diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5013945034627558,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333334e-08,
+      "loss": 0.2733,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666668e-07,
+      "loss": 3.6534,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 4.2805,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 3.7941,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 0.2773,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 4.5235,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.833333333333334e-07,
+      "loss": 4.3333,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 3.5693,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-07,
+      "loss": 3.4003,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 3.4185,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.166666666666666e-07,
+      "loss": 4.212,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.07,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0833333333333335e-06,
+      "loss": 3.0956,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1666666666666668e-06,
+      "loss": 3.9265,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 2.9713,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 2.6545,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4166666666666667e-06,
+      "loss": 3.1152,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 2.3911,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5833333333333333e-06,
+      "loss": 3.8956,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 2.9585,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.75e-06,
+      "loss": 3.8622,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8333333333333333e-06,
+      "loss": 1.8076,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.916666666666667e-06,
+      "loss": 2.2738,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.2744,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 2.3213,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.166666666666667e-06,
+      "loss": 2.1399,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.25e-06,
+      "loss": 1.6026,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 1.5869,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4166666666666667e-06,
+      "loss": 2.3128,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.4336,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5833333333333337e-06,
+      "loss": 1.4416,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.4658,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 1.4329,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8333333333333335e-06,
+      "loss": 2.8689,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.916666666666667e-06,
+      "loss": 0.249,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 2.1244,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0833333333333336e-06,
+      "loss": 2.7212,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1666666666666667e-06,
+      "loss": 1.3314,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 1.3399,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.3113,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.416666666666667e-06,
+      "loss": 2.0238,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 1.3142,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5833333333333335e-06,
+      "loss": 1.3878,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 0.2486,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 1.2056,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.833333333333334e-06,
+      "loss": 2.6471,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.916666666666667e-06,
+      "loss": 3.4132,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.253,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.083333333333334e-06,
+      "loss": 1.2937,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.2808,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.25e-06,
+      "loss": 1.9313,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 2.4249,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.416666666666667e-06,
+      "loss": 2.6168,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-06,
+      "loss": 1.2657,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.583333333333333e-06,
+      "loss": 1.4319,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.9357,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75e-06,
+      "loss": 1.1517,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833333333333333e-06,
+      "loss": 1.2251,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9166666666666665e-06,
+      "loss": 3.4978,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.1468,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0833333333333335e-06,
+      "loss": 2.6464,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1666666666666675e-06,
+      "loss": 1.2364,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 1.8725,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.2226,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.416666666666667e-06,
+      "loss": 1.9008,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.128,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.583333333333334e-06,
+      "loss": 1.9102,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 2.5699,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.75e-06,
+      "loss": 1.2163,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.8087,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.916666666666667e-06,
+      "loss": 2.5603,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 1.1676,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.083333333333333e-06,
+      "loss": 1.204,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.166666666666667e-06,
+      "loss": 1.2058,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-06,
+      "loss": 1.8695,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 1.8776,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.416666666666667e-06,
+      "loss": 0.3054,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.7631,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5833333333333335e-06,
+      "loss": 1.2755,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.1133,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.1567,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.833333333333334e-06,
+      "loss": 3.9493,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.916666666666667e-06,
+      "loss": 1.8723,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-06,
+      "loss": 1.2388,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.083333333333335e-06,
+      "loss": 2.5947,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.166666666666667e-06,
+      "loss": 0.2525,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.25e-06,
+      "loss": 3.2717,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.7705,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.416666666666668e-06,
+      "loss": 1.1246,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.1178,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.583333333333333e-06,
+      "loss": 0.2253,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 1.0702,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.75e-06,
+      "loss": 4.0154,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.833333333333333e-06,
+      "loss": 1.8741,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.916666666666667e-06,
+      "loss": 1.8483,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.8521,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.083333333333334e-06,
+      "loss": 1.1687,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.166666666666668e-06,
+      "loss": 1.656,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.25e-06,
+      "loss": 1.8079,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.8594,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.416666666666667e-06,
+      "loss": 1.8086,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-06,
+      "loss": 2.4845,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.583333333333333e-06,
+      "loss": 1.7178,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.8514,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 2.5422,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.833333333333334e-06,
+      "loss": 1.1952,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.916666666666667e-06,
+      "loss": 1.0688,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-06,
+      "loss": 1.8586,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.083333333333333e-06,
+      "loss": 1.0846,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.6773,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 1.8339,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.2793,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.416666666666667e-06,
+      "loss": 3.1819,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-06,
+      "loss": 2.4113,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.583333333333335e-06,
+      "loss": 1.15,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 1.6571,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.75e-06,
+      "loss": 1.695,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.833333333333333e-06,
+      "loss": 2.9362,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.916666666666668e-06,
+      "loss": 1.0743,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.0604,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0083333333333334e-05,
+      "loss": 1.0613,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0166666666666667e-05,
+      "loss": 1.6974,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.025e-05,
+      "loss": 1.1052,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0333333333333335e-05,
+      "loss": 2.1601,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 1.3061,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.2658,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0583333333333334e-05,
+      "loss": 1.0739,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.2471,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.075e-05,
+      "loss": 1.163,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.1889,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0916666666666667e-05,
+      "loss": 1.0068,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.2237,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1083333333333335e-05,
+      "loss": 1.2262,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1166666666666668e-05,
+      "loss": 1.1646,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.125e-05,
+      "loss": 1.0797,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 1.178,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1416666666666667e-05,
+      "loss": 1.0394,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.15e-05,
+      "loss": 1.0973,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1583333333333335e-05,
+      "loss": 1.164,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.1136,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1750000000000001e-05,
+      "loss": 1.0954,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1833333333333334e-05,
+      "loss": 1.1291,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1916666666666667e-05,
+      "loss": 1.1993,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.2576,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2083333333333333e-05,
+      "loss": 1.0565,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2166666666666667e-05,
+      "loss": 1.0638,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2250000000000001e-05,
+      "loss": 1.1281,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2333333333333334e-05,
+      "loss": 0.9966,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2416666666666667e-05,
+      "loss": 1.1337,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 1.0915,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2583333333333334e-05,
+      "loss": 1.0956,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 1.1358,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.275e-05,
+      "loss": 1.0974,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2833333333333335e-05,
+      "loss": 1.1161,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2916666666666668e-05,
+      "loss": 1.0112,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.1871,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3083333333333334e-05,
+      "loss": 1.0356,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3166666666666667e-05,
+      "loss": 1.1635,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.325e-05,
+      "loss": 1.09,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1423,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3416666666666666e-05,
+      "loss": 1.1058,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.986,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3583333333333334e-05,
+      "loss": 1.0223,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3666666666666667e-05,
+      "loss": 1.0575,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.375e-05,
+      "loss": 1.0389,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3833333333333334e-05,
+      "loss": 1.0389,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3916666666666667e-05,
+      "loss": 1.1002,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 1.1286,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4083333333333336e-05,
+      "loss": 1.035,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 1.0429,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.425e-05,
+      "loss": 0.9578,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4333333333333334e-05,
+      "loss": 1.1465,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4416666666666667e-05,
+      "loss": 1.0799,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.45e-05,
+      "loss": 1.0368,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4583333333333333e-05,
+      "loss": 1.0976,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 1.0457,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4750000000000003e-05,
+      "loss": 1.0486,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4833333333333336e-05,
+      "loss": 1.0423,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4916666666666669e-05,
+      "loss": 1.1189,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.11,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5083333333333333e-05,
+      "loss": 1.0984,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5166666666666667e-05,
+      "loss": 0.9857,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.525e-05,
+      "loss": 1.0689,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 1.0268,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5416666666666668e-05,
+      "loss": 0.9905,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.55e-05,
+      "loss": 1.1431,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5583333333333334e-05,
+      "loss": 1.1216,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5666666666666667e-05,
+      "loss": 1.0236,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575e-05,
+      "loss": 1.0599,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.2684,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5916666666666666e-05,
+      "loss": 1.017,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9917,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6083333333333336e-05,
+      "loss": 0.9694,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.616666666666667e-05,
+      "loss": 1.0455,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 1.0947,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6333333333333335e-05,
+      "loss": 1.017,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6416666666666668e-05,
+      "loss": 0.995,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.65e-05,
+      "loss": 1.1055,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6583333333333334e-05,
+      "loss": 1.1576,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.0576,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.675e-05,
+      "loss": 1.1473,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6833333333333334e-05,
+      "loss": 0.9862,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6916666666666667e-05,
+      "loss": 1.0702,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7e-05,
+      "loss": 1.0479,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7083333333333333e-05,
+      "loss": 1.0792,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7166666666666666e-05,
+      "loss": 1.0719,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7250000000000003e-05,
+      "loss": 1.022,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.0798,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.741666666666667e-05,
+      "loss": 1.0166,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.0467,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7583333333333335e-05,
+      "loss": 1.0388,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7666666666666668e-05,
+      "loss": 0.2999,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.775e-05,
+      "loss": 1.0556,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7833333333333334e-05,
+      "loss": 1.0102,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7916666666666667e-05,
+      "loss": 1.125,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 1.0733,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8083333333333334e-05,
+      "loss": 1.028,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8166666666666667e-05,
+      "loss": 1.0154,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.825e-05,
+      "loss": 0.2705,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 1.0066,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8416666666666666e-05,
+      "loss": 0.9838,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 1.1033,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8583333333333336e-05,
+      "loss": 0.9692,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.9666,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.0283,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8833333333333335e-05,
+      "loss": 1.0085,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8916666666666668e-05,
+      "loss": 0.9763,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9e-05,
+      "loss": 0.952,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9083333333333338e-05,
+      "loss": 1.0198,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.9513,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.925e-05,
+      "loss": 0.9542,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 1.0777,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9416666666666667e-05,
+      "loss": 1.015,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.95e-05,
+      "loss": 0.9848,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9583333333333333e-05,
+      "loss": 0.2441,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9666666666666666e-05,
+      "loss": 0.9842,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 1.0888,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9833333333333335e-05,
+      "loss": 0.9908,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.991666666666667e-05,
+      "loss": 1.0282,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9818,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999991756255e-05,
+      "loss": 1.0074,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996702502134e-05,
+      "loss": 1.0693,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992580630307e-05,
+      "loss": 1.0196,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986810010702e-05,
+      "loss": 1.0727,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997939064427e-05,
+      "loss": 1.078,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999970322532233e-05,
+      "loss": 1.0488,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995960567609e-05,
+      "loss": 1.0414,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999472400776e-05,
+      "loss": 1.0718,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993322573881e-05,
+      "loss": 0.9522,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999917562662026e-05,
+      "loss": 1.0529,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999900250849835e-05,
+      "loss": 0.2755,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999881290305082e-05,
+      "loss": 0.3159,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999860681030904e-05,
+      "loss": 1.0674,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999838423030695e-05,
+      "loss": 1.0243,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981451630812e-05,
+      "loss": 0.9946,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999788960867126e-05,
+      "loss": 1.0447,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999761756711925e-05,
+      "loss": 1.0192,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999732903847003e-05,
+      "loss": 0.984,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999702402277115e-05,
+      "loss": 0.9638,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999967025200729e-05,
+      "loss": 0.9567,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999636453042832e-05,
+      "loss": 1.1301,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999960100538931e-05,
+      "loss": 1.0007,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999563909052574e-05,
+      "loss": 1.0247,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999952516403873e-05,
+      "loss": 1.0475,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999948477035418e-05,
+      "loss": 1.0171,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999442728005572e-05,
+      "loss": 0.9975,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999399036999843e-05,
+      "loss": 1.0139,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999353697344196e-05,
+      "loss": 1.0798,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999306709046103e-05,
+      "loss": 1.0258,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999258072113314e-05,
+      "loss": 1.01,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999207786553852e-05,
+      "loss": 0.9809,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999155852376e-05,
+      "loss": 1.0111,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999910226958833e-05,
+      "loss": 1.009,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999904703819967e-05,
+      "loss": 1.0015,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998990158219123e-05,
+      "loss": 1.0904,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998931629656074e-05,
+      "loss": 1.0218,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999887145252017e-05,
+      "loss": 1.084,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998809626821335e-05,
+      "loss": 0.9943,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998746152569754e-05,
+      "loss": 1.0045,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998681029775905e-05,
+      "loss": 1.0465,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998614258450518e-05,
+      "loss": 1.2029,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99985458386046e-05,
+      "loss": 1.0391,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998475770249438e-05,
+      "loss": 1.0135,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998404053396583e-05,
+      "loss": 1.0883,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998330688057853e-05,
+      "loss": 1.0174,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999825567424535e-05,
+      "loss": 1.009,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999817901197144e-05,
+      "loss": 1.0094,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998100701248762e-05,
+      "loss": 0.9831,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998020742090233e-05,
+      "loss": 1.0266,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999793913450903e-05,
+      "loss": 0.9834,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997855878518606e-05,
+      "loss": 0.9584,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997770974132695e-05,
+      "loss": 1.0323,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999768442136529e-05,
+      "loss": 1.0186,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997596220230666e-05,
+      "loss": 0.9653,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997506370743362e-05,
+      "loss": 1.0537,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997414872918194e-05,
+      "loss": 1.104,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997321726770247e-05,
+      "loss": 1.0933,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997226932314874e-05,
+      "loss": 1.0358,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999713048956771e-05,
+      "loss": 0.9466,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997032398544654e-05,
+      "loss": 1.001,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999693265926188e-05,
+      "loss": 0.9707,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999683127173583e-05,
+      "loss": 1.0481,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996728235983224e-05,
+      "loss": 1.0201,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996623552021045e-05,
+      "loss": 0.9115,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999651721986656e-05,
+      "loss": 1.0548,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999640923953729e-05,
+      "loss": 1.0164,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996299611051046e-05,
+      "loss": 1.0992,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99961883344259e-05,
+      "loss": 1.1012,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996075409680205e-05,
+      "loss": 1.0311,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999596083683257e-05,
+      "loss": 1.0584,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999584461590189e-05,
+      "loss": 1.0884,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995726746907327e-05,
+      "loss": 1.1096,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999560722986831e-05,
+      "loss": 1.0446,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995486064804556e-05,
+      "loss": 0.9567,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995363251736027e-05,
+      "loss": 0.9763,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995238790682988e-05,
+      "loss": 0.934,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995112681665943e-05,
+      "loss": 0.9934,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994984924705693e-05,
+      "loss": 1.0341,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994855519823307e-05,
+      "loss": 1.0041,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999472446704011e-05,
+      "loss": 1.0394,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999459176637771e-05,
+      "loss": 0.9275,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994457417857998e-05,
+      "loss": 1.0308,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999432142150311e-05,
+      "loss": 1.0108,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994183777335477e-05,
+      "loss": 0.975,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994044485377793e-05,
+      "loss": 0.9849,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390354565302e-05,
+      "loss": 1.0323,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993760958184397e-05,
+      "loss": 1.0451,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993616722995433e-05,
+      "loss": 1.0106,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999347084010991e-05,
+      "loss": 0.8634,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993323309551877e-05,
+      "loss": 0.9933,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993174131345665e-05,
+      "loss": 1.0161,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993023305515863e-05,
+      "loss": 1.0288,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999287083208734e-05,
+      "loss": 1.0365,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992716711085236e-05,
+      "loss": 0.9835,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992560942534963e-05,
+      "loss": 1.0146,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99924035264622e-05,
+      "loss": 0.9506,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992244462892903e-05,
+      "loss": 1.0162,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992083751853296e-05,
+      "loss": 1.0103,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991921393369878e-05,
+      "loss": 0.9291,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999175738746942e-05,
+      "loss": 1.0065,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999159173417896e-05,
+      "loss": 0.9692,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991424433525806e-05,
+      "loss": 1.0685,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991255485537547e-05,
+      "loss": 1.0158,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991084890242038e-05,
+      "loss": 0.9994,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990912647667404e-05,
+      "loss": 0.9829,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990738757842044e-05,
+      "loss": 0.9684,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999056322079463e-05,
+      "loss": 1.063,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99903860365541e-05,
+      "loss": 1.0067,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999020720514967e-05,
+      "loss": 0.9581,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999002672661082e-05,
+      "loss": 0.9727,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989844600967315e-05,
+      "loss": 0.9065,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989660828249175e-05,
+      "loss": 0.9952,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989475408486708e-05,
+      "loss": 1.0114,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989288341710474e-05,
+      "loss": 1.045,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989099627951327e-05,
+      "loss": 0.2972,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998890926724037e-05,
+      "loss": 0.9989,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988717259609e-05,
+      "loss": 0.2801,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998852360508886e-05,
+      "loss": 1.0095,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988328303711896e-05,
+      "loss": 0.9707,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988131355510296e-05,
+      "loss": 1.048,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987932760516538e-05,
+      "loss": 1.0911,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998773251876336e-05,
+      "loss": 0.9877,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998753063028378e-05,
+      "loss": 0.9665,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987327095111085e-05,
+      "loss": 0.9336,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998712191327883e-05,
+      "loss": 1.0115,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986915084820847e-05,
+      "loss": 0.994,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986706609771236e-05,
+      "loss": 0.9339,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998649648816437e-05,
+      "loss": 0.9783,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986284720034894e-05,
+      "loss": 0.9433,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986071305417715e-05,
+      "loss": 0.9274,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985856244348034e-05,
+      "loss": 1.1452,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985639536861297e-05,
+      "loss": 0.9767,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985421182993236e-05,
+      "loss": 1.1309,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985201182779853e-05,
+      "loss": 0.9838,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984979536257428e-05,
+      "loss": 0.2794,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984756243462492e-05,
+      "loss": 1.0742,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998453130443187e-05,
+      "loss": 0.9872,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984304719202647e-05,
+      "loss": 1.0201,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984076487812175e-05,
+      "loss": 1.0381,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998384661029809e-05,
+      "loss": 0.9768,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983615086698295e-05,
+      "loss": 0.2901,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983381917050954e-05,
+      "loss": 0.9645,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983147101394516e-05,
+      "loss": 0.9789,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982910639767697e-05,
+      "loss": 0.9793,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982672532209487e-05,
+      "loss": 0.9411,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998243277875913e-05,
+      "loss": 0.9933,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982191379456176e-05,
+      "loss": 1.0326,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981948334340405e-05,
+      "loss": 1.0699,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981703643451902e-05,
+      "loss": 1.0102,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981457306831008e-05,
+      "loss": 0.9871,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981209324518335e-05,
+      "loss": 1.0026,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998095969655477e-05,
+      "loss": 1.0114,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980708422981473e-05,
+      "loss": 0.9901,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998045550383987e-05,
+      "loss": 1.0554,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980200939171662e-05,
+      "loss": 0.9758,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979944729018817e-05,
+      "loss": 0.9567,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979686873423586e-05,
+      "loss": 1.0077,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979427372428472e-05,
+      "loss": 1.0439,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997916622607627e-05,
+      "loss": 0.9862,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997890343441003e-05,
+      "loss": 0.933,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978638997473083e-05,
+      "loss": 0.9781,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997837291530903e-05,
+      "loss": 0.9921,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978105187961732e-05,
+      "loss": 0.986,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997783581547534e-05,
+      "loss": 0.2929,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977564797894264e-05,
+      "loss": 0.9835,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977292135263187e-05,
+      "loss": 0.9732,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977017827627062e-05,
+      "loss": 1.0172,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976741875031123e-05,
+      "loss": 0.9816,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997646427752086e-05,
+      "loss": 0.9242,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976185035142048e-05,
+      "loss": 1.0357,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997590414794072e-05,
+      "loss": 1.0057,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975621615963196e-05,
+      "loss": 0.9522,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975337439256046e-05,
+      "loss": 0.957,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975051617866137e-05,
+      "loss": 0.9894,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974764151840585e-05,
+      "loss": 0.9574,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997447504122679e-05,
+      "loss": 0.9394,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997418428607242e-05,
+      "loss": 1.0156,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973891886425408e-05,
+      "loss": 0.9567,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973597842333967e-05,
+      "loss": 0.9186,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973302153846577e-05,
+      "loss": 0.9455,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997300482101199e-05,
+      "loss": 0.9996,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997270584387923e-05,
+      "loss": 0.977,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972405222497587e-05,
+      "loss": 0.9709,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972102956916628e-05,
+      "loss": 1.0286,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971799047186188e-05,
+      "loss": 1.0035,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997149349335638e-05,
+      "loss": 0.3135,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971186295477575e-05,
+      "loss": 0.9547,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970877453600423e-05,
+      "loss": 0.99,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970566967775846e-05,
+      "loss": 1.0914,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970254838055037e-05,
+      "loss": 0.954,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969941064489452e-05,
+      "loss": 0.2809,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969625647130835e-05,
+      "loss": 0.9372,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996930858603118e-05,
+      "loss": 0.9868,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968989881242766e-05,
+      "loss": 0.9391,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968669532818143e-05,
+      "loss": 0.984,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968347540810125e-05,
+      "loss": 0.9573,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99680239052718e-05,
+      "loss": 0.9549,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996769862625653e-05,
+      "loss": 0.9554,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996737170381794e-05,
+      "loss": 0.9956,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996704313800994e-05,
+      "loss": 0.9913,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966712928886697e-05,
+      "loss": 0.9268,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996638107650265e-05,
+      "loss": 1.0003,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996604758091252e-05,
+      "loss": 1.1636,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965712442171287e-05,
+      "loss": 0.9307,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965375660334213e-05,
+      "loss": 0.2806,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996503723545682e-05,
+      "loss": 1.0007,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996469716759491e-05,
+      "loss": 1.0599,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996435545680454e-05,
+      "loss": 1.0122,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964012103142067e-05,
+      "loss": 0.8953,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996366710666409e-05,
+      "loss": 0.9816,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963320467427496e-05,
+      "loss": 0.9289,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996297218548943e-05,
+      "loss": 0.3136,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996262226090732e-05,
+      "loss": 0.8827,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996227069373886e-05,
+      "loss": 1.0178,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961917484042012e-05,
+      "loss": 0.9106,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961562631875013e-05,
+      "loss": 0.9125,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996120613729637e-05,
+      "loss": 0.9418,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996084800036486e-05,
+      "loss": 0.9186,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996048822113953e-05,
+      "loss": 0.9463,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99601267996797e-05,
+      "loss": 0.9807,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959763736044953e-05,
+      "loss": 0.9904,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959399030295158e-05,
+      "loss": 0.858,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995903268249044e-05,
+      "loss": 1.0022,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958664692691202e-05,
+      "loss": 0.2872,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958295060958113e-05,
+      "loss": 0.9801,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957923787352124e-05,
+      "loss": 0.9799,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957550871934442e-05,
+      "loss": 0.9612,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957176314766554e-05,
+      "loss": 1.0198,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956800115910216e-05,
+      "loss": 0.9679,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956422275427455e-05,
+      "loss": 0.323,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995604279338056e-05,
+      "loss": 0.2387,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955661669832106e-05,
+      "loss": 0.9701,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955278904844926e-05,
+      "loss": 0.9116,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995489449848213e-05,
+      "loss": 0.9595,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954508450807098e-05,
+      "loss": 0.9385,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995412076188348e-05,
+      "loss": 1.0246,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995373143177519e-05,
+      "loss": 1.0276,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995334046054643e-05,
+      "loss": 0.9691,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995294784826165e-05,
+      "loss": 0.9621,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952553594985586e-05,
+      "loss": 1.0289,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952157700783243e-05,
+      "loss": 0.9447,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995176016571989e-05,
+      "loss": 0.9497,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951360989861077e-05,
+      "loss": 0.8977,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950960173272614e-05,
+      "loss": 0.9264,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950557716020583e-05,
+      "loss": 1.0022,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995015361817134e-05,
+      "loss": 0.9006,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994974787979151e-05,
+      "loss": 0.9399,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949340500948e-05,
+      "loss": 1.0261,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948931481707962e-05,
+      "loss": 0.9993,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948520822138837e-05,
+      "loss": 0.9387,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948108522308337e-05,
+      "loss": 0.9137,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947694582284437e-05,
+      "loss": 0.8491,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994727900213538e-05,
+      "loss": 1.0324,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946861781929697e-05,
+      "loss": 1.0395,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946442921736172e-05,
+      "loss": 0.9995,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994602242162386e-05,
+      "loss": 1.0388,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945600281662088e-05,
+      "loss": 0.2804,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994517650192047e-05,
+      "loss": 1.0413,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944751082468864e-05,
+      "loss": 1.0426,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994432402337742e-05,
+      "loss": 1.0382,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943895324716542e-05,
+      "loss": 0.8986,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943464986556914e-05,
+      "loss": 0.959,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943033008969488e-05,
+      "loss": 0.9648,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942599392025488e-05,
+      "loss": 0.9985,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942164135796405e-05,
+      "loss": 0.9893,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941727240354e-05,
+      "loss": 1.0282,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941288705770314e-05,
+      "loss": 0.9617,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994084853211764e-05,
+      "loss": 0.8942,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994040671946856e-05,
+      "loss": 0.9434,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939963267895912e-05,
+      "loss": 0.9724,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939518177472813e-05,
+      "loss": 0.946,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939071448272642e-05,
+      "loss": 0.9692,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938623080369063e-05,
+      "loss": 0.9908,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938173073835992e-05,
+      "loss": 0.9807,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993772142874763e-05,
+      "loss": 0.9881,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937268145178437e-05,
+      "loss": 0.9241,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993681322320315e-05,
+      "loss": 0.8792,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936356662896777e-05,
+      "loss": 0.8797,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993589846433459e-05,
+      "loss": 0.9488,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935438627592135e-05,
+      "loss": 1.0937,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934977152745226e-05,
+      "loss": 0.8891,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993451403986995e-05,
+      "loss": 0.2767,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934049289042665e-05,
+      "loss": 0.942,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933582900339993e-05,
+      "loss": 1.0235,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933114873838832e-05,
+      "loss": 0.9775,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932645209616347e-05,
+      "loss": 0.9371,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932173907749978e-05,
+      "loss": 0.9576,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993170096831742e-05,
+      "loss": 0.9517,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931226391396663e-05,
+      "loss": 0.9998,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993075017706594e-05,
+      "loss": 0.9766,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930272325403778e-05,
+      "loss": 0.9955,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929792836488954e-05,
+      "loss": 0.9166,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929311710400528e-05,
+      "loss": 1.0329,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928828947217824e-05,
+      "loss": 0.9513,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992834454702044e-05,
+      "loss": 0.9246,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927858509888235e-05,
+      "loss": 0.9944,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927370835901355e-05,
+      "loss": 0.9975,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926881525140195e-05,
+      "loss": 0.9744,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926390577685434e-05,
+      "loss": 0.278,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925897993618016e-05,
+      "loss": 0.9053,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992540377301916e-05,
+      "loss": 0.9273,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924907915970345e-05,
+      "loss": 0.9657,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924410422553332e-05,
+      "loss": 1.0401,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923911292850136e-05,
+      "loss": 0.9772,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923410526943056e-05,
+      "loss": 0.9817,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992290812491466e-05,
+      "loss": 0.8818,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922404086847774e-05,
+      "loss": 0.9505,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921898412825503e-05,
+      "loss": 0.254,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992139110293122e-05,
+      "loss": 0.9839,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920882157248575e-05,
+      "loss": 0.9874,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992037157586147e-05,
+      "loss": 0.9393,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919859358854094e-05,
+      "loss": 0.89,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919345506310896e-05,
+      "loss": 0.9329,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991883001831659e-05,
+      "loss": 0.9325,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918312894956182e-05,
+      "loss": 0.9348,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917794136314924e-05,
+      "loss": 0.9126,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917273742478347e-05,
+      "loss": 0.9662,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916751713532255e-05,
+      "loss": 0.9953,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916228049562707e-05,
+      "loss": 0.9063,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915702750656053e-05,
+      "loss": 0.952,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915175816898897e-05,
+      "loss": 1.028,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914647248378117e-05,
+      "loss": 0.8946,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914117045180862e-05,
+      "loss": 0.98,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913585207394548e-05,
+      "loss": 0.9432,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991305173510686e-05,
+      "loss": 0.9135,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991251662840576e-05,
+      "loss": 0.9949,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991197988737947e-05,
+      "loss": 0.9068,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911441512116482e-05,
+      "loss": 1.006,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910901502705566e-05,
+      "loss": 0.2856,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991035985923575e-05,
+      "loss": 0.9849,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909816581796348e-05,
+      "loss": 1.0137,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990927167047692e-05,
+      "loss": 0.9499,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908725125367317e-05,
+      "loss": 0.9896,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908176946557646e-05,
+      "loss": 1.023,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907627134138292e-05,
+      "loss": 0.9434,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99070756881999e-05,
+      "loss": 0.9464,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906522608833395e-05,
+      "loss": 0.9646,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990596789612996e-05,
+      "loss": 0.9748,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990541155018106e-05,
+      "loss": 0.9603,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904853571078417e-05,
+      "loss": 0.915,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904293958914032e-05,
+      "loss": 0.93,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903732713780168e-05,
+      "loss": 0.9793,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903169835769362e-05,
+      "loss": 0.9358,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902605324974414e-05,
+      "loss": 0.9684,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902039181488407e-05,
