.401626182031661e-08,
+      "loss": 0.6804,
+      "step": 6120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.348586643165954e-08,
+      "loss": 0.7753,
+      "step": 6121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.295808094807009e-08,
+      "loss": 0.7249,
+      "step": 6122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.243290550803526e-08,
+      "loss": 0.7133,
+      "step": 6123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.191034024935815e-08,
+      "loss": 0.7769,
+      "step": 6124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1390385309157966e-08,
+      "loss": 0.7092,
+      "step": 6125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.087304082386557e-08,
+      "loss": 0.6931,
+      "step": 6126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.035830692923016e-08,
+      "loss": 0.6937,
+      "step": 6127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.98461837603148e-08,
+      "loss": 0.7124,
+      "step": 6128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9336671451498675e-08,
+      "loss": 0.7114,
+      "step": 6129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.882977013647261e-08,
+      "loss": 0.7324,
+      "step": 6130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.832547994824466e-08,
+      "loss": 0.71,
+      "step": 6131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.782380101913897e-08,
+      "loss": 0.7659,
+      "step": 6132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.732473348079247e-08,
+      "loss": 0.6866,
+      "step": 6133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.682827746415597e-08,
+      "loss": 0.7864,
+      "step": 6134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.633443309949859e-08,
+      "loss": 0.7479,
+      "step": 6135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.584320051640223e-08,
+      "loss": 0.7412,
+      "step": 6136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.535457984376046e-08,
+      "loss": 0.7524,
+      "step": 6137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.486857120978849e-08,
+      "loss": 0.71,
+      "step": 6138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.438517474200876e-08,
+      "loss": 0.7366,
+      "step": 6139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3904390567263145e-08,
+      "loss": 0.7651,
+      "step": 6140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3426218811705166e-08,
+      "loss": 0.2141,
+      "step": 6141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.295065960080669e-08,
+      "loss": 0.7297,
+      "step": 6142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2477713059349e-08,
+      "loss": 0.7908,
+      "step": 6143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.200737931143062e-08,
+      "loss": 0.7341,
+      "step": 6144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.153965848046615e-08,
+      "loss": 0.8369,
+      "step": 6145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.107455068918187e-08,
+      "loss": 0.7688,
+      "step": 6146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0612056059617935e-08,
+      "loss": 0.74,
+      "step": 6147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0152174713131706e-08,
+      "loss": 0.7305,
+      "step": 6148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9694906770393336e-08,
+      "loss": 0.7407,
+      "step": 6149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9240252351385736e-08,
+      "loss": 0.717,
+      "step": 6150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8788211575409016e-08,
+      "loss": 0.7671,
+      "step": 6151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.833878456107498e-08,
+      "loss": 0.8018,
+      "step": 6152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7891971426310405e-08,
+      "loss": 0.7275,
+      "step": 6153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.744777228835705e-08,
+      "loss": 0.7039,
+      "step": 6154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.700618726376948e-08,
+      "loss": 0.717,
+      "step": 6155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.656721646841721e-08,
+      "loss": 0.7677,
+      "step": 6156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.613086001748478e-08,
+      "loss": 0.7893,
+      "step": 6157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.569711802546727e-08,
+      "loss": 0.7024,
+      "step": 6158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.526599060617697e-08,
+      "loss": 0.7286,
+      "step": 6159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.483747787273895e-08,
+      "loss": 0.7759,
+      "step": 6160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4411579937592144e-08,
+      "loss": 0.7122,
+      "step": 6161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3988296912490503e-08,
+      "loss": 0.7739,
+      "step": 6162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.356762890849963e-08,
+      "loss": 0.7141,
+      "step": 6163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3149576036001216e-08,
+      "loss": 0.7621,
+      "step": 6164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2734138404689755e-08,
+      "loss": 0.7399,
+      "step": 6165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.232131612357359e-08,
+      "loss": 0.7375,
+      "step": 6166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1911109300974965e-08,
+      "loss": 0.7656,
+      "step": 6167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.150351804452889e-08,
+      "loss": 0.7419,
+      "step": 6168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1098542461184266e-08,
+      "loss": 0.7319,
+      "step": 6169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0696182657206085e-08,
+      "loss": 0.7061,
+      "step": 6170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.02964387381699e-08,
+      "loss": 0.6311,
