diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24032 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 4000,
+  "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": 4.99875e-05,
+      "loss": 1.4697,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9975e-05,
+      "loss": 1.0811,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99625e-05,
+      "loss": 1.3613,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995e-05,
+      "loss": 1.1719,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99375e-05,
+      "loss": 1.1777,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992500000000001e-05,
+      "loss": 1.3584,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99125e-05,
+      "loss": 1.0566,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99e-05,
+      "loss": 1.4434,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9887500000000006e-05,
+      "loss": 1.2295,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9875000000000006e-05,
+      "loss": 1.3047,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98625e-05,
+      "loss": 1.2568,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9850000000000006e-05,
+      "loss": 1.2852,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9837500000000005e-05,
+      "loss": 1.3584,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9825000000000005e-05,
+      "loss": 1.3174,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98125e-05,
+      "loss": 1.2148,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.3477,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9787500000000004e-05,
+      "loss": 1.6338,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9775000000000004e-05,
+      "loss": 1.4756,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9762500000000003e-05,
+      "loss": 1.1299,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975e-05,
+      "loss": 1.2607,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97375e-05,
+      "loss": 1.2324,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9725e-05,
+      "loss": 1.5986,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97125e-05,
+      "loss": 1.2881,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97e-05,
+      "loss": 1.2285,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96875e-05,
+      "loss": 1.2188,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967500000000001e-05,
+      "loss": 1.3916,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96625e-05,
+      "loss": 1.2383,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965e-05,
+      "loss": 1.3721,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96375e-05,
+      "loss": 1.2656,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962500000000001e-05,
+      "loss": 0.9907,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96125e-05,
+      "loss": 1.4502,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96e-05,
+      "loss": 1.3096,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9587500000000006e-05,
+      "loss": 1.3252,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9575000000000006e-05,
+      "loss": 1.1709,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95625e-05,
+      "loss": 1.4229,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9550000000000005e-05,
+      "loss": 1.3633,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9537500000000005e-05,
+      "loss": 1.0879,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9525000000000004e-05,
+      "loss": 1.2695,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95125e-05,
+      "loss": 1.0879,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 1.1084,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9487500000000003e-05,
+      "loss": 1.2812,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9475e-05,
+      "loss": 1.2246,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94625e-05,
+      "loss": 1.2803,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945e-05,
+      "loss": 1.4619,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94375e-05,
+      "loss": 1.25,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9425e-05,
+      "loss": 1.4443,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94125e-05,
+      "loss": 1.2559,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94e-05,
+      "loss": 1.2451,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93875e-05,
+      "loss": 1.209,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937500000000001e-05,
+      "loss": 1.3916,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93625e-05,
+      "loss": 1.4688,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935e-05,
+      "loss": 1.2725,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93375e-05,
+      "loss": 1.3896,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9325000000000006e-05,
+      "loss": 1.3613,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93125e-05,
+      "loss": 1.292,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93e-05,
+      "loss": 1.3525,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9287500000000005e-05,
+      "loss": 1.4473,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9275000000000005e-05,
+      "loss": 1.1826,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.92625e-05,
+      "loss": 1.2129,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9250000000000004e-05,
+      "loss": 1.3174,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9237500000000004e-05,
+      "loss": 1.4639,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9225000000000004e-05,
+      "loss": 1.3818,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9212500000000004e-05,
+      "loss": 1.3838,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92e-05,
+      "loss": 1.1934,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91875e-05,
+      "loss": 1.2314,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9175e-05,
+      "loss": 1.2461,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91625e-05,
+      "loss": 1.1143,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915e-05,
+      "loss": 1.3955,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91375e-05,
+      "loss": 1.3193,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9125e-05,
+      "loss": 1.2363,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91125e-05,
+      "loss": 1.4629,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91e-05,
+      "loss": 1.0479,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90875e-05,
+      "loss": 1.3818,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907500000000001e-05,
+      "loss": 1.2168,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90625e-05,
+      "loss": 1.3486,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905e-05,
+      "loss": 1.0703,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9037500000000006e-05,
+      "loss": 1.2959,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9025000000000006e-05,
+      "loss": 1.3691,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90125e-05,
+      "loss": 1.0488,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 1.3555,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8987500000000005e-05,
+      "loss": 1.1963,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8975000000000005e-05,
+      "loss": 1.1504,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89625e-05,
+      "loss": 1.4912,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8950000000000004e-05,
+      "loss": 1.2725,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8937500000000004e-05,
+      "loss": 1.4023,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8925e-05,
+      "loss": 1.5596,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89125e-05,
+      "loss": 1.2705,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89e-05,
+      "loss": 1.0928,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88875e-05,
+      "loss": 1.2275,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8875e-05,
+      "loss": 1.2451,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88625e-05,
+      "loss": 1.3652,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.885e-05,
+      "loss": 1.4785,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88375e-05,
+      "loss": 1.6387,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8825e-05,
+      "loss": 1.166,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88125e-05,
+      "loss": 1.1338,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88e-05,
+      "loss": 1.1611,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87875e-05,
+      "loss": 1.3193,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8775000000000007e-05,
+      "loss": 1.4756,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87625e-05,
+      "loss": 1.5615,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.875e-05,
+      "loss": 1.0479,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8737500000000006e-05,
+      "loss": 1.1396,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8725000000000005e-05,
+      "loss": 1.2549,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.87125e-05,
+      "loss": 1.2256,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.87e-05,
+      "loss": 1.2373,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8687500000000004e-05,
+      "loss": 1.4336,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8675000000000004e-05,
+      "loss": 1.2549,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86625e-05,
+      "loss": 1.1465,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8650000000000003e-05,
+      "loss": 1.5664,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86375e-05,
+      "loss": 1.2871,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8625e-05,
+      "loss": 1.4307,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86125e-05,
+      "loss": 1.5811,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-05,
+      "loss": 1.3457,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.85875e-05,
+      "loss": 1.0977,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8575e-05,
+      "loss": 1.252,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.85625e-05,
+      "loss": 1.4561,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.855e-05,
+      "loss": 0.9707,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.85375e-05,
+      "loss": 1.4014,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8525e-05,
+      "loss": 1.1699,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.85125e-05,
+      "loss": 1.1719,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.85e-05,
+      "loss": 1.123,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84875e-05,
+      "loss": 1.1445,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8475000000000006e-05,
+      "loss": 1.2402,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84625e-05,
+      "loss": 1.5352,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.845e-05,
+      "loss": 1.377,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 1.3467,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8425000000000005e-05,
+      "loss": 1.2383,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8412500000000004e-05,
+      "loss": 1.5068,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 1.5156,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8387500000000004e-05,
+      "loss": 1.2002,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8375000000000004e-05,
+      "loss": 1.2275,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.83625e-05,
+      "loss": 1.2041,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.835e-05,
+      "loss": 1.1094,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.83375e-05,
+      "loss": 1.3887,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8325e-05,
+      "loss": 1.3037,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.83125e-05,
+      "loss": 0.9478,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.83e-05,
+      "loss": 1.1729,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.82875e-05,
+      "loss": 1.2959,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8275e-05,
+      "loss": 1.251,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.826250000000001e-05,
+      "loss": 1.3135,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.825e-05,
+      "loss": 0.8872,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.82375e-05,
+      "loss": 1.1777,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.822500000000001e-05,
+      "loss": 1.0889,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8212500000000006e-05,
+      "loss": 1.2588,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.82e-05,
+      "loss": 1.3223,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.81875e-05,
+      "loss": 1.3848,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8175000000000005e-05,
+      "loss": 1.5635,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8162500000000005e-05,
+      "loss": 0.9692,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.815e-05,
+      "loss": 1.166,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8137500000000005e-05,
+      "loss": 1.0859,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8125000000000004e-05,
+      "loss": 1.1338,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8112500000000004e-05,
+      "loss": 1.3945,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 1.3887,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80875e-05,
+      "loss": 1.2529,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8075e-05,
+      "loss": 1.2939,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80625e-05,
+      "loss": 1.1787,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.805e-05,
+      "loss": 1.4326,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80375e-05,
+      "loss": 1.3193,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8025e-05,
+      "loss": 1.3359,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80125e-05,
+      "loss": 1.4834,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8e-05,
+      "loss": 1.3623,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79875e-05,
+      "loss": 1.3428,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7975e-05,
+      "loss": 1.1758,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.796250000000001e-05,
+      "loss": 1.1289,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.795e-05,
+      "loss": 0.9331,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79375e-05,
+      "loss": 1.3311,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7925000000000006e-05,
+      "loss": 1.3555,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7912500000000006e-05,
+      "loss": 1.3799,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79e-05,
+      "loss": 1.2207,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.78875e-05,
+      "loss": 1.2969,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7875000000000005e-05,
+      "loss": 1.3379,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7862500000000005e-05,
+      "loss": 1.3076,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.785e-05,
+      "loss": 1.2881,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7837500000000004e-05,
+      "loss": 1.1455,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7825000000000004e-05,
+      "loss": 1.3721,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7812500000000003e-05,
+      "loss": 1.2178,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.78e-05,
+      "loss": 1.0518,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77875e-05,
+      "loss": 1.5059,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7775e-05,
+      "loss": 1.3643,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77625e-05,
+      "loss": 1.1416,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.775e-05,
+      "loss": 1.3154,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.77375e-05,
+      "loss": 1.21,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7725e-05,
+      "loss": 1.334,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.771250000000001e-05,
+      "loss": 1.3164,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.77e-05,
+      "loss": 1.1533,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76875e-05,
+      "loss": 1.3809,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7675e-05,
+      "loss": 1.2109,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7662500000000007e-05,
+      "loss": 1.1982,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765e-05,
+      "loss": 1.1748,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76375e-05,
+      "loss": 1.292,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7625000000000006e-05,
+      "loss": 1.374,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7612500000000005e-05,
+      "loss": 1.1572,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76e-05,
+      "loss": 1.4795,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7587500000000005e-05,
+      "loss": 1.1904,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7575000000000004e-05,
+      "loss": 1.2695,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7562500000000004e-05,
+      "loss": 1.4336,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755e-05,
+      "loss": 1.6367,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7537500000000004e-05,
+      "loss": 1.2051,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7525e-05,
+      "loss": 1.2051,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75125e-05,
+      "loss": 1.1826,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75e-05,
+      "loss": 1.4717,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74875e-05,
+      "loss": 1.2285,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7475e-05,
+      "loss": 1.1729,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74625e-05,
+      "loss": 1.4365,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.745e-05,
+      "loss": 1.3457,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74375e-05,
+      "loss": 1.2188,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7425e-05,
+      "loss": 1.3076,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.741250000000001e-05,
+      "loss": 1.3574,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74e-05,
+      "loss": 1.3506,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.73875e-05,
+      "loss": 1.1006,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7375e-05,
+      "loss": 1.2354,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7362500000000006e-05,
+      "loss": 1.6055,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.735e-05,
+      "loss": 1.3994,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.73375e-05,
+      "loss": 1.1934,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7325000000000005e-05,
+      "loss": 1.2441,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7312500000000005e-05,
+      "loss": 1.3662,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.73e-05,
+      "loss": 1.0879,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7287500000000004e-05,
+      "loss": 1.2012,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7275000000000004e-05,
+      "loss": 1.4014,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7262500000000004e-05,
+      "loss": 1.2744,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 1.4756,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.72375e-05,
+      "loss": 1.2383,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7225e-05,
+      "loss": 1.2773,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.72125e-05,
+      "loss": 1.4775,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.72e-05,
+      "loss": 1.2236,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71875e-05,
+      "loss": 1.3848,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7175e-05,
+      "loss": 1.3799,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71625e-05,
+      "loss": 1.583,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.715e-05,
+      "loss": 1.2686,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71375e-05,
+      "loss": 1.2754,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7125e-05,
+      "loss": 1.3945,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.711250000000001e-05,
+      "loss": 1.5869,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71e-05,
+      "loss": 1.2168,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70875e-05,
+      "loss": 1.4951,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7075e-05,
+      "loss": 1.3994,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7062500000000006e-05,
+      "loss": 1.0771,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.705e-05,
+      "loss": 1.543,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70375e-05,
+      "loss": 1.3281,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7025000000000005e-05,
+      "loss": 1.4795,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7012500000000004e-05,
+      "loss": 1.5107,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7e-05,
+      "loss": 1.168,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6987500000000004e-05,