+      "loss": 1.0243,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901471405404673e-05,
+      "loss": 0.8915,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900901996816833e-05,
+      "loss": 0.9103,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990033095581876e-05,
+      "loss": 1.0244,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989975828250461e-05,
+      "loss": 0.8969,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899183976968802e-05,
+      "loss": 0.9156,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898608039306023e-05,
+      "loss": 0.9317,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898030469611234e-05,
+      "loss": 0.9525,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989745126797966e-05,
+      "loss": 1.0082,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896870434506788e-05,
+      "loss": 0.9533,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896287969288396e-05,
+      "loss": 0.9754,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989570387242051e-05,
+      "loss": 0.8874,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895118143999438e-05,
+      "loss": 0.3109,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989453078412175e-05,
+      "loss": 0.9803,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893941792884284e-05,
+      "loss": 1.0182,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989335117038415e-05,
+      "loss": 1.0133,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892758916718736e-05,
+      "loss": 0.8936,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989216503198568e-05,
+      "loss": 0.2698,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891569516282898e-05,
+      "loss": 0.9728,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890972369708584e-05,
+      "loss": 0.9086,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890373592361183e-05,
+      "loss": 0.9531,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889773184339425e-05,
+      "loss": 1.0069,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98891711457423e-05,
+      "loss": 0.9539,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988856747666907e-05,
+      "loss": 0.8911,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988796217721926e-05,
+      "loss": 1.0322,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988735524749268e-05,
+      "loss": 0.2678,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886746687589386e-05,
+      "loss": 1.0108,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988613649760972e-05,
+      "loss": 0.9125,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885524677654283e-05,
+      "loss": 0.9532,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988491122782395e-05,
+      "loss": 0.978,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884296148219867e-05,
+      "loss": 0.9841,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883679438943444e-05,
+      "loss": 0.9296,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988306110009636e-05,
+      "loss": 0.973,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988244113178056e-05,
+      "loss": 0.9333,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881819534098267e-05,
+      "loss": 0.9333,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881196307151966e-05,
+      "loss": 0.9736,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880571451044405e-05,
+      "loss": 0.8079,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879944965878615e-05,
+      "loss": 1.0267,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879316851757885e-05,
+      "loss": 0.9419,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878687108785774e-05,
+      "loss": 0.9253,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987805573706611e-05,
+      "loss": 0.9217,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877422736702997e-05,
+      "loss": 1.0117,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876788107800792e-05,
+      "loss": 0.9089,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876151850464133e-05,
+      "loss": 0.9369,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875513964797925e-05,
+      "loss": 0.973,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874874450907338e-05,
+      "loss": 0.9836,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987423330889781e-05,
+      "loss": 0.9159,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987359053887505e-05,
+      "loss": 0.9242,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987294614094504e-05,
+      "loss": 0.8919,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987230011521401e-05,
+      "loss": 1.0233,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871652461788492e-05,
+      "loss": 0.9478,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871003180775257e-05,
+      "loss": 1.0094,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987035227228136e-05,
+      "loss": 0.9349,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869699736414115e-05,
+      "loss": 0.9374,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986904557328111e-05,
+      "loss": 0.9285,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98683897829902e-05,
+      "loss": 0.93,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867732365649514e-05,
+      "loss": 0.9177,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867073321367435e-05,
+      "loss": 0.8986,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866412650252624e-05,
+      "loss": 0.9132,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865750352414016e-05,
+      "loss": 0.9344,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865086427960804e-05,
+      "loss": 1.0348,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864420877002447e-05,
+      "loss": 1.0015,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986375369964869e-05,
+      "loss": 0.8729,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986308489600952e-05,
+      "loss": 0.9208,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986241446619521e-05,
+      "loss": 1.0125,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861742410316304e-05,
+      "loss": 0.9866,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861068728483603e-05,
+      "loss": 0.9884,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860393420808175e-05,
+      "loss": 0.9319,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859716487401368e-05,
+      "loss": 1.0037,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985903792837479e-05,
+      "loss": 0.8868,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858357743840314e-05,
+      "loss": 0.943,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985767593391009e-05,
+      "loss": 1.039,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985699249869653e-05,
+      "loss": 1.0101,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985630743831232e-05,
+      "loss": 1.0542,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98556207528704e-05,
+      "loss": 0.9968,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985493244248399e-05,
+      "loss": 0.9067,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985424250726658e-05,
+      "loss": 0.8572,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985355094733192e-05,
+      "loss": 0.9206,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985285776279403e-05,
+      "loss": 0.8677,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98521629537672e-05,
+      "loss": 0.9976,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985146652036599e-05,
+      "loss": 1.0021,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850768462705214e-05,
+      "loss": 0.8783,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850068780899973e-05,
+      "loss": 0.8896,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849367475065625e-05,
+      "loss": 0.9432,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98486645453178e-05,
+      "loss": 0.9431,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847959991772384e-05,
+      "loss": 0.3091,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847253814545555e-05,
+      "loss": 0.9824,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984654601375373e-05,
+      "loss": 0.9877,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845836589513614e-05,
+      "loss": 0.9248,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845125541942175e-05,
+      "loss": 0.9791,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844412871156648e-05,
+      "loss": 0.9996,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843698577274525e-05,
+      "loss": 0.977,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842982660413585e-05,
+      "loss": 1.0114,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984226512069186e-05,
+      "loss": 0.9898,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841545958227654e-05,
+      "loss": 0.9071,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840825173139538e-05,
+      "loss": 0.9199,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840102765546357e-05,
+      "loss": 0.94,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839378735567218e-05,
+      "loss": 0.9703,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838653083321488e-05,
+      "loss": 0.9608,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983792580892881e-05,
+      "loss": 0.9936,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837196912509102e-05,
+      "loss": 0.9028,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983646639418253e-05,
+      "loss": 0.9553,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835734254069545e-05,
+      "loss": 0.9522,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835000492290855e-05,
+      "loss": 1.0227,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834265108967443e-05,
+      "loss": 0.9238,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983352810422055e-05,
+      "loss": 0.8704,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832789478171692e-05,
+      "loss": 0.8974,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983204923094265e-05,
+      "loss": 0.9605,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831307362655473e-05,
+      "loss": 0.9148,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830563873432476e-05,
+      "loss": 0.9764,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982981876339624e-05,
+      "loss": 0.9283,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829072032669617e-05,
+      "loss": 0.9412,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828323681375723e-05,
+      "loss": 0.9674,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827573709637944e-05,
+      "loss": 0.9689,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826822117579928e-05,
+      "loss": 0.9283,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826068905325598e-05,
+      "loss": 0.2724,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982531407299914e-05,
+      "loss": 0.9733,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824557620724997e-05,
+      "loss": 1.0395,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98237995486279e-05,
+      "loss": 1.0153,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823039856832837e-05,
+      "loss": 0.9161,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822278545465055e-05,
+      "loss": 0.9314,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982151561465008e-05,
+      "loss": 0.9253,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820751064513693e-05,
+      "loss": 0.9396,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981998489518196e-05,
+      "loss": 0.9364,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819217106781193e-05,
+      "loss": 0.9418,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818447699437987e-05,
+      "loss": 0.9005,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817676673279194e-05,
+      "loss": 0.9155,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816904028431942e-05,
+      "loss": 0.928,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816129765023618e-05,
+      "loss": 0.8617,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981535388318188e-05,
+      "loss": 0.9916,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814576383034647e-05,
+      "loss": 1.0249,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981379726471011e-05,
+      "loss": 0.9907,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813016528336732e-05,
+      "loss": 0.3139,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812234174043234e-05,
+      "loss": 0.9386,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811450201958602e-05,
+      "loss": 0.8361,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810664612212102e-05,
+      "loss": 1.0057,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980987740493325e-05,
+      "loss": 0.987,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980908858025184e-05,
+      "loss": 0.9128,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808298138297925e-05,
+      "loss": 0.8921,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980750607920184e-05,
+      "loss": 0.8561,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806712403094164e-05,
+      "loss": 0.9338,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805917110105765e-05,
+      "loss": 1.0201,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805120200367757e-05,
+      "loss": 0.2792,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804321674011533e-05,
+      "loss": 0.9631,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803521531168752e-05,
+      "loss": 0.9499,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980271977197134e-05,
+      "loss": 0.969,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980191639655148e-05,
+      "loss": 0.9584,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801111405041634e-05,
+      "loss": 0.8634,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800304797574522e-05,
+      "loss": 0.9481,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799496574283136e-05,
+      "loss": 1.0203,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979868673530073e-05,
+      "loss": 0.9477,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797875280760824e-05,
+      "loss": 1.0307,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979706221079721e-05,
+      "loss": 1.0363,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796247525543944e-05,
+      "loss": 0.9691,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795431225135342e-05,
+      "loss": 0.9235,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794613309705997e-05,
+      "loss": 0.9655,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793793779390754e-05,
+      "loss": 1.016,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792972634324744e-05,
+      "loss": 0.8841,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979214987464335e-05,
+      "loss": 0.8543,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791325500482217e-05,
+      "loss": 0.8827,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979049951197727e-05,
+      "loss": 0.9302,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789671909264697e-05,
+      "loss": 0.9235,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788842692480945e-05,
+      "loss": 0.959,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788011861762726e-05,
+      "loss": 0.9233,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787179417247032e-05,
+      "loss": 0.8636,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786345359071107e-05,
+      "loss": 0.9071,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785509687372468e-05,
+      "loss": 0.9661,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784672402288893e-05,
+      "loss": 0.9116,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783833503958436e-05,
+      "loss": 0.9469,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782992992519402e-05,
+      "loss": 0.9597,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782150868110378e-05,
+      "loss": 0.8779,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781307130870204e-05,
+      "loss": 1.0074,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780461780937995e-05,
+      "loss": 0.9938,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779614818453125e-05,
+      "loss": 0.9043,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778766243555235e-05,
+      "loss": 0.9015,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977791605638424e-05,
+      "loss": 0.9063,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977706425708031e-05,
+      "loss": 0.8594,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776210845783885e-05,
+      "loss": 0.8377,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775355822635675e-05,
+      "loss": 0.9486,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774499187776643e-05,
+      "loss": 0.9355,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977364094134804e-05,
+      "loss": 0.9078,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772781083491356e-05,
+      "loss": 0.2649,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977191961434837e-05,
+      "loss": 0.9004,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977105653406111e-05,
+      "loss": 0.9764,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977019184277188e-05,
+      "loss": 0.8928,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976932554062325e-05,
+      "loss": 0.9242,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976845762775804e-05,
+      "loss": 0.8661,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976758810431936e-05,
+      "loss": 0.9421,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766716970450556e-05,
+      "loss": 0.8986,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765844226295274e-05,
+      "loss": 0.9729,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764969871997397e-05,
+      "loss": 0.8989,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764093907701086e-05,
+      "loss": 1.0016,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763216333550768e-05,
+      "loss": 0.9357,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762337149691132e-05,
+      "loss": 0.9132,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976145635626713e-05,
+      "loss": 0.991,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760573953423983e-05,
+      "loss": 0.9537,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975968994130719e-05,
+      "loss": 0.9399,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975880432006248e-05,
+      "loss": 0.8805,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757917089835887e-05,
+      "loss": 0.8882,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757028250773686e-05,
+      "loss": 0.9333,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756137803022427e-05,
+      "loss": 1.0314,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755245746728922e-05,
+      "loss": 0.9391,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754352082040244e-05,
+      "loss": 0.9552,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753456809103743e-05,
+      "loss": 0.9595,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752559928067026e-05,
+      "loss": 0.8783,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975166143907796e-05,
+      "loss": 0.8508,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975076134228469e-05,
+      "loss": 0.9192,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749859637835613e-05,
+      "loss": 0.9714,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748956325879406e-05,
+      "loss": 0.9413,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748051406564994e-05,
+      "loss": 0.9262,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747144880041584e-05,
+      "loss": 0.9802,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746236746458626e-05,
+      "loss": 0.9088,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745327005965862e-05,
+      "loss": 0.9921,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744415658713282e-05,
+      "loss": 0.9619,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974350270485114e-05,
+      "loss": 1.0331,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974258814452996e-05,
+      "loss": 0.831,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974167197790054e-05,
+      "loss": 0.9593,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740754205113917e-05,
+      "loss": 1.0863,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973983482632142e-05,
+      "loss": 0.8722,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738913841674624e-05,
+      "loss": 0.8694,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737991251325384e-05,
+      "loss": 0.8893,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737067055425805e-05,
+      "loss": 0.9818,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973614125412827e-05,
+      "loss": 0.9717,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735213847585415e-05,
+      "loss": 0.9264,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973428483595015e-05,
+      "loss": 0.9322,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733354219375645e-05,
+      "loss": 0.8868,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732421998015332e-05,
+      "loss": 0.8935,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731488172022915e-05,
+      "loss": 0.9773,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730552741552357e-05,
+      "loss": 0.9291,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729615706757884e-05,
+      "loss": 0.8823,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728677067793993e-05,
+      "loss": 0.9348,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727736824815442e-05,
+      "loss": 0.2995,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726794977977254e-05,
+      "loss": 0.932,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972585152743471e-05,
+      "loss": 0.926,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972490647334337e-05,
+      "loss": 0.9101,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723959815859046e-05,
+      "loss": 0.3043,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723011555137815e-05,
+      "loss": 0.9524,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972206169133603e-05,
+      "loss": 0.8792,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972111022461029e-05,
+      "loss": 0.8971,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720157155117466e-05,
+      "loss": 0.9062,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971920248301471e-05,
+      "loss": 0.9249,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971824620845941e-05,
+      "loss": 0.9331,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717288331609237e-05,
+      "loss": 0.8889,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716328852622117e-05,
+      "loss": 0.9164,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715367771656252e-05,
+      "loss": 0.8731,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971440508887009e-05,
+      "loss": 0.8796,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713440804422365e-05,
+      "loss": 0.8468,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712474918472052e-05,
+      "loss": 0.9824,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711507431178403e-05,
+      "loss": 0.8885,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710538342700943e-05,
+      "loss": 0.9675,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709567653199437e-05,
+      "loss": 0.9308,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708595362833935e-05,
+      "loss": 0.9224,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707621471764742e-05,
+      "loss": 0.9161,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706645980152427e-05,
+      "loss": 0.9734,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705668888157826e-05,
+      "loss": 0.9342,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704690195942035e-05,
+      "loss": 1.0019,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703709903666414e-05,
+      "loss": 0.9467,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970272801149259e-05,
+      "loss": 0.9305,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701744519582456e-05,
+      "loss": 0.9747,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700759428098164e-05,
+      "loss": 0.9893,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699772737202126e-05,
+      "loss": 0.9429,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969878444705703e-05,
+      "loss": 0.8986,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697794557825812e-05,
+      "loss": 1.0291,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696803069671684e-05,
+      "loss": 0.9016,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695809982758122e-05,
+      "loss": 0.9003,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969481529724885e-05,
+      "loss": 0.9651,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693819013307877e-05,
+      "loss": 0.9199,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692821131099463e-05,
+      "loss": 1.0212,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969182165078813e-05,
+      "loss": 0.9153,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969082057253867e-05,
+      "loss": 0.9728,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689817896516133e-05,
+      "loss": 0.9139,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688813622885843e-05,
+      "loss": 0.9228,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968780775181337e-05,
+      "loss": 0.9807,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968680028346456e-05,
+      "loss": 0.9352,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968579121800552e-05,
+      "loss": 0.9626,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968478055560262e-05,
+      "loss": 0.9014,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683768296422495e-05,
+      "loss": 0.8971,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682754440632036e-05,
+      "loss": 0.9301,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96817389883984e-05,
+      "loss": 0.9667,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680721939889022e-05,
+      "loss": 0.9527,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679703295271578e-05,
+      "loss": 0.9531,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678683054714017e-05,
+      "loss": 0.9295,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677661218384555e-05,
+      "loss": 1.0495,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676637786451665e-05,
+      "loss": 0.8808,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675612759084088e-05,
+      "loss": 0.9531,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674586136450816e-05,
+      "loss": 0.9434,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673557918721127e-05,
+      "loss": 0.9636,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967252810606454e-05,
+      "loss": 0.9151,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671496698650846e-05,
+      "loss": 0.9411,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670463696650097e-05,
+      "loss": 0.8734,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966942910023261e-05,
+      "loss": 0.9078,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966839290956897e-05,
+      "loss": 0.8817,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966735512483001e-05,
+      "loss": 0.9314,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966631574618684e-05,
+      "loss": 0.8885,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665274773810827e-05,
+      "loss": 0.9115,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96642322078736e-05,
+      "loss": 0.904,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663188048547052e-05,
+      "loss": 1.0927,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662142296003335e-05,
+      "loss": 0.8933,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661094950414874e-05,
+      "loss": 0.9232,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966004601195435e-05,
+      "loss": 0.3132,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658995480794697e-05,
+      "loss": 0.9308,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657943357109132e-05,
+      "loss": 0.8885,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965688964107112e-05,
+      "loss": 0.8552,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655834332854392e-05,
+      "loss": 0.9389,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965477743263294e-05,
+      "loss": 0.8971,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653718940581025e-05,
+      "loss": 0.9512,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965265885687316e-05,
+      "loss": 0.9656,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965159718168413e-05,
+      "loss": 1.0162,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650533915188977e-05,
+      "loss": 0.8686,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964946905756301e-05,
+      "loss": 0.905,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648402608981797e-05,
+      "loss": 0.8737,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964733456962116e-05,
+      "loss": 0.9727,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646264939657206e-05,
+      "loss": 0.9724,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645193719266278e-05,
+      "loss": 0.9643,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644120908624997e-05,
+      "loss": 0.8273,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643046507910246e-05,
+      "loss": 0.9607,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641970517299166e-05,
+      "loss": 0.9866,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640892936969156e-05,
+      "loss": 0.8859,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639813767097886e-05,
+      "loss": 0.8901,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638733007863284e-05,
+      "loss": 0.9916,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637650659443535e-05,
+      "loss": 1.0045,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96365667220171e-05,
+      "loss": 0.9038,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963548119576269e-05,
+      "loss": 0.8681,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634394080859273e-05,
+      "loss": 0.9227,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96333053774861e-05,
+      "loss": 0.919,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632215085822658e-05,
+      "loss": 0.9801,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963112320604872e-05,
+      "loss": 0.8892,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96300297383443e-05,
+      "loss": 0.8807,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962893468288969e-05,
+      "loss": 0.3157,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627838039865436e-05,
+      "loss": 0.9673,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962673980945234e-05,
+      "loss": 0.8849,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625639991831485e-05,
+      "loss": 0.9723,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624538587184197e-05,
+      "loss": 0.9464,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962343559569207e-05,
+      "loss": 0.9233,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622331017536955e-05,
+      "loss": 0.9464,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621224852900975e-05,
+      "loss": 0.8288,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962011710196651e-05,
+      "loss": 0.922,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619007764916195e-05,
+      "loss": 0.9587,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617896841932938e-05,
+      "loss": 0.2984,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616784333199896e-05,
+      "loss": 0.8398,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615670238900497e-05,
+      "loss": 0.97,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614554559218427e-05,
+      "loss": 0.9402,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613437294337636e-05,
+      "loss": 0.9843,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612318444442326e-05,
+      "loss": 0.9767,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611198009716976e-05,
+      "loss": 0.9619,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610075990346313e-05,
+      "loss": 0.9155,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608952386515327e-05,
+      "loss": 0.9092,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607827198409275e-05,
+      "loss": 0.9215,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606700426213674e-05,
+      "loss": 0.9309,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96055720701143e-05,
+      "loss": 0.9576,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604442130297187e-05,
+      "loss": 0.9449,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603310606948638e-05,
+      "loss": 0.894,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960217750025521e-05,
+      "loss": 0.9724,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601042810403725e-05,
+      "loss": 0.9458,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599906537581266e-05,
+      "loss": 0.9027,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598768681975173e-05,
+      "loss": 0.9963,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959762924377305e-05,
+      "loss": 0.8311,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596488223162765e-05,
+      "loss": 0.8941,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959534562033244e-05,
+      "loss": 0.9415,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959420143547047e-05,
+      "loss": 0.8973,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959305566876549e-05,
+      "loss": 0.9387,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591908320406414e-05,
+      "loss": 0.9213,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959075939058241e-05,
+      "loss": 0.9064,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958960887948291e-05,
+      "loss": 0.9435,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588456787297602e-05,
+      "loss": 0.9617,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958730311421644e-05,
+      "loss": 0.9049,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586147860429635e-05,
+      "loss": 0.9918,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584991026127655e-05,
+      "loss": 0.8929,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583832611501238e-05,
+      "loss": 0.9623,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582672616741374e-05,
+      "loss": 0.9231,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581511042039318e-05,
+      "loss": 0.8548,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580347887586586e-05,
+      "loss": 0.9293,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957918315357495e-05,
+      "loss": 0.9197,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957801684019645e-05,
+      "loss": 0.9743,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957684894764338e-05,
+      "loss": 0.9839,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575679476108294e-05,
+      "loss": 0.8587,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957450842578401e-05,
+      "loss": 0.9073,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573335796863603e-05,
+      "loss": 0.8838,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572161589540415e-05,
+      "loss": 0.9002,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957098580400804e-05,
+      "loss": 0.932,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569808440460333e-05,
+      "loss": 0.895,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568629499091413e-05,
+      "loss": 0.9126,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567448980095663e-05,
+      "loss": 0.943,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566266883667718e-05,
+      "loss": 0.9699,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956508321000247e-05,
+      "loss": 0.9716,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563897959295085e-05,
+      "loss": 0.8734,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562711131740977e-05,
+      "loss": 0.9126,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956152272753583e-05,
+      "loss": 0.8385,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560332746875574e-05,
+      "loss": 0.3207,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955914118995641e-05,
+      "loss": 0.8766,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557948056974796e-05,
+      "loss": 0.9055,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955675334812745e-05,
+      "loss": 0.9285,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555557063611353e-05,
+      "loss": 0.9275,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554359203623734e-05,
+      "loss": 0.9362,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553159768362095e-05,
+      "loss": 0.7883,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551958758024194e-05,
+      "loss": 0.9144,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955075617280805e-05,
+      "loss": 0.9254,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954955201291193e-05,
+      "loss": 0.9231,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548346278534374e-05,
+      "loss": 0.8591,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954713896987418e-05,
+      "loss": 0.8869,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545930087130404e-05,
+      "loss": 0.9983,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544719630502352e-05,
+      "loss": 0.9183,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543507600189606e-05,
+      "loss": 0.9228,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542293996391993e-05,
+      "loss": 0.8128,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541078819309615e-05,
+      "loss": 0.8592,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539862069142817e-05,
+      "loss": 0.3464,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953864374609221e-05,
+      "loss": 0.8972,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953742385035867e-05,
+      "loss": 0.9146,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536202382143322e-05,
+      "loss": 0.9085,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534979341647562e-05,
+      "loss": 0.8842,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953375472907303e-05,
+      "loss": 1.0317,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532528544621643e-05,
+      "loss": 0.9583,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953130078849556e-05,
+      "loss": 0.9497,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530071460897212e-05,
+      "loss": 0.8158,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528840562029282e-05,
+      "loss": 0.809,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527608092094715e-05,
+      "loss": 0.936,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526374051296714e-05,
+      "loss": 0.879,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525138439838745e-05,
+      "loss": 0.9464,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523901257924525e-05,
+      "loss": 1.0425,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952266250575803e-05,
+      "loss": 0.8846,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952142218354351e-05,
+      "loss": 0.9457,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952018029148546e-05,
+      "loss": 0.9855,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951893682978863e-05,
+      "loss": 0.8703,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517691798658042e-05,
+      "loss": 0.9758,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516445198298966e-05,
+      "loss": 0.9154,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951519702891694e-05,
+      "loss": 0.7719,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513947290717752e-05,
+      "loss": 0.9052,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512695983907454e-05,
+      "loss": 0.3112,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511443108692357e-05,
+      "loss": 0.9113,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510188665279025e-05,
+      "loss": 0.8801,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508932653874283e-05,
+      "loss": 0.9546,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507675074685224e-05,
+      "loss": 0.8506,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950641592791918e-05,
+      "loss": 0.8547,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505155213783765e-05,
+      "loss": 0.9193,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503892932486828e-05,
+      "loss": 0.9408,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502629084236493e-05,
+      "loss": 0.9347,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501363669241136e-05,
+      "loss": 0.8569,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500096687709393e-05,
+      "loss": 0.9292,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498828139850158e-05,
+      "loss": 0.2922,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497558025872577e-05,
+      "loss": 0.9046,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496286345986067e-05,
+      "loss": 0.8258,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495013100400294e-05,
+      "loss": 0.9,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949373828932518e-05,
+      "loss": 0.944,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949246191297092e-05,
+      "loss": 0.9195,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491183971547943e-05,
+      "loss": 0.875,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489904465266957e-05,
+      "loss": 0.978,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948862339433892e-05,
+      "loss": 0.8926,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487340758975045e-05,
+      "loss": 0.9562,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486056559386816e-05,
+      "loss": 0.8732,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948477079578595e-05,
+      "loss": 0.9502,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948348346838445e-05,
+      "loss": 0.8941,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948219457739456e-05,
+      "loss": 0.9416,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480904123028784e-05,
+      "loss": 1.012,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479612105499887e-05,
+      "loss": 0.8815,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947831852502089e-05,
+      "loss": 0.9108,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947702338180507e-05,
+      "loss": 0.8876,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947572667606597e-05,
+      "loss": 0.8237,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474428408017378e-05,
+      "loss": 0.983,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473128577873346e-05,
+      "loss": 0.9645,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471827185848186e-05,
+      "loss": 0.8612,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470524232156466e-05,
+      "loss": 0.86,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469219717013002e-05,
+      "loss": 0.8609,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467913640632888e-05,
+      "loss": 0.8782,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466606003231455e-05,
+      "loss": 0.9209,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94652968050243e-05,
+      "loss": 0.9426,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463986046227284e-05,
+      "loss": 1.0122,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462673727056512e-05,
+      "loss": 0.8999,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946135984772835e-05,
+      "loss": 0.9087,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946004440845943e-05,
+      "loss": 0.8996,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945872740946663e-05,
+      "loss": 0.9182,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457408850967093e-05,
+      "loss": 0.9187,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456088733178216e-05,
+      "loss": 0.9221,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454767056317654e-05,
+      "loss": 0.9155,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453443820603314e-05,
+      "loss": 0.9135,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945211902625337e-05,
+      "loss": 0.8985,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450792673486245e-05,
+      "loss": 0.8542,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449464762520617e-05,
+      "loss": 0.8615,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448135293575433e-05,
+      "loss": 0.9658,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446804266869886e-05,
+      "loss": 0.9383,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445471682623425e-05,
+      "loss": 0.9691,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444137541055767e-05,
+      "loss": 0.9141,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944280184238687e-05,
+      "loss": 0.8768,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441464586836964e-05,
+      "loss": 0.912,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440125774626523e-05,
+      "loss": 0.9165,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943878540597629e-05,
+      "loss": 0.942,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437443481107256e-05,
+      "loss": 0.9582,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436100000240668e-05,
+      "loss": 0.8661,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434754963598033e-05,
+      "loss": 0.9932,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433408371401116e-05,
+      "loss": 0.9753,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432060223871937e-05,
+      "loss": 0.8763,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430710521232767e-05,
+      "loss": 0.926,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429359263706143e-05,
+      "loss": 0.9239,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942800645151485e-05,
+      "loss": 0.8952,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426652084881934e-05,
+      "loss": 0.9006,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425296164030696e-05,
+      "loss": 0.9864,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423938689184694e-05,
+      "loss": 0.8977,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422579660567743e-05,
+      "loss": 0.8505,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942121907840391e-05,
+      "loss": 0.9223,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419856942917517e-05,
+      "loss": 0.9093,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418493254333157e-05,
+      "loss": 0.3057,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417128012875657e-05,
+      "loss": 0.8954,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415761218770116e-05,
+      "loss": 0.9312,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414392872241884e-05,
+      "loss": 0.9246,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413022973516568e-05,
+      "loss": 0.9284,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941165152282003e-05,
+      "loss": 0.9142,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410278520378383e-05,
+      "loss": 0.9045,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408903966418004e-05,
+      "loss": 0.9205,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407527861165523e-05,
+      "loss": 0.9648,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406150204847827e-05,
+      "loss": 0.8735,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940477099769205e-05,
+      "loss": 0.907,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94033902399256e-05,
+      "loss": 0.931,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402007931776118e-05,
+      "loss": 0.9745,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400624073471513e-05,
+      "loss": 0.9248,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399238665239957e-05,
+      "loss": 0.9696,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397851707309864e-05,
+      "loss": 0.9006,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396463199909908e-05,
+      "loss": 0.8761,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939507314326902e-05,
+      "loss": 0.9031,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393681537616384e-05,
+      "loss": 0.8457,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939228838318144e-05,
+      "loss": 0.9728,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939089368019389e-05,
+      "loss": 0.9842,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938949742888368e-05,
+      "loss": 0.8716,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388099629481017e-05,
+      "loss": 0.9354,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386700282216364e-05,
+      "loss": 0.9763,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938529938732044e-05,
+      "loss": 0.9212,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383896945024215e-05,
+      "loss": 0.9162,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938249295555892e-05,
+      "loss": 0.9028,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381087419156034e-05,
+      "loss": 0.8663,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379680336047297e-05,
+      "loss": 0.9677,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93782717064647e-05,
+      "loss": 0.9494,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937686153064049e-05,
+      "loss": 1.0308,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937544980880717e-05,
+      "loss": 0.8686,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374036541197498e-05,
+      "loss": 0.9192,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372621728044492e-05,
+      "loss": 0.8913,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937120536958141e-05,
+      "loss": 0.3002,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369787466041776e-05,
+      "loss": 0.9217,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368368017659368e-05,
+      "loss": 0.9644,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366947024668223e-05,
+      "loss": 0.9397,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365524487302617e-05,
+      "loss": 0.8708,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93641004057971e-05,
+      "loss": 0.9336,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362674780386463e-05,
+      "loss": 0.9217,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361247611305754e-05,
+      "loss": 0.9162,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935981889879028e-05,
+      "loss": 0.9036,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358388643075597e-05,
+      "loss": 0.9102,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935695684439752e-05,
+      "loss": 0.8164,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935552350299212e-05,
+      "loss": 0.9119,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354088619095716e-05,
+      "loss": 0.9182,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352652192944883e-05,
+      "loss": 0.9387,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351214224776452e-05,
+      "loss": 0.854,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934977471482751e-05,
+      "loss": 0.9125,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348333663335393e-05,
+      "loss": 0.9009,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93468910705377e-05,