+      "step": 6171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.989931080896624e-08,
+      "loss": 0.7476,
+      "step": 6172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.950479897379954e-08,
+      "loss": 0.7593,
+      "step": 6173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9112903336188104e-08,
+      "loss": 0.7484,
+      "step": 6174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8723623998960782e-08,
+      "loss": 0.712,
+      "step": 6175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8336961064262535e-08,
+      "loss": 0.729,
+      "step": 6176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.795291463355332e-08,
+      "loss": 0.7365,
+      "step": 6177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7571484807602523e-08,
+      "loss": 0.6696,
+      "step": 6178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.719267168649564e-08,
+      "loss": 0.7268,
+      "step": 6179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6816475369632055e-08,
+      "loss": 0.7373,
+      "step": 6180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6442895955721694e-08,
+      "loss": 0.7678,
+      "step": 6181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6071933542790583e-08,
+      "loss": 0.646,
+      "step": 6182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5703588228177535e-08,
+      "loss": 0.6461,
+      "step": 6183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5337860108531897e-08,
+      "loss": 0.6494,
+      "step": 6184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4974749279820242e-08,
+      "loss": 0.7708,
+      "step": 6185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.461425583732191e-08,
+      "loss": 0.7239,
+      "step": 6186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.425637987562568e-08,
+      "loss": 0.743,
+      "step": 6187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3901121488636435e-08,
+      "loss": 0.7537,
+      "step": 6188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3548480769572947e-08,
+      "loss": 0.7289,
+      "step": 6189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3198457810964525e-08,
+      "loss": 0.7758,
+      "step": 6190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2851052704657704e-08,
+      "loss": 0.6346,
+      "step": 6191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2506265541806238e-08,
+      "loss": 0.7241,
+      "step": 6192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2164096412882197e-08,
+      "loss": 0.6936,
+      "step": 6193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.182454540766821e-08,
+      "loss": 0.6919,
+      "step": 6194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1487612615260777e-08,
+      "loss": 0.6874,
+      "step": 6195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1153298124069187e-08,
+      "loss": 0.7128,
+      "step": 6196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.082160202181438e-08,
+      "loss": 0.7069,
+      "step": 6197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0492524395531177e-08,
+      "loss": 0.2477,
+      "step": 6198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0166065331569397e-08,
+      "loss": 0.6467,
+      "step": 6199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9842224915589404e-08,
+      "loss": 0.7568,
+      "step": 6200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.952100323256323e-08,
+      "loss": 0.7294,
+      "step": 6201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9202400366780116e-08,
+      "loss": 0.7333,
+      "step": 6202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8886416401837636e-08,
+      "loss": 0.7522,
+      "step": 6203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857305142064725e-08,
+      "loss": 0.6947,
+      "step": 6204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8262305505437615e-08,
+      "loss": 0.7031,
+      "step": 6205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.795417873774241e-08,
+      "loss": 0.2535,
+      "step": 6206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7648671198415846e-08,
+      "loss": 0.7471,
+      "step": 6207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.734578296761824e-08,
+      "loss": 0.6942,
+      "step": 6208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7045514124828245e-08,
+      "loss": 0.7261,
+      "step": 6209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6747864748832833e-08,
+      "loss": 0.6982,
+      "step": 6210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6452834917732864e-08,
+      "loss": 0.6836,
+      "step": 6211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.616042470894419e-08,
+      "loss": 0.7659,
+      "step": 6212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5870634199193214e-08,
+      "loss": 0.7355,
+      "step": 6213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5583463464518e-08,
+      "loss": 0.7354,
+      "step": 6214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.529891258027161e-08,
+      "loss": 0.7362,
+      "step": 6215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5016981621116534e-08,
+      "loss": 0.709,
+      "step": 6216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4737670661032488e-08,
+      "loss": 0.7666,
+      "step": 6217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4460979773307515e-08,
+      "loss": 0.7939,
+      "step": 6218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4186909030543539e-08,
+      "loss": 0.7688,
+      "step": 6219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3915458504655255e-08,
+      "loss": 0.683,
+      "step": 6220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3646628266870133e-08,
+      "loss": 0.7637,
+      "step": 6221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.338041838772619e-08,
+      "loss": 0.8112,
+      "step": 6222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3116828937077553e-08,
+      "loss": 0.7208,