+      "loss": 1.042,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6975000000000003e-05,
+      "loss": 1.3467,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.69625e-05,
+      "loss": 1.3809,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.695e-05,
+      "loss": 1.2803,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.69375e-05,
+      "loss": 1.5029,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6925e-05,
+      "loss": 1.3438,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.69125e-05,
+      "loss": 1.2627,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.69e-05,
+      "loss": 1.2676,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.68875e-05,
+      "loss": 1.5713,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6875e-05,
+      "loss": 1.4014,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.68625e-05,
+      "loss": 1.4834,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.685000000000001e-05,
+      "loss": 1.0732,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.68375e-05,
+      "loss": 1.3906,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6825e-05,
+      "loss": 1.1523,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6812500000000006e-05,
+      "loss": 1.2881,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 1.3252,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.67875e-05,
+      "loss": 1.2266,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6775000000000005e-05,
+      "loss": 1.1689,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6762500000000005e-05,
+      "loss": 1.3545,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 1.2266,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.67375e-05,
+      "loss": 1.2725,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6725000000000004e-05,
+      "loss": 1.5791,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6712500000000004e-05,
+      "loss": 1.1914,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 1.2676,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.66875e-05,
+      "loss": 1.5557,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6675e-05,
+      "loss": 1.4473,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.66625e-05,
+      "loss": 1.3516,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.665e-05,
+      "loss": 1.4951,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.66375e-05,
+      "loss": 1.3242,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6625e-05,
+      "loss": 1.3662,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.66125e-05,
+      "loss": 1.4111,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 1.3701,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65875e-05,
+      "loss": 1.1309,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6575e-05,
+      "loss": 1.3066,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65625e-05,
+      "loss": 1.7031,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.655000000000001e-05,
+      "loss": 1.2695,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65375e-05,
+      "loss": 1.542,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6525e-05,
+      "loss": 1.1553,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6512500000000006e-05,
+      "loss": 1.0557,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 1.2744,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.64875e-05,
+      "loss": 1.3486,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6475000000000005e-05,
+      "loss": 1.2207,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6462500000000005e-05,
+      "loss": 1.3057,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6450000000000004e-05,
+      "loss": 1.3584,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.64375e-05,
+      "loss": 1.3965,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6425000000000004e-05,
+      "loss": 1.5166,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.64125e-05,
+      "loss": 1.3701,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.64e-05,
+      "loss": 1.417,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.63875e-05,
+      "loss": 1.3125,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6375e-05,
+      "loss": 1.1885,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.63625e-05,
+      "loss": 1.2764,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.635e-05,
+      "loss": 1.333,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.63375e-05,
+      "loss": 1.6396,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6325e-05,
+      "loss": 1.3408,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.63125e-05,
+      "loss": 1.3779,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 1.54,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.62875e-05,
+      "loss": 1.1768,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6275e-05,
+      "loss": 1.25,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6262500000000006e-05,
+      "loss": 1.2793,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 1.332,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.62375e-05,
+      "loss": 1.0869,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6225e-05,
+      "loss": 1.4512,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6212500000000005e-05,
+      "loss": 1.2168,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 1.2324,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61875e-05,
+      "loss": 1.3076,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6175000000000004e-05,
+      "loss": 1.1533,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6162500000000004e-05,
+      "loss": 1.1777,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6150000000000004e-05,
+      "loss": 1.3359,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61375e-05,
+      "loss": 1.2051,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6125e-05,
+      "loss": 1.2773,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61125e-05,
+      "loss": 1.04,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61e-05,
+      "loss": 1.1514,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.60875e-05,
+      "loss": 1.1631,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6075e-05,
+      "loss": 1.3125,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.60625e-05,
+      "loss": 1.4102,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.605e-05,
+      "loss": 1.043,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.60375e-05,
+      "loss": 1.1963,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6025e-05,
+      "loss": 1.208,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.60125e-05,
+      "loss": 1.4785,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.1035,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.59875e-05,
+      "loss": 1.2012,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5975e-05,
+      "loss": 1.3223,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5962500000000006e-05,
+      "loss": 1.3545,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5950000000000006e-05,
+      "loss": 1.1611,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.59375e-05,
+      "loss": 1.0781,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5925e-05,
+      "loss": 1.3135,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5912500000000005e-05,
+      "loss": 1.1562,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 1.4414,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.58875e-05,
+      "loss": 1.1416,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5875000000000004e-05,
+      "loss": 1.4316,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5862500000000003e-05,
+      "loss": 1.1533,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.585e-05,
+      "loss": 1.3535,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.58375e-05,
+      "loss": 1.0029,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5825e-05,
+      "loss": 1.459,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.58125e-05,
+      "loss": 1.2197,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.58e-05,
+      "loss": 1.3389,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.57875e-05,
+      "loss": 1.0547,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5775e-05,
+      "loss": 1.1631,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.57625e-05,
+      "loss": 1.3916,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.575e-05,
+      "loss": 1.1836,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.57375e-05,
+      "loss": 1.3574,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5725e-05,
+      "loss": 1.2012,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.57125e-05,
+      "loss": 1.4775,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 1.1055,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.56875e-05,
+      "loss": 1.2266,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5675e-05,
+      "loss": 1.4209,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5662500000000005e-05,
+      "loss": 1.2461,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5650000000000005e-05,
+      "loss": 1.1318,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.56375e-05,
+      "loss": 1.2803,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5625e-05,
+      "loss": 1.5684,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5612500000000004e-05,
+      "loss": 1.0918,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 1.293,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55875e-05,
+      "loss": 1.4717,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5575e-05,
+      "loss": 1.0234,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55625e-05,
+      "loss": 1.2119,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.555e-05,
+      "loss": 1.0752,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55375e-05,
+      "loss": 1.0225,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5525e-05,
+      "loss": 1.2529,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55125e-05,
+      "loss": 1.2646,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55e-05,
+      "loss": 1.5488,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.54875e-05,
+      "loss": 1.0527,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5475e-05,
+      "loss": 1.5684,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.54625e-05,
+      "loss": 1.4316,
+      "step": 363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.545000000000001e-05,
+      "loss": 1.2217,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.54375e-05,
+      "loss": 1.2725,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5425e-05,
+      "loss": 1.3457,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.54125e-05,
+      "loss": 1.2158,
+      "step": 367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 1.3408,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.53875e-05,
+      "loss": 1.2266,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5375e-05,
+      "loss": 1.1924,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5362500000000005e-05,
+      "loss": 1.4707,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5350000000000005e-05,
+      "loss": 1.376,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.53375e-05,
+      "loss": 1.3232,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5325000000000004e-05,
+      "loss": 1.2373,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 1.5205,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.53e-05,
+      "loss": 1.4629,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.52875e-05,
+      "loss": 1.3467,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5275e-05,
+      "loss": 1.5684,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.52625e-05,
+      "loss": 1.0811,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.525e-05,
+      "loss": 1.5781,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.52375e-05,
+      "loss": 1.4307,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5225e-05,
+      "loss": 1.1064,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.52125e-05,
+      "loss": 1.2891,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.52e-05,
+      "loss": 1.4316,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.518750000000001e-05,
+      "loss": 1.127,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5175e-05,
+      "loss": 1.1426,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51625e-05,
+      "loss": 0.9878,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5150000000000006e-05,
+      "loss": 1.292,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5137500000000006e-05,
+      "loss": 1.3682,
+      "step": 389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5125e-05,
+      "loss": 1.4287,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51125e-05,
+      "loss": 1.2559,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 1.0283,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5087500000000005e-05,
+      "loss": 1.2031,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5075e-05,
+      "loss": 1.2061,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5062500000000004e-05,
+      "loss": 1.0732,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5050000000000004e-05,
+      "loss": 1.4512,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5037500000000004e-05,
+      "loss": 1.4785,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5025000000000003e-05,
+      "loss": 1.2822,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.50125e-05,
+      "loss": 1.3174,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5e-05,
+      "loss": 1.2061,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49875e-05,
+      "loss": 1.3193,
+      "step": 401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4975e-05,
+      "loss": 1.2627,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49625e-05,
+      "loss": 1.249,
+      "step": 403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.495e-05,
+      "loss": 1.2891,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49375e-05,
+      "loss": 1.4824,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4925e-05,
+      "loss": 1.1689,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49125e-05,
+      "loss": 1.1357,
+      "step": 407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49e-05,
+      "loss": 1.5459,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.488750000000001e-05,
+      "loss": 1.2158,
+      "step": 409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4875e-05,
+      "loss": 1.1572,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.48625e-05,
+      "loss": 1.5752,
+      "step": 411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4850000000000006e-05,
+      "loss": 1.2451,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4837500000000006e-05,
+      "loss": 0.9365,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4825e-05,
+      "loss": 1.1738,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4812500000000005e-05,
+      "loss": 1.252,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 1.3584,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4787500000000004e-05,
+      "loss": 1.2285,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4775e-05,
+      "loss": 1.4404,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4762500000000004e-05,
+      "loss": 1.1992,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 1.2236,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.47375e-05,
+      "loss": 1.209,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4725e-05,
+      "loss": 1.209,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.47125e-05,
+      "loss": 1.4463,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.47e-05,
+      "loss": 1.4355,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46875e-05,
+      "loss": 1.4053,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4675e-05,
+      "loss": 1.1777,
+      "step": 426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46625e-05,
+      "loss": 1.4658,
+      "step": 427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.465e-05,
+      "loss": 1.1074,
+      "step": 428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.463750000000001e-05,
+      "loss": 1.3467,
+      "step": 429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4625e-05,
+      "loss": 1.0342,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46125e-05,
+      "loss": 1.2686,
+      "step": 431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46e-05,
+      "loss": 1.3926,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4587500000000006e-05,
+      "loss": 1.126,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4575e-05,
+      "loss": 1.3516,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.45625e-05,
+      "loss": 1.3867,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4550000000000005e-05,
+      "loss": 1.3232,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4537500000000005e-05,
+      "loss": 1.2852,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4525e-05,
+      "loss": 1.1914,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4512500000000005e-05,
+      "loss": 1.5039,
+      "step": 439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 1.3691,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4487500000000004e-05,
+      "loss": 1.0654,
+      "step": 441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4475e-05,
+      "loss": 1.3164,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.44625e-05,
+      "loss": 1.0352,
+      "step": 443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.445e-05,
+      "loss": 1.376,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.44375e-05,
+      "loss": 1.5088,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4425e-05,
+      "loss": 1.1826,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.44125e-05,
+      "loss": 1.4941,
+      "step": 447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.44e-05,
+      "loss": 1.1572,
+      "step": 448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.43875e-05,
+      "loss": 1.2783,
+      "step": 449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4375e-05,
+      "loss": 1.4111,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.43625e-05,
+      "loss": 1.3936,
+      "step": 451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.435e-05,
+      "loss": 1.1445,
+      "step": 452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.433750000000001e-05,
+      "loss": 1.4277,
+      "step": 453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4325e-05,
+      "loss": 1.3135,
+      "step": 454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.43125e-05,
+      "loss": 1.3232,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.43e-05,
+      "loss": 1.2715,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4287500000000006e-05,
+      "loss": 1.2764,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4275e-05,
+      "loss": 1.2959,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.42625e-05,
+      "loss": 1.4072,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 1.2031,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4237500000000005e-05,
+      "loss": 1.4219,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4225e-05,
+      "loss": 1.583,
+      "step": 462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4212500000000004e-05,
+      "loss": 1.2705,
+      "step": 463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 1.4443,
+      "step": 464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4187500000000003e-05,
+      "loss": 1.1699,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4174999999999996e-05,
+      "loss": 1.541,
+      "step": 466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.41625e-05,
+      "loss": 1.3848,
+      "step": 467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.415e-05,
+      "loss": 1.292,
+      "step": 468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.41375e-05,
+      "loss": 1.0996,
+      "step": 469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4125e-05,
+      "loss": 1.4707,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.41125e-05,
+      "loss": 1.5186,
+      "step": 471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.41e-05,
+      "loss": 1.1777,
+      "step": 472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40875e-05,
+      "loss": 1.5342,
+      "step": 473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4075e-05,
+      "loss": 1.2832,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40625e-05,
+      "loss": 1.3271,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.405e-05,
+      "loss": 1.2939,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4037500000000007e-05,
+      "loss": 1.3086,
+      "step": 477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4025e-05,
+      "loss": 1.2793,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40125e-05,
+      "loss": 1.2842,
+      "step": 479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.4668,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3987500000000005e-05,
+      "loss": 1.1406,