+      "loss": 1.0821,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934544693667227e-05,
+      "loss": 0.9346,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344001261977213e-05,
+      "loss": 0.9298,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934255404669088e-05,
+      "loss": 0.9907,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341105291051878e-05,
+      "loss": 0.9366,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339654995299073e-05,
+      "loss": 0.9493,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338203159671584e-05,
+      "loss": 0.8859,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933674978440878e-05,
+      "loss": 0.9721,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933529486975029e-05,
+      "loss": 0.8608,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333838415935985e-05,
+      "loss": 0.9042,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933238042320601e-05,
+      "loss": 0.9011,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933092089180074e-05,
+      "loss": 0.8733,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329459821960816e-05,
+      "loss": 0.9051,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327997213927136e-05,
+      "loss": 0.8461,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326533067940847e-05,
+      "loss": 0.8214,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325067384243353e-05,
+      "loss": 0.9334,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323600163076298e-05,
+      "loss": 0.9089,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322131404681602e-05,
+      "loss": 0.8825,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932066110930142e-05,
+      "loss": 0.8931,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319189277178165e-05,
+      "loss": 0.9441,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931771590855451e-05,
+      "loss": 0.9406,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931624100367337e-05,
+      "loss": 0.9361,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931476456277793e-05,
+      "loss": 0.9132,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931328658611161e-05,
+      "loss": 0.3172,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931180707391809e-05,
+      "loss": 0.9462,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310326026441314e-05,
+      "loss": 0.9182,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930884344392546e-05,
+      "loss": 0.9547,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307359326614975e-05,
+      "loss": 0.8617,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305873674754544e-05,
+      "loss": 0.8242,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304386488589125e-05,
+      "loss": 0.8882,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302897768363908e-05,
+      "loss": 0.8927,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301407514324352e-05,
+      "loss": 0.9924,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929991572671616e-05,
+      "loss": 0.9475,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929842240578529e-05,
+      "loss": 0.8645,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929692755177796e-05,
+      "loss": 0.9383,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295431164940618e-05,
+      "loss": 0.337,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293933245519993e-05,
+      "loss": 0.9205,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292433793763054e-05,
+      "loss": 0.8592,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929093280991702e-05,
+      "loss": 0.8438,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289430294229367e-05,
+      "loss": 1.0105,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928792624694782e-05,
+      "loss": 0.9026,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286420668320356e-05,
+      "loss": 0.9224,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284913558595214e-05,
+      "loss": 0.943,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283404918020878e-05,
+      "loss": 0.9906,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928189474684608e-05,
+      "loss": 0.8441,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280383045319813e-05,
+      "loss": 0.8553,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278869813691316e-05,
+      "loss": 0.9827,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277355052210084e-05,
+      "loss": 0.8078,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275838761125866e-05,
+      "loss": 0.848,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927432094068866e-05,
+      "loss": 0.9143,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272801591148708e-05,
+      "loss": 0.9756,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271280712756522e-05,
+      "loss": 1.086,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269758305762853e-05,
+      "loss": 0.8933,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926823437041871e-05,
+      "loss": 0.8054,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266708906975346e-05,
+      "loss": 0.8496,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926518191568428e-05,
+      "loss": 0.8702,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263653396797268e-05,
+      "loss": 0.9041,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926212335056633e-05,
+      "loss": 0.8655,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260591777243726e-05,
+      "loss": 0.896,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259058677081975e-05,
+      "loss": 0.8901,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257524050333854e-05,
+      "loss": 0.9162,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255987897252373e-05,
+      "loss": 0.926,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254450218090814e-05,
+      "loss": 0.8794,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92529110131027e-05,
+      "loss": 0.9237,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251370282541803e-05,
+      "loss": 0.9561,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249828026662154e-05,
+      "loss": 0.8768,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248284245718034e-05,
+      "loss": 0.9907,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924673893996397e-05,
+      "loss": 0.8553,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924519210965475e-05,
+      "loss": 0.8374,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92436437550454e-05,
+      "loss": 1.0235,
+      "step": 1204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242093876391208e-05,
+      "loss": 0.9086,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240542473947712e-05,
+      "loss": 0.948,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92389895479707e-05,
+      "loss": 0.9219,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92374350987162e-05,
+      "loss": 0.8812,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235879126440517e-05,
+      "loss": 0.9084,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234321631400182e-05,
+      "loss": 0.9586,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232762613851993e-05,
+      "loss": 0.8706,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231202074052986e-05,
+      "loss": 0.9051,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922964001226046e-05,
+      "loss": 0.8993,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228076428731954e-05,
+      "loss": 0.9076,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226511323725272e-05,
+      "loss": 0.8951,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224944697498452e-05,
+      "loss": 0.9377,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223376550309802e-05,
+      "loss": 0.9583,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922180688241786e-05,
+      "loss": 0.8305,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220235694081428e-05,
+      "loss": 0.912,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218662985559554e-05,
+      "loss": 0.9522,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217088757111543e-05,
+      "loss": 0.9183,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921551300899694e-05,
+      "loss": 0.8637,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213935741475557e-05,
+      "loss": 0.8477,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212356954807433e-05,
+      "loss": 0.8914,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210776649252875e-05,
+      "loss": 0.9558,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920919482507243e-05,
+      "loss": 0.9504,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207611482526917e-05,
+      "loss": 0.8495,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206026621877377e-05,
+      "loss": 0.8407,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204440243385117e-05,
+      "loss": 0.828,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202852347311688e-05,
+      "loss": 0.8839,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201262933918895e-05,
+      "loss": 0.3262,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199672003468795e-05,
+      "loss": 0.8617,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198079556223695e-05,
+      "loss": 0.9293,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196485592446143e-05,
+      "loss": 0.9861,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919489011239895e-05,
+      "loss": 0.8967,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919329311634516e-05,
+      "loss": 0.8369,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191694604548094e-05,
+      "loss": 0.8183,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919009457727129e-05,
+      "loss": 0.9016,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918849303477856e-05,
+      "loss": 0.9574,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918688997733396e-05,
+      "loss": 0.8829,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918528540520179e-05,
+      "loss": 0.9038,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183679318646612e-05,
+      "loss": 0.9541,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182071717933218e-05,
+      "loss": 0.9835,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180462603326667e-05,
+      "loss": 0.9339,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178851975092257e-05,
+      "loss": 0.906,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177239833495545e-05,
+      "loss": 0.8594,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917562617880233e-05,
+      "loss": 0.9554,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174011011278667e-05,
+      "loss": 0.9973,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172394331190855e-05,
+      "loss": 0.8728,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917077613880544e-05,
+      "loss": 0.8928,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169156434389228e-05,
+      "loss": 0.9214,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916753521820926e-05,
+      "loss": 0.7688,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165912490532838e-05,
+      "loss": 0.8979,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916428825162751e-05,
+      "loss": 0.8522,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916266250176107e-05,
+      "loss": 0.9196,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161035241201564e-05,
+      "loss": 0.9217,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159406470217288e-05,
+      "loss": 0.8904,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157776189076784e-05,
+      "loss": 0.8949,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156144398048847e-05,
+      "loss": 0.8667,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154511097402512e-05,
+      "loss": 0.8616,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152876287407073e-05,
+      "loss": 0.8458,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151239968332074e-05,
+      "loss": 0.9629,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149602140447298e-05,
+      "loss": 0.9111,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147962804022776e-05,
+      "loss": 0.9453,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146321959328805e-05,
+      "loss": 0.8637,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144679606635912e-05,
+      "loss": 0.9522,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143035746214883e-05,
+      "loss": 0.9649,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141390378336745e-05,
+      "loss": 0.8881,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913974350327278e-05,
+      "loss": 0.9413,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913809512129452e-05,
+      "loss": 0.8647,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913644523267374e-05,
+      "loss": 0.831,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913479383768246e-05,
+      "loss": 0.8499,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133140936592957e-05,
+      "loss": 0.8031,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131486529677755e-05,
+      "loss": 0.9062,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129830617209625e-05,
+      "loss": 0.8969,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128173199461582e-05,
+      "loss": 0.9215,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126514276706892e-05,
+      "loss": 0.298,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912485384921907e-05,
+      "loss": 0.8316,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123191917271883e-05,
+      "loss": 0.9133,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912152848113934e-05,
+      "loss": 0.8669,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119863541095697e-05,
+      "loss": 0.804,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911819709741546e-05,
+      "loss": 0.8911,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911652915037339e-05,
+      "loss": 0.8935,
+      "step": 1283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911485970024449e-05,
+      "loss": 0.9462,
+      "step": 1284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113188747304e-05,
+      "loss": 0.9679,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111516291827423e-05,
+      "loss": 0.8614,
+      "step": 1286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109842334090512e-05,
+      "loss": 0.8963,
+      "step": 1287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108166874369253e-05,
+      "loss": 0.9211,
+      "step": 1288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910648991293989e-05,
+      "loss": 0.9942,
+      "step": 1289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910481145007891e-05,
+      "loss": 0.9006,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910313148606305e-05,
+      "loss": 0.9188,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101450021169298e-05,
+      "loss": 0.876,
+      "step": 1292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099767055674882e-05,
+      "loss": 0.3232,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098082589857278e-05,
+      "loss": 0.8513,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096396623994215e-05,
+      "loss": 0.8153,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909470915836367e-05,
+      "loss": 0.8275,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093020193243855e-05,
+      "loss": 0.8842,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091329728913245e-05,
+      "loss": 0.8004,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089637765650556e-05,
+      "loss": 0.9188,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087944303734744e-05,
+      "loss": 0.8934,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086249343445024e-05,
+      "loss": 0.9819,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084552885060846e-05,
+      "loss": 0.912,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908285492886192e-05,
+      "loss": 0.8817,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081155475128197e-05,
+      "loss": 0.9377,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079454524139868e-05,
+      "loss": 0.9739,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907775207617738e-05,
+      "loss": 0.819,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076048131521428e-05,
+      "loss": 0.8661,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907434269045294e-05,
+      "loss": 0.9012,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072635753253112e-05,
+      "loss": 0.8739,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070927320203365e-05,
+      "loss": 0.8897,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069217391585385e-05,
+      "loss": 0.838,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067505967681092e-05,
+      "loss": 0.9427,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065793048772656e-05,
+      "loss": 0.8811,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064078635142494e-05,
+      "loss": 0.346,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062362727073273e-05,
+      "loss": 0.8364,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060645324847904e-05,
+      "loss": 0.9948,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905892642874954e-05,
+      "loss": 0.9661,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057206039061586e-05,
+      "loss": 0.9054,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055484156067688e-05,
+      "loss": 0.8805,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053760780051744e-05,
+      "loss": 0.9165,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052035911297895e-05,
+      "loss": 0.9693,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905030955009053e-05,
+      "loss": 0.788,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048581696714276e-05,
+      "loss": 0.8319,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046852351454022e-05,
+      "loss": 0.9174,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904512151459489e-05,
+      "loss": 0.9089,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043389186422246e-05,
+      "loss": 0.9,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041655367221717e-05,
+      "loss": 0.8536,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039920057279158e-05,
+      "loss": 0.8411,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038183256880685e-05,
+      "loss": 0.9162,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036444966312652e-05,
+      "loss": 0.8944,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034705185861647e-05,
+      "loss": 0.8646,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032963915814534e-05,
+      "loss": 0.8918,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031221156458396e-05,
+      "loss": 0.8884,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902947690808057e-05,
+      "loss": 0.8041,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902773117096864e-05,
+      "loss": 0.8595,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025983945410434e-05,
+      "loss": 0.8453,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024235231694024e-05,
+      "loss": 0.9908,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022485030107734e-05,
+      "loss": 0.8309,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020733340940122e-05,
+      "loss": 0.8526,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018980164480003e-05,
+      "loss": 0.888,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017225501016424e-05,
+      "loss": 0.8911,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90154693508387e-05,
+      "loss": 0.8783,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013711714236356e-05,
+      "loss": 0.9267,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90119525914992e-05,
+      "loss": 0.8678,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010191982917258e-05,
+      "loss": 0.9081,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008429888780808e-05,
+      "loss": 0.8357,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006666309380382e-05,
+      "loss": 0.9465,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004901245006748e-05,
+      "loss": 0.8185,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003134695950917e-05,
+      "loss": 0.8494,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001366662504154e-05,
+      "loss": 0.8632,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899959714495796e-05,
+      "loss": 0.951,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997826143604086e-05,
+      "loss": 0.8934,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996053658734522e-05,
+      "loss": 0.836,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899427969064151e-05,
+      "loss": 0.9577,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992504239617532e-05,
+      "loss": 0.8078,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990727305955312e-05,
+      "loss": 0.8939,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988948889947826e-05,
+      "loss": 0.8698,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987168991888293e-05,
+      "loss": 0.9208,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985387612070166e-05,
+      "loss": 0.8897,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898360475078715e-05,
+      "loss": 0.7866,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89818204083332e-05,
+      "loss": 0.954,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898003458500251e-05,
+      "loss": 0.8411,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897824728108951e-05,
+      "loss": 0.9476,
+      "step": 1363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976458496888885e-05,
+      "loss": 0.8657,
+      "step": 1364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974668232695562e-05,
+      "loss": 0.8797,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897287648880471e-05,
+      "loss": 0.9259,
+      "step": 1366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897108326551174e-05,
+      "loss": 0.894,
+      "step": 1367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969288563112313e-05,
+      "loss": 0.8983,
+      "step": 1368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967492381902333e-05,
+      "loss": 0.8011,
+      "step": 1369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896569472217794e-05,
+      "loss": 0.8896,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896389558423552e-05,
+      "loss": 0.8775,
+      "step": 1371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896209496837171e-05,
+      "loss": 0.8126,
+      "step": 1372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896029287488339e-05,
+      "loss": 0.8363,
+      "step": 1373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895848930406768e-05,
+      "loss": 0.9155,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956684256221934e-05,
+      "loss": 0.9071,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954877731643764e-05,
+      "loss": 0.844,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953069730631025e-05,
+      "loss": 1.0246,
+      "step": 1377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951260253481806e-05,
+      "loss": 0.8975,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949449300494444e-05,
+      "loss": 0.8328,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947636871967527e-05,
+      "loss": 0.7947,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894582296819987e-05,
+      "loss": 0.908,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944007589490546e-05,
+      "loss": 0.8662,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942190736138857e-05,
+      "loss": 0.2949,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894037240844437e-05,
+      "loss": 0.9578,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893855260670687e-05,
+      "loss": 0.9988,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936731331226402e-05,
+      "loss": 0.8873,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893490858230325e-05,
+      "loss": 0.8784,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933084360237935e-05,
+      "loss": 0.8384,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931258665331225e-05,
+      "loss": 0.9295,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892943149788414e-05,
+      "loss": 0.9243,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927602858197923e-05,
+      "loss": 0.9353,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892577274657408e-05,
+      "loss": 0.9083,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892394116331434e-05,
+      "loss": 0.9396,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922108108720695e-05,
+      "loss": 0.9679,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920273583095365e-05,
+      "loss": 0.9179,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891843758674082e-05,
+      "loss": 0.8945,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916600119959765e-05,
+      "loss": 0.9517,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914761183055157e-05,
+      "loss": 0.8398,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912920776330187e-05,
+      "loss": 0.918,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911078900088295e-05,
+      "loss": 0.9323,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909235554633156e-05,
+      "loss": 0.9035,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907390740268697e-05,
+      "loss": 0.8782,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905544457299076e-05,
+      "loss": 0.835,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89036967060287e-05,
+      "loss": 0.3044,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901847486762223e-05,
+      "loss": 0.8281,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899996799804526e-05,
+      "loss": 0.9246,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898144645460744e-05,
+      "loss": 0.8614,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896291024036257e-05,
+      "loss": 0.9086,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889443593583667e-05,
+      "loss": 0.8953,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892579381167847e-05,
+      "loss": 0.9412,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890721360335886e-05,
+      "loss": 0.9359,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888861873647127e-05,
+      "loss": 0.9208,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887000921408156e-05,
+      "loss": 0.8774,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885138503925793e-05,
+      "loss": 0.8341,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883274621507106e-05,
+      "loss": 0.9177,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881409274459403e-05,
+      "loss": 0.8073,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887954246309023e-05,
+      "loss": 0.965,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877674187707383e-05,
+      "loss": 0.8172,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887580444861889e-05,
+      "loss": 0.8652,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873933246133022e-05,
+      "loss": 0.9439,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872060580558295e-05,
+      "loss": 0.8729,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870186452203466e-05,
+      "loss": 0.934,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868310861377532e-05,
+      "loss": 0.3378,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886643380838973e-05,
+      "loss": 0.9798,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864555293549536e-05,
+      "loss": 0.9665,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862675317166676e-05,
+      "loss": 0.8704,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860793879551103e-05,
+      "loss": 0.9199,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858910981013025e-05,
+      "loss": 0.8606,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857026621862883e-05,
+      "loss": 0.8438,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885514080241136e-05,
+      "loss": 0.7969,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853253522969383e-05,
+      "loss": 0.9048,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851364783848117e-05,
+      "loss": 0.8601,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884947458535896e-05,
+      "loss": 0.8405,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847582927813568e-05,
+      "loss": 0.8823,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884568981152382e-05,
+      "loss": 0.3171,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884379523680185e-05,
+      "loss": 0.8168,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841899203960022e-05,
+      "loss": 0.9251,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840001713310943e-05,
+      "loss": 0.9574,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838102765167464e-05,
+      "loss": 0.8614,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883620235984268e-05,
+      "loss": 0.8926,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834300497649905e-05,
+      "loss": 0.9005,
+      "step": 1441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883239717890272e-05,
+      "loss": 0.8612,
+      "step": 1442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830492403914932e-05,
+      "loss": 0.8559,
+      "step": 1443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882858617300059e-05,
+      "loss": 0.932,
+      "step": 1444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826678486473987e-05,
+      "loss": 0.947,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824769344649644e-05,
+      "loss": 0.9361,
+      "step": 1446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822858747842337e-05,
+      "loss": 0.8603,
+      "step": 1447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820946696367077e-05,
+      "loss": 0.808,
+      "step": 1448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881903319053911e-05,
+      "loss": 0.8406,
+      "step": 1449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881711823067392e-05,
+      "loss": 0.8534,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881520181708725e-05,
+      "loss": 0.9384,
+      "step": 1451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813283950095056e-05,
+      "loss": 0.9328,
+      "step": 1452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811364630013554e-05,
+      "loss": 0.8444,
+      "step": 1453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809443857159183e-05,
+      "loss": 0.3026,
+      "step": 1454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807521631848635e-05,
+      "loss": 0.9128,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880559795439884e-05,
+      "loss": 0.849,
+      "step": 1456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880367282512696e-05,
+      "loss": 0.8589,
+      "step": 1457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88017462443504e-05,
+      "loss": 0.798,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799818212386808e-05,
+      "loss": 0.9017,
+      "step": 1459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797888729554064e-05,
+      "loss": 0.9671,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795957796170296e-05,
+      "loss": 0.9258,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794025412553863e-05,
+      "loss": 0.8972,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792091579023365e-05,
+      "loss": 0.9316,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879015629589765e-05,
+      "loss": 0.9771,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878821956349579e-05,
+      "loss": 0.7723,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878628138213711e-05,
+      "loss": 0.8105,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878434175214116e-05,
+      "loss": 0.8789,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782400673827742e-05,
+      "loss": 0.9083,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780458147516886e-05,
+      "loss": 0.9821,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778514173528873e-05,
+      "loss": 0.8969,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877656875218421e-05,
+      "loss": 0.886,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774621883803653e-05,
+      "loss": 0.3018,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772673568708184e-05,
+      "loss": 0.899,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770723807219037e-05,
+      "loss": 0.996,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876877259965768e-05,
+      "loss": 0.7686,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766819946345812e-05,
+      "loss": 0.8952,
+      "step": 1476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764865847605384e-05,
+      "loss": 0.9415,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876291030375857e-05,
+      "loss": 0.8373,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760953315127793e-05,
+      "loss": 0.9711,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758994882035718e-05,
+      "loss": 0.9029,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757035004805234e-05,
+      "loss": 1.0134,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755073683759475e-05,
+      "loss": 0.9614,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875311091922182e-05,
+      "loss": 0.8904,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875114671151587e-05,
+      "loss": 0.9189,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749181060965482e-05,
+      "loss": 0.9589,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874721396789474e-05,
+      "loss": 0.8641,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874524543262797e-05,
+      "loss": 0.8931,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874327545548973e-05,
+      "loss": 0.8504,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741304036804822e-05,
+      "loss": 0.8655,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739331176898284e-05,
+      "loss": 0.8463,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737356876095387e-05,
+      "loss": 0.8325,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735381134721652e-05,
+      "loss": 0.9008,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733403953102818e-05,
+      "loss": 0.919,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873142533156488e-05,
+      "loss": 0.8557,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872944527043407e-05,
+      "loss": 0.821,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727463770036834e-05,
+      "loss": 0.86,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872548083069988e-05,
+      "loss": 0.8887,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723496452750146e-05,
+      "loss": 0.8834,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721510636514805e-05,
+      "loss": 0.8431,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871952338232127e-05,
+      "loss": 0.9394,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717534690497184e-05,
+      "loss": 0.8465,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715544561370436e-05,
+      "loss": 0.8454,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871355299526915e-05,
+      "loss": 0.9173,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711559992521684e-05,
+      "loss": 0.8724,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709565553456632e-05,
+      "loss": 0.8825,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870756967840283e-05,
+      "loss": 0.8711,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705572367689348e-05,
+      "loss": 0.8483,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703573621645484e-05,
+      "loss": 0.8549,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870157344060079e-05,
+      "loss": 0.8959,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869957182488505e-05,
+      "loss": 0.8503,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697568774828265e-05,
+      "loss": 0.9244,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86955642907607e-05,
+      "loss": 0.8729,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869355837301284e-05,
+      "loss": 0.8351,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869155102191541e-05,
+      "loss": 0.9459,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689542237799373e-05,
+      "loss": 0.986,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687532020995925e-05,
+      "loss": 1.0134,
+      "step": 1516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685520371836504e-05,
+      "loss": 0.8687,
+      "step": 1517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868350729065278e-05,
+      "loss": 0.9386,
+      "step": 1518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681492777776656e-05,
+      "loss": 0.8617,
+      "step": 1519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679476833540275e-05,
+      "loss": 0.8192,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867745945827602e-05,
+      "loss": 0.8365,
+      "step": 1521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675440652316502e-05,
+      "loss": 0.8326,
+      "step": 1522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673420415994568e-05,
+      "loss": 0.9552,
+      "step": 1523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671398749643315e-05,
+      "loss": 0.9532,
+      "step": 1524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669375653596056e-05,
+      "loss": 0.856,
+      "step": 1525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667351128186347e-05,
+      "loss": 0.8539,
+      "step": 1526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866532517374799e-05,
+      "loss": 0.9189,
+      "step": 1527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866329779061501e-05,
+      "loss": 0.8937,
+      "step": 1528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866126897912167e-05,
+      "loss": 0.8981,
+      "step": 1529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865923873960247e-05,
+      "loss": 0.8569,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865720707239215e-05,
+      "loss": 0.8748,
+      "step": 1531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655173977825676e-05,
+      "loss": 0.9523,
+      "step": 1532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653139456238257e-05,
+      "loss": 0.8699,
+      "step": 1533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651103507965333e-05,
+      "loss": 0.8899,
+      "step": 1534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864906613334258e-05,
+      "loss": 0.9057,
+      "step": 1535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647027332705912e-05,
+      "loss": 0.8347,
+      "step": 1536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644987106391478e-05,
+      "loss": 0.898,
+      "step": 1537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642945454735656e-05,
+      "loss": 1.0053,
+      "step": 1538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640902378075062e-05,
+      "loss": 0.9749,
+      "step": 1539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638857876746556e-05,
+      "loss": 0.8253,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636811951087215e-05,
+      "loss": 0.8314,
+      "step": 1541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634764601434367e-05,
+      "loss": 0.8366,
+      "step": 1542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632715828125566e-05,
+      "loss": 0.8699,
+      "step": 1543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630665631498606e-05,
+      "loss": 0.8721,
+      "step": 1544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862861401189151e-05,
+      "loss": 0.9199,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862656096964254e-05,
+      "loss": 0.9144,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624506505090192e-05,
+      "loss": 0.8671,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622450618573193e-05,
+      "loss": 0.8838,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862039331043051e-05,
+      "loss": 0.9545,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861833458100134e-05,
+      "loss": 0.8634,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616274430625115e-05,
+      "loss": 0.9124,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614212859641502e-05,
+      "loss": 0.85,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612149868390407e-05,
+      "loss": 0.9005,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610085457211958e-05,
+      "loss": 0.9439,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860801962644653e-05,
+      "loss": 0.8582,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605952376434726e-05,
+      "loss": 0.8334,
+      "step": 1556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860388370751738e-05,
+      "loss": 0.8596,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860181362003557e-05,
+      "loss": 0.9218,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599742114330595e-05,
+      "loss": 0.8305,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597669190743998e-05,
+      "loss": 0.9595,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595594849617552e-05,
+      "loss": 0.8275,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593519091293265e-05,
+      "loss": 0.9411,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591441916113368e-05,
+      "loss": 0.885,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858936332442035e-05,
+      "loss": 0.8489,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587283316556905e-05,
+      "loss": 0.8811,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585201892865984e-05,
+      "loss": 0.8905,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583119053690756e-05,
+      "loss": 0.8986,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581034799374632e-05,
+      "loss": 0.9254,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857894913026125e-05,
+      "loss": 0.8523,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576862046694488e-05,
+      "loss": 0.9597,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857477354901845e-05,
+      "loss": 0.9295,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857268363757748e-05,
+      "loss": 0.8361,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570592312716147e-05,
+      "loss": 0.9635,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856849957477927e-05,
+      "loss": 0.9119,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566405424111873e-05,
+      "loss": 0.8848,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564309861059236e-05,
+      "loss": 0.9379,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562212885966874e-05,
+      "loss": 0.7578,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560114499180508e-05,
+      "loss": 0.8825,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558014701046123e-05,
+      "loss": 0.9206,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555913491909914e-05,
+      "loss": 0.8464,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553810872118325e-05,
+      "loss": 0.8642,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855170684201802e-05,
+      "loss": 0.8771,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549601401955906e-05,
+      "loss": 0.8727,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854749455227911e-05,
+      "loss": 0.8605,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854538629333501e-05,
+      "loss": 0.9227,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854327662547119e-05,
+      "loss": 0.9057,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854116554903549e-05,
+      "loss": 0.8833,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539053064375973e-05,
+      "loss": 0.8692,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536939171840934e-05,
+      "loss": 0.9598,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534823871778903e-05,
+      "loss": 0.8738,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532707164538636e-05,
+      "loss": 0.8943,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853058905046913e-05,
+      "loss": 0.9378,
+      "step": 1592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528469529919603e-05,
+      "loss": 0.8076,
+      "step": 1593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852634860323951e-05,
+      "loss": 0.9718,
+      "step": 1594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524226270778546e-05,
+      "loss": 0.8198,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522102532886627e-05,
+      "loss": 0.8775,
+      "step": 1596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85199773899139e-05,
+      "loss": 0.952,
+      "step": 1597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517850842210752e-05,
+      "loss": 0.8933,
+      "step": 1598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515722890127795e-05,
+      "loss": 0.8908,
+      "step": 1599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513593534015878e-05,
+      "loss": 0.8694,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511462774226078e-05,
+      "loss": 0.8736,
+      "step": 1601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509330611109702e-05,
+      "loss": 0.9631,
+      "step": 1602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507197045018286e-05,
+      "loss": 1.0248,
+      "step": 1603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505062076303614e-05,
+      "loss": 0.803,
+      "step": 1604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502925705317675e-05,
+      "loss": 0.3172,
+      "step": 1605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500787932412708e-05,
+      "loss": 0.8291,
+      "step": 1606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498648757941186e-05,
+      "loss": 0.7907,
+      "step": 1607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849650818225579e-05,
+      "loss": 0.8904,
+      "step": 1608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849436620570946e-05,
+      "loss": 0.8853,
+      "step": 1609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492222828655347e-05,
+      "loss": 0.9097,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490078051446844e-05,
+      "loss": 0.8688,
+      "step": 1611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848793187443757e-05,
+      "loss": 0.9282,
+      "step": 1612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485784297981374e-05,
+      "loss": 0.851,
+      "step": 1613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848363532243234e-05,
+      "loss": 0.906,
+      "step": 1614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481484948144777e-05,
+      "loss": 0.82,
+      "step": 1615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847933317547323e-05,
+      "loss": 0.8589,
+      "step": 1616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477180004772473e-05,
+      "loss": 0.8984,
+      "step": 1617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475025436397507e-05,
+      "loss": 0.3115,
+      "step": 1618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847286947070357e-05,
+      "loss": 0.8921,
+      "step": 1619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470712108046122e-05,
+      "loss": 0.8957,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468553348780862e-05,
+      "loss": 0.9103,
+      "step": 1621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466393193263714e-05,
+      "loss": 0.8694,
+      "step": 1622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846423164185083e-05,
+      "loss": 0.946,
+      "step": 1623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462068694898603e-05,
+      "loss": 0.8728,
+      "step": 1624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459904352763642e-05,
+      "loss": 0.8819,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457738615802797e-05,
+      "loss": 0.9042,
+      "step": 1626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455571484373142e-05,
+      "loss": 0.3333,
+      "step": 1627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845340295883198e-05,
+      "loss": 0.9018,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451233039536846e-05,
+      "loss": 0.91,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844906172684551e-05,
+      "loss": 0.3127,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446889021115967e-05,
+      "loss": 0.8598,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844471492270644e-05,
+      "loss": 0.8655,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844253943197538e-05,
+      "loss": 0.9155,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440362549281476e-05,
+      "loss": 0.7843,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438184274983638e-05,
+      "loss": 0.875,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436004609441007e-05,
+      "loss": 0.9236,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843382355301296e-05,
+      "loss": 0.9336,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84316411060591e-05,
+      "loss": 0.3299,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429457268939252e-05,
+      "loss": 0.912,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427272042013477e-05,
+      "loss": 0.9231,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425085425642062e-05,
+      "loss": 0.978,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842289742018553e-05,
+      "loss": 0.8449,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420708026004626e-05,
+      "loss": 0.9572,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841851724346033e-05,
+      "loss": 0.9638,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841632507291384e-05,
+      "loss": 0.287,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414131514726596e-05,
+      "loss": 0.8683,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411936569260257e-05,
+      "loss": 1.0142,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409740236876715e-05,
+      "loss": 0.3469,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840754251793809e-05,
+      "loss": 0.9554,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840534341280673e-05,
+      "loss": 0.8817,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840314292184522e-05,
+      "loss": 0.852,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400941045416352e-05,
+      "loss": 0.8559,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839873778388317e-05,
+      "loss": 0.9404,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396533137608935e-05,
+      "loss": 0.896,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839432710695714e-05,
+      "loss": 0.8827,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392119692291495e-05,
+      "loss": 0.9227,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389910893975953e-05,
+      "loss": 0.8425,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387700712374692e-05,
+      "loss": 0.8924,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385489147852117e-05,
+      "loss": 0.8405,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383276200772848e-05,
+      "loss": 0.8351,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381061871501756e-05,
+      "loss": 0.9598,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378846160403924e-05,
+      "loss": 0.8219,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376629067844668e-05,
+      "loss": 1.0074,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374410594189525e-05,
+      "loss": 0.9386,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372190739804278e-05,
+      "loss": 0.9168,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369969505054915e-05,
+      "loss": 0.914,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367746890307667e-05,