+      "step": 6223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.285585998408556e-08,
+      "loss": 0.7607,
+      "step": 6224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2597511597229884e-08,
+      "loss": 0.738,
+      "step": 6225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2341783844297406e-08,
+      "loss": 0.6672,
+      "step": 6226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2088676792391119e-08,
+      "loss": 0.7339,
+      "step": 6227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1838190507923453e-08,
+      "loss": 0.7676,
+      "step": 6228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1590325056620722e-08,
+      "loss": 0.7415,
+      "step": 6229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.134508050352201e-08,
+      "loss": 0.7644,
+      "step": 6230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1102456912978066e-08,
+      "loss": 0.7437,
+      "step": 6231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0862454348650187e-08,
+      "loss": 0.7654,
+      "step": 6232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0625072873515774e-08,
+      "loss": 0.7749,
+      "step": 6233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0390312549860559e-08,
+      "loss": 0.7283,
+      "step": 6234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0158173439285269e-08,
+      "loss": 0.6406,
+      "step": 6235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.928655602702287e-09,
+      "loss": 0.7429,
+      "step": 6236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.70175910033455e-09,
+      "loss": 0.7025,
+      "step": 6237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.477483991717773e-09,
+      "loss": 0.6425,
+      "step": 6238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.255830335702653e-09,
+      "loss": 0.7344,
+      "step": 6239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.036798190448226e-09,
+      "loss": 0.7357,
+      "step": 6240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.820387613426296e-09,
+      "loss": 0.6677,
+      "step": 6241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.60659866142366e-09,
+      "loss": 0.751,
+      "step": 6242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.395431390536558e-09,
+      "loss": 0.6415,
+      "step": 6243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.186885856173998e-09,
+      "loss": 0.7241,
+      "step": 6244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.980962113056655e-09,
+      "loss": 0.7701,
+      "step": 6245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.777660215216865e-09,
+      "loss": 0.7211,
+      "step": 6246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.57698021600195e-09,
+      "loss": 0.7379,
+      "step": 6247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3789221680686804e-09,
+      "loss": 0.7341,
+      "step": 6248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.183486123384375e-09,
+      "loss": 0.7255,
+      "step": 6249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.990672133232457e-09,
+      "loss": 0.7485,
+      "step": 6250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.800480248204677e-09,
+      "loss": 0.65,
+      "step": 6251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.612910518206672e-09,
+      "loss": 0.7471,
+      "step": 6252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4279629924557384e-09,
+      "loss": 0.7478,
+      "step": 6253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.245637719481945e-09,
+      "loss": 0.7104,
+      "step": 6254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.065934747123691e-09,
+      "loss": 0.7522,
+      "step": 6255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.888854122536592e-09,
+      "loss": 0.7371,
+      "step": 6256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.714395892184588e-09,
+      "loss": 0.712,
+      "step": 6257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.542560101844396e-09,
+      "loss": 0.7366,
+      "step": 6258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.373346796605506e-09,
+      "loss": 0.6781,
+      "step": 6259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2067560208668435e-09,
+      "loss": 0.702,
+      "step": 6260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0427878183412216e-09,
+      "loss": 0.7358,
+      "step": 6261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8814422320542235e-09,
+      "loss": 0.2708,
+      "step": 6262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.722719304340873e-09,
+      "loss": 0.7437,
+      "step": 6263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.566619076848966e-09,
+      "loss": 0.7247,
+      "step": 6264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.413141590539072e-09,
+      "loss": 0.7087,
+      "step": 6265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.262286885682309e-09,
+      "loss": 0.7256,
+      "step": 6266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1140550018614614e-09,
+      "loss": 0.6725,
+      "step": 6267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.96844597797208e-09,
+      "loss": 0.6582,
+      "step": 6268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.82545985222138e-09,
+      "loss": 0.6887,
+      "step": 6269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6850966621282402e-09,
+      "loss": 0.6901,
+      "step": 6270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.547356444523198e-09,
+      "loss": 0.7622,
+      "step": 6271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.412239235547343e-09,
+      "loss": 0.7612,
+      "step": 6272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2797450706556487e-09,
+      "loss": 0.7937,
+      "step": 6273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1498739846136383e-09,
+      "loss": 0.7617,
+      "step": 6274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0226260114973873e-09,