+      "step": 481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3975e-05,
+      "loss": 1.4277,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.39625e-05,
+      "loss": 1.2314,
+      "step": 483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3950000000000004e-05,
+      "loss": 1.3584,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3937500000000004e-05,
+      "loss": 1.1367,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3925e-05,
+      "loss": 1.2666,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3912500000000004e-05,
+      "loss": 1.1221,
+      "step": 487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.39e-05,
+      "loss": 1.4053,
+      "step": 488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.38875e-05,
+      "loss": 1.6719,
+      "step": 489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3875e-05,
+      "loss": 1.001,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.38625e-05,
+      "loss": 1.3086,
+      "step": 491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.385e-05,
+      "loss": 1.2441,
+      "step": 492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.38375e-05,
+      "loss": 1.3564,
+      "step": 493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3825e-05,
+      "loss": 1.207,
+      "step": 494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.38125e-05,
+      "loss": 1.1133,
+      "step": 495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.38e-05,
+      "loss": 1.1826,
+      "step": 496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.37875e-05,
+      "loss": 1.0186,
+      "step": 497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3775e-05,
+      "loss": 1.457,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.37625e-05,
+      "loss": 1.2441,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.375e-05,
+      "loss": 1.4072,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3737500000000006e-05,
+      "loss": 1.5332,
+      "step": 501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3725000000000006e-05,
+      "loss": 1.3984,
+      "step": 502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.37125e-05,
+      "loss": 1.6494,
+      "step": 503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 1.4189,
+      "step": 504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3687500000000005e-05,
+      "loss": 1.4746,
+      "step": 505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3675000000000005e-05,
+      "loss": 1.3496,
+      "step": 506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36625e-05,
+      "loss": 1.3467,
+      "step": 507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3650000000000004e-05,
+      "loss": 1.2197,
+      "step": 508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3637500000000004e-05,
+      "loss": 1.2158,
+      "step": 509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3625e-05,
+      "loss": 1.4854,
+      "step": 510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36125e-05,
+      "loss": 1.3701,
+      "step": 511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36e-05,
+      "loss": 1.1279,
+      "step": 512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.35875e-05,
+      "loss": 1.1768,
+      "step": 513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3575e-05,
+      "loss": 1.1729,
+      "step": 514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.35625e-05,
+      "loss": 1.126,
+      "step": 515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.355e-05,
+      "loss": 1.0459,
+      "step": 516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.35375e-05,
+      "loss": 1.4668,
+      "step": 517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.352500000000001e-05,
+      "loss": 1.3389,
+      "step": 518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.35125e-05,
+      "loss": 1.2412,
+      "step": 519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.35e-05,
+      "loss": 1.4082,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34875e-05,
+      "loss": 1.2871,
+      "step": 521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3475000000000006e-05,
+      "loss": 1.207,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34625e-05,
+      "loss": 1.3369,
+      "step": 523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.345e-05,
+      "loss": 1.4238,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3437500000000006e-05,
+      "loss": 1.1016,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3425000000000005e-05,
+      "loss": 1.458,
+      "step": 526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34125e-05,
+      "loss": 1.3125,
+      "step": 527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 1.1562,
+      "step": 528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3387500000000004e-05,
+      "loss": 1.2529,
+      "step": 529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3375000000000004e-05,
+      "loss": 1.3555,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3362500000000004e-05,
+      "loss": 1.501,
+      "step": 531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.335e-05,
+      "loss": 1.4688,
+      "step": 532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.33375e-05,
+      "loss": 1.1377,
+      "step": 533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3325e-05,
+      "loss": 1.3271,
+      "step": 534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.33125e-05,
+      "loss": 1.2324,
+      "step": 535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.33e-05,
+      "loss": 1.374,
+      "step": 536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.32875e-05,
+      "loss": 1.1279,
+      "step": 537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3275e-05,
+      "loss": 1.2637,
+      "step": 538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.32625e-05,
+      "loss": 1.3164,
+      "step": 539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.325e-05,
+      "loss": 1.416,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.32375e-05,
+      "loss": 1.1104,
+      "step": 541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.322500000000001e-05,
+      "loss": 1.2178,
+      "step": 542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.32125e-05,
+      "loss": 1.6201,
+      "step": 543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.32e-05,
+      "loss": 0.8984,
+      "step": 544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3187500000000006e-05,
+      "loss": 1.2275,
+      "step": 545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3175000000000006e-05,
+      "loss": 1.2236,
+      "step": 546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.31625e-05,
+      "loss": 1.3203,
+      "step": 547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.315e-05,
+      "loss": 1.417,
+      "step": 548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3137500000000005e-05,
+      "loss": 1.2295,
+      "step": 549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3125000000000005e-05,
+      "loss": 1.3818,
+      "step": 550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.31125e-05,
+      "loss": 1.165,
+      "step": 551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 1.1357,
+      "step": 552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3087500000000004e-05,
+      "loss": 1.3281,
+      "step": 553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3075000000000003e-05,
+      "loss": 0.978,
+      "step": 554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.30625e-05,
+      "loss": 1.1201,
+      "step": 555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.305e-05,
+      "loss": 1.0879,
+      "step": 556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.30375e-05,
+      "loss": 1.3799,
+      "step": 557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3025e-05,
+      "loss": 1.1758,
+      "step": 558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.30125e-05,
+      "loss": 1.1475,
+      "step": 559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3e-05,
+      "loss": 1.5498,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29875e-05,
+      "loss": 1.542,
+      "step": 561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2975e-05,
+      "loss": 1.4219,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29625e-05,
+      "loss": 0.9238,
+      "step": 563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.295e-05,
+      "loss": 1.376,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29375e-05,
+      "loss": 1.168,
+      "step": 565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2925000000000007e-05,
+      "loss": 1.3701,
+      "step": 566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29125e-05,
+      "loss": 1.4023,
+      "step": 567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29e-05,
+      "loss": 1.166,
+      "step": 568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2887500000000006e-05,
+      "loss": 1.4551,
+      "step": 569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2875000000000005e-05,
+      "loss": 1.1826,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.28625e-05,
+      "loss": 1.2969,
+      "step": 571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.285e-05,
+      "loss": 1.3623,
+      "step": 572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2837500000000004e-05,
+      "loss": 1.501,
+      "step": 573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2825000000000004e-05,
+      "loss": 1.4375,
+      "step": 574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.28125e-05,
+      "loss": 1.1934,
+      "step": 575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 1.5117,
+      "step": 576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.27875e-05,
+      "loss": 1.2471,
+      "step": 577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2775e-05,
+      "loss": 1.3838,
+      "step": 578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.27625e-05,
+      "loss": 1.0117,
+      "step": 579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.275e-05,
+      "loss": 1.3994,
+      "step": 580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.27375e-05,
+      "loss": 1.2607,
+      "step": 581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2725e-05,
+      "loss": 1.4033,
+      "step": 582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.27125e-05,
+      "loss": 1.2227,
+      "step": 583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.27e-05,
+      "loss": 1.3828,
+      "step": 584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26875e-05,
+      "loss": 1.4297,
+      "step": 585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2675e-05,
+      "loss": 1.3857,
+      "step": 586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26625e-05,
+      "loss": 1.4033,
+      "step": 587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.265e-05,
+      "loss": 1.4385,
+      "step": 588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26375e-05,
+      "loss": 1.2217,
+      "step": 589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2625000000000006e-05,
+      "loss": 1.2363,
+      "step": 590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26125e-05,
+      "loss": 1.1133,
+      "step": 591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26e-05,
+      "loss": 1.2666,
+      "step": 592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2587500000000005e-05,
+      "loss": 1.1016,
+      "step": 593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2575000000000005e-05,
+      "loss": 1.3369,
+      "step": 594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.25625e-05,
+      "loss": 1.2334,
+      "step": 595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2550000000000004e-05,
+      "loss": 1.1436,
+      "step": 596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2537500000000004e-05,
+      "loss": 1.4053,
+      "step": 597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2525000000000004e-05,
+      "loss": 1.3672,
+      "step": 598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2512499999999997e-05,
+      "loss": 1.1807,
+      "step": 599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.25e-05,
+      "loss": 1.29,
+      "step": 600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.24875e-05,
+      "loss": 1.2959,
+      "step": 601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2475e-05,
+      "loss": 1.3135,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.24625e-05,
+      "loss": 1.0537,
+      "step": 603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.245e-05,
+      "loss": 1.3936,
+      "step": 604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.24375e-05,
+      "loss": 1.4893,
+      "step": 605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2425e-05,
+      "loss": 1.3486,
+      "step": 606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.24125e-05,
+      "loss": 1.3818,
+      "step": 607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.24e-05,
+      "loss": 1.1475,
+      "step": 608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23875e-05,
+      "loss": 1.5996,
+      "step": 609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.237500000000001e-05,
+      "loss": 1.2461,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23625e-05,
+      "loss": 1.2969,
+      "step": 611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.235e-05,
+      "loss": 1.3809,
+      "step": 612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23375e-05,
+      "loss": 1.3838,
+      "step": 613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2325000000000006e-05,
+      "loss": 1.4326,
+      "step": 614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23125e-05,
+      "loss": 1.3594,
+      "step": 615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23e-05,
+      "loss": 1.4141,
+      "step": 616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2287500000000005e-05,
+      "loss": 1.4209,
+      "step": 617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2275000000000004e-05,
+      "loss": 0.9873,
+      "step": 618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.22625e-05,
+      "loss": 1.2969,
+      "step": 619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2250000000000004e-05,
+      "loss": 1.335,
+      "step": 620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2237500000000003e-05,
+      "loss": 1.4209,
+      "step": 621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2225e-05,
+      "loss": 1.3428,
+      "step": 622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2212499999999996e-05,
+      "loss": 1.3164,
+      "step": 623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.22e-05,
+      "loss": 1.3496,
+      "step": 624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21875e-05,
+      "loss": 1.5195,
+      "step": 625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2175e-05,
+      "loss": 1.1328,
+      "step": 626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21625e-05,
+      "loss": 1.1055,
+      "step": 627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.215e-05,
+      "loss": 1.2881,
+      "step": 628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21375e-05,
+      "loss": 1.4551,
+      "step": 629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2125e-05,
+      "loss": 1.252,
+      "step": 630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.211250000000001e-05,
+      "loss": 1.3984,
+      "step": 631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21e-05,
+      "loss": 1.2285,
+      "step": 632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20875e-05,
+      "loss": 1.1992,
+      "step": 633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2075000000000006e-05,
+      "loss": 1.2168,
+      "step": 634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2062500000000006e-05,
+      "loss": 1.0996,
+      "step": 635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205e-05,
+      "loss": 1.2461,
+      "step": 636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20375e-05,
+      "loss": 1.1836,
+      "step": 637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2025000000000005e-05,
+      "loss": 1.3779,
+      "step": 638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2012500000000005e-05,
+      "loss": 1.0488,
+      "step": 639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2e-05,
+      "loss": 1.1758,
+      "step": 640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1987500000000004e-05,
+      "loss": 1.2461,
+      "step": 641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1975000000000004e-05,
+      "loss": 1.501,
+      "step": 642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1962500000000004e-05,
+      "loss": 1.3477,
+      "step": 643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.195e-05,
+      "loss": 0.9019,
+      "step": 644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.19375e-05,
+      "loss": 0.9839,
+      "step": 645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1925e-05,
+      "loss": 1.0068,
+      "step": 646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.19125e-05,
+      "loss": 1.3271,
+      "step": 647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.19e-05,
+      "loss": 1.4053,
+      "step": 648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.18875e-05,
+      "loss": 1.2344,
+      "step": 649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1875e-05,
+      "loss": 1.2715,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.18625e-05,
+      "loss": 1.4316,
+      "step": 651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.185e-05,
+      "loss": 1.416,
+      "step": 652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.18375e-05,
+      "loss": 1.4365,
+      "step": 653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1825e-05,
+      "loss": 1.1396,
+      "step": 654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.181250000000001e-05,
+      "loss": 1.1924,
+      "step": 655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.18e-05,
+      "loss": 1.3936,
+      "step": 656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.17875e-05,
+      "loss": 1.3506,
+      "step": 657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1775000000000006e-05,
+      "loss": 1.0088,
+      "step": 658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1762500000000005e-05,
+      "loss": 1.1816,
+      "step": 659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.175e-05,
+      "loss": 1.6641,
+      "step": 660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1737500000000005e-05,
+      "loss": 1.2246,
+      "step": 661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1725000000000005e-05,
+      "loss": 0.9917,
+      "step": 662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1712500000000004e-05,
+      "loss": 1.5635,
+      "step": 663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.17e-05,
+      "loss": 1.458,
+      "step": 664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1687500000000004e-05,
+      "loss": 1.3379,
+      "step": 665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1675e-05,
+      "loss": 1.1113,
+      "step": 666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16625e-05,
+      "loss": 1.5508,
+      "step": 667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.165e-05,
+      "loss": 1.2432,
+      "step": 668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16375e-05,
+      "loss": 1.3447,
+      "step": 669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1625e-05,
+      "loss": 1.4277,
+      "step": 670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16125e-05,
+      "loss": 1.2617,
+      "step": 671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16e-05,
+      "loss": 1.3252,
+      "step": 672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15875e-05,
+      "loss": 1.5605,
+      "step": 673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1575e-05,
+      "loss": 0.9463,
+      "step": 674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.156250000000001e-05,
+      "loss": 1.3691,
+      "step": 675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.155e-05,
+      "loss": 1.5068,
+      "step": 676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15375e-05,
+      "loss": 1.3721,
+      "step": 677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1525e-05,
+      "loss": 1.4199,
+      "step": 678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1512500000000006e-05,
+      "loss": 1.502,
+      "step": 679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15e-05,
+      "loss": 1.2637,
+      "step": 680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.14875e-05,
+      "loss": 1.0732,
+      "step": 681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1475000000000005e-05,
+      "loss": 1.3799,
+      "step": 682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1462500000000005e-05,
+      "loss": 1.0322,
+      "step": 683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.145e-05,
+      "loss": 1.5371,