+      "loss": 1.0125,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365522895928984e-05,
+      "loss": 0.9271,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836329752228555e-05,
+      "loss": 0.9025,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361070769744268e-05,
+      "loss": 0.8698,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358842638672278e-05,
+      "loss": 0.8452,
+      "step": 1671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356613129436943e-05,
+      "loss": 0.898,
+      "step": 1672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354382242405853e-05,
+      "loss": 0.8992,
+      "step": 1673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352149977946826e-05,
+      "loss": 0.8206,
+      "step": 1674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83499163364279e-05,
+      "loss": 0.7802,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347681318217354e-05,
+      "loss": 0.8723,
+      "step": 1676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345444923683685e-05,
+      "loss": 0.9287,
+      "step": 1677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343207153195615e-05,
+      "loss": 0.9546,
+      "step": 1678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83409680071221e-05,
+      "loss": 0.8133,
+      "step": 1679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338727485832317e-05,
+      "loss": 0.8208,
+      "step": 1680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833648558969567e-05,
+      "loss": 0.7691,
+      "step": 1681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334242319081793e-05,
+      "loss": 0.9409,
+      "step": 1682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331997674360543e-05,
+      "loss": 0.9019,
+      "step": 1683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329751655902014e-05,
+      "loss": 0.8235,
+      "step": 1684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832750426407651e-05,
+      "loss": 0.9059,
+      "step": 1685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325255499254567e-05,
+      "loss": 0.8242,
+      "step": 1686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832300536180696e-05,
+      "loss": 0.3466,
+      "step": 1687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320753852104673e-05,
+      "loss": 0.9444,
+      "step": 1688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318500970518924e-05,
+      "loss": 0.8853,
+      "step": 1689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316246717421157e-05,
+      "loss": 0.828,
+      "step": 1690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313991093183043e-05,
+      "loss": 0.9122,
+      "step": 1691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311734098176478e-05,
+      "loss": 0.9049,
+      "step": 1692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309475732773583e-05,
+      "loss": 0.9963,
+      "step": 1693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307215997346703e-05,
+      "loss": 0.9116,
+      "step": 1694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830495489226842e-05,
+      "loss": 0.8442,
+      "step": 1695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830269241791152e-05,
+      "loss": 0.9262,
+      "step": 1696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300428574649044e-05,
+      "loss": 0.9318,
+      "step": 1697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829816336285423e-05,
+      "loss": 0.9082,
+      "step": 1698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829589678290056e-05,
+      "loss": 0.9439,
+      "step": 1699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829362883516174e-05,
+      "loss": 0.9299,
+      "step": 1700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291359520011687e-05,
+      "loss": 0.9149,
+      "step": 1701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289088837824565e-05,
+      "loss": 0.9207,
+      "step": 1702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286816788974747e-05,
+      "loss": 0.8757,
+      "step": 1703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284543373836838e-05,
+      "loss": 0.9458,
+      "step": 1704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282268592785666e-05,
+      "loss": 0.9238,
+      "step": 1705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279992446196285e-05,
+      "loss": 0.8834,
+      "step": 1706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277714934443977e-05,
+      "loss": 0.8511,
+      "step": 1707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275436057904246e-05,
+      "loss": 0.9059,
+      "step": 1708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273155816952822e-05,
+      "loss": 0.8637,
+      "step": 1709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270874211965655e-05,
+      "loss": 0.9081,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826859124331893e-05,
+      "loss": 0.8618,
+      "step": 1711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266306911389048e-05,
+      "loss": 0.8832,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264021216552637e-05,
+      "loss": 0.8542,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261734159186556e-05,
+      "loss": 0.8938,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825944573966788e-05,
+      "loss": 0.9551,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825715595837391e-05,
+      "loss": 0.9594,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254864815682173e-05,
+      "loss": 0.8986,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252572311970426e-05,
+      "loss": 0.8875,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825027844761664e-05,
+      "loss": 0.9096,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247983222999024e-05,
+      "loss": 0.9304,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245686638495993e-05,
+      "loss": 0.8511,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82433886944862e-05,
+      "loss": 0.8065,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824108939134852e-05,
+      "loss": 0.8207,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823878872946205e-05,
+      "loss": 0.8377,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236486709206107e-05,
+      "loss": 0.8699,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823418333096024e-05,
+      "loss": 0.8395,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823187859510422e-05,
+      "loss": 0.9278,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822957250201804e-05,
+      "loss": 0.9507,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227265052081913e-05,
+      "loss": 0.9712,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822495624567628e-05,
+      "loss": 0.8175,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222646083181807e-05,
+      "loss": 0.918,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220334564979384e-05,
+      "loss": 0.8701,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821802169145012e-05,
+      "loss": 0.8349,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215707462975346e-05,
+      "loss": 0.9029,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213391879936626e-05,
+      "loss": 0.8051,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821107494271574e-05,
+      "loss": 0.8207,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820875665169469e-05,
+      "loss": 0.8294,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206437007255705e-05,
+      "loss": 0.8167,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204116009781243e-05,
+      "loss": 0.8801,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820179365965397e-05,
+      "loss": 0.9044,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199469957256785e-05,
+      "loss": 0.86,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197144902972808e-05,
+      "loss": 1.0523,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194818497185385e-05,
+      "loss": 0.9433,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819249074027808e-05,
+      "loss": 0.9335,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190161632634682e-05,
+      "loss": 0.8493,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81878311746392e-05,
+      "loss": 0.8749,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818549936667587e-05,
+      "loss": 0.7922,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818316620912915e-05,
+      "loss": 0.9169,
+      "step": 1748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818083170238372e-05,
+      "loss": 0.8696,
+      "step": 1749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178495846824474e-05,
+      "loss": 0.8834,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176158642836545e-05,
+      "loss": 0.8671,
+      "step": 1751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817382009080528e-05,
+      "loss": 0.3367,
+      "step": 1752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171480191116236e-05,
+      "loss": 0.9987,
+      "step": 1753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169138944155215e-05,
+      "loss": 0.9004,
+      "step": 1754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166796350308222e-05,
+      "loss": 0.9438,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164452409961498e-05,
+      "loss": 0.832,
+      "step": 1756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81621071235015e-05,
+      "loss": 0.9511,
+      "step": 1757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81597604913149e-05,
+      "loss": 0.9138,
+      "step": 1758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157412513788606e-05,
+      "loss": 0.9572,
+      "step": 1759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155063191309734e-05,
+      "loss": 0.8457,
+      "step": 1760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152712524265638e-05,
+      "loss": 0.848,
+      "step": 1761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815036051304387e-05,
+      "loss": 0.8237,
+      "step": 1762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148007158032233e-05,
+      "loss": 0.9243,
+      "step": 1763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814565245961873e-05,
+      "loss": 0.896,
+      "step": 1764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143296418191586e-05,
+      "loss": 0.8734,
+      "step": 1765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140939034139257e-05,
+      "loss": 0.9174,
+      "step": 1766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813858030785042e-05,
+      "loss": 0.9472,
+      "step": 1767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136220239713965e-05,
+      "loss": 0.8567,
+      "step": 1768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813385883011901e-05,
+      "loss": 0.8627,
+      "step": 1769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131496079454892e-05,
+      "loss": 0.9191,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129131988111174e-05,
+      "loss": 0.3201,
+      "step": 1771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126766556477625e-05,
+      "loss": 0.8632,
+      "step": 1772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124399784944253e-05,
+      "loss": 0.9321,
+      "step": 1773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122031673901278e-05,
+      "loss": 0.8958,
+      "step": 1774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811966222373914e-05,
+      "loss": 0.8833,
+      "step": 1775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117291434848503e-05,
+      "loss": 0.9072,
+      "step": 1776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114919307620252e-05,
+      "loss": 0.861,
+      "step": 1777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112545842445488e-05,
+      "loss": 0.9378,
+      "step": 1778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110171039715537e-05,
+      "loss": 0.9005,
+      "step": 1779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107794899821948e-05,
+      "loss": 0.8585,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105417423156482e-05,
+      "loss": 0.3384,
+      "step": 1781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810303861011113e-05,
+      "loss": 0.9505,
+      "step": 1782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810065846107809e-05,
+      "loss": 0.9265,
+      "step": 1783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80982769764498e-05,
+      "loss": 0.313,
+      "step": 1784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80958941566189e-05,
+      "loss": 0.8656,
+      "step": 1785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809351000197826e-05,
+      "loss": 0.9349,
+      "step": 1786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091124512920963e-05,
+      "loss": 0.9105,
+      "step": 1787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088737689840322e-05,
+      "loss": 0.8679,
+      "step": 1788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808634953312986e-05,
+      "loss": 0.8637,
+      "step": 1789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083960043183328e-05,
+      "loss": 0.8079,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081569220394688e-05,
+      "loss": 0.3198,
+      "step": 1791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807917706515813e-05,
+      "loss": 0.8464,
+      "step": 1792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807678357786806e-05,
+      "loss": 0.8341,
+      "step": 1793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074388758919103e-05,
+      "loss": 0.8146,
+      "step": 1794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071992608706103e-05,
+      "loss": 0.842,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806959512762413e-05,
+      "loss": 0.9405,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067196316068465e-05,
+      "loss": 0.8457,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064796174434614e-05,
+      "loss": 0.8563,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062394703118294e-05,
+      "loss": 0.9103,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059991902515457e-05,
+      "loss": 0.866,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057587773022253e-05,
+      "loss": 0.9149,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055182315035077e-05,
+      "loss": 0.8623,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052775528950516e-05,
+      "loss": 0.8784,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050367415165392e-05,
+      "loss": 0.8893,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047957974076745e-05,
+      "loss": 0.8602,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804554720608183e-05,
+      "loss": 0.9046,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043135111578125e-05,
+      "loss": 0.8689,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040721690963317e-05,
+      "loss": 0.88,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038306944635327e-05,
+      "loss": 0.8421,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035890872992278e-05,
+      "loss": 0.3752,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033473476432525e-05,
+      "loss": 0.8472,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031054755354636e-05,
+      "loss": 0.8719,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028634710157392e-05,
+      "loss": 0.8229,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80262133412398e-05,
+      "loss": 0.9199,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023790649001087e-05,
+      "loss": 0.8816,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021366633840694e-05,
+      "loss": 0.3501,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018941296158273e-05,
+      "loss": 0.8469,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801651463635371e-05,
+      "loss": 0.8562,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014086654827093e-05,
+      "loss": 0.3392,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801165735197874e-05,
+      "loss": 0.8052,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009226728209178e-05,
+      "loss": 0.8414,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006794783919162e-05,
+      "loss": 0.8878,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004361519509654e-05,
+      "loss": 0.9912,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800192693538184e-05,
+      "loss": 0.8568,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999491031937118e-05,
+      "loss": 0.8932,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997053809577112e-05,
+      "loss": 0.8591,
+      "step": 1826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994615268703655e-05,
+      "loss": 0.9461,
+      "step": 1827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992175409718807e-05,
+      "loss": 0.8708,
+      "step": 1828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989734233024837e-05,
+      "loss": 0.8849,
+      "step": 1829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987291739024227e-05,
+      "loss": 0.8796,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984847928119692e-05,
+      "loss": 0.8441,
+      "step": 1831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798240280071415e-05,
+      "loss": 0.8519,
+      "step": 1832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979956357210745e-05,
+      "loss": 0.8983,
+      "step": 1833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977508598012834e-05,
+      "loss": 0.8534,
+      "step": 1834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975059523523985e-05,
+      "loss": 0.8715,
+      "step": 1835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972609134147992e-05,
+      "loss": 0.8849,
+      "step": 1836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970157430288866e-05,
+      "loss": 0.8696,
+      "step": 1837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796770441235083e-05,
+      "loss": 0.7821,
+      "step": 1838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965250080738322e-05,
+      "loss": 0.9416,
+      "step": 1839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962794435856003e-05,
+      "loss": 0.8097,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960337478108743e-05,
+      "loss": 0.8275,
+      "step": 1841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957879207901637e-05,
+      "loss": 0.8313,
+      "step": 1842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955419625639995e-05,
+      "loss": 0.8471,
+      "step": 1843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795295873172933e-05,
+      "loss": 0.8526,
+      "step": 1844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950496526575392e-05,
+      "loss": 0.8545,
+      "step": 1845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794803301058413e-05,
+      "loss": 0.9178,
+      "step": 1846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794556818416172e-05,
+      "loss": 0.8871,
+      "step": 1847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943102047714548e-05,
+      "loss": 0.3248,
+      "step": 1848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794063460164922e-05,
+      "loss": 0.9109,
+      "step": 1849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793816584637255e-05,
+      "loss": 0.923,
+      "step": 1850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793569578229158e-05,
+      "loss": 0.8356,
+      "step": 1851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933224409813563e-05,
+      "loss": 0.8511,
+      "step": 1852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793075172934596e-05,
+      "loss": 0.7699,
+      "step": 1853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792827774129646e-05,
+      "loss": 0.9008,
+      "step": 1854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925802446072957e-05,
+      "loss": 0.8684,
+      "step": 1855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792332584408356e-05,
+      "loss": 0.8849,
+      "step": 1856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920847935736614e-05,
+      "loss": 0.8074,
+      "step": 1857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918368721440652e-05,
+      "loss": 0.8413,
+      "step": 1858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915888201604436e-05,
+      "loss": 0.8773,
+      "step": 1859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913406376636947e-05,
+      "loss": 0.8519,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910923246947368e-05,
+      "loss": 0.9088,
+      "step": 1861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908438812945106e-05,
+      "loss": 0.8402,
+      "step": 1862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790595307503979e-05,
+      "loss": 0.8376,
+      "step": 1863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903466033641244e-05,
+      "loss": 0.8369,
+      "step": 1864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900977689159525e-05,
+      "loss": 0.8401,
+      "step": 1865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78984880420049e-05,
+      "loss": 0.8454,
+      "step": 1866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895997092587848e-05,
+      "loss": 0.8408,
+      "step": 1867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789350484131906e-05,
+      "loss": 0.8757,
+      "step": 1868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891011288609454e-05,
+      "loss": 0.9945,
+      "step": 1869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788851643487014e-05,
+      "loss": 0.9252,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886020280512476e-05,
+      "loss": 0.8157,
+      "step": 1871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883522825948e-05,
+      "loss": 0.8838,
+      "step": 1872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881024071588483e-05,
+      "loss": 0.9116,
+      "step": 1873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878524017845914e-05,
+      "loss": 0.8342,
+      "step": 1874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876022665132483e-05,
+      "loss": 0.89,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873520013860595e-05,
+      "loss": 0.9194,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871016064442886e-05,
+      "loss": 0.9841,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868510817292188e-05,
+      "loss": 0.9119,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866004272821552e-05,
+      "loss": 0.8916,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786349643144425e-05,
+      "loss": 0.8682,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860987293573754e-05,
+      "loss": 0.8682,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858476859623766e-05,
+      "loss": 0.8999,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855965130008188e-05,
+      "loss": 0.7925,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853452105141146e-05,
+      "loss": 0.8525,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785093778543697e-05,
+      "loss": 0.8664,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848422171310208e-05,
+      "loss": 0.9253,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845905263175625e-05,
+      "loss": 0.9248,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843387061448195e-05,
+      "loss": 0.8681,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840867566543107e-05,
+      "loss": 0.8462,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783834677887576e-05,
+      "loss": 0.826,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835824698861774e-05,
+      "loss": 0.9639,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833301326916967e-05,
+      "loss": 0.9218,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830776663457387e-05,
+      "loss": 0.8721,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782825070889929e-05,
+      "loss": 0.8853,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825723463659134e-05,
+      "loss": 0.9141,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823194928153604e-05,
+      "loss": 0.7869,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782066510279959e-05,
+      "loss": 0.911,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78181339880142e-05,
+      "loss": 0.3124,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815601584214747e-05,
+      "loss": 0.7928,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813067891818764e-05,
+      "loss": 0.3244,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781053291124399e-05,
+      "loss": 0.8919,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780799664290838e-05,
+      "loss": 0.8646,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805459087230108e-05,
+      "loss": 0.9021,
+      "step": 1903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802920244627543e-05,
+      "loss": 0.8443,
+      "step": 1904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800380115519284e-05,
+      "loss": 0.8317,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797838700324132e-05,
+      "loss": 0.8228,
+      "step": 1906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77952959994611e-05,
+      "loss": 0.8789,
+      "step": 1907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779275201334942e-05,
+      "loss": 0.8348,
+      "step": 1908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779020674240853e-05,
+      "loss": 0.9797,
+      "step": 1909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778766018705808e-05,
+      "loss": 0.8568,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778511234771793e-05,
+      "loss": 0.8382,
+      "step": 1911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782563224808165e-05,
+      "loss": 0.7756,
+      "step": 1912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780012818749062e-05,
+      "loss": 0.9242,
+      "step": 1913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777461129961123e-05,
+      "loss": 0.8502,
+      "step": 1914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774908158865054e-05,
+      "loss": 0.8359,
+      "step": 1915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772353905881782e-05,
+      "loss": 0.838,
+      "step": 1916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769798371432433e-05,
+      "loss": 0.9744,
+      "step": 1917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776724155593835e-05,
+      "loss": 0.8939,
+      "step": 1918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776468345982109e-05,
+      "loss": 0.8756,
+      "step": 1919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762124083502425e-05,
+      "loss": 0.8821,
+      "step": 1920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759563427404325e-05,
+      "loss": 0.8765,
+      "step": 1921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757001491948976e-05,
+      "loss": 0.9251,
+      "step": 1922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754438277558783e-05,
+      "loss": 0.8735,
+      "step": 1923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775187378465635e-05,
+      "loss": 0.9421,
+      "step": 1924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749308013664503e-05,
+      "loss": 0.8518,
+      "step": 1925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774674096500627e-05,
+      "loss": 0.8097,
+      "step": 1926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744172639104894e-05,
+      "loss": 0.8647,
+      "step": 1927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741603036383824e-05,
+      "loss": 0.876,
+      "step": 1928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773903215726673e-05,
+      "loss": 0.902,
+      "step": 1929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773646000217748e-05,
+      "loss": 0.8766,
+      "step": 1930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733886571540155e-05,
+      "loss": 0.8931,
+      "step": 1931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731311865779058e-05,
+      "loss": 0.8924,
+      "step": 1932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772873588531869e-05,
+      "loss": 0.8634,
+      "step": 1933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772615863058376e-05,
+      "loss": 0.8639,
+      "step": 1934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723580101999195e-05,
+      "loss": 0.8549,
+      "step": 1935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721000299990137e-05,
+      "loss": 0.8818,
+      "step": 1936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771841922498192e-05,
+      "loss": 0.8775,
+      "step": 1937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715836877400108e-05,
+      "loss": 0.9772,
+      "step": 1938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771325325767046e-05,
+      "loss": 0.8416,
+      "step": 1939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771066836621895e-05,
+      "loss": 0.8609,
+      "step": 1940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770808220347176e-05,
+      "loss": 0.8357,
+      "step": 1941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705494769855288e-05,
+      "loss": 0.9414,
+      "step": 1942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702906065796136e-05,
+      "loss": 0.8896,
+      "step": 1943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770031609172111e-05,
+      "loss": 0.876,
+      "step": 1944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769772484805724e-05,
+      "loss": 0.854,
+      "step": 1945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695132335231758e-05,
+      "loss": 0.9157,
+      "step": 1946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7692538553672096e-05,
+      "loss": 0.9931,
+      "step": 1947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689943503805906e-05,
+      "loss": 0.8502,
+      "step": 1948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687347186061054e-05,
+      "loss": 1.0265,
+      "step": 1949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684749600865597e-05,
+      "loss": 0.9051,
+      "step": 1950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682150748647815e-05,
+      "loss": 0.8607,
+      "step": 1951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76795506298362e-05,
+      "loss": 0.8701,
+      "step": 1952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676949244859435e-05,
+      "loss": 0.85,
+      "step": 1953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674346594146433e-05,
+      "loss": 0.8261,
+      "step": 1954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76717426781263e-05,
+      "loss": 0.816,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669137497228365e-05,
+      "loss": 0.8965,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666531051882146e-05,
+      "loss": 0.8658,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663923342517384e-05,
+      "loss": 0.9041,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661314369564024e-05,
+      "loss": 0.8822,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658704133452228e-05,
+      "loss": 0.888,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656092634612348e-05,
+      "loss": 0.9049,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653479873474957e-05,
+      "loss": 0.8527,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765086585047084e-05,
+      "loss": 0.8359,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648250566030975e-05,
+      "loss": 0.8984,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645634020586562e-05,
+      "loss": 0.8353,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643016214569005e-05,
+      "loss": 0.8901,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764039714840991e-05,
+      "loss": 0.889,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637776822541098e-05,
+      "loss": 0.8658,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635155237394597e-05,
+      "loss": 0.9204,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632532393402638e-05,
+      "loss": 0.8436,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762990829099766e-05,
+      "loss": 0.8883,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627282930612315e-05,
+      "loss": 0.8699,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762465631267946e-05,
+      "loss": 0.8732,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622028437632154e-05,
+      "loss": 0.813,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761939930590367e-05,
+      "loss": 0.8193,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616768917927484e-05,
+      "loss": 0.8583,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614137274137286e-05,
+      "loss": 0.879,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761150437496696e-05,
+      "loss": 0.8963,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608870220850615e-05,
+      "loss": 1.0276,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760623481222255e-05,
+      "loss": 0.9303,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603598149517277e-05,
+      "loss": 0.3146,
+      "step": 1981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760096023316952e-05,
+      "loss": 0.9152,
+      "step": 1982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598321063614202e-05,
+      "loss": 0.8753,
+      "step": 1983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759568064128646e-05,
+      "loss": 0.8467,
+      "step": 1984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593038966621626e-05,
+      "loss": 0.8599,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590396040055252e-05,
+      "loss": 0.8865,
+      "step": 1986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587751862023086e-05,
+      "loss": 0.8897,
+      "step": 1987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585106432961093e-05,
+      "loss": 0.844,
+      "step": 1988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582459753305432e-05,
+      "loss": 0.9033,
+      "step": 1989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579811823492476e-05,
+      "loss": 0.9123,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75771626439588e-05,
+      "loss": 0.8413,
+      "step": 1991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574512215141194e-05,
+      "loss": 0.8706,
+      "step": 1992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571860537476636e-05,
+      "loss": 0.9777,
+      "step": 1993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569207611402334e-05,
+      "loss": 0.8668,
+      "step": 1994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566553437355674e-05,
+      "loss": 0.8746,
+      "step": 1995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756389801577428e-05,
+      "loss": 0.87,
+      "step": 1996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756124134709595e-05,
+      "loss": 0.8898,
+      "step": 1997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755858343175871e-05,
+      "loss": 0.9292,
+      "step": 1998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755592427020078e-05,
+      "loss": 0.8405,
+      "step": 1999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755326386286059e-05,
+      "loss": 0.8821,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550602210176775e-05,
+      "loss": 0.9747,
+      "step": 2001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754793931258817e-05,
+      "loss": 0.8583,
+      "step": 2002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754527517053383e-05,
+      "loss": 0.9416,
+      "step": 2003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542609784452996e-05,
+      "loss": 0.8539,
+      "step": 2004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539943154785126e-05,
+      "loss": 0.8271,
+      "step": 2005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537275281969883e-05,
+      "loss": 0.8,
+      "step": 2006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534606166447127e-05,
+      "loss": 0.859,
+      "step": 2007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531935808656933e-05,
+      "loss": 0.8612,
+      "step": 2008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529264209039573e-05,
+      "loss": 0.8112,
+      "step": 2009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752659136803553e-05,
+      "loss": 0.8679,
+      "step": 2010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7523917286085486e-05,
+      "loss": 0.9083,
+      "step": 2011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752124196363033e-05,
+      "loss": 0.8954,
+      "step": 2012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751856540111115e-05,
+      "loss": 0.8793,
+      "step": 2013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515887598969256e-05,
+      "loss": 0.8382,
+      "step": 2014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513208557646145e-05,
+      "loss": 0.8088,
+      "step": 2015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751052827758352e-05,
+      "loss": 0.8817,
+      "step": 2016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507846759223296e-05,
+      "loss": 0.8672,
+      "step": 2017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505164003007585e-05,
+      "loss": 0.9074,
+      "step": 2018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502480009378707e-05,
+      "loss": 0.91,
+      "step": 2019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749979477877919e-05,
+      "loss": 0.8965,
+      "step": 2020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497108311651754e-05,
+      "loss": 0.8997,
+      "step": 2021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749442060843934e-05,
+      "loss": 0.8447,
+      "step": 2022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491731669585066e-05,
+      "loss": 0.9156,
+      "step": 2023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489041495532286e-05,
+      "loss": 0.8819,
+      "step": 2024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486350086724538e-05,
+      "loss": 0.8679,
+      "step": 2025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483657443605563e-05,
+      "loss": 0.9184,
+      "step": 2026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480963566619315e-05,
+      "loss": 0.7979,
+      "step": 2027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478268456209947e-05,
+      "loss": 0.9237,
+      "step": 2028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747557211282181e-05,
+      "loss": 0.9449,
+      "step": 2029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747287453689947e-05,
+      "loss": 0.9006,
+      "step": 2030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470175728887683e-05,
+      "loss": 0.9233,
+      "step": 2031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746747568923142e-05,
+      "loss": 0.9421,
+      "step": 2032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746477441837585e-05,
+      "loss": 0.8188,
+      "step": 2033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746207191676634e-05,
+      "loss": 0.9987,
+      "step": 2034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745936818484847e-05,
+      "loss": 0.8635,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456663223068014e-05,
+      "loss": 1.0003,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745395703187095e-05,
+      "loss": 0.8827,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7451249611703468e-05,
+      "loss": 0.8748,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448540963011948e-05,
+      "loss": 0.9627,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744583108624298e-05,
+      "loss": 0.8818,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443119981843358e-05,
+      "loss": 0.9536,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440407650260067e-05,
+      "loss": 0.931,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743769409194031e-05,
+      "loss": 0.8271,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434979307331482e-05,
+      "loss": 0.8599,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432263296881183e-05,
+      "loss": 0.8857,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742954606103721e-05,
+      "loss": 0.9295,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426827600247575e-05,
+      "loss": 0.8994,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742410791496048e-05,
+      "loss": 0.8823,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421387005624333e-05,
+      "loss": 0.8523,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418664872687744e-05,
+      "loss": 0.8768,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415941516599525e-05,
+      "loss": 0.8637,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413216937808687e-05,
+      "loss": 0.8448,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410491136764448e-05,
+      "loss": 0.8809,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407764113916222e-05,
+      "loss": 0.9682,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405035869713626e-05,
+      "loss": 0.8429,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740230640460648e-05,
+      "loss": 0.9133,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399575719044805e-05,
+      "loss": 0.8405,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396843813478825e-05,
+      "loss": 0.8219,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394110688358954e-05,
+      "loss": 0.9297,
+      "step": 2059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739137634413582e-05,
+      "loss": 0.857,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388640781260254e-05,
+      "loss": 0.8357,
+      "step": 2061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385904000183274e-05,
+      "loss": 0.8265,
+      "step": 2062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738316600135611e-05,
+      "loss": 0.8325,
+      "step": 2063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738042678523019e-05,
+      "loss": 0.9004,
+      "step": 2064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377686352257136e-05,
+      "loss": 0.9269,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374944702888782e-05,
+      "loss": 0.8603,
+      "step": 2066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372201837577155e-05,
+      "loss": 0.8939,
+      "step": 2067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736945775677449e-05,
+      "loss": 0.7859,
+      "step": 2068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736671246093321e-05,
+      "loss": 0.8301,
+      "step": 2069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736396595050595e-05,
+      "loss": 0.9136,
+      "step": 2070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361218225945532e-05,
+      "loss": 0.8848,
+      "step": 2071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358469287705e-05,
+      "loss": 0.9492,
+      "step": 2072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355719136237578e-05,
+      "loss": 0.8939,
+      "step": 2073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352967771996695e-05,
+      "loss": 0.9065,
+      "step": 2074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350215195435986e-05,
+      "loss": 0.8581,
+      "step": 2075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734746140700928e-05,
+      "loss": 0.8169,
+      "step": 2076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344706407170608e-05,
+      "loss": 0.9113,
+      "step": 2077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73419501963742e-05,
+      "loss": 0.8655,
+      "step": 2078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339192775074486e-05,
+      "loss": 0.9087,
+      "step": 2079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336434143726094e-05,
+      "loss": 0.8373,
+      "step": 2080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333674302783855e-05,
+      "loss": 0.8748,
+      "step": 2081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330913252702796e-05,
+      "loss": 0.9281,
+      "step": 2082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732815099393815e-05,
+      "loss": 0.7861,
+      "step": 2083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732538752694534e-05,
+      "loss": 0.8192,
+      "step": 2084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732262285217999e-05,
+      "loss": 0.8818,
+      "step": 2085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319856970097927e-05,
+      "loss": 0.8541,
+      "step": 2086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731708988115518e-05,
+      "loss": 0.8345,
+      "step": 2087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314321585807965e-05,
+      "loss": 0.8955,
+      "step": 2088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731155208451271e-05,
+      "loss": 0.3307,
+      "step": 2089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308781377726032e-05,
+      "loss": 0.8268,
+      "step": 2090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730600946590476e-05,
+      "loss": 0.8279,
+      "step": 2091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73032363495059e-05,
+      "loss": 0.9373,
+      "step": 2092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730046202898668e-05,
+      "loss": 0.829,
+      "step": 2093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729768650480451e-05,
+      "loss": 0.8562,
+      "step": 2094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294909777417007e-05,
+      "loss": 0.813,
+      "step": 2095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292131847281983e-05,
+      "loss": 0.8537,
+      "step": 2096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289352714857447e-05,
+      "loss": 0.9128,
+      "step": 2097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728657238060161e-05,
+      "loss": 0.8691,
+      "step": 2098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283790844972876e-05,
+      "loss": 0.8602,
+      "step": 2099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281008108429854e-05,
+      "loss": 0.881,
+      "step": 2100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727822417143135e-05,
+      "loss": 0.8372,
+      "step": 2101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727543903443636e-05,
+      "loss": 0.8697,
+      "step": 2102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272652697904083e-05,
+      "loss": 0.8499,
+      "step": 2103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726986516229392e-05,
+      "loss": 0.801,
+      "step": 2104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267076428065464e-05,
+      "loss": 0.8718,
+      "step": 2105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726428649567851e-05,
+      "loss": 0.8196,
+      "step": 2106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726149536559304e-05,
+      "loss": 0.8763,
+      "step": 2107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725870303826925e-05,
+      "loss": 0.9055,
+      "step": 2108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725590951416752e-05,
+      "loss": 0.8285,
+      "step": 2109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725311479374843e-05,
+      "loss": 0.8558,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250318877472766e-05,
+      "loss": 0.9533,
+      "step": 2111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72475217658015e-05,
+      "loss": 0.7733,
+      "step": 2112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244723459195803e-05,
+      "loss": 0.8826,
+      "step": 2113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7241923958117047e-05,
+      "loss": 0.812,
+      "step": 2114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239123263026805e-05,
+      "loss": 0.8891,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723632137438684e-05,
+      "loss": 0.9156,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233518292659105e-05,
+      "loss": 0.8253,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230714018305767e-05,
+      "loss": 0.8764,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227908551789178e-05,
+      "loss": 0.9059,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225101893571887e-05,
+      "loss": 0.8449,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222294044116637e-05,
+      "loss": 0.8391,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721948500388638e-05,
+      "loss": 0.9401,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216674773344255e-05,
+      "loss": 0.9349,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213863352953594e-05,
+      "loss": 0.8218,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7211050743177935e-05,
+      "loss": 0.903,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208236944481005e-05,
+      "loss": 0.8025,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205421957326727e-05,
+      "loss": 0.8074,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202605782179223e-05,
+      "loss": 0.8488,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719978841950281e-05,
+      "loss": 0.8867,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196969869761996e-05,
+      "loss": 0.8381,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194150133421498e-05,
+      "loss": 0.9198,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191329210946212e-05,
+      "loss": 0.8619,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718850710280124e-05,
+      "loss": 0.8827,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185683809451875e-05,
+      "loss": 0.8791,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718285933136361e-05,
+      "loss": 0.8572,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718003366900213e-05,
+      "loss": 0.3139,
+      "step": 2136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717720682283332e-05,
+      "loss": 0.9409,
+      "step": 2137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174378793323245e-05,
+      "loss": 0.8739,
+      "step": 2138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171549580938184e-05,
+      "loss": 0.9051,
+      "step": 2139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168719186144603e-05,
+      "loss": 0.8822,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716588760940916e-05,
+      "loss": 0.8515,
+      "step": 2141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163054851198712e-05,
+      "loss": 0.8021,
+      "step": 2142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160220911980313e-05,
+      "loss": 0.8268,
+      "step": 2143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157385792221203e-05,
+      "loss": 0.824,
+      "step": 2144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154549492388825e-05,
+      "loss": 0.9243,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151712012950815e-05,
+      "loss": 0.8089,
+      "step": 2146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148873354375e-05,
+      "loss": 0.9175,
+      "step": 2147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71460335171294e-05,
+      "loss": 0.8967,
+      "step": 2148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7143192501682243e-05,
+      "loss": 0.9016,
+      "step": 2149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714035030850193e-05,
+      "loss": 0.8486,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713750693805708e-05,
+      "loss": 0.9157,
+      "step": 2151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713466239081648e-05,
+      "loss": 0.9081,
+      "step": 2152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131816667249128e-05,
+      "loss": 0.9227,
+      "step": 2153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712896976782422e-05,
+      "loss": 0.3407,
+      "step": 2154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712612169301113e-05,
+      "loss": 0.9364,