+      "loss": 0.7634,
+      "step": 6275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8980011846979628e-09,
+      "loss": 0.6785,
+      "step": 6276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7759995369158744e-09,
+      "loss": 0.7776,
+      "step": 6277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6566211001621822e-09,
+      "loss": 0.6854,
+      "step": 6278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5398659057629394e-09,
+      "loss": 0.7556,
+      "step": 6279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.425733984351419e-09,
+      "loss": 0.7715,
+      "step": 6280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.314225365878109e-09,
+      "loss": 0.7888,
+      "step": 6281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.205340079599605e-09,
+      "loss": 0.2506,
+      "step": 6282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0990781540886073e-09,
+      "loss": 0.7877,
+      "step": 6283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9954396172272573e-09,
+      "loss": 0.2608,
+      "step": 6284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.894424496209357e-09,
+      "loss": 0.7225,
+      "step": 6285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7960328175403718e-09,
+      "loss": 0.6383,
+      "step": 6286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7002646070374273e-09,
+      "loss": 0.7263,
+      "step": 6287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6071198898315321e-09,
+      "loss": 0.7205,
+      "step": 6288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5165986903609153e-09,
+      "loss": 0.7786,
+      "step": 6289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4287010323776884e-09,
+      "loss": 0.6818,
+      "step": 6290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3434269389478449e-09,
+      "loss": 0.7118,
+      "step": 6291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2607764324457095e-09,
+      "loss": 0.6644,
+      "step": 6292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1807495345583786e-09,
+      "loss": 0.7217,
+      "step": 6293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1033462662835003e-09,
+      "loss": 0.7261,
+      "step": 6294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.028566647932605e-09,
+      "loss": 0.2356,
+      "step": 6295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.564106991266642e-10,
+      "loss": 0.2729,
+      "step": 6296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.868784387994212e-10,
+      "loss": 0.7727,
+      "step": 6297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.199698851940607e-10,
+      "loss": 0.7063,
+      "step": 6298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.556850558698703e-10,
+      "loss": 0.7014,
+      "step": 6299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.940239676922478e-10,
+      "loss": 0.7118,
+      "step": 6300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.349866368415835e-10,
+      "loss": 0.6903,
+      "step": 6301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.785730788099297e-10,
+      "loss": 0.7305,
+      "step": 6302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.247833083987796e-10,
+      "loss": 0.7322,
+      "step": 6303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.736173397212884e-10,
+      "loss": 0.78,
+      "step": 6304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.250751862056035e-10,
+      "loss": 0.6898,
+      "step": 6305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.79156860585983e-10,
+      "loss": 0.7883,
+      "step": 6306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3586237491389784e-10,
+      "loss": 0.2656,
+      "step": 6307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9519174054803976e-10,
+      "loss": 0.7446,
+      "step": 6308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5714496815987253e-10,
+      "loss": 0.6593,
+      "step": 6309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.217220677336318e-10,
+      "loss": 0.6901,
+      "step": 6310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8892304856299449e-10,
+      "loss": 0.677,
+      "step": 6311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.587479192555197e-10,
+      "loss": 0.7598,
+      "step": 6312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.311966877282078e-10,
+      "loss": 0.6876,
+      "step": 6313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0626936120972098e-10,
+      "loss": 0.7874,
+      "step": 6314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.396594624260346e-11,
+      "loss": 0.7142,
+      "step": 6315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.428644867773059e-11,
+      "loss": 0.267,
+      "step": 6316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7230873678749677e-11,
+      "loss": 0.7473,
+      "step": 6317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.279922572096972e-11,
+      "loss": 0.781,
+      "step": 6318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0991508592471676e-11,
+      "loss": 0.7585,
+      "step": 6319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.180772539077779e-11,
+      "loss": 0.8018,
+      "step": 6320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.247878525072026e-12,
+      "loss": 0.7246,
+      "step": 6321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3119697173102907e-12,
+      "loss": 0.4335,
+      "step": 6322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5225,
+      "step": 6323
+    },
+    {
+      "epoch": 1.0,
+      "step": 6323,
+      "total_flos": 4.063301084548956e+19,
+      "train_loss": 0.7561503465692586,
+      "train_runtime": 69621.1004,
+      "train_samples_per_second": 11.624,
+      "train_steps_per_second": 0.091
+    }
+  ],
+  "max_steps": 6323,
+  "num_train_epochs": 1,
+  "total_flos": 4.063301084548956e+19,
+  "trial_name": null,
+  "trial_params": null
+}