+      "step": 684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1437500000000004e-05,
+      "loss": 1.5928,
+      "step": 685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1425000000000004e-05,
+      "loss": 1.373,
+      "step": 686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1412500000000004e-05,
+      "loss": 1.4736,
+      "step": 687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.14e-05,
+      "loss": 1.1123,
+      "step": 688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13875e-05,
+      "loss": 1.2568,
+      "step": 689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1375e-05,
+      "loss": 1.6328,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13625e-05,
+      "loss": 1.1602,
+      "step": 691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.135e-05,
+      "loss": 1.25,
+      "step": 692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13375e-05,
+      "loss": 1.1611,
+      "step": 693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1325e-05,
+      "loss": 1.208,
+      "step": 694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13125e-05,
+      "loss": 1.4609,
+      "step": 695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13e-05,
+      "loss": 1.1875,
+      "step": 696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.12875e-05,
+      "loss": 1.2285,
+      "step": 697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1275e-05,
+      "loss": 1.3574,
+      "step": 698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.126250000000001e-05,
+      "loss": 1.1133,
+      "step": 699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.125e-05,
+      "loss": 1.1914,
+      "step": 700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.12375e-05,
+      "loss": 1.3809,
+      "step": 701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1225e-05,
+      "loss": 1.4746,
+      "step": 702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1212500000000006e-05,
+      "loss": 1.4023,
+      "step": 703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.12e-05,
+      "loss": 1.0762,
+      "step": 704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11875e-05,
+      "loss": 1.4287,
+      "step": 705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1175000000000005e-05,
+      "loss": 1.3496,
+      "step": 706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1162500000000004e-05,
+      "loss": 1.1621,
+      "step": 707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.115e-05,
+      "loss": 1.1758,
+      "step": 708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1137500000000004e-05,
+      "loss": 0.918,
+      "step": 709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1125000000000004e-05,
+      "loss": 1.2559,
+      "step": 710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11125e-05,
+      "loss": 1.2812,
+      "step": 711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11e-05,
+      "loss": 1.3105,
+      "step": 712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.10875e-05,
+      "loss": 1.5576,
+      "step": 713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1075e-05,
+      "loss": 1.123,
+      "step": 714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.10625e-05,
+      "loss": 1.1084,
+      "step": 715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.105e-05,
+      "loss": 1.4131,
+      "step": 716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.10375e-05,
+      "loss": 1.1846,
+      "step": 717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1025e-05,
+      "loss": 1.2266,
+      "step": 718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.10125e-05,
+      "loss": 1.1758,
+      "step": 719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1e-05,
+      "loss": 1.4951,
+      "step": 720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09875e-05,
+      "loss": 1.1445,
+      "step": 721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0975e-05,
+      "loss": 1.2363,
+      "step": 722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0962500000000006e-05,
+      "loss": 1.4482,
+      "step": 723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.095e-05,
+      "loss": 1.1211,
+      "step": 724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09375e-05,
+      "loss": 1.5625,
+      "step": 725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0925000000000005e-05,
+      "loss": 1.375,
+      "step": 726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0912500000000005e-05,
+      "loss": 1.1699,
+      "step": 727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09e-05,
+      "loss": 1.5117,
+      "step": 728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.08875e-05,
+      "loss": 1.167,
+      "step": 729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0875000000000004e-05,
+      "loss": 1.1299,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0862500000000004e-05,
+      "loss": 1.4971,
+      "step": 731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.085e-05,
+      "loss": 1.2285,
+      "step": 732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.08375e-05,
+      "loss": 1.1797,
+      "step": 733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0825e-05,
+      "loss": 1.1592,
+      "step": 734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.08125e-05,
+      "loss": 1.209,
+      "step": 735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.08e-05,
+      "loss": 1.3438,
+      "step": 736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.07875e-05,
+      "loss": 1.3096,
+      "step": 737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0775e-05,
+      "loss": 1.2148,
+      "step": 738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.07625e-05,
+      "loss": 1.4014,
+      "step": 739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.075e-05,
+      "loss": 1.458,
+      "step": 740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.07375e-05,
+      "loss": 1.0186,
+      "step": 741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0725e-05,
+      "loss": 1.3369,
+      "step": 742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.07125e-05,
+      "loss": 1.5264,
+      "step": 743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.07e-05,
+      "loss": 1.3477,
+      "step": 744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.06875e-05,
+      "loss": 1.3721,
+      "step": 745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0675e-05,
+      "loss": 1.2207,
+      "step": 746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0662500000000006e-05,
+      "loss": 1.3633,
+      "step": 747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.065e-05,
+      "loss": 1.2295,
+      "step": 748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.06375e-05,
+      "loss": 1.4932,
+      "step": 749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 1.2959,
+      "step": 750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0612500000000005e-05,
+      "loss": 1.5557,
+      "step": 751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 1.1582,
+      "step": 752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05875e-05,
+      "loss": 1.2197,
+      "step": 753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0575000000000004e-05,
+      "loss": 1.2529,
+      "step": 754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0562500000000003e-05,
+      "loss": 1.3789,
+      "step": 755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.055e-05,
+      "loss": 1.166,
+      "step": 756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05375e-05,
+      "loss": 1.2803,
+      "step": 757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0525e-05,
+      "loss": 1.2871,
+      "step": 758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05125e-05,
+      "loss": 1.4541,
+      "step": 759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05e-05,
+      "loss": 1.1025,
+      "step": 760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.04875e-05,
+      "loss": 1.1543,
+      "step": 761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0475e-05,
+      "loss": 1.2422,
+      "step": 762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.04625e-05,
+      "loss": 1.3203,
+      "step": 763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.045000000000001e-05,
+      "loss": 0.9521,
+      "step": 764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.04375e-05,
+      "loss": 1.1641,
+      "step": 765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0425e-05,
+      "loss": 1.4248,
+      "step": 766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0412500000000007e-05,
+      "loss": 1.5068,
+      "step": 767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 1.1973,
+      "step": 768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.03875e-05,
+      "loss": 1.2461,
+      "step": 769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0375e-05,
+      "loss": 1.3662,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0362500000000005e-05,
+      "loss": 1.374,
+      "step": 771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0350000000000005e-05,
+      "loss": 1.4375,
+      "step": 772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.03375e-05,
+      "loss": 1.0625,
+      "step": 773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0325000000000004e-05,
+      "loss": 1.4922,
+      "step": 774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0312500000000004e-05,
+      "loss": 1.0986,
+      "step": 775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 1.4023,
+      "step": 776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0287500000000003e-05,
+      "loss": 1.0986,
+      "step": 777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0275e-05,
+      "loss": 1.4121,
+      "step": 778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.02625e-05,
+      "loss": 1.2002,
+      "step": 779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.025e-05,
+      "loss": 1.2021,
+      "step": 780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.02375e-05,
+      "loss": 1.3281,
+      "step": 781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0225e-05,
+      "loss": 1.1572,
+      "step": 782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.02125e-05,
+      "loss": 1.5605,
+      "step": 783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.02e-05,
+      "loss": 1.1787,
+      "step": 784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01875e-05,
+      "loss": 0.9858,
+      "step": 785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0175e-05,
+      "loss": 1.3232,
+      "step": 786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01625e-05,
+      "loss": 1.0088,
+      "step": 787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.015000000000001e-05,
+      "loss": 1.332,
+      "step": 788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01375e-05,
+      "loss": 1.2041,
+      "step": 789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0125e-05,
+      "loss": 1.1553,
+      "step": 790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0112500000000006e-05,
+      "loss": 1.375,
+      "step": 791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0100000000000006e-05,
+      "loss": 1.4023,
+      "step": 792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.00875e-05,
+      "loss": 1.1025,
+      "step": 793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0075e-05,
+      "loss": 1.3975,
+      "step": 794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0062500000000005e-05,
+      "loss": 1.3955,
+      "step": 795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0050000000000004e-05,
+      "loss": 1.2715,
+      "step": 796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.00375e-05,
+      "loss": 1.4873,
+      "step": 797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0025000000000004e-05,
+      "loss": 1.249,
+      "step": 798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0012500000000004e-05,
+      "loss": 1.2549,
+      "step": 799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4e-05,
+      "loss": 1.3174,
+      "step": 800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.99875e-05,
+      "loss": 1.292,
+      "step": 801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9975e-05,
+      "loss": 1.2256,
+      "step": 802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.99625e-05,
+      "loss": 1.5977,
+      "step": 803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.995e-05,
+      "loss": 1.3779,
+      "step": 804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.99375e-05,
+      "loss": 1.4854,
+      "step": 805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9925e-05,
+      "loss": 0.9644,
+      "step": 806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.99125e-05,
+      "loss": 1.3379,
+      "step": 807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.99e-05,
+      "loss": 1.2783,
+      "step": 808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.98875e-05,
+      "loss": 1.1953,
+      "step": 809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9875e-05,
+      "loss": 1.5205,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.98625e-05,
+      "loss": 1.1602,
+      "step": 811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9850000000000006e-05,
+      "loss": 1.2871,
+      "step": 812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.98375e-05,
+      "loss": 1.1621,
+      "step": 813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9825e-05,
+      "loss": 1.5674,
+      "step": 814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9812500000000005e-05,
+      "loss": 1.1426,
+      "step": 815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.3682,
+      "step": 816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.97875e-05,
+      "loss": 1.2705,
+      "step": 817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9775e-05,
+      "loss": 1.1943,
+      "step": 818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9762500000000004e-05,
+      "loss": 1.4277,
+      "step": 819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9750000000000004e-05,
+      "loss": 1.3457,
+      "step": 820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.97375e-05,
+      "loss": 1.1816,
+      "step": 821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9725e-05,
+      "loss": 1.2783,
+      "step": 822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.97125e-05,
+      "loss": 1.293,
+      "step": 823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.97e-05,
+      "loss": 1.1768,
+      "step": 824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96875e-05,
+      "loss": 1.1641,
+      "step": 825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9675e-05,
+      "loss": 1.2686,
+      "step": 826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96625e-05,
+      "loss": 1.4355,
+      "step": 827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.965e-05,
+      "loss": 1.291,
+      "step": 828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96375e-05,
+      "loss": 1.3135,
+      "step": 829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9625e-05,
+      "loss": 1.5332,
+      "step": 830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96125e-05,
+      "loss": 1.374,
+      "step": 831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 1.3359,
+      "step": 832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.95875e-05,
+      "loss": 1.3896,
+      "step": 833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9575e-05,
+      "loss": 1.4287,
+      "step": 834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.95625e-05,
+      "loss": 1.251,
+      "step": 835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9550000000000006e-05,
+      "loss": 1.082,
+      "step": 836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.95375e-05,
+      "loss": 1.2666,
+      "step": 837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9525e-05,
+      "loss": 1.1162,
+      "step": 838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9512500000000005e-05,
+      "loss": 1.5312,
+      "step": 839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 1.1943,
+      "step": 840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94875e-05,
+      "loss": 1.458,
+      "step": 841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9475000000000004e-05,
+      "loss": 1.4961,
+      "step": 842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9462500000000004e-05,
+      "loss": 1.1055,
+      "step": 843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9450000000000003e-05,
+      "loss": 1.3672,
+      "step": 844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9437499999999996e-05,
+      "loss": 1.1396,
+      "step": 845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9425e-05,
+      "loss": 1.3252,
+      "step": 846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94125e-05,
+      "loss": 1.0371,
+      "step": 847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94e-05,
+      "loss": 1.0215,
+      "step": 848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.93875e-05,
+      "loss": 1.127,
+      "step": 849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9375e-05,
+      "loss": 1.3994,
+      "step": 850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.93625e-05,
+      "loss": 1.1406,
+      "step": 851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.935e-05,
+      "loss": 1.2197,
+      "step": 852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.93375e-05,
+      "loss": 1.1094,
+      "step": 853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9325e-05,
+      "loss": 1.2803,
+      "step": 854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.93125e-05,
+      "loss": 1.3066,
+      "step": 855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 1.0352,
+      "step": 856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.92875e-05,
+      "loss": 1.2617,
+      "step": 857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9275e-05,
+      "loss": 1.5791,
+      "step": 858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.92625e-05,
+      "loss": 1.2324,
+      "step": 859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9250000000000005e-05,
+      "loss": 1.3047,
+      "step": 860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.92375e-05,
+      "loss": 1.3955,
+      "step": 861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9225e-05,
+      "loss": 1.2402,
+      "step": 862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9212500000000004e-05,
+      "loss": 1.2021,
+      "step": 863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 1.3643,
+      "step": 864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91875e-05,
+      "loss": 1.2568,
+      "step": 865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9175000000000004e-05,
+      "loss": 1.2119,
+      "step": 866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91625e-05,
+      "loss": 1.5898,
+      "step": 867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.915e-05,
+      "loss": 1.123,
+      "step": 868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9137499999999996e-05,
+      "loss": 1.3604,
+      "step": 869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9125e-05,
+      "loss": 1.4951,
+      "step": 870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91125e-05,
+      "loss": 1.3623,
+      "step": 871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91e-05,
+      "loss": 1.1826,
+      "step": 872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90875e-05,
+      "loss": 1.1748,
+      "step": 873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9075e-05,
+      "loss": 1.3809,
+      "step": 874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90625e-05,
+      "loss": 0.939,
+      "step": 875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.905e-05,
+      "loss": 1.5186,
+      "step": 876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.903750000000001e-05,
+      "loss": 1.0059,
+      "step": 877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9025e-05,
+      "loss": 1.3555,
+      "step": 878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90125e-05,
+      "loss": 1.1816,
+      "step": 879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.2344,
+      "step": 880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8987500000000006e-05,
+      "loss": 1.3164,
+      "step": 881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8975e-05,
+      "loss": 1.3145,
+      "step": 882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8962500000000005e-05,
+      "loss": 1.374,
+      "step": 883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8950000000000005e-05,
+      "loss": 1.2471,
+      "step": 884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8937500000000005e-05,
+      "loss": 1.3438,
+      "step": 885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8925e-05,
+      "loss": 1.4355,