+      "step": 2155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712327244327944e-05,
+      "loss": 0.8741,
+      "step": 2156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712042201909891e-05,
+      "loss": 0.9461,
+      "step": 2157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117570420939516e-05,
+      "loss": 0.8595,
+      "step": 2158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114717649271405e-05,
+      "loss": 0.8071,
+      "step": 2159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111863704564936e-05,
+      "loss": 0.9051,
+      "step": 2160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109008587290647e-05,
+      "loss": 0.8213,
+      "step": 2161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106152297919276e-05,
+      "loss": 0.8676,
+      "step": 2162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103294836921752e-05,
+      "loss": 0.8364,
+      "step": 2163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100436204769204e-05,
+      "loss": 0.8661,
+      "step": 2164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097576401932944e-05,
+      "loss": 0.891,
+      "step": 2165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094715428884482e-05,
+      "loss": 0.8742,
+      "step": 2166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091853286095525e-05,
+      "loss": 0.748,
+      "step": 2167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708898997403796e-05,
+      "loss": 0.9495,
+      "step": 2168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086125493183884e-05,
+      "loss": 0.8253,
+      "step": 2169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708325984400557e-05,
+      "loss": 0.823,
+      "step": 2170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70803930269755e-05,
+      "loss": 0.829,
+      "step": 2171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077525042566335e-05,
+      "loss": 0.86,
+      "step": 2172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074655891250938e-05,
+      "loss": 0.9298,
+      "step": 2173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071785573502353e-05,
+      "loss": 1.0005,
+      "step": 2174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068914089793827e-05,
+      "loss": 0.8628,
+      "step": 2175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70660414405988e-05,
+      "loss": 0.8277,
+      "step": 2176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063167626390893e-05,
+      "loss": 0.988,
+      "step": 2177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060292647643927e-05,
+      "loss": 0.8764,
+      "step": 2178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705741650483192e-05,
+      "loss": 0.8643,
+      "step": 2179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054539198429066e-05,
+      "loss": 0.8818,
+      "step": 2180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051660728909766e-05,
+      "loss": 0.9058,
+      "step": 2181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048781096748608e-05,
+      "loss": 0.8789,
+      "step": 2182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704590030242037e-05,
+      "loss": 0.8087,
+      "step": 2183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043018346400024e-05,
+      "loss": 0.8767,
+      "step": 2184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704013522916273e-05,
+      "loss": 0.865,
+      "step": 2185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037250951183844e-05,
+      "loss": 0.859,
+      "step": 2186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034365512938906e-05,
+      "loss": 0.8334,
+      "step": 2187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031478914903662e-05,
+      "loss": 0.8668,
+      "step": 2188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028591157554027e-05,
+      "loss": 1.0375,
+      "step": 2189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702570224136613e-05,
+      "loss": 0.9484,
+      "step": 2190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022812166816277e-05,
+      "loss": 0.8929,
+      "step": 2191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019920934380968e-05,
+      "loss": 0.8102,
+      "step": 2192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017028544536892e-05,
+      "loss": 0.8963,
+      "step": 2193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701413499776094e-05,
+      "loss": 0.8569,
+      "step": 2194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011240294530178e-05,
+      "loss": 0.8239,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700834443532187e-05,
+      "loss": 0.8241,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700544742061348e-05,
+      "loss": 0.3685,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002549250882637e-05,
+      "loss": 0.8354,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999649926607193e-05,
+      "loss": 0.8465,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699674944826516e-05,
+      "loss": 0.8262,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699384781633476e-05,
+      "loss": 0.9682,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69909450312944e-05,
+      "loss": 0.9254,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988041093622676e-05,
+      "loss": 0.8314,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985136003798376e-05,
+      "loss": 0.8468,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698222976230047e-05,
+      "loss": 0.8521,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979322369608134e-05,
+      "loss": 0.915,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976413826200715e-05,
+      "loss": 0.921,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973504132557766e-05,
+      "loss": 0.8433,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970593289159014e-05,
+      "loss": 0.8594,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967681296484393e-05,
+      "loss": 0.9412,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964768155014012e-05,
+      "loss": 0.8762,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961853865228176e-05,
+      "loss": 0.8224,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958938427607377e-05,
+      "loss": 0.8244,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956021842632302e-05,
+      "loss": 0.8757,
+      "step": 2214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695310411078382e-05,
+      "loss": 0.8238,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6950185232542994e-05,
+      "loss": 0.7759,
+      "step": 2216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947265208391064e-05,
+      "loss": 0.358,
+      "step": 2217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694434403880948e-05,
+      "loss": 0.9032,
+      "step": 2218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941421724279866e-05,
+      "loss": 0.8487,
+      "step": 2219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693849826528404e-05,
+      "loss": 0.8183,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935573662304e-05,
+      "loss": 0.9277,
+      "step": 2221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932647915821946e-05,
+      "loss": 0.8492,
+      "step": 2222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929721026320264e-05,
+      "loss": 0.351,
+      "step": 2223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926792994281514e-05,
+      "loss": 0.8224,
+      "step": 2224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692386382018846e-05,
+      "loss": 0.8684,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920933504524048e-05,
+      "loss": 0.7848,
+      "step": 2226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691800204777142e-05,
+      "loss": 0.8495,
+      "step": 2227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691506945041389e-05,
+      "loss": 0.76,
+      "step": 2228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691213571293498e-05,
+      "loss": 0.877,
+      "step": 2229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690920083581838e-05,
+      "loss": 0.8698,
+      "step": 2230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906264819547984e-05,
+      "loss": 0.8591,
+      "step": 2231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903327664607863e-05,
+      "loss": 0.8757,
+      "step": 2232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900389371482286e-05,
+      "loss": 0.8396,
+      "step": 2233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689744994065569e-05,
+      "loss": 0.8561,
+      "step": 2234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894509372612735e-05,
+      "loss": 0.8991,
+      "step": 2235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689156766783823e-05,
+      "loss": 0.9666,
+      "step": 2236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888624826817193e-05,
+      "loss": 0.8674,
+      "step": 2237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885680850034827e-05,
+      "loss": 0.7695,
+      "step": 2238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882735737976516e-05,
+      "loss": 0.3532,
+      "step": 2239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879789491127837e-05,
+      "loss": 0.8108,
+      "step": 2240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876842109974552e-05,
+      "loss": 0.8436,
+      "step": 2241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6873893595002612e-05,
+      "loss": 0.8034,
+      "step": 2242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870943946698153e-05,
+      "loss": 0.7896,
+      "step": 2243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867993165547493e-05,
+      "loss": 0.3369,
+      "step": 2244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865041252037145e-05,
+      "loss": 0.8428,
+      "step": 2245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686208820665381e-05,
+      "loss": 0.8717,
+      "step": 2246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685913402988436e-05,
+      "loss": 0.8592,
+      "step": 2247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6856178722215876e-05,
+      "loss": 0.8579,
+      "step": 2248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685322228413561e-05,
+      "loss": 0.83,
+      "step": 2249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850264716131e-05,
+      "loss": 0.8261,
+      "step": 2250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684730601868968e-05,
+      "loss": 0.8113,
+      "step": 2251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844346192299464e-05,
+      "loss": 0.874,
+      "step": 2252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684138523744835e-05,
+      "loss": 0.8729,
+      "step": 2253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838423154624534e-05,
+      "loss": 0.776,
+      "step": 2254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835459944316378e-05,
+      "loss": 0.8822,
+      "step": 2255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832495607012448e-05,
+      "loss": 0.9256,
+      "step": 2256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829530143201485e-05,
+      "loss": 0.9186,
+      "step": 2257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826563553372423e-05,
+      "loss": 0.8233,
+      "step": 2258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823595838014373e-05,
+      "loss": 0.8024,
+      "step": 2259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682062699761664e-05,
+      "loss": 0.8259,
+      "step": 2260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817657032668715e-05,
+      "loss": 0.8272,
+      "step": 2261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814685943660266e-05,
+      "loss": 0.88,
+      "step": 2262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811713731081153e-05,
+      "loss": 0.8958,
+      "step": 2263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808740395421414e-05,
+      "loss": 0.9122,
+      "step": 2264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805765937171285e-05,
+      "loss": 0.8391,
+      "step": 2265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802790356821174e-05,
+      "loss": 0.8209,
+      "step": 2266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799813654861687e-05,
+      "loss": 0.8003,
+      "step": 2267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796835831783597e-05,
+      "loss": 0.8384,
+      "step": 2268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679385688807788e-05,
+      "loss": 0.8104,
+      "step": 2269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790876824235685e-05,
+      "loss": 1.0001,
+      "step": 2270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678789564074835e-05,
+      "loss": 0.8082,
+      "step": 2271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784913338107402e-05,
+      "loss": 0.8806,
+      "step": 2272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678192991680454e-05,
+      "loss": 0.8947,
+      "step": 2273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778945377331663e-05,
+      "loss": 0.8667,
+      "step": 2274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6775959720180847e-05,
+      "loss": 0.8053,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677297294584434e-05,
+      "loss": 0.8338,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769985054814598e-05,
+      "loss": 0.8091,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676699604758425e-05,
+      "loss": 0.8574,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67640059246461e-05,
+      "loss": 0.8632,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761014686493145e-05,
+      "loss": 0.8733,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758022333618575e-05,
+      "loss": 0.8503,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675502886651574e-05,
+      "loss": 0.8415,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67520342856782e-05,
+      "loss": 0.9196,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749038591599678e-05,
+      "loss": 0.8276,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746041784774095e-05,
+      "loss": 0.7896,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743043865695546e-05,
+      "loss": 0.873,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740044834858308e-05,
+      "loss": 0.8682,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737044692756853e-05,
+      "loss": 0.9252,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734043439885826e-05,
+      "loss": 0.9087,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673104107674006e-05,
+      "loss": 0.9305,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728037603814566e-05,
+      "loss": 0.7993,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725033021604545e-05,
+      "loss": 0.8267,
+      "step": 2292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722027330605376e-05,
+      "loss": 0.9266,
+      "step": 2293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671902053131262e-05,
+      "loss": 0.8865,
+      "step": 2294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716012624222028e-05,
+      "loss": 0.8906,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713003609829518e-05,
+      "loss": 0.9039,
+      "step": 2296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670999348863121e-05,
+      "loss": 0.8946,
+      "step": 2297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67069822611234e-05,
+      "loss": 0.8362,
+      "step": 2298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703969927802556e-05,
+      "loss": 0.8557,
+      "step": 2299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670095648916534e-05,
+      "loss": 0.8404,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669794194570859e-05,
+      "loss": 0.8839,
+      "step": 2301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669492629792933e-05,
+      "loss": 0.9168,
+      "step": 2302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669190954632477e-05,
+      "loss": 0.819,
+      "step": 2303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668889169139228e-05,
+      "loss": 0.8145,
+      "step": 2304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685872733629448e-05,
+      "loss": 0.8289,
+      "step": 2305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6682852673534014e-05,
+      "loss": 0.7832,
+      "step": 2306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679831511603913e-05,
+      "loss": 0.8528,
+      "step": 2307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667680924833726e-05,
+      "loss": 0.848,
+      "step": 2308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673785884232346e-05,
+      "loss": 0.8246,
+      "step": 2309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667076141978765e-05,
+      "loss": 0.8085,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6667735855501832e-05,
+      "loss": 0.8369,
+      "step": 2311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666470919187373e-05,
+      "loss": 0.9743,
+      "step": 2312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661681429402365e-05,
+      "loss": 0.8995,
+      "step": 2313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658652568586938e-05,
+      "loss": 0.8541,
+      "step": 2314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655622609926837e-05,
+      "loss": 0.8085,
+      "step": 2315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652591553921617e-05,
+      "loss": 0.8429,
+      "step": 2316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664955940107103e-05,
+      "loss": 0.8241,
+      "step": 2317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646526151875005e-05,
+      "loss": 0.915,
+      "step": 2318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643491806833637e-05,
+      "loss": 0.8208,
+      "step": 2319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640456366447223e-05,
+      "loss": 0.917,
+      "step": 2320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663741983121623e-05,
+      "loss": 0.8226,
+      "step": 2321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634382201641304e-05,
+      "loss": 0.8882,
+      "step": 2322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663134347822327e-05,
+      "loss": 0.8806,
+      "step": 2323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662830366146315e-05,
+      "loss": 0.8197,
+      "step": 2324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662526275186212e-05,
+      "loss": 0.9151,
+      "step": 2325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622220749921555e-05,
+      "loss": 0.8152,
+      "step": 2326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619177656143006e-05,
+      "loss": 0.8992,
+      "step": 2327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66161334710282e-05,
+      "loss": 0.8643,
+      "step": 2328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661308819507905e-05,
+      "loss": 0.8256,
+      "step": 2329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661004182879764e-05,
+      "loss": 0.8181,
+      "step": 2330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606994372686246e-05,
+      "loss": 0.8702,
+      "step": 2331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603945827247313e-05,
+      "loss": 0.8968,
+      "step": 2332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660089619298347e-05,
+      "loss": 0.7957,
+      "step": 2333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659784547039753e-05,
+      "loss": 0.8759,
+      "step": 2334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594793659992473e-05,
+      "loss": 0.3268,
+      "step": 2335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659174076227147e-05,
+      "loss": 0.8386,
+      "step": 2336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588686777737867e-05,
+      "loss": 0.8044,
+      "step": 2337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585631706895186e-05,
+      "loss": 0.8303,
+      "step": 2338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658257555024714e-05,
+      "loss": 0.8685,
+      "step": 2339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579518308297603e-05,
+      "loss": 0.3168,
+      "step": 2340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576459981550643e-05,
+      "loss": 0.8368,
+      "step": 2341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65734005705105e-05,
+      "loss": 0.8404,
+      "step": 2342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6570340075681593e-05,
+      "loss": 0.9295,
+      "step": 2343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567278497568522e-05,
+      "loss": 0.9035,
+      "step": 2344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564215836676066e-05,
+      "loss": 0.8168,
+      "step": 2345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561152093509173e-05,
+      "loss": 0.7497,
+      "step": 2346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655808726857299e-05,
+      "loss": 0.846,
+      "step": 2347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655502136237282e-05,
+      "loss": 0.9286,
+      "step": 2348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551954375414158e-05,
+      "loss": 0.3236,
+      "step": 2349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654888630820267e-05,
+      "loss": 1.0055,
+      "step": 2350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545817161244207e-05,
+      "loss": 0.8048,
+      "step": 2351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542746935044793e-05,
+      "loss": 0.8041,
+      "step": 2352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539675630110634e-05,
+      "loss": 0.8334,
+      "step": 2353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65366032469481e-05,
+      "loss": 0.8705,
+      "step": 2354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6533529786063766e-05,
+      "loss": 0.8605,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530455247964356e-05,
+      "loss": 0.861,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652737963315679e-05,
+      "loss": 0.9307,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652430294214816e-05,
+      "loss": 0.8103,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652122517544573e-05,
+      "loss": 0.8819,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651814633355695e-05,
+      "loss": 0.8712,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651506641698945e-05,
+      "loss": 0.871,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511985426251014e-05,
+      "loss": 0.8697,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508903361849638e-05,
+      "loss": 0.8466,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650582022429347e-05,
+      "loss": 0.7967,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650273601409084e-05,
+      "loss": 0.8922,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499650731750256e-05,
+      "loss": 0.7704,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649656437778041e-05,
+      "loss": 0.7935,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493476952690158e-05,
+      "loss": 0.7946,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490388456988544e-05,
+      "loss": 0.869,
+      "step": 2369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648729889118478e-05,
+      "loss": 0.8443,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484208255788255e-05,
+      "loss": 0.8724,
+      "step": 2371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481116551308545e-05,
+      "loss": 0.8919,
+      "step": 2372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647802377825539e-05,
+      "loss": 0.8394,
+      "step": 2373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647492993713871e-05,
+      "loss": 0.8719,
+      "step": 2374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6471835028468604e-05,
+      "loss": 0.9007,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646873905275534e-05,
+      "loss": 0.8915,
+      "step": 2376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465642010509374e-05,
+      "loss": 0.8644,
+      "step": 2377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6462543902241326e-05,
+      "loss": 0.7935,
+      "step": 2378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459444728461993e-05,
+      "loss": 0.8189,
+      "step": 2379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645634448968236e-05,
+      "loss": 0.8776,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453243186413572e-05,
+      "loss": 0.7944,
+      "step": 2381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645014081916696e-05,
+      "loss": 0.8813,
+      "step": 2382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644703738845402e-05,
+      "loss": 0.91,
+      "step": 2383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443932894786433e-05,
+      "loss": 0.8665,
+      "step": 2384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440827338676055e-05,
+      "loss": 0.9673,
+      "step": 2385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437720720634915e-05,
+      "loss": 0.8928,
+      "step": 2386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643461304117521e-05,
+      "loss": 0.9491,
+      "step": 2387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431504300809322e-05,
+      "loss": 0.81,
+      "step": 2388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428394500049807e-05,
+      "loss": 0.8141,
+      "step": 2389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425283639409387e-05,
+      "loss": 0.9184,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642217171940097e-05,
+      "loss": 0.9105,
+      "step": 2391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641905874053763e-05,
+      "loss": 0.8832,
+      "step": 2392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415944703332625e-05,
+      "loss": 0.9137,
+      "step": 2393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412829608299373e-05,
+      "loss": 0.8026,
+      "step": 2394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640971345595148e-05,
+      "loss": 0.8767,
+      "step": 2395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406596246802723e-05,
+      "loss": 0.8857,
+      "step": 2396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403477981367047e-05,
+      "loss": 0.9278,
+      "step": 2397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400358660158578e-05,
+      "loss": 0.8325,
+      "step": 2398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397238283691615e-05,
+      "loss": 0.911,
+      "step": 2399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394116852480623e-05,
+      "loss": 0.883,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390994367040257e-05,
+      "loss": 0.7993,
+      "step": 2401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387870827885334e-05,
+      "loss": 0.8839,
+      "step": 2402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384746235530845e-05,
+      "loss": 0.7954,
+      "step": 2403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381620590491957e-05,
+      "loss": 0.8475,
+      "step": 2404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637849389328401e-05,
+      "loss": 0.3797,
+      "step": 2405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637536614442252e-05,
+      "loss": 0.8452,
+      "step": 2406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372237344423173e-05,
+      "loss": 0.8315,
+      "step": 2407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636910749380183e-05,
+      "loss": 0.8473,
+      "step": 2408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365976593074523e-05,
+      "loss": 0.8612,
+      "step": 2409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636284464275746e-05,
+      "loss": 0.8834,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635971164336702e-05,
+      "loss": 0.952,
+      "step": 2411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356577595419763e-05,
+      "loss": 0.8092,
+      "step": 2412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353442499432404e-05,
+      "loss": 0.8418,
+      "step": 2413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350306355921853e-05,
+      "loss": 0.8783,
+      "step": 2414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634716916540517e-05,
+      "loss": 0.9364,
+      "step": 2415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344030928399607e-05,
+      "loss": 0.8577,
+      "step": 2416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6340891645422575e-05,
+      "loss": 0.8536,
+      "step": 2417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633775131699167e-05,
+      "loss": 0.8391,
+      "step": 2418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334609943624643e-05,
+      "loss": 0.804,
+      "step": 2419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331467525839437e-05,
+      "loss": 0.9682,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328324064154153e-05,
+      "loss": 0.8588,
+      "step": 2421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632517955908707e-05,
+      "loss": 0.8616,
+      "step": 2422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322034011156644e-05,
+      "loss": 0.8012,
+      "step": 2423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318887420881485e-05,
+      "loss": 0.8191,
+      "step": 2424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315739788780398e-05,
+      "loss": 0.8257,
+      "step": 2425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631259111537234e-05,
+      "loss": 0.8449,
+      "step": 2426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309441401176453e-05,
+      "loss": 0.8691,
+      "step": 2427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306290646712043e-05,
+      "loss": 0.9063,
+      "step": 2428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303138852498594e-05,
+      "loss": 0.9636,
+      "step": 2429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299986019055753e-05,
+      "loss": 0.8521,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629683214690335e-05,
+      "loss": 1.0336,
+      "step": 2431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629367723656137e-05,
+      "loss": 0.7916,
+      "step": 2432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629052128854999e-05,
+      "loss": 0.9075,
+      "step": 2433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287364303389537e-05,
+      "loss": 0.801,
+      "step": 2434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628420628160052e-05,
+      "loss": 0.7703,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281047223703625e-05,
+      "loss": 0.874,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627788713021969e-05,
+      "loss": 0.8586,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274726001669742e-05,
+      "loss": 0.811,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271563838574968e-05,
+      "loss": 0.8122,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268400641456738e-05,
+      "loss": 0.8887,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265236410836577e-05,
+      "loss": 0.8493,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262071147236182e-05,
+      "loss": 0.8217,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258904851177434e-05,
+      "loss": 0.8752,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255737523182377e-05,
+      "loss": 0.8947,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252569163773217e-05,
+      "loss": 0.8953,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249399773472342e-05,
+      "loss": 0.8807,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62462293528023e-05,
+      "loss": 0.8077,
+      "step": 2447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624305790228582e-05,
+      "loss": 0.8352,
+      "step": 2448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623988542244579e-05,
+      "loss": 0.9163,
+      "step": 2449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236711913805273e-05,
+      "loss": 0.9425,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62335373768875e-05,
+      "loss": 0.8645,
+      "step": 2451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230361812215878e-05,
+      "loss": 0.8299,
+      "step": 2452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227185220313973e-05,
+      "loss": 0.8485,
+      "step": 2453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224007601705522e-05,
+      "loss": 0.3311,
+      "step": 2454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220828956914444e-05,
+      "loss": 0.8342,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621764928646481e-05,
+      "loss": 0.8167,
+      "step": 2456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621446859088087e-05,
+      "loss": 0.8559,
+      "step": 2457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211286870687044e-05,
+      "loss": 0.9278,
+      "step": 2458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620810412640791e-05,
+      "loss": 0.8482,
+      "step": 2459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6204920358568233e-05,
+      "loss": 0.9187,
+      "step": 2460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620173556769293e-05,
+      "loss": 0.9466,
+      "step": 2461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619854975430709e-05,
+      "loss": 0.9053,
+      "step": 2462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619536291893598e-05,
+      "loss": 0.8298,
+      "step": 2463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619217506210503e-05,
+      "loss": 0.8627,
+      "step": 2464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618898618433983e-05,
+      "loss": 0.7528,
+      "step": 2465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185796286166154e-05,
+      "loss": 0.8535,
+      "step": 2466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182605368109928e-05,
+      "loss": 0.8281,
+      "step": 2467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179413430697265e-05,
+      "loss": 0.8312,
+      "step": 2468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176220474454424e-05,
+      "loss": 0.8626,
+      "step": 2469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617302649990785e-05,
+      "loss": 0.8662,
+      "step": 2470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169831507584152e-05,
+      "loss": 0.8331,
+      "step": 2471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166635498010092e-05,
+      "loss": 0.9277,
+      "step": 2472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163438471712627e-05,
+      "loss": 0.8295,
+      "step": 2473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6160240429218856e-05,
+      "loss": 0.8616,
+      "step": 2474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157041371056057e-05,
+      "loss": 0.8807,
+      "step": 2475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615384129775168e-05,
+      "loss": 0.9132,
+      "step": 2476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615064020983333e-05,
+      "loss": 0.9043,
+      "step": 2477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614743810782879e-05,
+      "loss": 0.8715,
+      "step": 2478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614423499226601e-05,
+      "loss": 0.9527,
+      "step": 2479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614103086367309e-05,
+      "loss": 0.904,
+      "step": 2480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137825722578327e-05,
+      "loss": 0.7649,
+      "step": 2481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134619569510156e-05,
+      "loss": 0.9205,
+      "step": 2482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131412404997195e-05,
+      "loss": 0.8882,
+      "step": 2483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128204229568233e-05,
+      "loss": 0.8918,
+      "step": 2484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61249950437522e-05,
+      "loss": 0.8287,
+      "step": 2485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6121784848078225e-05,
+      "loss": 0.9042,
+      "step": 2486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118573643075587e-05,
+      "loss": 0.892,
+      "step": 2487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115361429273724e-05,
+      "loss": 0.8166,
+      "step": 2488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611214820720226e-05,
+      "loss": 0.8784,
+      "step": 2489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108933977390973e-05,
+      "loss": 0.3223,
+      "step": 2490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61057187403698e-05,
+      "loss": 0.8513,
+      "step": 2491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610250249666886e-05,
+      "loss": 0.8364,
+      "step": 2492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099285246818434e-05,
+      "loss": 0.7718,
+      "step": 2493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096066991348957e-05,
+      "loss": 0.8378,
+      "step": 2494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092847730791043e-05,
+      "loss": 0.8741,
+      "step": 2495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608962746567547e-05,
+      "loss": 0.8197,
+      "step": 2496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086406196533174e-05,
+      "loss": 0.386,
+      "step": 2497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083183923895264e-05,
+      "loss": 0.7817,
+      "step": 2498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079960648293016e-05,
+      "loss": 0.9401,
+      "step": 2499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6076736370257857e-05,
+      "loss": 0.8972,
+      "step": 2500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6073511090321398e-05,
+      "loss": 0.8368,
+      "step": 2501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070284809015405e-05,
+      "loss": 0.824,
+      "step": 2502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606705752687181e-05,
+      "loss": 0.8266,
+      "step": 2503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063829244422707e-05,
+      "loss": 0.8381,
+      "step": 2504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6060599962200365e-05,
+      "loss": 0.8401,
+      "step": 2505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605736968073721e-05,
+      "loss": 0.8348,
+      "step": 2506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605413840056583e-05,
+      "loss": 0.8473,
+      "step": 2507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605090612221899e-05,
+      "loss": 0.8854,
+      "step": 2508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6047672846229603e-05,
+      "loss": 0.7992,
+      "step": 2509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604443857313076e-05,
+      "loss": 0.8846,
+      "step": 2510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604120330345571e-05,
+      "loss": 0.8222,
+      "step": 2511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603796703773787e-05,
+      "loss": 0.8368,
+      "step": 2512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6034729776510817e-05,
+      "loss": 0.8026,
+      "step": 2513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603149152030829e-05,
+      "loss": 0.8797,
+      "step": 2514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028252269664207e-05,
+      "loss": 0.7877,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602501202511263e-05,
+      "loss": 0.9653,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021770787187794e-05,
+      "loss": 0.806,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60185285564241e-05,
+      "loss": 0.7885,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015285333356116e-05,
+      "loss": 0.8241,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012041118518558e-05,
+      "loss": 0.8578,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6008795912446325e-05,
+      "loss": 0.8027,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005549715674463e-05,
+      "loss": 0.8487,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600230252873819e-05,
+      "loss": 0.8879,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999054352172886e-05,
+      "loss": 0.8255,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995805186514098e-05,
+      "loss": 0.7929,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5992555032297526e-05,
+      "loss": 0.8274,
+      "step": 2526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598930389005904e-05,
+      "loss": 0.8957,
+      "step": 2527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598605176033467e-05,
+      "loss": 0.807,
+      "step": 2528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598279864366062e-05,
+      "loss": 0.8123,
+      "step": 2529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597954454057324e-05,
+      "loss": 0.8972,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976289451609044e-05,
+      "loss": 0.8626,
+      "step": 2531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597303337730473e-05,
+      "loss": 0.8449,
+      "step": 2532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969776318197127e-05,
+      "loss": 0.8972,
+      "step": 2533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596651827482325e-05,
+      "loss": 0.8018,
+      "step": 2534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963259247720272e-05,
+      "loss": 0.822,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5959999237425517e-05,
+      "loss": 0.7419,
+      "step": 2536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956738244476483e-05,
+      "loss": 0.8276,
+      "step": 2537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953476269410826e-05,
+      "loss": 0.8426,
+      "step": 2538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950213312766364e-05,
+      "loss": 0.856,
+      "step": 2539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5946949375081075e-05,
+      "loss": 0.3626,
+      "step": 2540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943684456893103e-05,
+      "loss": 0.3549,
+      "step": 2541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940418558740744e-05,
+      "loss": 0.8365,
+      "step": 2542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937151681162473e-05,
+      "loss": 0.8071,
+      "step": 2543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593388382469691e-05,
+      "loss": 0.8768,
+      "step": 2544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930614989882843e-05,
+      "loss": 0.9938,
+      "step": 2545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592734517725922e-05,
+      "loss": 0.8742,
+      "step": 2546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924074387365153e-05,
+      "loss": 0.8777,
+      "step": 2547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920802620739914e-05,
+      "loss": 0.3314,
+      "step": 2548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591752987792293e-05,
+      "loss": 0.822,
+      "step": 2549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914256159453804e-05,
+      "loss": 0.8768,
+      "step": 2550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591098146587228e-05,
+      "loss": 0.961,
+      "step": 2551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907705797718276e-05,
+      "loss": 0.8726,
+      "step": 2552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904429155531867e-05,
+      "loss": 0.8109,
+      "step": 2553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901151539853293e-05,
+      "loss": 0.8494,
+      "step": 2554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897872951222946e-05,
+      "loss": 0.7802,
+      "step": 2555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894593390181386e-05,
+      "loss": 0.8298,
+      "step": 2556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589131285726933e-05,
+      "loss": 0.7947,
+      "step": 2557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888031353027655e-05,
+      "loss": 0.8837,
+      "step": 2558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884748877997397e-05,
+      "loss": 0.8255,
+      "step": 2559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881465432719752e-05,
+      "loss": 0.8265,
+      "step": 2560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878181017736082e-05,
+      "loss": 0.8361,
+      "step": 2561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874895633587904e-05,
+      "loss": 0.887,
+      "step": 2562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5871609280816898e-05,
+      "loss": 0.999,
+      "step": 2563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868321959964895e-05,
+      "loss": 0.8474,
+      "step": 2564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865033671573896e-05,
+      "loss": 0.8827,
+      "step": 2565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861744416186052e-05,
+      "loss": 0.8411,
+      "step": 2566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858454194343685e-05,
+      "loss": 0.8185,
+      "step": 2567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855163006589265e-05,
+      "loss": 0.804,
+      "step": 2568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585187085346543e-05,
+      "loss": 0.9824,
+      "step": 2569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848577735514974e-05,
+      "loss": 0.8096,
+      "step": 2570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845283653280844e-05,
+      "loss": 0.876,
+      "step": 2571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841988607306156e-05,
+      "loss": 0.9681,
+      "step": 2572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583869259813418e-05,
+      "loss": 0.8237,
+      "step": 2573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5835395626308345e-05,
+      "loss": 0.868,
+      "step": 2574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832097692372238e-05,
+      "loss": 0.8591,
+      "step": 2575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828798796869607e-05,
+      "loss": 0.8142,
+      "step": 2576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825498940344354e-05,
+      "loss": 0.9061,
+      "step": 2577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822198123340547e-05,
+      "loss": 0.8425,
+      "step": 2578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818896346402402e-05,
+      "loss": 0.8634,
+      "step": 2579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815593610074304e-05,
+      "loss": 0.2878,
+      "step": 2580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581228991490079e-05,
+      "loss": 0.8945,
+      "step": 2581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808985261426558e-05,
+      "loss": 0.8224,
+      "step": 2582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805679650196456e-05,
+      "loss": 0.9175,
+      "step": 2583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802373081755504e-05,
+      "loss": 0.3813,
+      "step": 2584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579906555664887e-05,
+      "loss": 0.8091,
+      "step": 2585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579575707542188e-05,
+      "loss": 0.908,
+      "step": 2586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579244763862002e-05,
+      "loss": 0.8352,
+      "step": 2587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789137246788933e-05,
+      "loss": 0.8658,
+      "step": 2588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785825900474427e-05,
+      "loss": 0.8048,
+      "step": 2589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782513600222443e-05,
+      "loss": 0.8265,
+      "step": 2590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779200346579112e-05,
+      "loss": 0.833,
+      "step": 2591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57758861400907e-05,
+      "loss": 0.8423,
+      "step": 2592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772570981303634e-05,
+      "loss": 0.8401,
+      "step": 2593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576925487076451e-05,
+      "loss": 0.8473,
+      "step": 2594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576593780902006e-05,
+      "loss": 0.8445,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762619796617186e-05,
+      "loss": 0.879,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759300834102952e-05,
+      "loss": 0.9064,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755980922024567e-05,
+      "loss": 0.8702,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752660060929407e-05,
+      "loss": 0.3602,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749338251364985e-05,
+      "loss": 0.8397,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746015493878998e-05,
+      "loss": 0.852,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742691789019277e-05,
+      "loss": 0.8003,
+      "step": 2602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573936713733382e-05,
+      "loss": 0.7999,
+      "step": 2603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736041539370783e-05,
+      "loss": 0.8115,
+      "step": 2604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732714995678465e-05,
+      "loss": 0.9445,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729387506805335e-05,
+      "loss": 0.7997,
+      "step": 2606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726059073300015e-05,
+      "loss": 0.8633,
+      "step": 2607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722729695711274e-05,
+      "loss": 0.8204,
+      "step": 2608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719399374588044e-05,
+      "loss": 0.8303,
+      "step": 2609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5716068110479418e-05,
+      "loss": 0.8339,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712735903934627e-05,
+      "loss": 0.8566,
+      "step": 2611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709402755503076e-05,
+      "loss": 0.8806,
+      "step": 2612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706068665734314e-05,
+      "loss": 0.8288,
+      "step": 2613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702733635178052e-05,
+      "loss": 0.896,
+      "step": 2614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699397664384152e-05,
+      "loss": 0.8787,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5696060753902628e-05,
+      "loss": 0.3308,
+      "step": 2616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692722904283657e-05,
+      "loss": 0.8333,
+      "step": 2617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568938411607757e-05,
+      "loss": 0.7819,
+      "step": 2618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686044389834838e-05,
+      "loss": 0.8838,
+      "step": 2619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682703726106105e-05,
+      "loss": 0.9241,
+      "step": 2620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679362125442165e-05,
+      "loss": 0.8149,
+      "step": 2621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567601958839396e-05,
+      "loss": 0.803,
+      "step": 2622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672676115512594e-05,
+      "loss": 0.8187,
+      "step": 2623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669331707349317e-05,
+      "loss": 0.8537,
+      "step": 2624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566598636445554e-05,
+      "loss": 0.795,
+      "step": 2625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662640087382827e-05,
+      "loss": 0.9088,
+      "step": 2626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659292876682894e-05,
+      "loss": 0.8009,
+      "step": 2627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565594473290761e-05,