+      "step": 886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8912500000000004e-05,
+      "loss": 1.3486,
+      "step": 887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 1.5176,
+      "step": 888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88875e-05,
+      "loss": 1.3564,
+      "step": 889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8875e-05,
+      "loss": 1.248,
+      "step": 890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88625e-05,
+      "loss": 1.2598,
+      "step": 891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.885e-05,
+      "loss": 1.4619,
+      "step": 892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88375e-05,
+      "loss": 1.3936,
+      "step": 893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8825e-05,
+      "loss": 1.5,
+      "step": 894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88125e-05,
+      "loss": 1.4307,
+      "step": 895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88e-05,
+      "loss": 1.2354,
+      "step": 896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.878750000000001e-05,
+      "loss": 1.376,
+      "step": 897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8775e-05,
+      "loss": 1.0195,
+      "step": 898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.87625e-05,
+      "loss": 1.3574,
+      "step": 899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.875e-05,
+      "loss": 1.251,
+      "step": 900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8737500000000006e-05,
+      "loss": 1.2549,
+      "step": 901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8725e-05,
+      "loss": 1.3887,
+      "step": 902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.87125e-05,
+      "loss": 1.3408,
+      "step": 903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8700000000000006e-05,
+      "loss": 1.2666,
+      "step": 904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8687500000000005e-05,
+      "loss": 1.2754,
+      "step": 905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8675e-05,
+      "loss": 1.6475,
+      "step": 906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8662500000000005e-05,
+      "loss": 1.127,
+      "step": 907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8650000000000004e-05,
+      "loss": 1.3506,
+      "step": 908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8637500000000004e-05,
+      "loss": 1.3965,
+      "step": 909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8625e-05,
+      "loss": 1.2754,
+      "step": 910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8612500000000003e-05,
+      "loss": 1.0889,
+      "step": 911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.86e-05,
+      "loss": 1.3887,
+      "step": 912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85875e-05,
+      "loss": 1.2588,
+      "step": 913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8575e-05,
+      "loss": 1.0518,
+      "step": 914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85625e-05,
+      "loss": 1.1377,
+      "step": 915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.855e-05,
+      "loss": 1.0254,
+      "step": 916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85375e-05,
+      "loss": 1.2764,
+      "step": 917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8525e-05,
+      "loss": 1.1592,
+      "step": 918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85125e-05,
+      "loss": 1.1914,
+      "step": 919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85e-05,
+      "loss": 1.333,
+      "step": 920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.848750000000001e-05,
+      "loss": 1.4053,
+      "step": 921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8475e-05,
+      "loss": 1.6387,
+      "step": 922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.84625e-05,
+      "loss": 1.2471,
+      "step": 923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.845e-05,
+      "loss": 1.4365,
+      "step": 924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8437500000000006e-05,
+      "loss": 1.2227,
+      "step": 925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8425e-05,
+      "loss": 1.2432,
+      "step": 926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.84125e-05,
+      "loss": 1.2246,
+      "step": 927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 1.1934,
+      "step": 928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8387500000000005e-05,
+      "loss": 1.584,
+      "step": 929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8375e-05,
+      "loss": 1.1484,
+      "step": 930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8362500000000004e-05,
+      "loss": 1.3975,
+      "step": 931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8350000000000004e-05,
+      "loss": 1.2852,
+      "step": 932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8337500000000004e-05,
+      "loss": 1.1152,
+      "step": 933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8324999999999996e-05,
+      "loss": 1.3564,
+      "step": 934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.83125e-05,
+      "loss": 1.1904,
+      "step": 935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.83e-05,
+      "loss": 1.4736,
+      "step": 936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.82875e-05,
+      "loss": 1.0049,
+      "step": 937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8275e-05,
+      "loss": 1.1719,
+      "step": 938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.82625e-05,
+      "loss": 1.5039,
+      "step": 939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.825e-05,
+      "loss": 1.2793,
+      "step": 940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.82375e-05,
+      "loss": 1.3203,
+      "step": 941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8225e-05,
+      "loss": 1.1084,
+      "step": 942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.82125e-05,
+      "loss": 1.2969,
+      "step": 943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.82e-05,
+      "loss": 0.9868,
+      "step": 944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.818750000000001e-05,
+      "loss": 1.3506,
+      "step": 945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8175e-05,
+      "loss": 1.3828,
+      "step": 946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.81625e-05,
+      "loss": 1.3701,
+      "step": 947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8150000000000006e-05,
+      "loss": 1.4082,
+      "step": 948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8137500000000005e-05,
+      "loss": 1.5078,
+      "step": 949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8125e-05,
+      "loss": 1.3682,
+      "step": 950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.81125e-05,
+      "loss": 1.3926,
+      "step": 951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 1.1982,
+      "step": 952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8087500000000004e-05,
+      "loss": 1.0508,
+      "step": 953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8075e-05,
+      "loss": 1.624,
+      "step": 954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8062500000000004e-05,
+      "loss": 1.501,
+      "step": 955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.805e-05,
+      "loss": 1.2305,
+      "step": 956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.80375e-05,
+      "loss": 1.1494,
+      "step": 957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8025e-05,
+      "loss": 1.5332,
+      "step": 958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.80125e-05,
+      "loss": 1.2236,
+      "step": 959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8e-05,
+      "loss": 1.1191,
+      "step": 960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79875e-05,
+      "loss": 1.1709,
+      "step": 961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7975e-05,
+      "loss": 1.1953,
+      "step": 962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79625e-05,
+      "loss": 1.2744,
+      "step": 963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.795e-05,
+      "loss": 1.2168,
+      "step": 964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79375e-05,
+      "loss": 1.1836,
+      "step": 965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7925e-05,
+      "loss": 1.3926,
+      "step": 966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79125e-05,
+      "loss": 1.3389,
+      "step": 967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79e-05,
+      "loss": 1.3604,
+      "step": 968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7887500000000006e-05,
+      "loss": 1.3701,
+      "step": 969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7875e-05,
+      "loss": 1.0205,
+      "step": 970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.78625e-05,
+      "loss": 1.2266,
+      "step": 971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7850000000000005e-05,
+      "loss": 1.3154,
+      "step": 972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7837500000000005e-05,
+      "loss": 1.2275,
+      "step": 973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7825e-05,
+      "loss": 1.252,
+      "step": 974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.78125e-05,
+      "loss": 1.4424,
+      "step": 975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.4238,
+      "step": 976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7787500000000004e-05,
+      "loss": 1.3125,
+      "step": 977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7775e-05,
+      "loss": 1.375,
+      "step": 978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.77625e-05,
+      "loss": 1.4473,
+      "step": 979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.775e-05,
+      "loss": 1.3359,
+      "step": 980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.77375e-05,
+      "loss": 1.0459,
+      "step": 981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7725e-05,
+      "loss": 1.2871,
+      "step": 982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.77125e-05,
+      "loss": 1.3369,
+      "step": 983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.77e-05,
+      "loss": 1.1621,
+      "step": 984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.76875e-05,
+      "loss": 1.4043,
+      "step": 985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7675e-05,
+      "loss": 1.5332,
+      "step": 986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.76625e-05,
+      "loss": 1.2207,
+      "step": 987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.765e-05,
+      "loss": 1.3369,
+      "step": 988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.76375e-05,
+      "loss": 1.4971,
+      "step": 989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7625e-05,
+      "loss": 1.1133,
+      "step": 990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.76125e-05,
+      "loss": 1.124,
+      "step": 991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.76e-05,
+      "loss": 1.3652,
+      "step": 992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7587500000000006e-05,
+      "loss": 1.3516,
+      "step": 993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7575e-05,
+      "loss": 1.1934,
+      "step": 994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.75625e-05,
+      "loss": 1.0791,
+      "step": 995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7550000000000005e-05,
+      "loss": 1.502,
+      "step": 996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7537500000000004e-05,
+      "loss": 1.2354,
+      "step": 997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7525e-05,
+      "loss": 1.0391,
+      "step": 998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7512500000000004e-05,
+      "loss": 1.3535,
+      "step": 999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.2197,
+      "step": 1000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74875e-05,
+      "loss": 1.2812,
+      "step": 1001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7475e-05,
+      "loss": 1.4512,
+      "step": 1002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74625e-05,
+      "loss": 1.1377,
+      "step": 1003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.745e-05,
+      "loss": 1.4092,
+      "step": 1004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74375e-05,
+      "loss": 1.1562,
+      "step": 1005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7425e-05,
+      "loss": 1.4639,
+      "step": 1006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74125e-05,
+      "loss": 1.2744,
+      "step": 1007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74e-05,
+      "loss": 1.3574,
+      "step": 1008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.73875e-05,
+      "loss": 1.2168,
+      "step": 1009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.737500000000001e-05,
+      "loss": 1.4785,
+      "step": 1010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.73625e-05,
+      "loss": 1.2695,
+      "step": 1011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.735e-05,
+      "loss": 1.085,
+      "step": 1012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7337500000000006e-05,
+      "loss": 1.0508,
+      "step": 1013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7325000000000006e-05,
+      "loss": 1.3018,
+      "step": 1014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.73125e-05,
+      "loss": 1.4287,
+      "step": 1015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.73e-05,
+      "loss": 1.3848,
+      "step": 1016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7287500000000005e-05,
+      "loss": 1.2256,
+      "step": 1017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7275000000000005e-05,
+      "loss": 1.1953,
+      "step": 1018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.72625e-05,
+      "loss": 1.3906,
+      "step": 1019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7250000000000004e-05,
+      "loss": 1.1787,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7237500000000004e-05,
+      "loss": 1.4307,
+      "step": 1021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7225000000000004e-05,
+      "loss": 1.1855,
+      "step": 1022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.72125e-05,
+      "loss": 1.3223,
+      "step": 1023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.72e-05,
+      "loss": 1.1582,
+      "step": 1024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.71875e-05,
+      "loss": 1.4531,
+      "step": 1025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7175e-05,
+      "loss": 1.1982,
+      "step": 1026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.71625e-05,
+      "loss": 1.2109,
+      "step": 1027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.715e-05,
+      "loss": 1.2783,
+      "step": 1028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.71375e-05,
+      "loss": 1.2783,
+      "step": 1029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7125e-05,
+      "loss": 1.3389,
+      "step": 1030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.71125e-05,
+      "loss": 1.0391,
+      "step": 1031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.71e-05,
+      "loss": 1.3242,
+      "step": 1032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.70875e-05,
+      "loss": 1.4072,
+      "step": 1033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.707500000000001e-05,
+      "loss": 1.3311,
+      "step": 1034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.70625e-05,
+      "loss": 1.1104,
+      "step": 1035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.705e-05,
+      "loss": 1.3896,
+      "step": 1036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7037500000000006e-05,
+      "loss": 1.4521,
+      "step": 1037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7025000000000005e-05,
+      "loss": 1.3955,
+      "step": 1038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.70125e-05,
+      "loss": 1.4727,
+      "step": 1039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7e-05,
+      "loss": 1.2686,
+      "step": 1040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6987500000000005e-05,
+      "loss": 1.2793,
+      "step": 1041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6975000000000004e-05,
+      "loss": 1.0898,
+      "step": 1042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.69625e-05,
+      "loss": 1.2764,
+      "step": 1043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6950000000000004e-05,
+      "loss": 1.3643,
+      "step": 1044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.69375e-05,
+      "loss": 1.1055,
+      "step": 1045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6925e-05,
+      "loss": 1.1992,
+      "step": 1046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.69125e-05,
+      "loss": 1.2285,
+      "step": 1047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.69e-05,
+      "loss": 1.2432,
+      "step": 1048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.68875e-05,
+      "loss": 1.3779,
+      "step": 1049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6875e-05,
+      "loss": 1.5156,
+      "step": 1050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.68625e-05,
+      "loss": 1.2217,
+      "step": 1051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.685e-05,
+      "loss": 1.375,
+      "step": 1052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.68375e-05,
+      "loss": 1.3506,
+      "step": 1053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6825e-05,
+      "loss": 1.2012,
+      "step": 1054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.68125e-05,
+      "loss": 1.0918,
+      "step": 1055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.68e-05,
+      "loss": 1.2852,
+      "step": 1056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.67875e-05,
+      "loss": 1.3252,
+      "step": 1057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6775000000000006e-05,
+      "loss": 1.4424,
+      "step": 1058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.67625e-05,
+      "loss": 1.0098,
+      "step": 1059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.675e-05,
+      "loss": 1.3145,
+      "step": 1060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6737500000000005e-05,
+      "loss": 1.3613,
+      "step": 1061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6725000000000005e-05,
+      "loss": 1.25,
+      "step": 1062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.67125e-05,
+      "loss": 1.2168,
+      "step": 1063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 1.3916,
+      "step": 1064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6687500000000004e-05,
+      "loss": 1.2998,
+      "step": 1065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6675000000000004e-05,
+      "loss": 1.2207,
+      "step": 1066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.66625e-05,
+      "loss": 1.1035,
+      "step": 1067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.665e-05,
+      "loss": 1.2275,
+      "step": 1068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.66375e-05,
+      "loss": 0.9653,
+      "step": 1069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6625e-05,
+      "loss": 1.5322,
+      "step": 1070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.66125e-05,
+      "loss": 1.1826,
+      "step": 1071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.66e-05,
+      "loss": 1.3877,
+      "step": 1072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65875e-05,
+      "loss": 1.4268,
+      "step": 1073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6575e-05,
+      "loss": 1.5703,
+      "step": 1074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65625e-05,
+      "loss": 1.1982,
+      "step": 1075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.655e-05,
+      "loss": 1.1094,
+      "step": 1076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65375e-05,
+      "loss": 1.1924,
+      "step": 1077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.652500000000001e-05,
+      "loss": 1.0469,
+      "step": 1078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65125e-05,
+      "loss": 1.3115,
+      "step": 1079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65e-05,
+      "loss": 1.3916,
+      "step": 1080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.64875e-05,
+      "loss": 1.1943,
+      "step": 1081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6475000000000006e-05,
+      "loss": 1.3232,
+      "step": 1082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.64625e-05,
+      "loss": 1.1318,
+      "step": 1083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.645e-05,
+      "loss": 1.3672,
+      "step": 1084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6437500000000005e-05,
+      "loss": 1.1924,
+      "step": 1085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6425000000000004e-05,
+      "loss": 0.9907,
+      "step": 1086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.64125e-05,
+      "loss": 1.2354,
+      "step": 1087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 1.1055,
+      "step": 1088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6387500000000004e-05,
+      "loss": 1.1699,
+      "step": 1089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6375e-05,
+      "loss": 1.3926,
+      "step": 1090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6362499999999996e-05,
+      "loss": 1.2627,
+      "step": 1091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.635e-05,
+      "loss": 1.3018,
+      "step": 1092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.63375e-05,
+      "loss": 1.1504,
+      "step": 1093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6325e-05,
+      "loss": 1.29,
+      "step": 1094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.63125e-05,
+      "loss": 1.3232,
+      "step": 1095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.63e-05,
+      "loss": 1.2324,
+      "step": 1096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.62875e-05,
+      "loss": 1.0781,
+      "step": 1097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6275e-05,
+      "loss": 1.0801,
+      "step": 1098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.62625e-05,
+      "loss": 1.4609,
+      "step": 1099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.625e-05,
+      "loss": 1.0732,
+      "step": 1100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.62375e-05,
+      "loss": 1.248,
+      "step": 1101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6225000000000006e-05,
+      "loss": 1.0537,
+      "step": 1102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.62125e-05,
+      "loss": 1.3418,
+      "step": 1103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.62e-05,
+      "loss": 1.457,
+      "step": 1104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61875e-05,
+      "loss": 1.166,
+      "step": 1105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6175000000000005e-05,
+      "loss": 1.3857,
+      "step": 1106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61625e-05,
+      "loss": 1.0391,
+      "step": 1107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.615e-05,
+      "loss": 1.5898,
+      "step": 1108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6137500000000004e-05,
+      "loss": 1.498,
+      "step": 1109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6125000000000004e-05,
+      "loss": 1.3594,
+      "step": 1110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61125e-05,
+      "loss": 1.2568,
+      "step": 1111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61e-05,
+      "loss": 1.1445,
+      "step": 1112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.60875e-05,
+      "loss": 1.2266,
+      "step": 1113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6075e-05,
+      "loss": 1.2451,
+      "step": 1114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.60625e-05,
+      "loss": 1.0137,
+      "step": 1115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.605e-05,
+      "loss": 1.4746,
+      "step": 1116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.60375e-05,
+      "loss": 1.2393,
+      "step": 1117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6025e-05,
+      "loss": 1.8584,
+      "step": 1118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.60125e-05,
+      "loss": 1.252,
+      "step": 1119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6e-05,
+      "loss": 1.4043,
+      "step": 1120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59875e-05,
+      "loss": 1.4785,
+      "step": 1121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5975e-05,
+      "loss": 1.1904,
+      "step": 1122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59625e-05,
+      "loss": 1.2168,
+      "step": 1123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.595e-05,
+      "loss": 1.293,
+      "step": 1124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59375e-05,
+      "loss": 1.3711,
+      "step": 1125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5925000000000006e-05,
+      "loss": 1.2607,
+      "step": 1126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5912500000000006e-05,
+      "loss": 1.1016,
+      "step": 1127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59e-05,
+      "loss": 1.208,
+      "step": 1128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5887500000000005e-05,
+      "loss": 1.2686,
+      "step": 1129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5875000000000005e-05,
+      "loss": 1.0693,
+      "step": 1130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5862500000000004e-05,
+      "loss": 1.1738,
+      "step": 1131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.585e-05,
+      "loss": 1.2383,
+      "step": 1132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5837500000000004e-05,
+      "loss": 1.373,
+      "step": 1133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5825000000000003e-05,
+      "loss": 1.0381,
+      "step": 1134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.58125e-05,
+      "loss": 1.4277,
+      "step": 1135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.58e-05,
+      "loss": 1.2695,
+      "step": 1136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.57875e-05,
+      "loss": 1.1982,
+      "step": 1137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5775e-05,
+      "loss": 1.5889,
+      "step": 1138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.57625e-05,
+      "loss": 1.084,
+      "step": 1139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.575e-05,
+      "loss": 1.2432,
+      "step": 1140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.57375e-05,
+      "loss": 1.1885,
+      "step": 1141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5725e-05,
+      "loss": 1.3057,
+      "step": 1142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.571250000000001e-05,
+      "loss": 1.3643,
+      "step": 1143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.57e-05,
+      "loss": 1.1201,
+      "step": 1144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56875e-05,
+      "loss": 1.3076,
+      "step": 1145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5675e-05,
+      "loss": 1.6758,
+      "step": 1146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5662500000000006e-05,
+      "loss": 1.3701,
+      "step": 1147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.565e-05,
+      "loss": 1.3486,
+      "step": 1148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56375e-05,
+      "loss": 1.2627,
+      "step": 1149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5625000000000005e-05,
+      "loss": 1.3076,
+      "step": 1150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5612500000000005e-05,
+      "loss": 1.0928,
+      "step": 1151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56e-05,
+      "loss": 1.2246,
+      "step": 1152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5587500000000004e-05,
+      "loss": 1.2949,
+      "step": 1153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5575000000000004e-05,
+      "loss": 1.0293,
+      "step": 1154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5562500000000004e-05,
+      "loss": 1.3213,
+      "step": 1155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.555e-05,
+      "loss": 1.1367,
+      "step": 1156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.55375e-05,
+      "loss": 1.1465,
+      "step": 1157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5525e-05,
+      "loss": 1.2266,
+      "step": 1158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.55125e-05,
+      "loss": 1.2197,
+      "step": 1159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.55e-05,
+      "loss": 1.0557,
+      "step": 1160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54875e-05,
+      "loss": 1.0645,
+      "step": 1161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5475e-05,
+      "loss": 1.6475,
+      "step": 1162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54625e-05,
+      "loss": 1.1816,
+      "step": 1163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.545e-05,
+      "loss": 1.1074,
+      "step": 1164
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54375e-05,
+      "loss": 1.2773,
+      "step": 1165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5425e-05,
+      "loss": 1.1895,
+      "step": 1166
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.541250000000001e-05,
+      "loss": 1.1396,
+      "step": 1167
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54e-05,
+      "loss": 1.1953,
+      "step": 1168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.53875e-05,
+      "loss": 0.8403,
+      "step": 1169
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5375e-05,
+      "loss": 1.3359,
+      "step": 1170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5362500000000006e-05,
+      "loss": 1.4443,
+      "step": 1171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.535e-05,
+      "loss": 1.1865,
+      "step": 1172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.53375e-05,
+      "loss": 1.2529,
+      "step": 1173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5325000000000005e-05,
+      "loss": 1.293,
+      "step": 1174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5312500000000005e-05,
+      "loss": 1.2588,
+      "step": 1175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.53e-05,
+      "loss": 1.0117,
+      "step": 1176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5287500000000004e-05,
+      "loss": 1.2246,
+      "step": 1177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5275000000000004e-05,
+      "loss": 1.2227,
+      "step": 1178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.52625e-05,
+      "loss": 1.3555,
+      "step": 1179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.525e-05,
+      "loss": 1.3086,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.52375e-05,
+      "loss": 1.1807,
+      "step": 1181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5225e-05,
+      "loss": 1.3203,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.52125e-05,
+      "loss": 1.2793,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.52e-05,
+      "loss": 1.0938,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.51875e-05,
+      "loss": 1.3486,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5175e-05,
+      "loss": 1.1914,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.51625e-05,
+      "loss": 1.208,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.515e-05,
+      "loss": 1.2754,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.51375e-05,
+      "loss": 1.1104,
+      "step": 1189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5125e-05,
+      "loss": 1.2646,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5112500000000006e-05,
+      "loss": 1.4062,
+      "step": 1191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.51e-05,
+      "loss": 1.6143,
+      "step": 1192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.50875e-05,
+      "loss": 1.2139,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5075000000000006e-05,
+      "loss": 1.3545,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5062500000000005e-05,
+      "loss": 1.3096,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.505e-05,
+      "loss": 1.2559,
+      "step": 1196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.50375e-05,
+      "loss": 1.542,
+      "step": 1197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5025000000000004e-05,
+      "loss": 1.3379,
+      "step": 1198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5012500000000004e-05,
+      "loss": 1.2793,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5e-05,
+      "loss": 1.3584,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4987500000000003e-05,
+      "loss": 1.248,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4975e-05,
+      "loss": 1.3799,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.49625e-05,
+      "loss": 1.2178,
+      "step": 1203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.495e-05,
+      "loss": 1.2158,
+      "step": 1204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.49375e-05,
+      "loss": 1.4082,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4925e-05,
+      "loss": 1.2969,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.49125e-05,
+      "loss": 1.4219,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.49e-05,
+      "loss": 1.2041,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.48875e-05,
+      "loss": 1.2998,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4875e-05,
+      "loss": 1.1992,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.48625e-05,
+      "loss": 1.1016,
+      "step": 1211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.485e-05,
+      "loss": 1.3154,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.48375e-05,
+      "loss": 1.2705,
+      "step": 1213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4825e-05,
+      "loss": 1.3193,
+      "step": 1214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4812500000000006e-05,
+      "loss": 1.1455,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.48e-05,
+      "loss": 1.4912,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.47875e-05,
+      "loss": 1.4463,
+      "step": 1217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4775000000000005e-05,
+      "loss": 1.3506,
+      "step": 1218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4762500000000005e-05,
+      "loss": 1.4941,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.475e-05,
+      "loss": 1.4258,
+      "step": 1220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.47375e-05,
+      "loss": 1.0801,
+      "step": 1221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4725000000000004e-05,
+      "loss": 1.3447,
+      "step": 1222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4712500000000003e-05,
+      "loss": 1.0938,
+      "step": 1223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4699999999999996e-05,
+      "loss": 1.3096,
+      "step": 1224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46875e-05,
+      "loss": 1.1729,
+      "step": 1225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4675e-05,
+      "loss": 1.4648,
+      "step": 1226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46625e-05,
+      "loss": 1.2432,
+      "step": 1227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.465e-05,
+      "loss": 1.2061,
+      "step": 1228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46375e-05,
+      "loss": 1.126,
+      "step": 1229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4625e-05,
+      "loss": 1.085,
+      "step": 1230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46125e-05,
+      "loss": 1.0498,
+      "step": 1231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46e-05,
+      "loss": 0.9463,
+      "step": 1232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.45875e-05,
+      "loss": 1.3916,
+      "step": 1233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4575e-05,
+      "loss": 1.0703,
+      "step": 1234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.45625e-05,
+      "loss": 1.3281,
+      "step": 1235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.455e-05,
+      "loss": 1.5742,
+      "step": 1236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.45375e-05,
+      "loss": 1.2344,
+      "step": 1237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4525e-05,
+      "loss": 1.3252,
+      "step": 1238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4512500000000005e-05,
+      "loss": 1.3076,
+      "step": 1239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.45e-05,
+      "loss": 1.3262,
+      "step": 1240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.44875e-05,
+      "loss": 1.2783,
+      "step": 1241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4475000000000005e-05,
+      "loss": 1.1035,
+      "step": 1242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4462500000000004e-05,
+      "loss": 1.3584,
+      "step": 1243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.445e-05,
+      "loss": 1.2373,
+      "step": 1244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4437500000000004e-05,
+      "loss": 1.335,
+      "step": 1245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4425e-05,
+      "loss": 1.3457,
+      "step": 1246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.44125e-05,
+      "loss": 1.0176,
+      "step": 1247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 1.5439,
+      "step": 1248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43875e-05,
+      "loss": 1.2422,
+      "step": 1249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4375e-05,
+      "loss": 1.498,
+      "step": 1250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43625e-05,
+      "loss": 1.4229,
+      "step": 1251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.435e-05,
+      "loss": 1.498,
+      "step": 1252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43375e-05,
+      "loss": 1.4043,
+      "step": 1253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4325e-05,
+      "loss": 1.2324,
+      "step": 1254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43125e-05,
+      "loss": 1.3486,
+      "step": 1255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 1.1729,
+      "step": 1256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.42875e-05,
+      "loss": 1.2363,
+      "step": 1257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4275e-05,
+      "loss": 1.4473,
+      "step": 1258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4262500000000006e-05,
+      "loss": 1.3438,
+      "step": 1259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4250000000000006e-05,
+      "loss": 1.2549,
+      "step": 1260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.42375e-05,
+      "loss": 1.3105,
+      "step": 1261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4225e-05,
+      "loss": 1.2568,
+      "step": 1262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4212500000000005e-05,
+      "loss": 1.2285,
+      "step": 1263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 1.458,
+      "step": 1264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.41875e-05,
+      "loss": 1.6045,
+      "step": 1265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4175000000000004e-05,
+      "loss": 1.3115,
+      "step": 1266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4162500000000004e-05,
+      "loss": 1.3838,
+      "step": 1267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.415e-05,
+      "loss": 1.2646,
+      "step": 1268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.41375e-05,
+      "loss": 1.3936,
+      "step": 1269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4125e-05,
+      "loss": 1.2324,
+      "step": 1270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.41125e-05,
+      "loss": 0.9497,
+      "step": 1271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.41e-05,
+      "loss": 0.8579,
+      "step": 1272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.40875e-05,
+      "loss": 1.3271,
+      "step": 1273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4075e-05,
+      "loss": 1.4785,
+      "step": 1274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.40625e-05,
+      "loss": 1.168,
+      "step": 1275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.405e-05,
+      "loss": 1.3164,
+      "step": 1276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.40375e-05,
+      "loss": 1.3291,
+      "step": 1277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4025e-05,
+      "loss": 1.2637,
+      "step": 1278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.40125e-05,
+      "loss": 1.2568,
+      "step": 1279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.2607,
+      "step": 1280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.39875e-05,
+      "loss": 1.3164,
+      "step": 1281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3975e-05,
+      "loss": 1.2266,
+      "step": 1282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3962500000000006e-05,
+      "loss": 1.3506,
+      "step": 1283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3950000000000005e-05,
+      "loss": 1.3555,
+      "step": 1284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.39375e-05,
+      "loss": 1.3594,
+      "step": 1285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3925e-05,
+      "loss": 1.2793,
+      "step": 1286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3912500000000004e-05,
+      "loss": 1.4092,
+      "step": 1287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 1.0908,
+      "step": 1288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.38875e-05,
+      "loss": 1.4971,
+      "step": 1289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3875000000000003e-05,