+      "loss": 0.8105,
+      "step": 2628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565259565660901e-05,
+      "loss": 0.8614,
+      "step": 2629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564924564833926e-05,
+      "loss": 0.8697,
+      "step": 2630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56458947086507e-05,
+      "loss": 0.8028,
+      "step": 2631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642542838095814e-05,
+      "loss": 0.8396,
+      "step": 2632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639190037227237e-05,
+      "loss": 0.8693,
+      "step": 2633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563583630659777e-05,
+      "loss": 0.8736,
+      "step": 2634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563248164676035e-05,
+      "loss": 0.8927,
+      "step": 2635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562912605826808e-05,
+      "loss": 0.842,
+      "step": 2636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625769541674216e-05,
+      "loss": 0.8649,
+      "step": 2637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622412097532162e-05,
+      "loss": 0.8438,
+      "step": 2638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619053726395468e-05,
+      "loss": 0.8084,
+      "step": 2639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615694428817857e-05,
+      "loss": 0.8996,
+      "step": 2640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612334205353183e-05,
+      "loss": 0.7602,
+      "step": 2641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608973056555474e-05,
+      "loss": 0.9077,
+      "step": 2642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605610982978885e-05,
+      "loss": 0.8586,
+      "step": 2643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602247985177746e-05,
+      "loss": 0.9116,
+      "step": 2644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559888406370653e-05,
+      "loss": 0.8375,
+      "step": 2645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5595519219119863e-05,
+      "loss": 0.8316,
+      "step": 2646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592153451972523e-05,
+      "loss": 0.8246,
+      "step": 2647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588786762819445e-05,
+      "loss": 0.8379,
+      "step": 2648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55854191522157e-05,
+      "loss": 0.8627,
+      "step": 2649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5582050620716533e-05,
+      "loss": 0.9476,
+      "step": 2650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578681168877328e-05,
+      "loss": 0.872,
+      "step": 2651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557531079725362e-05,
+      "loss": 0.9455,
+      "step": 2652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571939506401103e-05,
+      "loss": 0.917,
+      "step": 2653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568567296875614e-05,
+      "loss": 0.8022,
+      "step": 2654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565194169233146e-05,
+      "loss": 0.8047,
+      "step": 2655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561820124029847e-05,
+      "loss": 0.9411,
+      "step": 2656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555844516182201e-05,
+      "loss": 0.862,
+      "step": 2657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555069283166084e-05,
+      "loss": 0.8861,
+      "step": 2658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551692488618657e-05,
+      "loss": 0.7979,
+      "step": 2659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5548314778736487e-05,
+      "loss": 0.7857,
+      "step": 2660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544936154076478e-05,
+      "loss": 0.8281,
+      "step": 2661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554155661519567e-05,
+      "loss": 0.9053,
+      "step": 2662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538176162651265e-05,
+      "loss": 0.8223,
+      "step": 2663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534794797000625e-05,
+      "loss": 0.2939,
+      "step": 2664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531412518801237e-05,
+      "loss": 0.8041,
+      "step": 2665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552802932861077e-05,
+      "loss": 0.8411,
+      "step": 2666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552464522698701e-05,
+      "loss": 0.8651,
+      "step": 2667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521260214487926e-05,
+      "loss": 0.8271,
+      "step": 2668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517874291671613e-05,
+      "loss": 0.8146,
+      "step": 2669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514487459096326e-05,
+      "loss": 0.8027,
+      "step": 2670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551109971732047e-05,
+      "loss": 0.3479,
+      "step": 2671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55077110669026e-05,
+      "loss": 0.8565,
+      "step": 2672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504321508401417e-05,
+      "loss": 0.8888,
+      "step": 2673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550093104237577e-05,
+      "loss": 0.8681,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497539669384675e-05,
+      "loss": 0.7975,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5494147389987267e-05,
+      "loss": 0.8114,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549075420474286e-05,
+      "loss": 0.8615,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487360114210905e-05,
+      "loss": 0.9144,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483965118950992e-05,
+      "loss": 0.8355,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5480569219522884e-05,
+      "loss": 0.9862,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477172416486464e-05,
+      "loss": 0.8299,
+      "step": 2681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5473774710401796e-05,
+      "loss": 0.8881,
+      "step": 2682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470376101829065e-05,
+      "loss": 0.8568,
+      "step": 2683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466976591328622e-05,
+      "loss": 0.7994,
+      "step": 2684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546357617946096e-05,
+      "loss": 0.8138,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5460174866786718e-05,
+      "loss": 0.8064,
+      "step": 2686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456772653866688e-05,
+      "loss": 0.8271,
+      "step": 2687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453369541261814e-05,
+      "loss": 0.8042,
+      "step": 2688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544996552953318e-05,
+      "loss": 0.8581,
+      "step": 2689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544656061924203e-05,
+      "loss": 0.75,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443154810949736e-05,
+      "loss": 0.8465,
+      "step": 2691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543974810521783e-05,
+      "loss": 0.8783,
+      "step": 2692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436340502608005e-05,
+      "loss": 0.8959,
+      "step": 2693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432932003682083e-05,
+      "loss": 0.8574,
+      "step": 2694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429522609002034e-05,
+      "loss": 0.8627,
+      "step": 2695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426112319129987e-05,
+      "loss": 0.7194,
+      "step": 2696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422701134628216e-05,
+      "loss": 0.9116,
+      "step": 2697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419289056059137e-05,
+      "loss": 0.837,
+      "step": 2698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415876083985313e-05,
+      "loss": 0.8806,
+      "step": 2699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541246221896946e-05,
+      "loss": 0.7951,
+      "step": 2700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5409047461574443e-05,
+      "loss": 0.9218,
+      "step": 2701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540563181236326e-05,
+      "loss": 0.9003,
+      "step": 2702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540221527189907e-05,
+      "loss": 0.8188,
+      "step": 2703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398797840745182e-05,
+      "loss": 0.8241,
+      "step": 2704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539537951946504e-05,
+      "loss": 0.8893,
+      "step": 2705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391960308622233e-05,
+      "loss": 0.7863,
+      "step": 2706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388540208780512e-05,
+      "loss": 0.7856,
+      "step": 2707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538511922050376e-05,
+      "loss": 0.9353,
+      "step": 2708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381697344356014e-05,
+      "loss": 0.8833,
+      "step": 2709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537827458090145e-05,
+      "loss": 0.8301,
+      "step": 2710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374850930704407e-05,
+      "loss": 0.8881,
+      "step": 2711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371426394329352e-05,
+      "loss": 0.7727,
+      "step": 2712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368000972340907e-05,
+      "loss": 0.9616,
+      "step": 2713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536457466530384e-05,
+      "loss": 0.3423,
+      "step": 2714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361147473783057e-05,
+      "loss": 0.8259,
+      "step": 2715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535771939834362e-05,
+      "loss": 0.8215,
+      "step": 2716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535429043955073e-05,
+      "loss": 0.8356,
+      "step": 2717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535086059796974e-05,
+      "loss": 0.9014,
+      "step": 2718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534742987416614e-05,
+      "loss": 0.8377,
+      "step": 2719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343998268705575e-05,
+      "loss": 0.808,
+      "step": 2720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534056578215383e-05,
+      "loss": 0.9014,
+      "step": 2721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5337132415076834e-05,
+      "loss": 0.7963,
+      "step": 2722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333698168040664e-05,
+      "loss": 0.8706,
+      "step": 2723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533026304161154e-05,
+      "loss": 0.8751,
+      "step": 2724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532682703635583e-05,
+      "loss": 0.9007,
+      "step": 2725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532339015284004e-05,
+      "loss": 0.8292,
+      "step": 2726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319952391630834e-05,
+      "loss": 0.9472,
+      "step": 2727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316513753295006e-05,
+      "loss": 0.8864,
+      "step": 2728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313074238399503e-05,
+      "loss": 0.8826,
+      "step": 2729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530963384751142e-05,
+      "loss": 0.8066,
+      "step": 2730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306192581197983e-05,
+      "loss": 0.8361,
+      "step": 2731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5302750440026574e-05,
+      "loss": 0.8595,
+      "step": 2732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529930742456471e-05,
+      "loss": 0.859,
+      "step": 2733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295863535380072e-05,
+      "loss": 0.9378,
+      "step": 2734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292418773040457e-05,
+      "loss": 0.3057,
+      "step": 2735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288973138113826e-05,
+      "loss": 0.8129,
+      "step": 2736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285526631168273e-05,
+      "loss": 0.8371,
+      "step": 2737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528207925277205e-05,
+      "loss": 0.7961,
+      "step": 2738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278631003493534e-05,
+      "loss": 0.7999,
+      "step": 2739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275181883901256e-05,
+      "loss": 0.9581,
+      "step": 2740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271731894563896e-05,
+      "loss": 0.7633,
+      "step": 2741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268281036050267e-05,
+      "loss": 0.9028,
+      "step": 2742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264829308929324e-05,
+      "loss": 0.8339,
+      "step": 2743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5261376713770176e-05,
+      "loss": 0.8769,
+      "step": 2744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257923251142063e-05,
+      "loss": 0.7734,
+      "step": 2745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254468921614383e-05,
+      "loss": 0.7036,
+      "step": 2746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251013725756663e-05,
+      "loss": 0.9097,
+      "step": 2747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524755766413858e-05,
+      "loss": 0.7798,
+      "step": 2748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524410073732995e-05,
+      "loss": 0.8611,
+      "step": 2749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5240642945900734e-05,
+      "loss": 0.8963,
+      "step": 2750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237184290421035e-05,
+      "loss": 0.8261,
+      "step": 2751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52337247714611e-05,
+      "loss": 0.7004,
+      "step": 2752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230264389591312e-05,
+      "loss": 0.3168,
+      "step": 2753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226803145382207e-05,
+      "loss": 0.8419,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223341039404456e-05,
+      "loss": 0.8922,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521987807222887e-05,
+      "loss": 0.8776,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216414244426413e-05,
+      "loss": 0.86,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521294955656817e-05,
+      "loss": 0.8936,
+      "step": 2758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209484009225395e-05,
+      "loss": 0.872,
+      "step": 2759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206017602969463e-05,
+      "loss": 0.905,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202550338371897e-05,
+      "loss": 0.8684,
+      "step": 2761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199082216004361e-05,
+      "loss": 0.8681,
+      "step": 2762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195613236438669e-05,
+      "loss": 0.8742,
+      "step": 2763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519214340024676e-05,
+      "loss": 0.9247,
+      "step": 2764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188672708000725e-05,
+      "loss": 0.8664,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185201160272795e-05,
+      "loss": 0.917,
+      "step": 2766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518172875763534e-05,
+      "loss": 0.8016,
+      "step": 2767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178255500660877e-05,
+      "loss": 0.7846,
+      "step": 2768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174781389922055e-05,
+      "loss": 0.8284,
+      "step": 2769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171306425991663e-05,
+      "loss": 0.8363,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167830609442646e-05,
+      "loss": 0.792,
+      "step": 2771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164353940848068e-05,
+      "loss": 0.7646,
+      "step": 2772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516087642078115e-05,
+      "loss": 0.8149,
+      "step": 2773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515739804981525e-05,
+      "loss": 0.8455,
+      "step": 2774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515391882852386e-05,
+      "loss": 0.9005,
+      "step": 2775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150438757480616e-05,
+      "loss": 0.8863,
+      "step": 2776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146957837259297e-05,
+      "loss": 0.8415,
+      "step": 2777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143476068433817e-05,
+      "loss": 0.9067,
+      "step": 2778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139993451578236e-05,
+      "loss": 0.907,
+      "step": 2779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136509987266746e-05,
+      "loss": 0.8307,
+      "step": 2780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133025676073683e-05,
+      "loss": 0.8577,
+      "step": 2781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129540518573529e-05,
+      "loss": 0.8864,
+      "step": 2782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126054515340886e-05,
+      "loss": 0.8197,
+      "step": 2783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122567666950524e-05,
+      "loss": 0.8059,
+      "step": 2784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119079973977324e-05,
+      "loss": 0.8002,
+      "step": 2785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115591436996327e-05,
+      "loss": 0.9962,
+      "step": 2786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112102056582703e-05,
+      "loss": 0.894,
+      "step": 2787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108611833311759e-05,
+      "loss": 0.8299,
+      "step": 2788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510512076775895e-05,
+      "loss": 0.7927,
+      "step": 2789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101628860499863e-05,
+      "loss": 0.8381,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098136112110227e-05,
+      "loss": 0.8539,
+      "step": 2791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094642523165908e-05,
+      "loss": 0.8643,
+      "step": 2792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091148094242913e-05,
+      "loss": 0.3367,
+      "step": 2793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087652825917385e-05,
+      "loss": 0.7726,
+      "step": 2794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084156718765601e-05,
+      "loss": 0.8415,
+      "step": 2795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080659773363989e-05,
+      "loss": 0.7725,
+      "step": 2796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077161990289102e-05,
+      "loss": 0.8231,
+      "step": 2797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507366337011764e-05,
+      "loss": 0.8563,
+      "step": 2798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070163913426435e-05,
+      "loss": 0.7826,
+      "step": 2799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066663620792463e-05,
+      "loss": 0.9016,
+      "step": 2800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063162492792826e-05,
+      "loss": 0.8357,
+      "step": 2801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059660530004783e-05,
+      "loss": 0.8395,
+      "step": 2802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056157733005716e-05,
+      "loss": 0.9164,
+      "step": 2803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052654102373146e-05,
+      "loss": 0.7826,
+      "step": 2804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049149638684735e-05,
+      "loss": 0.9044,
+      "step": 2805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045644342518281e-05,
+      "loss": 0.8483,
+      "step": 2806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042138214451719e-05,
+      "loss": 0.8032,
+      "step": 2807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038631255063123e-05,
+      "loss": 0.8662,
+      "step": 2808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50351234649307e-05,
+      "loss": 0.8252,
+      "step": 2809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031614844632798e-05,
+      "loss": 0.8721,
+      "step": 2810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028105394747906e-05,
+      "loss": 0.8546,
+      "step": 2811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502459511585463e-05,
+      "loss": 0.7516,
+      "step": 2812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021084008531742e-05,
+      "loss": 0.7913,
+      "step": 2813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017572073358127e-05,
+      "loss": 0.7873,
+      "step": 2814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014059310912817e-05,
+      "loss": 0.79,
+      "step": 2815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501054572177498e-05,
+      "loss": 0.8301,
+      "step": 2816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007031306523916e-05,
+      "loss": 0.8958,
+      "step": 2817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003516065739066e-05,
+      "loss": 0.8532,
+      "step": 2818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8636,
+      "step": 2819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996483109886438e-05,
+      "loss": 0.7755,
+      "step": 2820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992965395978219e-05,
+      "loss": 0.7776,
+      "step": 2821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498944685885533e-05,
+      "loss": 0.8467,
+      "step": 2822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985927499097885e-05,
+      "loss": 0.9088,
+      "step": 2823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982407317286146e-05,
+      "loss": 0.9159,
+      "step": 2824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978886314000494e-05,
+      "loss": 0.843,
+      "step": 2825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497536448982146e-05,
+      "loss": 0.9743,
+      "step": 2826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4971841845329701e-05,
+      "loss": 0.859,
+      "step": 2827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968318381106013e-05,
+      "loss": 0.865,
+      "step": 2828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964794097731327e-05,
+      "loss": 0.8706,
+      "step": 2829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961268995786715e-05,
+      "loss": 0.8891,
+      "step": 2830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957743075853368e-05,
+      "loss": 0.7385,
+      "step": 2831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954216338512628e-05,
+      "loss": 0.8876,
+      "step": 2832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950688784345961e-05,
+      "loss": 0.8655,
+      "step": 2833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947160413934976e-05,
+      "loss": 0.8518,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943631227861412e-05,
+      "loss": 0.8736,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940101226707142e-05,
+      "loss": 0.8865,
+      "step": 2836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936570411054174e-05,
+      "loss": 0.8787,
+      "step": 2837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933038781484655e-05,
+      "loss": 0.8311,
+      "step": 2838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929506338580858e-05,
+      "loss": 0.7978,
+      "step": 2839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925973082925197e-05,
+      "loss": 0.8263,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922439015100212e-05,
+      "loss": 0.8606,
+      "step": 2841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918904135688586e-05,
+      "loss": 0.8476,
+      "step": 2842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491536844527313e-05,
+      "loss": 0.8435,
+      "step": 2843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911831944436796e-05,
+      "loss": 0.8185,
+      "step": 2844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490829463376266e-05,
+      "loss": 0.8296,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904756513833933e-05,
+      "loss": 0.9022,
+      "step": 2846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490121758523397e-05,
+      "loss": 0.8909,
+      "step": 2847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897677848546241e-05,
+      "loss": 0.8416,
+      "step": 2848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894137304354367e-05,
+      "loss": 0.9771,
+      "step": 2849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890595953242096e-05,
+      "loss": 0.8553,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887053795793305e-05,
+      "loss": 0.9038,
+      "step": 2851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883510832592003e-05,
+      "loss": 0.8665,
+      "step": 2852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879967064222343e-05,
+      "loss": 0.9099,
+      "step": 2853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48764224912686e-05,
+      "loss": 0.838,
+      "step": 2854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872877114315185e-05,
+      "loss": 0.8397,
+      "step": 2855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869330933946641e-05,
+      "loss": 0.8628,
+      "step": 2856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486578395074765e-05,
+      "loss": 0.8195,
+      "step": 2857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486223616530301e-05,
+      "loss": 0.9295,
+      "step": 2858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485868757819767e-05,
+      "loss": 0.8792,
+      "step": 2859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48551381900167e-05,
+      "loss": 0.8567,
+      "step": 2860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851588001345306e-05,
+      "loss": 0.7842,
+      "step": 2861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848037012768824e-05,
+      "loss": 0.9148,
+      "step": 2862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844485224872721e-05,
+      "loss": 0.8418,
+      "step": 2863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840932638242602e-05,
+      "loss": 0.8867,
+      "step": 2864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48373792534642e-05,
+      "loss": 0.8062,
+      "step": 2865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833825071123373e-05,
+      "loss": 0.7765,
+      "step": 2866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483027009180612e-05,
+      "loss": 0.8058,
+      "step": 2867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826714316098567e-05,
+      "loss": 0.8705,
+      "step": 2868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823157744586975e-05,
+      "loss": 0.8009,
+      "step": 2869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481960037785773e-05,
+      "loss": 0.846,
+      "step": 2870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816042216497356e-05,
+      "loss": 0.8728,
+      "step": 2871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812483261092496e-05,
+      "loss": 0.8843,
+      "step": 2872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808923512229945e-05,
+      "loss": 0.2989,
+      "step": 2873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805362970496607e-05,
+      "loss": 0.85,
+      "step": 2874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801801636479529e-05,
+      "loss": 0.7876,
+      "step": 2875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798239510765883e-05,
+      "loss": 0.8939,
+      "step": 2876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794676593942979e-05,
+      "loss": 0.8245,
+      "step": 2877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791112886598252e-05,
+      "loss": 0.8128,
+      "step": 2878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787548389319262e-05,
+      "loss": 0.8618,
+      "step": 2879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783983102693712e-05,
+      "loss": 0.8864,
+      "step": 2880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780417027309421e-05,
+      "loss": 0.8232,
+      "step": 2881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776850163754355e-05,
+      "loss": 0.78,
+      "step": 2882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773282512616591e-05,
+      "loss": 0.8678,
+      "step": 2883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476971407448435e-05,
+      "loss": 0.7596,
+      "step": 2884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766144849945978e-05,
+      "loss": 0.9126,
+      "step": 2885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762574839589952e-05,
+      "loss": 0.768,
+      "step": 2886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759004044004869e-05,
+      "loss": 0.3287,
+      "step": 2887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755432463779472e-05,
+      "loss": 0.8742,
+      "step": 2888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751860099502621e-05,
+      "loss": 0.897,
+      "step": 2889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474828695176331e-05,
+      "loss": 0.8525,
+      "step": 2890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744713021150665e-05,
+      "loss": 0.8831,
+      "step": 2891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474113830825393e-05,
+      "loss": 0.7754,
+      "step": 2892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737562813662492e-05,
+      "loss": 0.8229,
+      "step": 2893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473398653796586e-05,
+      "loss": 0.8337,
+      "step": 2894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730409481753665e-05,
+      "loss": 0.8365,
+      "step": 2895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472683164561568e-05,
+      "loss": 0.8266,
+      "step": 2896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723253030141801e-05,
+      "loss": 0.8319,
+      "step": 2897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719673635922047e-05,
+      "loss": 0.9578,
+      "step": 2898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716093463546575e-05,
+      "loss": 0.3513,
+      "step": 2899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712512513605666e-05,
+      "loss": 0.8704,
+      "step": 2900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708930786689723e-05,
+      "loss": 0.8613,
+      "step": 2901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705348283389287e-05,
+      "loss": 0.9751,
+      "step": 2902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701765004295026e-05,
+      "loss": 0.7962,
+      "step": 2903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698180949997725e-05,
+      "loss": 0.8379,
+      "step": 2904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694596121088309e-05,
+      "loss": 0.9357,
+      "step": 2905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4691010518157826e-05,
+      "loss": 0.8509,
+      "step": 2906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687424141797456e-05,
+      "loss": 0.8861,
+      "step": 2907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683836992598492e-05,
+      "loss": 0.8547,
+      "step": 2908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680249071152375e-05,
+      "loss": 0.8778,
+      "step": 2909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467666037805066e-05,
+      "loss": 0.8321,
+      "step": 2910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4673070913885031e-05,
+      "loss": 0.8013,
+      "step": 2911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4669480679247299e-05,
+      "loss": 0.8576,
+      "step": 2912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466588967472941e-05,
+      "loss": 0.3613,
+      "step": 2913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662297900923423e-05,
+      "loss": 0.3314,
+      "step": 2914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658705358421539e-05,
+      "loss": 0.7918,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4655112047816071e-05,
+      "loss": 0.8158,
+      "step": 2916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651517969699468e-05,
+      "loss": 0.8756,
+      "step": 2917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647923124664305e-05,
+      "loss": 0.8794,
+      "step": 2918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644327513303281e-05,
+      "loss": 0.8647,
+      "step": 2919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464073113620922e-05,
+      "loss": 0.9161,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637133993975079e-05,
+      "loss": 0.8503,
+      "step": 2921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633536087193929e-05,
+      "loss": 1.0003,
+      "step": 2922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629937416458981e-05,
+      "loss": 0.8424,
+      "step": 2923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626337982363566e-05,
+      "loss": 0.8452,
+      "step": 2924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622737785501136e-05,
+      "loss": 0.837,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619136826465277e-05,
+      "loss": 0.8051,
+      "step": 2926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615535105849689e-05,
+      "loss": 0.8179,
+      "step": 2927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611932624248216e-05,
+      "loss": 0.8569,
+      "step": 2928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608329382254812e-05,
+      "loss": 0.8222,
+      "step": 2929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604725380463559e-05,
+      "loss": 0.8457,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601120619468667e-05,
+      "loss": 0.7985,
+      "step": 2931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597515099864474e-05,
+      "loss": 0.8232,
+      "step": 2932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593908822245437e-05,
+      "loss": 0.8546,
+      "step": 2933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590301787206142e-05,
+      "loss": 0.7758,
+      "step": 2934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45866939953413e-05,
+      "loss": 0.8407,
+      "step": 2935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583085447245737e-05,
+      "loss": 0.8823,
+      "step": 2936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579476143514425e-05,
+      "loss": 0.861,
+      "step": 2937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575866084742435e-05,
+      "loss": 0.8175,
+      "step": 2938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572255271524985e-05,
+      "loss": 0.8936,
+      "step": 2939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568643704457404e-05,
+      "loss": 0.9433,
+      "step": 2940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565031384135144e-05,
+      "loss": 0.9139,
+      "step": 2941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561418311153794e-05,
+      "loss": 0.8101,
+      "step": 2942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557804486109053e-05,
+      "loss": 0.3206,
+      "step": 2943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554189909596753e-05,
+      "loss": 0.9239,
+      "step": 2944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550574582212846e-05,
+      "loss": 0.7283,
+      "step": 2945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546958504553408e-05,
+      "loss": 0.8456,
+      "step": 2946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454334167721464e-05,
+      "loss": 0.8967,
+      "step": 2947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539724100792867e-05,
+      "loss": 0.8675,
+      "step": 2948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536105775884534e-05,
+      "loss": 0.8222,
+      "step": 2949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4532486703086219e-05,
+      "loss": 0.3232,
+      "step": 2950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528866882994608e-05,
+      "loss": 0.8344,
+      "step": 2951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452524631620652e-05,
+      "loss": 0.8256,
+      "step": 2952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45216250033189e-05,
+      "loss": 0.784,
+      "step": 2953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518002944928807e-05,
+      "loss": 0.8678,
+      "step": 2954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451438014163343e-05,
+      "loss": 0.8358,
+      "step": 2955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510756594030077e-05,
+      "loss": 0.8225,
+      "step": 2956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507132302716183e-05,
+      "loss": 0.7782,
+      "step": 2957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503507268289297e-05,
+      "loss": 0.8311,
+      "step": 2958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499881491347099e-05,
+      "loss": 0.8705,
+      "step": 2959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496254972487393e-05,
+      "loss": 0.8245,
+      "step": 2960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492627712308094e-05,
+      "loss": 0.8303,
+      "step": 2961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448899971140725e-05,
+      "loss": 0.8774,
+      "step": 2962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485370970383028e-05,
+      "loss": 0.8958,
+      "step": 2963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481741489833713e-05,
+      "loss": 0.8317,
+      "step": 2964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478111270357717e-05,
+      "loss": 0.8881,
+      "step": 2965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474480312553576e-05,
+      "loss": 0.7389,
+      "step": 2966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470848617019934e-05,
+      "loss": 0.8348,
+      "step": 2967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467216184355577e-05,
+      "loss": 0.8447,
+      "step": 2968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4463583015159394e-05,
+      "loss": 0.8489,
+      "step": 2969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445994911003041e-05,
+      "loss": 0.8342,
+      "step": 2970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456314469567758e-05,
+      "loss": 0.8956,
+      "step": 2971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452679094370708e-05,
+      "loss": 0.8622,
+      "step": 2972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4449042985038633e-05,
+      "loss": 0.7859,
+      "step": 2973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445406142171042e-05,
+      "loss": 0.9077,
+      "step": 2974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441768566367554e-05,
+      "loss": 0.8065,
+      "step": 2975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438130258227916e-05,
+      "loss": 0.8468,
+      "step": 2976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434491218351998e-05,
+      "loss": 0.8364,
+      "step": 2977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443085144733978e-05,
+      "loss": 0.7746,
+      "step": 2978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427210945791372e-05,
+      "loss": 0.9385,
+      "step": 2979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423569714307005e-05,
+      "loss": 0.8083,
+      "step": 2980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419927753487018e-05,
+      "loss": 0.8505,
+      "step": 2981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4416285063931887e-05,
+      "loss": 0.8605,
+      "step": 2982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412641646242195e-05,
+      "loss": 0.7693,
+      "step": 2983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440899750101865e-05,
+      "loss": 0.8658,
+      "step": 2984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405352628862086e-05,
+      "loss": 0.8186,
+      "step": 2985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401707030373447e-05,
+      "loss": 0.7597,
+      "step": 2986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4398060706153796e-05,
+      "loss": 0.8387,
+      "step": 2987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394413656804329e-05,
+      "loss": 0.8607,
+      "step": 2988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4390765882926348e-05,
+      "loss": 0.8846,
+      "step": 2989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387117385121278e-05,
+      "loss": 0.7799,
+      "step": 2990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438346816399067e-05,
+      "loss": 0.8088,
+      "step": 2991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379818220136184e-05,
+      "loss": 0.8542,
+      "step": 2992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376167554159608e-05,
+      "loss": 0.8424,
+      "step": 2993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437251616666284e-05,
+      "loss": 0.8811,
+      "step": 2994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368864058247907e-05,
+      "loss": 0.8427,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365211229516951e-05,
+      "loss": 0.7984,
+      "step": 2996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361557681072224e-05,
+      "loss": 0.8004,
+      "step": 2997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357903413516111e-05,
+      "loss": 0.8466,
+      "step": 2998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435424842745111e-05,
+      "loss": 0.8919,
+      "step": 2999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350592723479832e-05,
+      "loss": 0.9017,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346936302205015e-05,
+      "loss": 0.8939,
+      "step": 3001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343279164229507e-05,
+      "loss": 0.7414,
+      "step": 3002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433962131015628e-05,
+      "loss": 0.7909,
+      "step": 3003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335962740588425e-05,
+      "loss": 0.8662,
+      "step": 3004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332303456129144e-05,
+      "loss": 0.8646,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328643457381762e-05,
+      "loss": 0.8102,
+      "step": 3006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324982744949724e-05,
+      "loss": 0.8926,
+      "step": 3007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321321319436586e-05,
+      "loss": 0.7355,
+      "step": 3008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317659181446029e-05,
+      "loss": 0.8283,
+      "step": 3009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4313996331581841e-05,
+      "loss": 0.8108,
+      "step": 3010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310332770447942e-05,
+      "loss": 0.8029,
+      "step": 3011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306668498648354e-05,
+      "loss": 0.825,
+      "step": 3012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430300351678723e-05,
+      "loss": 0.9051,
+      "step": 3013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429933782546883e-05,
+      "loss": 0.9133,
+      "step": 3014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295671425297535e-05,
+      "loss": 0.8929,
+      "step": 3015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292004316877843e-05,
+      "loss": 0.8271,
+      "step": 3016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288336500814366e-05,
+      "loss": 0.8055,
+      "step": 3017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284667977711837e-05,
+      "loss": 0.8088,
+      "step": 3018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280998748175104e-05,
+      "loss": 0.7439,
+      "step": 3019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277328812809128e-05,
+      "loss": 0.8897,
+      "step": 3020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273658172218992e-05,
+      "loss": 0.8638,
+      "step": 3021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426998682700989e-05,
+      "loss": 0.8796,
+      "step": 3022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266314777787136e-05,
+      "loss": 0.3061,
+      "step": 3023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426264202515616e-05,
+      "loss": 0.8552,
+      "step": 3024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258968569722505e-05,
+      "loss": 0.7944,
+      "step": 3025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255294412091831e-05,
+      "loss": 0.7832,
+      "step": 3026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251619552869915e-05,
+      "loss": 0.8644,
+      "step": 3027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247943992662653e-05,
+      "loss": 0.8535,
+      "step": 3028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244267732076046e-05,
+      "loss": 0.8017,
+      "step": 3029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424059077171622e-05,
+      "loss": 0.8183,
+      "step": 3030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236913112189417e-05,
+      "loss": 0.8577,
+      "step": 3031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233234754101985e-05,
+      "loss": 0.909,
+      "step": 3032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229555698060398e-05,
+      "loss": 0.8312,
+      "step": 3033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225875944671232e-05,
+      "loss": 0.8036,
+      "step": 3034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222195494541193e-05,
+      "loss": 0.8853,
+      "step": 3035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218514348277098e-05,
+      "loss": 0.7927,
+      "step": 3036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214832506485866e-05,
+      "loss": 0.3189,
+      "step": 3037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211149969774544e-05,
+      "loss": 0.8564,
+      "step": 3038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207466738750294e-05,
+      "loss": 0.8469,
+      "step": 3039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203782814020386e-05,
+      "loss": 0.8208,
+      "step": 3040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200098196192205e-05,
+      "loss": 0.9001,
+      "step": 3041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196412885873251e-05,
+      "loss": 0.8288,
+      "step": 3042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192726883671144e-05,
+      "loss": 0.8331,
+      "step": 3043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418904019019361e-05,
+      "loss": 0.804,
+      "step": 3044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418535280604849e-05,
+      "loss": 0.8168,
+      "step": 3045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181664731843746e-05,
+      "loss": 0.9164,
+      "step": 3046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177975968187448e-05,
+      "loss": 0.8977,
+      "step": 3047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4174286515687776e-05,
+      "loss": 0.7889,
+      "step": 3048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170596374953034e-05,
+      "loss": 0.7998,
+      "step": 3049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166905546591627e-05,
+      "loss": 0.7927,
+      "step": 3050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163214031212087e-05,
+      "loss": 0.8169,
+      "step": 3051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159521829423049e-05,
+      "loss": 0.8042,
+      "step": 3052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155828941833263e-05,
+      "loss": 0.379,
+      "step": 3053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152135369051596e-05,
+      "loss": 0.8667,
+      "step": 3054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148441111687024e-05,
+      "loss": 0.9205,
+      "step": 3055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144746170348637e-05,
+      "loss": 0.8521,
+      "step": 3056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141050545645642e-05,
+      "loss": 0.8769,
+      "step": 3057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137354238187349e-05,
+      "loss": 0.9149,
+      "step": 3058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133657248583186e-05,
+      "loss": 0.3442,
+      "step": 3059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129959577442702e-05,
+      "loss": 0.8749,
+      "step": 3060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126261225375542e-05,
+      "loss": 1.04,
+      "step": 3061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122562192991473e-05,
+      "loss": 0.8655,
+      "step": 3062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118862480900376e-05,
+      "loss": 0.9045,
+      "step": 3063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115162089712237e-05,
+      "loss": 0.7781,
+      "step": 3064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111461020037158e-05,
+      "loss": 0.8687,
+      "step": 3065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410775927248536e-05,
+      "loss": 0.8143,
+      "step": 3066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104056847667154e-05,
+      "loss": 0.8984,
+      "step": 3067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100353746192987e-05,
+      "loss": 0.8565,
+      "step": 3068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096649968673408e-05,
+      "loss": 0.8573,
+      "step": 3069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092945515719073e-05,
+      "loss": 0.7212,
+      "step": 3070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089240387940753e-05,
+      "loss": 0.8327,
+      "step": 3071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085534585949335e-05,
+      "loss": 0.3684,
+      "step": 3072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081828110355806e-05,
+      "loss": 0.8132,
+      "step": 3073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078120961771276e-05,
+      "loss": 0.8317,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407441314080696e-05,
+      "loss": 0.882,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070704648074185e-05,
+      "loss": 0.7453,
+      "step": 3076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066995484184384e-05,
+      "loss": 0.8596,
+      "step": 3077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063285649749111e-05,
+      "loss": 0.8216,
+      "step": 3078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059575145380024e-05,
+      "loss": 0.8189,
+      "step": 3079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055863971688886e-05,
+      "loss": 0.831,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052152129287581e-05,
+      "loss": 0.8536,
+      "step": 3081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048439618788101e-05,
+      "loss": 0.8287,
+      "step": 3082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044726440802538e-05,
+      "loss": 0.8976,
+      "step": 3083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041012595943113e-05,
+      "loss": 0.7821,
+      "step": 3084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037298084822135e-05,
+      "loss": 0.8446,
+      "step": 3085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033582908052038e-05,
+      "loss": 0.8796,
+      "step": 3086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029867066245363e-05,
+      "loss": 0.7968,
+      "step": 3087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026150560014757e-05,
+      "loss": 0.8334,
+      "step": 3088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402243338997298e-05,