+      "loss": 1.4951,
+      "step": 1290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.38625e-05,
+      "loss": 1.1885,
+      "step": 1291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.385e-05,
+      "loss": 1.251,
+      "step": 1292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.38375e-05,
+      "loss": 1.5693,
+      "step": 1293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3825e-05,
+      "loss": 1.4561,
+      "step": 1294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.38125e-05,
+      "loss": 1.251,
+      "step": 1295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.38e-05,
+      "loss": 1.2754,
+      "step": 1296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.37875e-05,
+      "loss": 1.4287,
+      "step": 1297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3775e-05,
+      "loss": 1.0977,
+      "step": 1298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.37625e-05,
+      "loss": 1.1836,
+      "step": 1299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 1.335,
+      "step": 1300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.37375e-05,
+      "loss": 1.3125,
+      "step": 1301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3725e-05,
+      "loss": 1.5264,
+      "step": 1302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.37125e-05,
+      "loss": 1.1992,
+      "step": 1303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3700000000000006e-05,
+      "loss": 1.1211,
+      "step": 1304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.36875e-05,
+      "loss": 1.3701,
+      "step": 1305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3675e-05,
+      "loss": 1.3584,
+      "step": 1306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3662500000000005e-05,
+      "loss": 1.1543,
+      "step": 1307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3650000000000005e-05,
+      "loss": 1.3994,
+      "step": 1308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.36375e-05,
+      "loss": 1.2881,
+      "step": 1309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3625000000000004e-05,
+      "loss": 1.0586,
+      "step": 1310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3612500000000004e-05,
+      "loss": 1.2305,
+      "step": 1311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 1.1982,
+      "step": 1312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3587499999999996e-05,
+      "loss": 1.084,
+      "step": 1313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3575e-05,
+      "loss": 1.2324,
+      "step": 1314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.35625e-05,
+      "loss": 1.4414,
+      "step": 1315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.355e-05,
+      "loss": 1.2803,
+      "step": 1316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.35375e-05,
+      "loss": 1.2393,
+      "step": 1317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3525e-05,
+      "loss": 1.2148,
+      "step": 1318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.35125e-05,
+      "loss": 1.4492,
+      "step": 1319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.35e-05,
+      "loss": 1.2227,
+      "step": 1320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34875e-05,
+      "loss": 1.5527,
+      "step": 1321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3475e-05,
+      "loss": 1.0098,
+      "step": 1322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34625e-05,
+      "loss": 1.5029,
+      "step": 1323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.345000000000001e-05,
+      "loss": 1.248,
+      "step": 1324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34375e-05,
+      "loss": 1.2939,
+      "step": 1325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3425e-05,
+      "loss": 1.4971,
+      "step": 1326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34125e-05,
+      "loss": 1.2881,
+      "step": 1327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 1.4365,
+      "step": 1328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33875e-05,
+      "loss": 1.3535,
+      "step": 1329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3375e-05,
+      "loss": 1.417,
+      "step": 1330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3362500000000005e-05,
+      "loss": 1.5557,
+      "step": 1331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3350000000000004e-05,
+      "loss": 1.3154,
+      "step": 1332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33375e-05,
+      "loss": 1.2412,
+      "step": 1333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3325000000000004e-05,
+      "loss": 1.3467,
+      "step": 1334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33125e-05,
+      "loss": 1.1523,
+      "step": 1335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33e-05,
+      "loss": 1.2891,
+      "step": 1336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3287499999999996e-05,
+      "loss": 1.4531,
+      "step": 1337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3275e-05,
+      "loss": 1.4043,
+      "step": 1338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.32625e-05,
+      "loss": 1.2266,
+      "step": 1339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.325e-05,
+      "loss": 1.4219,
+      "step": 1340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.32375e-05,
+      "loss": 1.2637,
+      "step": 1341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3225e-05,
+      "loss": 1.3193,
+      "step": 1342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.32125e-05,
+      "loss": 1.2754,
+      "step": 1343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.32e-05,
+      "loss": 1.2891,
+      "step": 1344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.31875e-05,
+      "loss": 1.4824,
+      "step": 1345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3175e-05,
+      "loss": 1.2432,
+      "step": 1346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.31625e-05,
+      "loss": 1.1191,
+      "step": 1347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3150000000000006e-05,
+      "loss": 1.0879,
+      "step": 1348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.31375e-05,
+      "loss": 1.1689,
+      "step": 1349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3125e-05,
+      "loss": 1.1328,
+      "step": 1350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.31125e-05,
+      "loss": 1.2676,
+      "step": 1351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3100000000000005e-05,
+      "loss": 1.1895,
+      "step": 1352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.30875e-05,
+      "loss": 1.4727,
+      "step": 1353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3075e-05,
+      "loss": 1.3643,
+      "step": 1354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3062500000000004e-05,
+      "loss": 1.082,
+      "step": 1355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3050000000000004e-05,
+      "loss": 1.2725,
+      "step": 1356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.30375e-05,
+      "loss": 1.0879,
+      "step": 1357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3025e-05,
+      "loss": 1.4346,
+      "step": 1358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.30125e-05,
+      "loss": 1.1455,
+      "step": 1359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3e-05,
+      "loss": 1.3105,
+      "step": 1360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29875e-05,
+      "loss": 1.2744,
+      "step": 1361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2975e-05,
+      "loss": 1.2441,
+      "step": 1362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29625e-05,
+      "loss": 1.4414,
+      "step": 1363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.295e-05,
+      "loss": 0.9292,
+      "step": 1364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29375e-05,
+      "loss": 1.0918,
+      "step": 1365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2925e-05,
+      "loss": 1.0869,
+      "step": 1366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29125e-05,
+      "loss": 1.2061,
+      "step": 1367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29e-05,
+      "loss": 1.332,
+      "step": 1368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.28875e-05,
+      "loss": 1.252,
+      "step": 1369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2875e-05,
+      "loss": 1.1064,
+      "step": 1370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.28625e-05,
+      "loss": 1.1572,
+      "step": 1371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2850000000000006e-05,
+      "loss": 1.4609,
+      "step": 1372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.28375e-05,
+      "loss": 1.4668,
+      "step": 1373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2825e-05,
+      "loss": 1.1318,
+      "step": 1374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 1.2207,
+      "step": 1375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 1.2959,
+      "step": 1376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2787500000000004e-05,
+      "loss": 1.1318,
+      "step": 1377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2775e-05,
+      "loss": 1.0918,
+      "step": 1378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2762500000000004e-05,
+      "loss": 1.1631,
+      "step": 1379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.275e-05,
+      "loss": 1.3369,
+      "step": 1380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.27375e-05,
+      "loss": 1.2891,
+      "step": 1381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2725e-05,
+      "loss": 1.0879,
+      "step": 1382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.27125e-05,
+      "loss": 0.9976,
+      "step": 1383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.27e-05,
+      "loss": 1.1064,
+      "step": 1384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.26875e-05,
+      "loss": 1.6338,
+      "step": 1385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2675e-05,
+      "loss": 1.2725,
+      "step": 1386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.26625e-05,
+      "loss": 1.0498,
+      "step": 1387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.265e-05,
+      "loss": 1.0576,
+      "step": 1388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.263750000000001e-05,
+      "loss": 1.3457,
+      "step": 1389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2625e-05,
+      "loss": 1.4082,
+      "step": 1390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.26125e-05,
+      "loss": 1.1846,
+      "step": 1391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.26e-05,
+      "loss": 1.5312,
+      "step": 1392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2587500000000006e-05,
+      "loss": 1.2881,
+      "step": 1393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2575e-05,
+      "loss": 1.0508,
+      "step": 1394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.25625e-05,
+      "loss": 1.4756,
+      "step": 1395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2550000000000005e-05,
+      "loss": 1.1729,
+      "step": 1396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2537500000000005e-05,
+      "loss": 1.3506,
+      "step": 1397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2525e-05,
+      "loss": 1.0508,
+      "step": 1398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2512500000000004e-05,
+      "loss": 1.3262,
+      "step": 1399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.4307,
+      "step": 1400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2487500000000004e-05,
+      "loss": 1.376,
+      "step": 1401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2474999999999997e-05,
+      "loss": 1.3711,
+      "step": 1402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.24625e-05,
+      "loss": 1.3867,
+      "step": 1403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.245e-05,
+      "loss": 1.4453,
+      "step": 1404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.24375e-05,
+      "loss": 1.0645,
+      "step": 1405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2425e-05,
+      "loss": 1.4248,
+      "step": 1406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.24125e-05,
+      "loss": 1.2725,
+      "step": 1407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.24e-05,
+      "loss": 1.3105,
+      "step": 1408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23875e-05,
+      "loss": 1.0283,
+      "step": 1409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2375e-05,
+      "loss": 0.8887,
+      "step": 1410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23625e-05,
+      "loss": 1.3027,
+      "step": 1411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.235e-05,
+      "loss": 1.3184,
+      "step": 1412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.233750000000001e-05,
+      "loss": 1.2373,
+      "step": 1413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2325e-05,
+      "loss": 1.0898,
+      "step": 1414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23125e-05,
+      "loss": 1.0283,
+      "step": 1415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 1.3369,
+      "step": 1416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2287500000000006e-05,
+      "loss": 1.1943,
+      "step": 1417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2275e-05,
+      "loss": 1.251,
+      "step": 1418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.22625e-05,
+      "loss": 1.3359,
+      "step": 1419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2250000000000005e-05,
+      "loss": 1.1748,
+      "step": 1420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2237500000000004e-05,
+      "loss": 0.979,
+      "step": 1421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2225e-05,
+      "loss": 1.3594,
+      "step": 1422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2212500000000004e-05,
+      "loss": 1.2715,
+      "step": 1423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 1.249,
+      "step": 1424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21875e-05,
+      "loss": 1.2686,
+      "step": 1425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2175e-05,
+      "loss": 1.1328,
+      "step": 1426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21625e-05,
+      "loss": 1.46,
+      "step": 1427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.215e-05,
+      "loss": 1.2178,
+      "step": 1428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21375e-05,
+      "loss": 1.3438,
+      "step": 1429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2125e-05,
+      "loss": 1.666,
+      "step": 1430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21125e-05,
+      "loss": 1.2178,
+      "step": 1431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21e-05,
+      "loss": 1.1406,
+      "step": 1432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.20875e-05,
+      "loss": 1.2627,
+      "step": 1433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2075e-05,
+      "loss": 1.3887,
+      "step": 1434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.20625e-05,
+      "loss": 1.1943,
+      "step": 1435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.205e-05,
+      "loss": 1.5547,
+      "step": 1436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2037500000000006e-05,
+      "loss": 1.0479,
+      "step": 1437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2025e-05,
+      "loss": 1.29,
+      "step": 1438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.20125e-05,
+      "loss": 1.3008,
+      "step": 1439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.4775,
+      "step": 1440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1987500000000005e-05,
+      "loss": 1.4639,
+      "step": 1441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1975e-05,
+      "loss": 1.0547,
+      "step": 1442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.19625e-05,
+      "loss": 1.4414,
+      "step": 1443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1950000000000004e-05,
+      "loss": 1.3252,
+      "step": 1444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1937500000000004e-05,
+      "loss": 1.2832,
+      "step": 1445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1925e-05,
+      "loss": 1.3672,
+      "step": 1446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.19125e-05,
+      "loss": 1.3545,
+      "step": 1447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.19e-05,
+      "loss": 1.1006,
+      "step": 1448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18875e-05,
+      "loss": 1.2559,
+      "step": 1449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1875e-05,
+      "loss": 1.3945,
+      "step": 1450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18625e-05,
+      "loss": 1.3105,
+      "step": 1451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.185e-05,
+      "loss": 1.2461,
+      "step": 1452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18375e-05,
+      "loss": 1.1504,
+      "step": 1453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1825e-05,
+      "loss": 1.6436,
+      "step": 1454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18125e-05,
+      "loss": 1.1611,
+      "step": 1455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18e-05,
+      "loss": 1.2119,
+      "step": 1456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.17875e-05,
+      "loss": 1.2656,
+      "step": 1457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1775e-05,
+      "loss": 1.542,
+      "step": 1458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.17625e-05,
+      "loss": 1.3623,
+      "step": 1459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.175e-05,
+      "loss": 1.125,
+      "step": 1460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1737500000000006e-05,
+      "loss": 1.0186,
+      "step": 1461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1725e-05,
+      "loss": 1.3447,
+      "step": 1462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.17125e-05,
+      "loss": 1.4268,
+      "step": 1463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1700000000000005e-05,
+      "loss": 1.1885,
+      "step": 1464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1687500000000005e-05,
+      "loss": 1.2012,
+      "step": 1465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1675e-05,
+      "loss": 1.1943,
+      "step": 1466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.16625e-05,
+      "loss": 1.1689,
+      "step": 1467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1650000000000004e-05,
+      "loss": 1.0049,
+      "step": 1468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.16375e-05,
+      "loss": 1.2061,
+      "step": 1469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1624999999999996e-05,
+      "loss": 1.2256,
+      "step": 1470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.16125e-05,
+      "loss": 1.3232,
+      "step": 1471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.16e-05,
+      "loss": 1.3203,
+      "step": 1472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15875e-05,
+      "loss": 1.3027,
+      "step": 1473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1575e-05,
+      "loss": 1.2002,
+      "step": 1474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15625e-05,
+      "loss": 1.2373,
+      "step": 1475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.155e-05,
+      "loss": 1.0684,
+      "step": 1476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15375e-05,
+      "loss": 1.2471,
+      "step": 1477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1525e-05,
+      "loss": 1.25,
+      "step": 1478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15125e-05,
+      "loss": 1.2461,
+      "step": 1479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15e-05,
+      "loss": 1.3096,
+      "step": 1480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1487500000000006e-05,
+      "loss": 1.207,
+      "step": 1481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1475e-05,
+      "loss": 1.168,
+      "step": 1482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.14625e-05,
+      "loss": 1.1729,
+      "step": 1483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.145e-05,
+      "loss": 1.4189,
+      "step": 1484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1437500000000005e-05,
+      "loss": 1.3213,
+      "step": 1485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1425e-05,
+      "loss": 1.3584,
+      "step": 1486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.14125e-05,
+      "loss": 1.3311,
+      "step": 1487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 1.1914,
+      "step": 1488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1387500000000004e-05,
+      "loss": 1.2715,
+      "step": 1489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1375e-05,
+      "loss": 1.1299,