+      "loss": 0.932,
+      "step": 3089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40187155567329e-05,
+      "loss": 0.804,
+      "step": 3090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014997060907492e-05,
+      "loss": 0.8116,
+      "step": 3091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011277903109845e-05,
+      "loss": 0.813,
+      "step": 3092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4007558083953155e-05,
+      "loss": 0.7967,
+      "step": 3093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400383760405072e-05,
+      "loss": 0.8002,
+      "step": 3094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000116464015967e-05,
+      "loss": 0.8398,
+      "step": 3095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996394664462408e-05,
+      "loss": 0.9202,
+      "step": 3096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992672206003677e-05,
+      "loss": 0.8711,
+      "step": 3097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988949089253517e-05,
+      "loss": 0.7815,
+      "step": 3098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398522531482577e-05,
+      "loss": 0.8313,
+      "step": 3099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39815008833344e-05,
+      "loss": 0.8317,
+      "step": 3100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977775795393467e-05,
+      "loss": 0.8743,
+      "step": 3101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974050051617148e-05,
+      "loss": 0.8814,
+      "step": 3102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970323652619725e-05,
+      "loss": 0.8846,
+      "step": 3103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396659659901558e-05,
+      "loss": 0.7426,
+      "step": 3104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396286889141922e-05,
+      "loss": 0.3676,
+      "step": 3105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395914053044525e-05,
+      "loss": 0.8512,
+      "step": 3106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955411516708377e-05,
+      "loss": 0.8282,
+      "step": 3107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951681850823427e-05,
+      "loss": 0.8171,
+      "step": 3108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947951533405324e-05,
+      "loss": 0.8375,
+      "step": 3109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944220565069104e-05,
+      "loss": 0.85,
+      "step": 3110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940488946429918e-05,
+      "loss": 0.8013,
+      "step": 3111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936756678103005e-05,
+      "loss": 0.7574,
+      "step": 3112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393302376070373e-05,
+      "loss": 0.8161,
+      "step": 3113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929290194847553e-05,
+      "loss": 0.9354,
+      "step": 3114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392555598115005e-05,
+      "loss": 0.7502,
+      "step": 3115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921821120226895e-05,
+      "loss": 0.7425,
+      "step": 3116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3918085612693873e-05,
+      "loss": 0.8393,
+      "step": 3117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914349459166876e-05,
+      "loss": 0.8396,
+      "step": 3118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910612660261906e-05,
+      "loss": 0.8442,
+      "step": 3119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906875216595061e-05,
+      "loss": 0.8517,
+      "step": 3120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903137128782554e-05,
+      "loss": 0.9024,
+      "step": 3121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899398397440704e-05,
+      "loss": 0.7899,
+      "step": 3122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895659023185932e-05,
+      "loss": 0.9285,
+      "step": 3123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891919006634767e-05,
+      "loss": 0.8477,
+      "step": 3124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888178348403845e-05,
+      "loss": 0.767,
+      "step": 3125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884437049109905e-05,
+      "loss": 0.856,
+      "step": 3126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880695109369792e-05,
+      "loss": 0.7888,
+      "step": 3127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876952529800463e-05,
+      "loss": 0.8613,
+      "step": 3128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873209311018974e-05,
+      "loss": 0.3205,
+      "step": 3129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869465453642484e-05,
+      "loss": 0.8809,
+      "step": 3130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865720958288263e-05,
+      "loss": 0.8389,
+      "step": 3131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861975825573686e-05,
+      "loss": 0.9239,
+      "step": 3132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858230056116228e-05,
+      "loss": 0.8364,
+      "step": 3133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854483650533476e-05,
+      "loss": 0.8805,
+      "step": 3134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850736609443119e-05,
+      "loss": 0.8169,
+      "step": 3135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846988933462944e-05,
+      "loss": 0.8433,
+      "step": 3136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843240623210855e-05,
+      "loss": 0.7866,
+      "step": 3137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383949167930485e-05,
+      "loss": 0.8282,
+      "step": 3138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383574210236304e-05,
+      "loss": 0.3388,
+      "step": 3139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831991893003629e-05,
+      "loss": 0.8343,
+      "step": 3140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828241051844942e-05,
+      "loss": 0.8021,
+      "step": 3141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824489579505391e-05,
+      "loss": 0.8265,
+      "step": 3142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820737476603506e-05,
+      "loss": 0.9331,
+      "step": 3143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816984743757904e-05,
+      "loss": 0.8082,
+      "step": 3144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813231381587326e-05,
+      "loss": 0.8507,
+      "step": 3145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809477390710609e-05,
+      "loss": 0.822,
+      "step": 3146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805722771746681e-05,
+      "loss": 0.7822,
+      "step": 3147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801967525314595e-05,
+      "loss": 0.7982,
+      "step": 3148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379821165203349e-05,
+      "loss": 0.8771,
+      "step": 3149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794455152522619e-05,
+      "loss": 0.8771,
+      "step": 3150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790698027401334e-05,
+      "loss": 0.8404,
+      "step": 3151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378694027728909e-05,
+      "loss": 0.7982,
+      "step": 3152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783181902805446e-05,
+      "loss": 0.8435,
+      "step": 3153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779422904570061e-05,
+      "loss": 0.8436,
+      "step": 3154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775663283202704e-05,
+      "loss": 0.8229,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771903039323238e-05,
+      "loss": 0.8649,
+      "step": 3156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3768142173551638e-05,
+      "loss": 0.8893,
+      "step": 3157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376438068650797e-05,
+      "loss": 0.8233,
+      "step": 3158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760618578812415e-05,
+      "loss": 0.9252,
+      "step": 3159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756855851085242e-05,
+      "loss": 0.8685,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753092503946838e-05,
+      "loss": 0.848,
+      "step": 3161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749328538017685e-05,
+      "loss": 0.8409,
+      "step": 3162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374556395391836e-05,
+      "loss": 0.8193,
+      "step": 3163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741798752269553e-05,
+      "loss": 0.8399,
+      "step": 3164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373803293369205e-05,
+      "loss": 0.9136,
+      "step": 3165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734266498806739e-05,
+      "loss": 0.8286,
+      "step": 3166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730499448234614e-05,
+      "loss": 0.7735,
+      "step": 3167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726731782596762e-05,
+      "loss": 0.8808,
+      "step": 3168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722963502514379e-05,
+      "loss": 0.9143,
+      "step": 3169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719194608608764e-05,
+      "loss": 0.7857,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715425101501306e-05,
+      "loss": 0.8878,
+      "step": 3171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711654981813504e-05,
+      "loss": 0.7726,
+      "step": 3172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370788425016696e-05,
+      "loss": 0.9675,
+      "step": 3173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704112907183368e-05,
+      "loss": 0.8363,
+      "step": 3174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700340953484534e-05,
+      "loss": 0.8685,
+      "step": 3175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369656838969235e-05,
+      "loss": 0.859,
+      "step": 3176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692795216428824e-05,
+      "loss": 0.9218,
+      "step": 3177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689021434316057e-05,
+      "loss": 0.9152,
+      "step": 3178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685247043976244e-05,
+      "loss": 0.8729,
+      "step": 3179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681472046031695e-05,
+      "loss": 0.8037,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677696441104808e-05,
+      "loss": 0.8973,
+      "step": 3181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673920229818092e-05,
+      "loss": 0.8438,
+      "step": 3182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670143412794138e-05,
+      "loss": 0.9039,
+      "step": 3183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666365990655658e-05,
+      "loss": 0.932,
+      "step": 3184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662587964025456e-05,
+      "loss": 0.8124,
+      "step": 3185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658809333526424e-05,
+      "loss": 0.7479,
+      "step": 3186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655030099781568e-05,
+      "loss": 0.8119,
+      "step": 3187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651250263413991e-05,
+      "loss": 0.8282,
+      "step": 3188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364746982504689e-05,
+      "loss": 0.8087,
+      "step": 3189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643688785303566e-05,
+      "loss": 0.8067,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639907144807418e-05,
+      "loss": 0.7617,
+      "step": 3191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363612490418194e-05,
+      "loss": 0.8153,
+      "step": 3192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632342064050734e-05,
+      "loss": 0.7916,
+      "step": 3193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628558625037491e-05,
+      "loss": 0.8498,
+      "step": 3194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624774587766007e-05,
+      "loss": 0.3664,
+      "step": 3195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620989952860175e-05,
+      "loss": 0.8359,
+      "step": 3196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617204720943989e-05,
+      "loss": 0.8667,
+      "step": 3197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361341889264153e-05,
+      "loss": 0.8021,
+      "step": 3198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609632468576997e-05,
+      "loss": 0.8057,
+      "step": 3199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605845449374666e-05,
+      "loss": 0.8112,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602057835658928e-05,
+      "loss": 0.8241,
+      "step": 3201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598269628054266e-05,
+      "loss": 0.8755,
+      "step": 3202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594480827185255e-05,
+      "loss": 0.8917,
+      "step": 3203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590691433676577e-05,
+      "loss": 0.3594,
+      "step": 3204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358690144815301e-05,
+      "loss": 0.7805,
+      "step": 3205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358311087123942e-05,
+      "loss": 0.7901,
+      "step": 3206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579319703560785e-05,
+      "loss": 0.7847,
+      "step": 3207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575527945742174e-05,
+      "loss": 0.8043,
+      "step": 3208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571735598408745e-05,
+      "loss": 0.7927,
+      "step": 3209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356794266218577e-05,
+      "loss": 0.8414,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564149137698599e-05,
+      "loss": 0.7671,
+      "step": 3211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560355025572697e-05,
+      "loss": 0.8053,
+      "step": 3212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556560326433617e-05,
+      "loss": 0.8094,
+      "step": 3213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552765040907008e-05,
+      "loss": 0.8605,
+      "step": 3214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548969169618616e-05,
+      "loss": 0.8954,
+      "step": 3215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545172713194288e-05,
+      "loss": 0.9125,
+      "step": 3216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354137567225996e-05,
+      "loss": 0.7962,
+      "step": 3217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537578047441675e-05,
+      "loss": 0.8896,
+      "step": 3218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353377983936556e-05,
+      "loss": 0.8214,
+      "step": 3219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529981048657846e-05,
+      "loss": 0.8188,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526181675944861e-05,
+      "loss": 0.8526,
+      "step": 3221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522381721853022e-05,
+      "loss": 0.8253,
+      "step": 3222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518581187008848e-05,
+      "loss": 0.8982,
+      "step": 3223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514780072038956e-05,
+      "loss": 0.8938,
+      "step": 3224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510978377570045e-05,
+      "loss": 0.7998,
+      "step": 3225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507176104228926e-05,
+      "loss": 0.9563,
+      "step": 3226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35033732526425e-05,
+      "loss": 0.8327,
+      "step": 3227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499569823437754e-05,
+      "loss": 0.8401,
+      "step": 3228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495765817241787e-05,
+      "loss": 0.8342,
+      "step": 3229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3491961234681776e-05,
+      "loss": 0.8366,
+      "step": 3230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3488156076385006e-05,
+      "loss": 0.7901,
+      "step": 3231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484350342978852e-05,
+      "loss": 0.8763,
+      "step": 3232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348054403509078e-05,
+      "loss": 0.847,
+      "step": 3233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476737153348363e-05,
+      "loss": 0.7837,
+      "step": 3234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347292969837925e-05,
+      "loss": 0.7821,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469121670811199e-05,
+      "loss": 0.8674,
+      "step": 3236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465313071272061e-05,
+      "loss": 0.8518,
+      "step": 3237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461503900389777e-05,
+      "loss": 0.8568,
+      "step": 3238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345769415879238e-05,
+      "loss": 0.9172,
+      "step": 3239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453883847108005e-05,
+      "loss": 0.8536,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450072965964878e-05,
+      "loss": 0.7745,
+      "step": 3241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344626151599131e-05,
+      "loss": 0.8498,
+      "step": 3242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344244949781572e-05,
+      "loss": 0.7622,
+      "step": 3243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438636912066615e-05,
+      "loss": 0.8396,
+      "step": 3244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434823759372591e-05,
+      "loss": 0.8368,
+      "step": 3245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343101004036234e-05,
+      "loss": 0.7619,
+      "step": 3246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427195755664656e-05,
+      "loss": 0.7659,
+      "step": 3247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342338090590841e-05,
+      "loss": 0.8092,
+      "step": 3248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419565491722579e-05,
+      "loss": 0.9094,
+      "step": 3249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415749513736231e-05,
+      "loss": 0.7997,
+      "step": 3250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411932972578521e-05,
+      "loss": 0.8752,
+      "step": 3251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408115868878702e-05,
+      "loss": 0.9616,
+      "step": 3252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3404298203266123e-05,
+      "loss": 0.8192,
+      "step": 3253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400479976370214e-05,
+      "loss": 0.8217,
+      "step": 3254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396661188820505e-05,
+      "loss": 0.8663,
+      "step": 3255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3392841841246628e-05,
+      "loss": 0.8267,
+      "step": 3256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389021934278289e-05,
+      "loss": 0.8892,
+      "step": 3257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385201468545295e-05,
+      "loss": 0.7939,
+      "step": 3258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381380444677548e-05,
+      "loss": 0.8418,
+      "step": 3259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377558863305035e-05,
+      "loss": 0.8207,
+      "step": 3260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373736725057847e-05,
+      "loss": 0.8179,
+      "step": 3261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369914030566147e-05,
+      "loss": 0.8509,
+      "step": 3262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336609078046021e-05,
+      "loss": 0.8261,
+      "step": 3263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362266975370393e-05,
+      "loss": 0.8316,
+      "step": 3264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358442615927144e-05,
+      "loss": 0.7825,
+      "step": 3265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354617702761004e-05,
+      "loss": 0.8223,
+      "step": 3266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350792236502601e-05,
+      "loss": 0.8217,
+      "step": 3267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346966217782664e-05,
+      "loss": 0.7399,
+      "step": 3268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343139647232008e-05,
+      "loss": 0.8143,
+      "step": 3269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339312525481533e-05,
+      "loss": 0.8576,
+      "step": 3270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335484853162241e-05,
+      "loss": 0.8069,
+      "step": 3271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333165663090522e-05,
+      "loss": 0.8512,
+      "step": 3272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327827859341637e-05,
+      "loss": 0.7879,
+      "step": 3273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323998539102774e-05,
+      "loss": 0.7867,
+      "step": 3274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320168670819981e-05,
+      "loss": 0.9188,
+      "step": 3275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316338255124708e-05,
+      "loss": 0.8338,
+      "step": 3276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33125072926485e-05,
+      "loss": 0.8698,
+      "step": 3277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330867578402298e-05,
+      "loss": 0.7973,
+      "step": 3278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330484372987987e-05,
+      "loss": 0.3252,
+      "step": 3279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301011130850983e-05,
+      "loss": 0.7833,
+      "step": 3280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297177987568211e-05,
+      "loss": 0.8152,
+      "step": 3281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293344300663549e-05,
+      "loss": 0.8029,
+      "step": 3282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289510070769074e-05,
+      "loss": 0.8298,
+      "step": 3283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285675298516956e-05,
+      "loss": 0.7983,
+      "step": 3284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281839984539448e-05,
+      "loss": 0.8024,
+      "step": 3285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278004129468903e-05,
+      "loss": 0.8676,
+      "step": 3286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327416773393775e-05,
+      "loss": 0.8261,
+      "step": 3287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270330798578523e-05,
+      "loss": 0.8034,
+      "step": 3288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326649332402383e-05,
+      "loss": 0.8376,
+      "step": 3289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262655310906375e-05,
+      "loss": 0.8173,
+      "step": 3290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258816759858957e-05,
+      "loss": 0.8325,
+      "step": 3291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254977671514444e-05,
+      "loss": 0.9375,
+      "step": 3292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251138046505814e-05,
+      "loss": 0.8299,
+      "step": 3293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247297885466127e-05,
+      "loss": 0.8902,
+      "step": 3294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324345718902852e-05,
+      "loss": 0.7894,
+      "step": 3295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239615957826234e-05,
+      "loss": 0.8076,
+      "step": 3296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323577419249259e-05,
+      "loss": 0.885,
+      "step": 3297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231931893661e-05,
+      "loss": 0.8778,
+      "step": 3298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322808906196496e-05,
+      "loss": 0.8718,
+      "step": 3299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322424569803806e-05,
+      "loss": 0.8132,
+      "step": 3300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220401802513974e-05,
+      "loss": 0.9022,
+      "step": 3301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216557376026457e-05,
+      "loss": 0.8545,
+      "step": 3302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212712419209366e-05,
+      "loss": 0.8793,
+      "step": 3303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208866932696639e-05,
+      "loss": 0.8317,
+      "step": 3304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320502091712229e-05,
+      "loss": 0.7951,
+      "step": 3305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201174373120443e-05,
+      "loss": 0.8618,
+      "step": 3306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197327301325288e-05,
+      "loss": 0.8142,
+      "step": 3307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193479702371113e-05,
+      "loss": 0.7842,
+      "step": 3308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3189631576892292e-05,
+      "loss": 0.9734,
+      "step": 3309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3185782925523283e-05,
+      "loss": 0.8396,
+      "step": 3310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181933748898629e-05,
+      "loss": 0.8329,
+      "step": 3311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317808404765297e-05,
+      "loss": 0.8069,
+      "step": 3312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317423382242102e-05,
+      "loss": 0.822,
+      "step": 3313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170383073837582e-05,
+      "loss": 0.8206,
+      "step": 3314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166531802537554e-05,
+      "loss": 0.3408,
+      "step": 3315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162680009155908e-05,
+      "loss": 0.8031,
+      "step": 3316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158827694327715e-05,
+      "loss": 0.8658,
+      "step": 3317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154974858688121e-05,
+      "loss": 0.8932,
+      "step": 3318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151121502872357e-05,
+      "loss": 0.8002,
+      "step": 3319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147267627515757e-05,
+      "loss": 0.9297,
+      "step": 3320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143413233253715e-05,
+      "loss": 0.8551,
+      "step": 3321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313955832072173e-05,
+      "loss": 0.7326,
+      "step": 3322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135702890555385e-05,
+      "loss": 0.8229,
+      "step": 3323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131846943390334e-05,
+      "loss": 0.8392,
+      "step": 3324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3127990479862333e-05,
+      "loss": 0.8817,
+      "step": 3325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3124133500607214e-05,
+      "loss": 0.8205,
+      "step": 3326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120276006260895e-05,
+      "loss": 0.8437,
+      "step": 3327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3116417997459384e-05,
+      "loss": 0.8043,
+      "step": 3328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3112559474838759e-05,
+      "loss": 0.751,
+      "step": 3329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3108700439035207e-05,
+      "loss": 0.7996,
+      "step": 3330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310484089068498e-05,
+      "loss": 0.7935,
+      "step": 3331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100980830424419e-05,
+      "loss": 0.8619,
+      "step": 3332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097120258889957e-05,
+      "loss": 0.8237,
+      "step": 3333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093259176718097e-05,
+      "loss": 0.8246,
+      "step": 3334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308939758454544e-05,
+      "loss": 0.8331,
+      "step": 3335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085535483008667e-05,
+      "loss": 0.8491,
+      "step": 3336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081672872744535e-05,
+      "loss": 0.8363,
+      "step": 3337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077809754389898e-05,
+      "loss": 0.8966,
+      "step": 3338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073946128581685e-05,
+      "loss": 0.8243,
+      "step": 3339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307008199595691e-05,
+      "loss": 0.8979,
+      "step": 3340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066217357152672e-05,
+      "loss": 0.3539,
+      "step": 3341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062352212806155e-05,
+      "loss": 0.859,
+      "step": 3342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058486563554622e-05,
+      "loss": 0.7268,
+      "step": 3343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054620410035421e-05,
+      "loss": 0.3619,
+      "step": 3344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050753752885986e-05,
+      "loss": 0.76,
+      "step": 3345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046886592743828e-05,
+      "loss": 0.8812,
+      "step": 3346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304301893024655e-05,
+      "loss": 0.8557,
+      "step": 3347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039150766031827e-05,
+      "loss": 0.3229,
+      "step": 3348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035282100737425e-05,
+      "loss": 0.9086,
+      "step": 3349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031412935001191e-05,
+      "loss": 0.7707,
+      "step": 3350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3027543269461051e-05,
+      "loss": 0.9208,
+      "step": 3351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023673104755017e-05,
+      "loss": 0.7978,
+      "step": 3352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019802441521181e-05,
+      "loss": 0.8333,
+      "step": 3353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301593128039772e-05,
+      "loss": 0.8601,
+      "step": 3354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012059622022888e-05,
+      "loss": 0.7488,
+      "step": 3355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008187467035027e-05,
+      "loss": 0.8037,
+      "step": 3356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004314816072558e-05,
+      "loss": 0.809,
+      "step": 3357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000441669773985e-05,
+      "loss": 0.8547,
+      "step": 3358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996568028777888e-05,
+      "loss": 0.7406,
+      "step": 3359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992693893722939e-05,
+      "loss": 0.8139,
+      "step": 3360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988819265247883e-05,
+      "loss": 0.8714,
+      "step": 3361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2984944143991544e-05,
+      "loss": 0.8404,
+      "step": 3362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981068530592844e-05,
+      "loss": 0.821,
+      "step": 3363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977192425690763e-05,
+      "loss": 0.8198,
+      "step": 3364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2973315829924379e-05,
+      "loss": 0.8382,
+      "step": 3365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969438743932848e-05,
+      "loss": 0.8815,
+      "step": 3366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965561168355394e-05,
+      "loss": 0.8378,
+      "step": 3367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961683103831342e-05,
+      "loss": 0.819,
+      "step": 3368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957804551000085e-05,
+      "loss": 0.7048,
+      "step": 3369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953925510501094e-05,
+      "loss": 0.7973,
+      "step": 3370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950045982973932e-05,
+      "loss": 0.8744,
+      "step": 3371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294616596905823e-05,
+      "loss": 0.7907,
+      "step": 3372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294228546939371e-05,
+      "loss": 0.7699,
+      "step": 3373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938404484620169e-05,
+      "loss": 0.8328,
+      "step": 3374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934523015377477e-05,
+      "loss": 0.8723,
+      "step": 3375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29306410623056e-05,
+      "loss": 0.7442,
+      "step": 3376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2926758626044568e-05,
+      "loss": 0.892,
+      "step": 3377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922875707234499e-05,
+      "loss": 0.8467,
+      "step": 3378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2918992306515592e-05,
+      "loss": 0.7667,
+      "step": 3379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915108424528119e-05,
+      "loss": 0.792,
+      "step": 3380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911224061912433e-05,
+      "loss": 0.7512,
+      "step": 3381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907339219308974e-05,
+      "loss": 0.8625,
+      "step": 3382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903453897358247e-05,
+      "loss": 0.8689,
+      "step": 3383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899568096700852e-05,
+      "loss": 0.8051,
+      "step": 3384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895681817977454e-05,
+      "loss": 0.7673,
+      "step": 3385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891795061828806e-05,
+      "loss": 0.9071,
+      "step": 3386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887907828895739e-05,
+      "loss": 0.3305,
+      "step": 3387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2884020119819152e-05,
+      "loss": 0.873,
+      "step": 3388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880131935240036e-05,
+      "loss": 0.8863,
+      "step": 3389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876243275799459e-05,
+      "loss": 0.8957,
+      "step": 3390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872354142138554e-05,
+      "loss": 0.7989,
+      "step": 3391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286846453489855e-05,
+      "loss": 0.7995,
+      "step": 3392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286457445472074e-05,
+      "loss": 0.9238,
+      "step": 3393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860683902246502e-05,
+      "loss": 0.34,
+      "step": 3394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856792878117293e-05,
+      "loss": 0.85,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285290138297464e-05,
+      "loss": 0.7226,
+      "step": 3396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849009417460159e-05,
+      "loss": 0.8494,
+      "step": 3397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845116982215537e-05,
+      "loss": 0.7997,
+      "step": 3398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841224077882532e-05,
+      "loss": 0.8186,
+      "step": 3399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837330705102992e-05,
+      "loss": 0.8463,
+      "step": 3400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833436864518838e-05,
+      "loss": 0.8603,
+      "step": 3401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829542556772059e-05,
+      "loss": 0.7559,
+      "step": 3402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825647782504737e-05,
+      "loss": 0.7822,
+      "step": 3403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821752542359016e-05,
+      "loss": 0.7897,
+      "step": 3404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817856836977127e-05,
+      "loss": 0.8367,
+      "step": 3405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813960667001373e-05,
+      "loss": 0.7973,
+      "step": 3406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810064033074137e-05,
+      "loss": 0.8206,
+      "step": 3407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806166935837872e-05,
+      "loss": 0.803,
+      "step": 3408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802269375935112e-05,
+      "loss": 0.8909,
+      "step": 3409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798371354008472e-05,
+      "loss": 0.9022,
+      "step": 3410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2794472870700631e-05,
+      "loss": 0.8239,
+      "step": 3411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790573926654355e-05,
+      "loss": 0.8245,
+      "step": 3412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786674522512482e-05,
+      "loss": 0.8689,
+      "step": 3413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782774658917927e-05,
+      "loss": 0.829,
+      "step": 3414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778874336513675e-05,
+      "loss": 0.8285,
+      "step": 3415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774973555942796e-05,
+      "loss": 0.8002,
+      "step": 3416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771072317848428e-05,
+      "loss": 0.8763,
+      "step": 3417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767170622873786e-05,
+      "loss": 0.8631,
+      "step": 3418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763268471662166e-05,
+      "loss": 0.8271,
+      "step": 3419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759365864856932e-05,
+      "loss": 0.8548,
+      "step": 3420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755462803101525e-05,
+      "loss": 0.7591,
+      "step": 3421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751559287039469e-05,
+      "loss": 0.8304,
+      "step": 3422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2747655317314344e-05,
+      "loss": 0.8982,
+      "step": 3423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743750894569823e-05,
+      "loss": 0.9225,
+      "step": 3424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273984601944965e-05,
+      "loss": 0.8001,
+      "step": 3425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735940692597637e-05,
+      "loss": 0.874,
+      "step": 3426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732034914657677e-05,
+      "loss": 0.8388,
+      "step": 3427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272812868627373e-05,
+      "loss": 0.7492,
+      "step": 3428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272422200808984e-05,
+      "loss": 0.842,
+      "step": 3429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720314880750118e-05,
+      "loss": 0.8636,
+      "step": 3430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271640730489875e-05,
+      "loss": 0.8502,
+      "step": 3431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712499281179998e-05,
+      "loss": 0.3592,
+      "step": 3432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708590810238202e-05,
+      "loss": 0.7839,
+      "step": 3433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704681892717763e-05,
+      "loss": 0.8373,
+      "step": 3434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700772529263167e-05,
+      "loss": 0.8475,
+      "step": 3435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269686272051897e-05,
+      "loss": 0.8053,
+      "step": 3436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26929524671298e-05,
+      "loss": 0.8468,
+      "step": 3437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689041769740365e-05,
+      "loss": 0.8256,
+      "step": 3438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2685130628995432e-05,
+      "loss": 0.8219,
+      "step": 3439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681219045539854e-05,
+      "loss": 0.8204,
+      "step": 3440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677307020018559e-05,
+      "loss": 0.8312,
+      "step": 3441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673394553076533e-05,
+      "loss": 0.824,
+      "step": 3442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669481645358847e-05,
+      "loss": 0.7948,
+      "step": 3443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266556829751064e-05,
+      "loss": 0.8583,
+      "step": 3444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661654510177126e-05,
+      "loss": 0.8619,
+      "step": 3445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657740284003593e-05,
+      "loss": 0.9344,
+      "step": 3446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653825619635391e-05,
+      "loss": 0.9326,
+      "step": 3447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649910517717957e-05,
+      "loss": 0.8534,
+      "step": 3448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645994978896794e-05,
+      "loss": 0.9479,
+      "step": 3449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642079003817469e-05,
+      "loss": 0.8428,
+      "step": 3450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638162593125634e-05,
+      "loss": 0.8623,
+      "step": 3451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634245747467002e-05,
+      "loss": 0.816,
+      "step": 3452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630328467487367e-05,
+      "loss": 0.7955,
+      "step": 3453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262641075383259e-05,
+      "loss": 0.827,
+      "step": 3454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26224926071486e-05,
+      "loss": 0.8667,
+      "step": 3455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618574028081402e-05,
+      "loss": 0.8827,
+      "step": 3456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614655017277078e-05,
+      "loss": 0.803,
+      "step": 3457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610735575381763e-05,
+      "loss": 0.3748,
+      "step": 3458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606815703041682e-05,
+      "loss": 0.8696,
+      "step": 3459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602895400903124e-05,
+      "loss": 0.7846,
+      "step": 3460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598974669612445e-05,
+      "loss": 0.7846,
+      "step": 3461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595053509816075e-05,
+      "loss": 0.8509,
+      "step": 3462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259113192216052e-05,
+      "loss": 0.8625,
+      "step": 3463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2587209907292347e-05,
+      "loss": 0.8745,
+      "step": 3464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583287465858197e-05,
+      "loss": 0.7897,
+      "step": 3465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579364598504782e-05,
+      "loss": 0.819,
+      "step": 3466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257544130587889e-05,
+      "loss": 0.3284,
+      "step": 3467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571517588627369e-05,
+      "loss": 0.8022,
+      "step": 3468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256759344739714e-05,
+      "loss": 0.8801,
+      "step": 3469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25636688828352e-05,
+      "loss": 0.828,
+      "step": 3470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255974389558861e-05,
+      "loss": 0.7489,
+      "step": 3471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555818486304497e-05,
+      "loss": 0.8624,
+      "step": 3472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255189265563007e-05,
+      "loss": 0.7689,
+      "step": 3473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547966404212592e-05,
+      "loss": 0.8203,
+      "step": 3474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254403973269941e-05,
+      "loss": 0.8739,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540112641737931e-05,
+      "loss": 0.3726,
+      "step": 3476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536185131975632e-05,
+      "loss": 0.7996,
+      "step": 3477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532257204060065e-05,
+      "loss": 0.8208,
+      "step": 3478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528328858638844e-05,
+      "loss": 0.7514,
+      "step": 3479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252440009635965e-05,
+      "loss": 0.7892,
+      "step": 3480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520470917870246e-05,
+      "loss": 0.8817,
+      "step": 3481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251654132381845e-05,
+      "loss": 0.8491,
+      "step": 3482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512611314852156e-05,
+      "loss": 0.8472,
+      "step": 3483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250868089161932e-05,
+      "loss": 0.3844,
+      "step": 3484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504750054767974e-05,
+      "loss": 0.8203,
+      "step": 3485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2500818804946211e-05,
+      "loss": 0.8331,
+      "step": 3486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24968871428022e-05,
+      "loss": 0.7888,
+      "step": 3487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492955068984165e-05,
+      "loss": 0.3297,
+      "step": 3488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489022584140418e-05,
+      "loss": 0.8654,
+      "step": 3489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2485089688919316e-05,
+      "loss": 0.8182,
+      "step": 3490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481156383969305e-05,
+      "loss": 0.8033,
+      "step": 3491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477222669938875e-05,
+      "loss": 0.7667,
+      "step": 3492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247328854747661e-05,
+      "loss": 0.863,
+      "step": 3493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246935401723114e-05,
+      "loss": 0.9224,
+      "step": 3494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465419079851176e-05,
+      "loss": 0.7671,
+      "step": 3495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461483735985484e-05,
+      "loss": 0.8298,
+      "step": 3496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245754798628291e-05,
+      "loss": 0.8316,
+      "step": 3497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453611831392354e-05,
+      "loss": 0.7899,
+      "step": 3498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449675271962791e-05,
+      "loss": 0.7972,
+      "step": 3499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445738308643267e-05,
+      "loss": 0.8226,
+      "step": 3500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244180094208288e-05,
+      "loss": 0.733,
+      "step": 3501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437863172930806e-05,
+      "loss": 0.842,
+      "step": 3502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433925001836288e-05,
+      "loss": 0.852,
+      "step": 3503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429986429448624e-05,
+      "loss": 0.8857,
+      "step": 3504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426047456417191e-05,
+      "loss": 0.6854,
+      "step": 3505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422108083391426e-05,
+      "loss": 0.8209,
+      "step": 3506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418168311020834e-05,
+      "loss": 0.8779,
+      "step": 3507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414228139954983e-05,
+      "loss": 0.8462,
+      "step": 3508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410287570843504e-05,
+      "loss": 0.8361,
+      "step": 3509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406346604336108e-05,
+      "loss": 0.852,
+      "step": 3510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402405241082552e-05,
+      "loss": 0.8243,
+      "step": 3511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398463481732672e-05,
+      "loss": 0.7487,
+      "step": 3512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2394521326936366e-05,
+      "loss": 0.7899,
+      "step": 3513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390578777343594e-05,
+      "loss": 0.8321,
+      "step": 3514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386635833604385e-05,
+      "loss": 0.8465,
+      "step": 3515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382692496368833e-05,
+      "loss": 0.8702,
+      "step": 3516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378748766287092e-05,
+      "loss": 0.8017,
+      "step": 3517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374804644009383e-05,
+      "loss": 0.7843,
+      "step": 3518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370860130186e-05,
+      "loss": 1.0097,
+      "step": 3519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366915225467287e-05,
+      "loss": 0.8145,
+      "step": 3520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236296993050366e-05,
+      "loss": 0.7932,
+      "step": 3521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359024245945603e-05,
+      "loss": 0.9305,
+      "step": 3522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355078172443661e-05,
+      "loss": 0.8148,
+      "step": 3523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351131710648441e-05,
+      "loss": 0.7771,
+      "step": 3524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347184861210612e-05,
+      "loss": 0.8344,
+      "step": 3525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343237624780914e-05,
+      "loss": 0.8594,
+      "step": 3526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339290002010146e-05,
+      "loss": 0.7727,
+      "step": 3527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335341993549175e-05,
+      "loss": 0.8054,
+      "step": 3528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331393600048924e-05,
+      "loss": 0.8378,
+      "step": 3529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327444822160386e-05,
+      "loss": 0.8225,
+      "step": 3530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323495660534615e-05,
+      "loss": 0.8407,
+      "step": 3531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231954611582273e-05,
+      "loss": 0.8386,
+      "step": 3532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315596188675906e-05,
+      "loss": 0.8334,
+      "step": 3533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311645879745398e-05,
+      "loss": 0.8477,
+      "step": 3534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307695189682502e-05,
+      "loss": 0.835,
+      "step": 3535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303744119138592e-05,
+      "loss": 0.7724,
+      "step": 3536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299792668765102e-05,
+      "loss": 0.8188,
+      "step": 3537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295840839213525e-05,
+      "loss": 1.0145,
+      "step": 3538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291888631135418e-05,
+      "loss": 0.7907,
+      "step": 3539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287936045182403e-05,
+      "loss": 0.8585,
+      "step": 3540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283983082006159e-05,
+      "loss": 0.7956,
+      "step": 3541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280029742258435e-05,
+      "loss": 0.8919,
+      "step": 3542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276076026591031e-05,
+      "loss": 0.866,
+      "step": 3543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227212193565582e-05,
+      "loss": 0.8204,
+      "step": 3544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268167470104734e-05,
+      "loss": 0.7614,
+      "step": 3545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2264212630589762e-05,
+      "loss": 0.7802,
+      "step": 3546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260257417762956e-05,
+      "loss": 0.9373,
+      "step": 3547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256301832276438e-05,
+      "loss": 0.7825,
+      "step": 3548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252345874782376e-05,
+      "loss": 0.7935,
+      "step": 3549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2248389545933017e-05,
+      "loss": 0.7679,
+      "step": 3550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244432846380652e-05,
+      "loss": 0.7558,
+      "step": 3551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2240475776777647e-05,
+      "loss": 0.758,
+      "step": 3552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236518337776424e-05,
+      "loss": 0.8375,
+      "step": 3553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223256053002946e-05,
+      "loss": 0.7908,
+      "step": 3554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228602354189303e-05,
+      "loss": 0.8638,
+      "step": 3555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2224643810908556e-05,
+      "loss": 0.8222,
+      "step": 3556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220684900839881e-05,
+      "loss": 0.8241,
+      "step": 3557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216725624636005e-05,
+      "loss": 0.8051,
+      "step": 3558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212765982949713e-05,
+      "loss": 0.9146,
+      "step": 3559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208805976433852e-05,
+      "loss": 0.8176,
+      "step": 3560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204845605741325e-05,
+      "loss": 0.8278,
+      "step": 3561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200884871525096e-05,
+      "loss": 0.8323,
+      "step": 3562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2196923774438195e-05,
+      "loss": 0.8927,
+      "step": 3563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192962315133708e-05,
+      "loss": 0.9818,
+      "step": 3564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189000494264775e-05,
+      "loss": 0.7883,
+      "step": 3565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185038312484606e-05,
+      "loss": 0.8125,
+      "step": 3566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181075770446464e-05,
+      "loss": 0.9,
+      "step": 3567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177112868803671e-05,
+      "loss": 0.8689,
+      "step": 3568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173149608209612e-05,
+      "loss": 0.844,
+      "step": 3569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169185989317724e-05,
+      "loss": 0.8463,
+      "step": 3570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165222012781517e-05,
+      "loss": 0.834,
+      "step": 3571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2161257679254546e-05,
+      "loss": 0.8091,
+      "step": 3572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215729298939043e-05,
+      "loss": 0.3048,
+      "step": 3573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153327943842845e-05,
+      "loss": 0.788,
+      "step": 3574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149362543265533e-05,
+      "loss": 0.8554,
+      "step": 3575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145396788312286e-05,
+      "loss": 0.8202,
+      "step": 3576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141430679636959e-05,
+      "loss": 0.8596,
+      "step": 3577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213746421789346e-05,
+      "loss": 0.7471,
+      "step": 3578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133497403735764e-05,
+      "loss": 0.313,
+      "step": 3579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2129530237817893e-05,
+      "loss": 0.8786,
+      "step": 3580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125562720793937e-05,
+      "loss": 0.782,
+      "step": 3581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121594853318041e-05,
+      "loss": 0.8246,
+      "step": 3582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117626636044406e-05,
+      "loss": 0.8512,
+      "step": 3583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211365806962729e-05,
+      "loss": 0.7502,
+      "step": 3584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109689154721012e-05,
+      "loss": 0.8691,
+      "step": 3585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105719891979941e-05,
+      "loss": 0.8398,
+      "step": 3586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101750282058515e-05,
+      "loss": 0.8769,
+      "step": 3587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097780325611223e-05,
+      "loss": 0.7879,
+      "step": 3588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093810023292606e-05,
+      "loss": 0.812,
+      "step": 3589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2089839375757274e-05,
+      "loss": 0.3351,
+      "step": 3590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085868383659882e-05,
+      "loss": 0.8754,
+      "step": 3591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2081897047655146e-05,
+      "loss": 0.7233,
+      "step": 3592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077925368397848e-05,
+      "loss": 0.8554,
+      "step": 3593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073953346542807e-05,
+      "loss": 0.7682,
+      "step": 3594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069980982744913e-05,
+      "loss": 0.7804,
+      "step": 3595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066008277659116e-05,
+      "loss": 0.7787,
+      "step": 3596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062035231940408e-05,
+      "loss": 0.8173,
+      "step": 3597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058061846243847e-05,
+      "loss": 0.8205,
+      "step": 3598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054088121224546e-05,
+      "loss": 0.7641,
+      "step": 3599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050114057537669e-05,
+      "loss": 0.7456,
+      "step": 3600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046139655838444e-05,
+      "loss": 0.872,
+      "step": 3601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042164916782144e-05,
+      "loss": 0.8504,
+      "step": 3602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038189841024108e-05,
+      "loss": 0.8313,
+      "step": 3603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034214429219728e-05,
+      "loss": 0.8538,
+      "step": 3604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030238682024444e-05,
+      "loss": 0.351,
+      "step": 3605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026262600093761e-05,
+      "loss": 0.8122,
+      "step": 3606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022286184083234e-05,
+      "loss": 0.8582,
+      "step": 3607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018309434648475e-05,
+      "loss": 0.7238,
+      "step": 3608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2014332352445148e-05,
+      "loss": 0.8235,
+      "step": 3609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010354938128976e-05,
+      "loss": 0.8667,
+      "step": 3610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006377192355734e-05,
+      "loss": 0.788,
+      "step": 3611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002399115781253e-05,
+      "loss": 0.9295,
+      "step": 3612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998420709061417e-05,
+      "loss": 0.8177,
+      "step": 3613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994441972852167e-05,
+      "loss": 0.8472,
+      "step": 3614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990462907809495e-05,
+      "loss": 0.8246,
+      "step": 3615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986483514589449e-05,
+      "loss": 0.8805,
+      "step": 3616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982503793848136e-05,
+      "loss": 0.8357,
+      "step": 3617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978523746241702e-05,
+      "loss": 0.8015,
+      "step": 3618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974543372426363e-05,
+      "loss": 0.7432,
+      "step": 3619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970562673058388e-05,
+      "loss": 0.3235,
+      "step": 3620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966581648794084e-05,
+      "loss": 0.7904,
+      "step": 3621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962600300289829e-05,
+      "loss": 0.7869,
+      "step": 3622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958618628202045e-05,
+      "loss": 0.8142,
+      "step": 3623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954636633187209e-05,
+      "loss": 0.8338,
+      "step": 3624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950654315901856e-05,
+      "loss": 0.8398,
+      "step": 3625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946671677002563e-05,
+      "loss": 0.7411,
+      "step": 3626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942688717145973e-05,
+      "loss": 0.7741,
+      "step": 3627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938705436988775e-05,
+      "loss": 0.324,
+      "step": 3628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1934721837187712e-05,
+      "loss": 0.7969,
+      "step": 3629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.193073791839958e-05,
+      "loss": 0.8234,
+      "step": 3630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926753681281224e-05,
+      "loss": 0.8258,
+      "step": 3631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192276912648955e-05,
+      "loss": 0.8261,
+      "step": 3632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918784254681506e-05,
+      "loss": 0.8482,
+      "step": 3633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19147990665141e-05,
+      "loss": 0.7678,
+      "step": 3634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191081356264439e-05,
+      "loss": 0.7824,
+      "step": 3635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906827743729483e-05,
+      "loss": 0.8211,
+      "step": 3636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902841610426543e-05,
+      "loss": 0.7794,
+      "step": 3637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898855163392782e-05,
+      "loss": 0.7971,
+      "step": 3638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894868403285467e-05,
+      "loss": 0.7988,
+      "step": 3639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189088133076191e-05,
+      "loss": 0.8281,
+      "step": 3640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886893946479483e-05,
+      "loss": 0.733,
+      "step": 3641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882906251095606e-05,
+      "loss": 0.8574,
+      "step": 3642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187891824526775e-05,
+      "loss": 0.8042,
+      "step": 3643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874929929653435e-05,
+      "loss": 0.9167,
+      "step": 3644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870941304910234e-05,
+      "loss": 0.8047,
+      "step": 3645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186695237169577e-05,
+      "loss": 0.807,
+      "step": 3646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862963130667724e-05,
+      "loss": 0.894,
+      "step": 3647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858973582483815e-05,
+      "loss": 0.8242,
+      "step": 3648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1854983727801823e-05,
+      "loss": 0.8284,
+      "step": 3649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850993567279576e-05,
+      "loss": 0.8749,
+      "step": 3650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847003101574946e-05,
+      "loss": 0.8096,
+      "step": 3651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843012331345866e-05,
+      "loss": 0.8076,
+      "step": 3652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1839021257250312e-05,
+      "loss": 0.8506,
+      "step": 3653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835029879946308e-05,
+      "loss": 0.8066,
+      "step": 3654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831038200091942e-05,
+      "loss": 0.8705,
+      "step": 3655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182704621834533e-05,
+      "loss": 0.8033,
+      "step": 3656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823053935364657e-05,
+      "loss": 0.8925,
+      "step": 3657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181906135180815e-05,
+      "loss": 0.8222,
+      "step": 3658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815068468334082e-05,
+      "loss": 0.804,
+      "step": 3659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811075285600783e-05,
+      "loss": 0.7625,
+      "step": 3660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1807081804266625e-05,
+      "loss": 0.8011,
+      "step": 3661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803088024990034e-05,
+      "loss": 0.845,
+      "step": 3662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179909394842949e-05,
+      "loss": 0.8696,
+      "step": 3663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795099575243506e-05,
+      "loss": 0.8992,
+      "step": 3664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791104906090658e-05,
+      "loss": 0.817,
+      "step": 3665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178710994162957e-05,
+      "loss": 0.8204,
+      "step": 3666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783114682518907e-05,
+      "loss": 0.7916,
+      "step": 3667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779119129417394e-05,
+      "loss": 0.3153,
+      "step": 3668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775123282983785e-05,
+      "loss": 0.8453,
+      "step": 3669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771127143876907e-05,
+      "loss": 0.7937,
+      "step": 3670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767130712755614e-05,
+      "loss": 0.8677,
+      "step": 3671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763133990278824e-05,
+      "loss": 0.342,
+      "step": 3672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759136977105489e-05,
+      "loss": 0.8866,
+      "step": 3673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755139673894624e-05,
+      "loss": 0.8694,
+      "step": 3674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175114208130528e-05,
+      "loss": 0.8701,
+      "step": 3675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747144199996561e-05,
+      "loss": 0.7999,
+      "step": 3676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743146030627614e-05,
+      "loss": 0.8233,
+      "step": 3677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739147573857642e-05,
+      "loss": 0.873,
+      "step": 3678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735148830345887e-05,
+      "loss": 0.7603,
+      "step": 3679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731149800751637e-05,
+      "loss": 0.7948,
+      "step": 3680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172715048573424e-05,
+      "loss": 0.7988,
+      "step": 3681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1723150885953081e-05,
+      "loss": 0.3282,
+      "step": 3682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719151002067588e-05,
+      "loss": 0.7917,
+      "step": 3683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715150834737245e-05,
+      "loss": 0.8794,
+      "step": 3684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711150384621583e-05,
+      "loss": 0.8281,
+      "step": 3685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707149652380169e-05,
+      "loss": 0.8148,
+      "step": 3686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703148638672625e-05,
+      "loss": 0.9372,
+      "step": 3687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699147344158624e-05,
+      "loss": 0.3091,
+      "step": 3688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695145769497871e-05,
+      "loss": 0.7628,
+      "step": 3689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691143915350127e-05,
+      "loss": 0.7777,
+      "step": 3690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16871417823752e-05,
+      "loss": 0.7727,
+      "step": 3691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683139371232943e-05,
+      "loss": 0.7535,
+      "step": 3692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679136682583247e-05,
+      "loss": 0.8671,
+      "step": 3693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675133717086058e-05,
+      "loss": 0.8444,
+      "step": 3694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671130475401366e-05,
+      "loss": 0.7741,
+      "step": 3695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1667126958189203e-05,
+      "loss": 0.8106,
+      "step": 3696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663123166109649e-05,
+      "loss": 0.7765,
+      "step": 3697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659119099822828e-05,
+      "loss": 0.7952,
+      "step": 3698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655114759988912e-05,
+      "loss": 0.8283,
+      "step": 3699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651110147268118e-05,
+      "loss": 0.7202,
+      "step": 3700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647105262320699e-05,
+      "loss": 0.7634,
+      "step": 3701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643100105806965e-05,
+      "loss": 0.7547,
+      "step": 3702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1639094678387268e-05,
+      "loss": 0.8234,
+      "step": 3703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635088980721996e-05,
+      "loss": 0.8482,
+      "step": 3704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1631083013471591e-05,
+      "loss": 0.7508,
+      "step": 3705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162707677729654e-05,
+      "loss": 0.7985,
+      "step": 3706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623070272857366e-05,
+      "loss": 0.8264,
+      "step": 3707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619063500814645e-05,
+      "loss": 0.8418,
+      "step": 3708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615056461828986e-05,
+      "loss": 0.7554,
+      "step": 3709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611049156561055e-05,
+      "loss": 0.7812,
+      "step": 3710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160704158567156e-05,
+      "loss": 0.7655,
+      "step": 3711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160303374982124e-05,
+      "loss": 0.8651,
+      "step": 3712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159902564967089e-05,
+      "loss": 0.8527,
+      "step": 3713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159501728588135e-05,
+      "loss": 0.8911,
+      "step": 3714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159100865911349e-05,
+      "loss": 0.7804,
+      "step": 3715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586999770028235e-05,
+      "loss": 0.833,
+      "step": 3716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582990619286555e-05,
+      "loss": 0.7829,
+      "step": 3717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157898120754945e-05,
+      "loss": 0.7828,
+      "step": 3718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574971535477984e-05,
+      "loss": 0.8268,
+      "step": 3719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570961603733238e-05,
+      "loss": 0.364,
+      "step": 3720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566951412976352e-05,
+      "loss": 0.7977,
+      "step": 3721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562940963868511e-05,
+      "loss": 0.9309,
+      "step": 3722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558930257070934e-05,
+      "loss": 0.8491,
+      "step": 3723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554919293244885e-05,
+      "loss": 0.8483,
+      "step": 3724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550908073051677e-05,
+      "loss": 0.8092,
+      "step": 3725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546896597152653e-05,
+      "loss": 0.7712,
+      "step": 3726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542884866209208e-05,
+      "loss": 0.7915,
+      "step": 3727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153887288088277e-05,
+      "loss": 0.7701,
+      "step": 3728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534860641834822e-05,
+      "loss": 0.9758,
+      "step": 3729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530848149726881e-05,
+      "loss": 0.8479,
+      "step": 3730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526835405220503e-05,
+      "loss": 0.7946,
+      "step": 3731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1522822408977288e-05,
+      "loss": 0.9248,
+      "step": 3732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518809161658881e-05,
+      "loss": 0.8814,
+      "step": 3733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514795663926965e-05,
+      "loss": 0.8167,
+      "step": 3734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510781916443268e-05,
+      "loss": 0.7684,
+      "step": 3735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150676791986955e-05,
+      "loss": 0.8158,
+      "step": 3736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502753674867621e-05,
+      "loss": 0.8694,
+      "step": 3737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498739182099335e-05,
+      "loss": 0.8171,
+      "step": 3738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149472444222657e-05,
+      "loss": 0.2947,
+      "step": 3739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490709455911264e-05,
+      "loss": 0.8494,
+      "step": 3740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486694223815386e-05,
+      "loss": 0.8562,
+      "step": 3741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482678746600944e-05,
+      "loss": 0.7648,
+      "step": 3742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478663024929992e-05,
+      "loss": 0.9485,
+      "step": 3743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147464705946462e-05,
+      "loss": 0.7973,
+      "step": 3744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470630850866966e-05,
+      "loss": 0.3191,
+      "step": 3745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466614399799195e-05,
+      "loss": 0.7609,
+      "step": 3746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146259770692352e-05,
+      "loss": 0.3132,
+      "step": 3747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458580772902197e-05,
+      "loss": 0.7933,
+      "step": 3748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145456359839751e-05,
+      "loss": 0.8734,
+      "step": 3749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1450546184071798e-05,
+      "loss": 0.8973,
+      "step": 3750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1446528530587428e-05,
+      "loss": 0.8038,
+      "step": 3751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442510638606813e-05,
+      "loss": 0.8253,
+      "step": 3752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438492508792395e-05,
+      "loss": 0.7911,
+      "step": 3753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434474141806672e-05,
+      "loss": 0.8238,
+      "step": 3754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143045553831217e-05,
+      "loss": 0.8624,
+      "step": 3755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426436698971449e-05,
+      "loss": 0.8807,
+      "step": 3756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422417624447121e-05,
+      "loss": 0.7788,
+      "step": 3757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141839831540183e-05,
+      "loss": 0.7826,
+      "step": 3758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141437877249826e-05,
+      "loss": 0.7943,
+      "step": 3759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410358996399129e-05,
+      "loss": 0.837,
+      "step": 3760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14063389877672e-05,
+      "loss": 0.8201,
+      "step": 3761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140231874726527e-05,
+      "loss": 0.8164,
+      "step": 3762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398298275556176e-05,
+      "loss": 0.7915,
+      "step": 3763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394277573302794e-05,
+      "loss": 0.844,
+      "step": 3764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390256641168038e-05,
+      "loss": 0.7913,
+      "step": 3765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386235479814856e-05,
+      "loss": 0.7628,
+      "step": 3766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382214089906236e-05,
+      "loss": 0.8544,
+      "step": 3767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378192472105208e-05,
+      "loss": 0.8794,
+      "step": 3768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137417062707483e-05,
+      "loss": 0.7038,
+      "step": 3769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1370148555478211e-05,
+      "loss": 0.8337,
+      "step": 3770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366126257978484e-05,
+      "loss": 0.7861,
+      "step": 3771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362103735238826e-05,
+      "loss": 0.7687,
+      "step": 3772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358080987922452e-05,
+      "loss": 0.8103,
+      "step": 3773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354058016692609e-05,
+      "loss": 0.8537,
+      "step": 3774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350034822212588e-05,
+      "loss": 0.3256,
+      "step": 3775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134601140514571e-05,
+      "loss": 0.3699,
+      "step": 3776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341987766155333e-05,
+      "loss": 0.7833,
+      "step": 3777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133796390590486e-05,
+      "loss": 0.8437,
+      "step": 3778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333939825057718e-05,
+      "loss": 0.8122,
+      "step": 3779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329915524277384e-05,
+      "loss": 0.8341,
+      "step": 3780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132589100422736e-05,
+      "loss": 0.8782,
+      "step": 3781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321866265571186e-05,
+      "loss": 0.766,
+      "step": 3782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317841308972448e-05,
+      "loss": 0.7813,
+      "step": 3783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313816135094752e-05,
+      "loss": 0.8121,
+      "step": 3784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309790744601753e-05,
+      "loss": 0.8009,
+      "step": 3785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305765138157139e-05,
+      "loss": 0.8149,
+      "step": 3786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301739316424623e-05,
+      "loss": 0.8289,
+      "step": 3787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1297713280067969e-05,
+      "loss": 0.8772,
+      "step": 3788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293687029750968e-05,
+      "loss": 0.8637,
+      "step": 3789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289660566137445e-05,
+      "loss": 0.78,
+      "step": 3790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285633889891267e-05,
+      "loss": 0.7396,
+      "step": 3791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281607001676332e-05,
+      "loss": 0.7388,
+      "step": 3792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277579902156565e-05,
+      "loss": 0.7785,
+      "step": 3793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1273552591995943e-05,
+      "loss": 0.8748,
+      "step": 3794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269525071858465e-05,
+      "loss": 0.858,
+      "step": 3795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265497342408166e-05,
+      "loss": 0.7894,
+      "step": 3796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261469404309116e-05,
+      "loss": 0.8377,
+      "step": 3797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257441258225428e-05,
+      "loss": 0.7888,
+      "step": 3798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253412904821236e-05,
+      "loss": 0.7909,
+      "step": 3799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249384344760716e-05,
+      "loss": 0.775,
+      "step": 3800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124535557870808e-05,
+      "loss": 0.8655,
+      "step": 3801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241326607327563e-05,
+      "loss": 0.7996,
+      "step": 3802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237297431283447e-05,
+      "loss": 0.8,
+      "step": 3803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1233268051240042e-05,
+      "loss": 0.9652,
+      "step": 3804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229238467861689e-05,
+      "loss": 0.8767,
+      "step": 3805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225208681812765e-05,
+      "loss": 0.7753,
+      "step": 3806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221178693757682e-05,
+      "loss": 0.7932,
+      "step": 3807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217148504360885e-05,
+      "loss": 0.9135,
+      "step": 3808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213118114286852e-05,
+      "loss": 0.7695,
+      "step": 3809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209087524200089e-05,
+      "loss": 0.807,
+      "step": 3810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120505673476514e-05,
+      "loss": 0.8786,
+      "step": 3811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201025746646583e-05,
+      "loss": 0.8371,
+      "step": 3812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1196994560509026e-05,
+      "loss": 0.8159,
+      "step": 3813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192963177017111e-05,
+      "loss": 0.898,
+      "step": 3814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188931596835509e-05,
+      "loss": 0.7404,
+      "step": 3815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184899820628932e-05,
+      "loss": 0.7475,
+      "step": 3816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180867849062111e-05,
+      "loss": 0.7918,
+      "step": 3817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176835682799825e-05,
+      "loss": 0.8002,
+      "step": 3818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117280332250687e-05,
+      "loss": 0.2952,
+      "step": 3819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168770768848084e-05,
+      "loss": 0.8518,
+      "step": 3820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1164738022488339e-05,
+      "loss": 0.8247,
+      "step": 3821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160705084092526e-05,
+      "loss": 0.7479,
+      "step": 3822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156671954325577e-05,
+      "loss": 0.8519,
+      "step": 3823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152638633852457e-05,
+      "loss": 0.8091,
+      "step": 3824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1148605123338157e-05,
+      "loss": 0.7951,
+      "step": 3825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11445714234477e-05,
+      "loss": 0.7759,
+      "step": 3826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140537534846145e-05,
+      "loss": 0.8745,
+      "step": 3827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136503458198577e-05,
+      "loss": 0.7951,
+      "step": 3828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132469194170117e-05,
+      "loss": 0.8652,
+      "step": 3829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128434743425908e-05,
+      "loss": 0.8679,
+      "step": 3830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124400106631134e-05,
+      "loss": 0.8564,
+      "step": 3831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112036528445101e-05,
+      "loss": 0.81,
+      "step": 3832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116330277550765e-05,
+      "loss": 0.856,
+      "step": 3833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1112295086595681e-05,
+      "loss": 0.7529,
+      "step": 3834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108259712251053e-05,
+      "loss": 0.7789,
+      "step": 3835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104224155182215e-05,
+      "loss": 0.8515,
+      "step": 3836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100188416054532e-05,
+      "loss": 0.804,
+      "step": 3837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096152495533393e-05,
+      "loss": 0.768,
+      "step": 3838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092116394284219e-05,
+      "loss": 0.8311,
+      "step": 3839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088080112972465e-05,
+      "loss": 0.8355,
+      "step": 3840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108404365226361e-05,
+      "loss": 0.8911,
+      "step": 3841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080007012823169e-05,
+      "loss": 0.8051,
+      "step": 3842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075970195316677e-05,
+      "loss": 0.8244,
+      "step": 3843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071933200409706e-05,
+      "loss": 0.8168,
+      "step": 3844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067896028767856e-05,
+      "loss": 0.8371,
+      "step": 3845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063858681056756e-05,
+      "loss": 0.7651,
+      "step": 3846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059821157942059e-05,
+      "loss": 0.8376,
+      "step": 3847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055783460089458e-05,
+      "loss": 0.76,
+      "step": 3848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1051745588164662e-05,
+      "loss": 0.7449,
+      "step": 3849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104770754283342e-05,
+      "loss": 0.8067,
+      "step": 3850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10436693247615e-05,
+      "loss": 0.7845,
+      "step": 3851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039630934614703e-05,
+      "loss": 0.8474,
+      "step": 3852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103559237305886e-05,
+      "loss": 0.8452,
+      "step": 3853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031553640759827e-05,
+      "loss": 0.8736,
+      "step": 3854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027514738383491e-05,
+      "loss": 0.8491,
+      "step": 3855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023475666595765e-05,
+      "loss": 0.8897,
+      "step": 3856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101943642606259e-05,
+      "loss": 0.8827,
+      "step": 3857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015397017449937e-05,
+      "loss": 0.8115,
+      "step": 3858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011357441423802e-05,
+      "loss": 0.8733,
+      "step": 3859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007317698650209e-05,
+      "loss": 0.7872,
+      "step": 3860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003277789795213e-05,
+      "loss": 0.8764,
+      "step": 3861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999237715524888e-05,
+      "loss": 0.7902,
+      "step": 3862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0995197476505347e-05,
+      "loss": 0.911,
+      "step": 3863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0991157073402723e-05,
+      "loss": 0.7507,
+      "step": 3864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987116506883171e-05,
+      "loss": 0.8627,
+      "step": 3865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983075777612886e-05,
+      "loss": 0.851,
+      "step": 3866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979034886258082e-05,
+      "loss": 0.7472,
+      "step": 3867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0974993833484994e-05,
+      "loss": 0.8188,
+      "step": 3868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.09709526199599e-05,
+      "loss": 0.796,
+      "step": 3869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966911246349087e-05,
+      "loss": 0.337,
+      "step": 3870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096286971331888e-05,
+      "loss": 0.7275,
+      "step": 3871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958828021535625e-05,
+      "loss": 0.8339,
+      "step": 3872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954786171665695e-05,
+      "loss": 0.9021,
+      "step": 3873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950744164375489e-05,
+      "loss": 0.8136,
+      "step": 3874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946702000331436e-05,
+      "loss": 0.7641,
+      "step": 3875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942659680199983e-05,
+      "loss": 0.9205,
+      "step": 3876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938617204647613e-05,
+      "loss": 0.8387,
+      "step": 3877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934574574340821e-05,
+      "loss": 0.836,
+      "step": 3878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093053178994614e-05,
+      "loss": 0.8147,
+      "step": 3879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0926488852130126e-05,
+      "loss": 0.8045,
+      "step": 3880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0922445761559351e-05,
+      "loss": 0.9113,
+      "step": 3881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918402518900424e-05,
+      "loss": 0.3476,
+      "step": 3882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914359124819976e-05,
+      "loss": 0.8165,
+      "step": 3883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910315579984656e-05,
+      "loss": 0.9263,
+      "step": 3884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906271885061149e-05,
+      "loss": 0.793,
+      "step": 3885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902228040716152e-05,
+      "loss": 0.9441,
+      "step": 3886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898184047616396e-05,
+      "loss": 0.7957,
+      "step": 3887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0894139906428639e-05,
+      "loss": 0.8575,
+      "step": 3888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890095617819649e-05,
+      "loss": 0.7805,
+      "step": 3889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886051182456235e-05,
+      "loss": 0.8787,
+      "step": 3890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882006601005221e-05,
+      "loss": 0.9235,
+      "step": 3891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877961874133458e-05,
+      "loss": 0.8267,
+      "step": 3892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873917002507816e-05,
+      "loss": 0.7959,
+      "step": 3893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0869871986795198e-05,
+      "loss": 0.7929,
+      "step": 3894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086582682766252e-05,
+      "loss": 0.719,
+      "step": 3895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0861781525776733e-05,
+      "loss": 0.7746,
+      "step": 3896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857736081804802e-05,
+      "loss": 0.8157,
+      "step": 3897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085369049641372e-05,
+      "loss": 0.9572,
+      "step": 3898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0849644770270502e-05,
+      "loss": 0.819,
+      "step": 3899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845598904042188e-05,
+      "loss": 0.7914,
+      "step": 3900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084155289839584e-05,
+      "loss": 0.8091,
+      "step": 3901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837506753998542e-05,
+      "loss": 0.8266,
+      "step": 3902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833460471517398e-05,
+      "loss": 0.8304,
+      "step": 3903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829414051619545e-05,
+      "loss": 0.8323,
+      "step": 3904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825367494972132e-05,
+      "loss": 0.7774,
+      "step": 3905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821320802242335e-05,
+      "loss": 0.8466,
+      "step": 3906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817273974097356e-05,
+      "loss": 0.7876,
+      "step": 3907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813227011204408e-05,
+      "loss": 0.7947,
+      "step": 3908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809179914230738e-05,
+      "loss": 0.8623,
+      "step": 3909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805132683843614e-05,
+      "loss": 0.884,
+      "step": 3910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801085320710314e-05,
+      "loss": 0.7599,
+      "step": 3911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797037825498156e-05,
+      "loss": 0.8183,
+      "step": 3912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792990198874462e-05,
+      "loss": 0.7113,
+      "step": 3913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788942441506592e-05,
+      "loss": 0.8417,
+      "step": 3914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0784894554061912e-05,
+      "loss": 0.8779,
+      "step": 3915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780846537207824e-05,
+      "loss": 0.7944,
+      "step": 3916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776798391611742e-05,
+      "loss": 0.8691,
+      "step": 3917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772750117941099e-05,
+      "loss": 0.7438,
+      "step": 3918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768701716863358e-05,
+      "loss": 0.9124,
+      "step": 3919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764653189046002e-05,
+      "loss": 0.831,
+      "step": 3920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760604535156525e-05,
+      "loss": 0.7547,
+      "step": 3921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075655575586245e-05,
+      "loss": 0.7281,
+      "step": 3922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752506851831322e-05,
+      "loss": 0.85,
+      "step": 3923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07484578237307e-05,
+      "loss": 0.7567,
+      "step": 3924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744408672228173e-05,
+      "loss": 0.7497,
+      "step": 3925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740359397991335e-05,
+      "loss": 0.8388,
+      "step": 3926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073631000168782e-05,
+      "loss": 0.8036,
+      "step": 3927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732260483985266e-05,
+      "loss": 0.794,
+      "step": 3928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728210845551335e-05,
+      "loss": 0.3228,
+      "step": 3929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724161087053714e-05,
+      "loss": 0.8612,
+      "step": 3930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720111209160108e-05,
+      "loss": 0.8111,
+      "step": 3931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0716061212538238e-05,
+      "loss": 0.7333,
+      "step": 3932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712011097855846e-05,
+      "loss": 0.8126,
+      "step": 3933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707960865780697e-05,
+      "loss": 0.8763,
+      "step": 3934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703910516980569e-05,
+      "loss": 0.8088,
+      "step": 3935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699860052123268e-05,
+      "loss": 0.7835,
+      "step": 3936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695809471876607e-05,
+      "loss": 0.9187,
+      "step": 3937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691758776908428e-05,
+      "loss": 0.8371,
+      "step": 3938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687707967886594e-05,
+      "loss": 0.753,
+      "step": 3939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683657045478974e-05,
+      "loss": 0.8137,
+      "step": 3940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679606010353467e-05,
+      "loss": 0.7612,
+      "step": 3941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675554863177989e-05,
+      "loss": 0.8628,
+      "step": 3942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671503604620467e-05,
+      "loss": 0.8058,
+      "step": 3943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667452235348861e-05,
+      "loss": 0.8512,
+      "step": 3944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663400756031128e-05,
+      "loss": 0.8,
+      "step": 3945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659349167335264e-05,
+      "loss": 0.707,
+      "step": 3946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655297469929273e-05,
+      "loss": 0.8049,
+      "step": 3947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651245664481176e-05,
+      "loss": 0.3303,
+      "step": 3948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0647193751659012e-05,
+      "loss": 0.7926,
+      "step": 3949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643141732130848e-05,
+      "loss": 0.7327,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0639089606564753e-05,
+      "loss": 0.773,
+      "step": 3951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635037375628826e-05,
+      "loss": 0.8552,
+      "step": 3952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063098503999117e-05,
+      "loss": 0.8329,
+      "step": 3953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062693260031992e-05,
+      "loss": 0.9049,
+      "step": 3954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062288005728322e-05,
+      "loss": 0.3356,
+      "step": 3955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618827411549233e-05,
+      "loss": 0.9267,
+      "step": 3956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614774663786139e-05,
+      "loss": 0.8336,
+      "step": 3957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610721814662135e-05,
+      "loss": 0.7559,
+      "step": 3958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606668864845429e-05,
+      "loss": 0.8342,
+      "step": 3959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602615815004258e-05,
+      "loss": 0.77,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598562665806862e-05,
+      "loss": 0.7329,
+      "step": 3961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594509417921505e-05,
+      "loss": 0.3477,
+      "step": 3962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0590456072016473e-05,
+      "loss": 0.8143,
+      "step": 3963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058640262876005e-05,
+      "loss": 0.8305,
+      "step": 3964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582349088820552e-05,
+      "loss": 0.8183,
+      "step": 3965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0578295452866309e-05,
+      "loss": 0.7919,
+      "step": 3966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574241721565658e-05,
+      "loss": 0.7921,
+      "step": 3967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570187895586961e-05,
+      "loss": 0.8049,
+      "step": 3968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566133975598592e-05,
+      "loss": 0.8433,
+      "step": 3969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562079962268943e-05,
+      "loss": 0.7898,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558025856266412e-05,
+      "loss": 0.8129,
+      "step": 3971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553971658259425e-05,
+      "loss": 0.8541,
+      "step": 3972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549917368916418e-05,
+      "loss": 0.7662,
+      "step": 3973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545862988905837e-05,
+      "loss": 0.7573,
+      "step": 3974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541808518896153e-05,
+      "loss": 0.8445,
+      "step": 3975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537753959555844e-05,
+      "loss": 0.7604,
+      "step": 3976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05336993115534e-05,
+      "loss": 0.8036,
+      "step": 3977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529644575557336e-05,
+      "loss": 0.8418,
+      "step": 3978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052558975223618e-05,
+      "loss": 0.7891,
+      "step": 3979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521534842258464e-05,
+      "loss": 0.8772,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051747984629274e-05,
+      "loss": 0.7457,
+      "step": 3981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0513424765007583e-05,
+      "loss": 0.6914,
+      "step": 3982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509369599071563e-05,
+      "loss": 0.351,
+      "step": 3983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505314349153289e-05,
+      "loss": 0.85,
+      "step": 3984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0501259015921357e-05,
+      "loss": 0.8274,
+      "step": 3985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497203600044394e-05,
+      "loss": 0.8026,
+      "step": 3986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049314810219104e-05,
+      "loss": 0.7982,
+      "step": 3987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048909252302994e-05,
+      "loss": 0.8075,
+      "step": 3988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485036863229758e-05,
+      "loss": 0.8499,
+      "step": 3989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0480981123459175e-05,
+      "loss": 0.9009,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476925304386873e-05,
+      "loss": 0.7887,
+      "step": 3991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472869406681565e-05,
+      "loss": 0.8383,
+      "step": 3992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468813431011956e-05,
+      "loss": 0.8191,
+      "step": 3993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464757378046779e-05,
+      "loss": 0.7691,
+      "step": 3994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046070124845478e-05,
+      "loss": 0.7764,
+      "step": 3995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456645042904703e-05,
+      "loss": 0.8116,
+      "step": 3996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452588762065323e-05,
+      "loss": 0.8891,
+      "step": 3997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448532406605416e-05,
+      "loss": 0.7772,
+      "step": 3998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444475977193772e-05,
+      "loss": 0.7697,
+      "step": 3999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04404194744992e-05,
+      "loss": 0.8177,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7977,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 8.392163750542574e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}