+      "step": 1490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.13625e-05,
+      "loss": 1.5498,
+      "step": 1491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.135e-05,
+      "loss": 1.2441,
+      "step": 1492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.13375e-05,
+      "loss": 1.4102,
+      "step": 1493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1324999999999996e-05,
+      "loss": 1.4072,
+      "step": 1494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.13125e-05,
+      "loss": 1.4434,
+      "step": 1495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.13e-05,
+      "loss": 1.3252,
+      "step": 1496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.12875e-05,
+      "loss": 1.3496,
+      "step": 1497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1275e-05,
+      "loss": 1.208,
+      "step": 1498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.12625e-05,
+      "loss": 1.5273,
+      "step": 1499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.125e-05,
+      "loss": 1.4863,
+      "step": 1500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.12375e-05,
+      "loss": 1.2764,
+      "step": 1501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.122500000000001e-05,
+      "loss": 1.0059,
+      "step": 1502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.12125e-05,
+      "loss": 1.2568,
+      "step": 1503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.12e-05,
+      "loss": 1.2119,
+      "step": 1504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1187500000000006e-05,
+      "loss": 1.1816,
+      "step": 1505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1175000000000006e-05,
+      "loss": 1.1943,
+      "step": 1506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.11625e-05,
+      "loss": 0.8667,
+      "step": 1507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.115e-05,
+      "loss": 1.2979,
+      "step": 1508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1137500000000005e-05,
+      "loss": 1.1182,
+      "step": 1509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1125000000000004e-05,
+      "loss": 1.0293,
+      "step": 1510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.11125e-05,
+      "loss": 1.3057,
+      "step": 1511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1100000000000004e-05,
+      "loss": 1.1123,
+      "step": 1512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1087500000000003e-05,
+      "loss": 1.1914,
+      "step": 1513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1075e-05,
+      "loss": 1.3799,
+      "step": 1514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.10625e-05,
+      "loss": 1.4736,
+      "step": 1515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.105e-05,
+      "loss": 1.2646,
+      "step": 1516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.10375e-05,
+      "loss": 1.3818,
+      "step": 1517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1025e-05,
+      "loss": 1.0898,
+      "step": 1518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.10125e-05,
+      "loss": 1.2568,
+      "step": 1519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1e-05,
+      "loss": 1.1348,
+      "step": 1520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09875e-05,
+      "loss": 1.4785,
+      "step": 1521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0975e-05,
+      "loss": 1.291,
+      "step": 1522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09625e-05,
+      "loss": 0.9766,
+      "step": 1523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.095e-05,
+      "loss": 1.3779,
+      "step": 1524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09375e-05,
+      "loss": 1.2949,
+      "step": 1525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0925000000000006e-05,
+      "loss": 1.3965,
+      "step": 1526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09125e-05,
+      "loss": 1.3711,
+      "step": 1527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09e-05,
+      "loss": 1.4717,
+      "step": 1528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0887500000000005e-05,
+      "loss": 0.8662,
+      "step": 1529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0875000000000005e-05,
+      "loss": 1.1465,
+      "step": 1530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.08625e-05,
+      "loss": 1.2754,
+      "step": 1531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0850000000000004e-05,
+      "loss": 1.3926,
+      "step": 1532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0837500000000004e-05,
+      "loss": 1.1953,
+      "step": 1533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0825000000000004e-05,
+      "loss": 1.4141,
+      "step": 1534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.08125e-05,
+      "loss": 1.3281,
+      "step": 1535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.08e-05,
+      "loss": 1.1963,
+      "step": 1536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.07875e-05,
+      "loss": 1.2715,
+      "step": 1537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0775e-05,
+      "loss": 1.4609,
+      "step": 1538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.07625e-05,
+      "loss": 1.3545,
+      "step": 1539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.075e-05,
+      "loss": 1.4775,
+      "step": 1540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.07375e-05,
+      "loss": 1.1377,
+      "step": 1541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0725e-05,
+      "loss": 1.3535,
+      "step": 1542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.07125e-05,
+      "loss": 1.4141,
+      "step": 1543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.07e-05,
+      "loss": 1.0811,
+      "step": 1544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06875e-05,
+      "loss": 0.9536,
+      "step": 1545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.067500000000001e-05,
+      "loss": 1.2568,
+      "step": 1546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06625e-05,
+      "loss": 1.1855,
+      "step": 1547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.065e-05,
+      "loss": 1.2666,
+      "step": 1548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06375e-05,
+      "loss": 1.3457,
+      "step": 1549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0625000000000006e-05,
+      "loss": 1.4102,
+      "step": 1550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06125e-05,
+      "loss": 1.3125,
+      "step": 1551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06e-05,
+      "loss": 1.2148,
+      "step": 1552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0587500000000005e-05,
+      "loss": 1.2871,
+      "step": 1553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0575000000000005e-05,
+      "loss": 1.2646,
+      "step": 1554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.05625e-05,
+      "loss": 1.2393,
+      "step": 1555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0550000000000004e-05,
+      "loss": 1.1348,
+      "step": 1556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0537500000000004e-05,
+      "loss": 1.2793,
+      "step": 1557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0525e-05,
+      "loss": 1.1709,
+      "step": 1558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.05125e-05,
+      "loss": 1.374,
+      "step": 1559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.05e-05,
+      "loss": 1.3232,
+      "step": 1560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0487500000000002e-05,
+      "loss": 1.2529,
+      "step": 1561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0475000000000002e-05,
+      "loss": 1.3037,
+      "step": 1562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04625e-05,
+      "loss": 1.3252,
+      "step": 1563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.045e-05,
+      "loss": 1.2754,
+      "step": 1564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04375e-05,
+      "loss": 1.4697,
+      "step": 1565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0425000000000004e-05,
+      "loss": 1.2637,
+      "step": 1566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04125e-05,
+      "loss": 1.3057,
+      "step": 1567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04e-05,
+      "loss": 1.3232,
+      "step": 1568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0387500000000003e-05,
+      "loss": 1.2803,
+      "step": 1569
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0375000000000003e-05,
+      "loss": 1.4551,
+      "step": 1570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.03625e-05,
+      "loss": 1.2129,
+      "step": 1571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.035e-05,
+      "loss": 1.3301,
+      "step": 1572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0337500000000002e-05,
+      "loss": 1.1367,
+      "step": 1573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0325000000000002e-05,
+      "loss": 1.4189,
+      "step": 1574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0312499999999998e-05,
+      "loss": 1.4287,
+      "step": 1575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.03e-05,
+      "loss": 1.0947,
+      "step": 1576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.02875e-05,
+      "loss": 1.2764,
+      "step": 1577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0275000000000004e-05,
+      "loss": 1.2432,
+      "step": 1578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.02625e-05,
+      "loss": 1.1133,
+      "step": 1579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.025e-05,
+      "loss": 1.0732,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0237500000000003e-05,
+      "loss": 1.3018,
+      "step": 1581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0225000000000003e-05,
+      "loss": 1.4414,
+      "step": 1582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.02125e-05,
+      "loss": 1.3701,
+      "step": 1583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.02e-05,
+      "loss": 1.3135,
+      "step": 1584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0187500000000002e-05,
+      "loss": 1.0068,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0175e-05,
+      "loss": 0.9097,
+      "step": 1586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0162499999999998e-05,
+      "loss": 1.2061,
+      "step": 1587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.015e-05,
+      "loss": 1.1377,
+      "step": 1588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.01375e-05,
+      "loss": 1.2725,
+      "step": 1589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0125000000000004e-05,
+      "loss": 1.2842,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.01125e-05,
+      "loss": 1.1221,
+      "step": 1591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.01e-05,
+      "loss": 1.2246,
+      "step": 1592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0087500000000003e-05,
+      "loss": 1.1611,
+      "step": 1593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0075000000000003e-05,
+      "loss": 1.3584,
+      "step": 1594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.00625e-05,
+      "loss": 1.54,
+      "step": 1595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0050000000000002e-05,
+      "loss": 1.3818,
+      "step": 1596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.00375e-05,
+      "loss": 1.4404,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0025000000000005e-05,
+      "loss": 1.0508,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0012499999999998e-05,
+      "loss": 1.4141,
+      "step": 1599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3e-05,
+      "loss": 1.1309,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.99875e-05,
+      "loss": 1.082,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9975000000000004e-05,
+      "loss": 1.3789,
+      "step": 1602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.99625e-05,
+      "loss": 1.2129,
+      "step": 1603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.995e-05,
+      "loss": 1.2021,
+      "step": 1604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9937500000000003e-05,
+      "loss": 1.1006,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9925000000000002e-05,
+      "loss": 1.2061,
+      "step": 1606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.99125e-05,
+      "loss": 1.124,
+      "step": 1607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 1.6035,
+      "step": 1608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.98875e-05,
+      "loss": 1.1455,
+      "step": 1609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9875000000000004e-05,
+      "loss": 1.3486,
+      "step": 1610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9862499999999997e-05,
+      "loss": 1.2529,
+      "step": 1611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.985e-05,
+      "loss": 1.2188,
+      "step": 1612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.98375e-05,
+      "loss": 1.2236,
+      "step": 1613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9825000000000003e-05,
+      "loss": 1.2773,
+      "step": 1614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.98125e-05,
+      "loss": 1.2949,
+      "step": 1615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.98e-05,
+      "loss": 1.3809,
+      "step": 1616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9787500000000002e-05,
+      "loss": 1.3115,
+      "step": 1617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9775000000000002e-05,
+      "loss": 0.8994,
+      "step": 1618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.97625e-05,
+      "loss": 1.168,
+      "step": 1619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.975e-05,
+      "loss": 1.4424,
+      "step": 1620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.97375e-05,
+      "loss": 1.5215,
+      "step": 1621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9725000000000004e-05,
+      "loss": 1.1514,
+      "step": 1622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9712499999999997e-05,
+      "loss": 1.1387,
+      "step": 1623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.97e-05,
+      "loss": 1.3037,
+      "step": 1624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96875e-05,
+      "loss": 1.543,
+      "step": 1625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9675000000000003e-05,
+      "loss": 1.5391,
+      "step": 1626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9662500000000003e-05,
+      "loss": 1.4121,
+      "step": 1627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.965e-05,
+      "loss": 1.1865,
+      "step": 1628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9637500000000002e-05,
+      "loss": 1.4648,
+      "step": 1629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9625000000000002e-05,
+      "loss": 1.2002,
+      "step": 1630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9612500000000005e-05,
+      "loss": 1.3477,
+      "step": 1631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96e-05,
+      "loss": 1.2383,
+      "step": 1632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.95875e-05,
+      "loss": 1.4541,
+      "step": 1633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9575000000000004e-05,
+      "loss": 1.1016,
+      "step": 1634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9562500000000004e-05,
+      "loss": 1.4229,
+      "step": 1635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.955e-05,
+      "loss": 1.4551,
+      "step": 1636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.95375e-05,
+      "loss": 1.1631,
+      "step": 1637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9525000000000003e-05,
+      "loss": 1.208,
+      "step": 1638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9512500000000002e-05,
+      "loss": 1.1729,
+      "step": 1639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.95e-05,
+      "loss": 1.2285,
+      "step": 1640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9487500000000002e-05,
+      "loss": 0.978,
+      "step": 1641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9475e-05,
+      "loss": 1.3906,
+      "step": 1642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9462500000000005e-05,
+      "loss": 1.5371,
+      "step": 1643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.945e-05,
+      "loss": 1.3916,
+      "step": 1644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.94375e-05,
+      "loss": 1.2686,
+      "step": 1645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9425000000000004e-05,
+      "loss": 0.9917,
+      "step": 1646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9412500000000003e-05,
+      "loss": 1.1689,
+      "step": 1647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.94e-05,
+      "loss": 1.3965,
+      "step": 1648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9387500000000003e-05,
+      "loss": 1.2627,
+      "step": 1649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9375000000000003e-05,
+      "loss": 1.4766,
+      "step": 1650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9362500000000002e-05,
+      "loss": 1.2158,
+      "step": 1651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.935e-05,
+      "loss": 1.1758,
+      "step": 1652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.93375e-05,
+      "loss": 1.3076,
+      "step": 1653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9325e-05,
+      "loss": 1.0146,
+      "step": 1654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9312500000000004e-05,
+      "loss": 1.3262,
+      "step": 1655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.93e-05,
+      "loss": 1.1084,
+      "step": 1656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.92875e-05,
+      "loss": 1.2158,
+      "step": 1657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9275000000000003e-05,
+      "loss": 1.293,
+      "step": 1658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9262500000000003e-05,
+      "loss": 1.1807,
+      "step": 1659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.925e-05,
+      "loss": 0.9556,
+      "step": 1660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9237500000000003e-05,
+      "loss": 1.3984,
+      "step": 1661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9225000000000002e-05,
+      "loss": 1.458,
+      "step": 1662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9212500000000005e-05,
+      "loss": 1.3164,
+      "step": 1663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 1.3955,
+      "step": 1664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.91875e-05,
+      "loss": 1.1465,
+      "step": 1665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9175e-05,
+      "loss": 1.1953,
+      "step": 1666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9162500000000004e-05,
+      "loss": 1.5879,
+      "step": 1667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.915e-05,
+      "loss": 1.1846,
+      "step": 1668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.91375e-05,
+      "loss": 1.25,
+      "step": 1669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9125000000000003e-05,
+      "loss": 1.3564,
+      "step": 1670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9112500000000003e-05,
+      "loss": 1.3164,
+      "step": 1671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.91e-05,
+      "loss": 1.249,
+      "step": 1672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9087500000000002e-05,
+      "loss": 1.2715,
+      "step": 1673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9075000000000002e-05,
+      "loss": 1.0967,
+      "step": 1674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9062500000000005e-05,
+      "loss": 1.3467,
+      "step": 1675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9049999999999998e-05,
+      "loss": 1.1611,
+      "step": 1676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.90375e-05,
+      "loss": 1.2773,
+      "step": 1677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9025e-05,
+      "loss": 1.1904,
+      "step": 1678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9012500000000004e-05,
+      "loss": 1.3096,
+      "step": 1679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9e-05,
+      "loss": 1.3877,
+      "step": 1680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.89875e-05,
+      "loss": 0.9272,
+      "step": 1681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8975000000000003e-05,
+      "loss": 1.2822,
+      "step": 1682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8962500000000003e-05,
+      "loss": 1.0537,
+      "step": 1683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.895e-05,
+      "loss": 1.3184,
+      "step": 1684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8937500000000002e-05,
+      "loss": 1.3125,
+      "step": 1685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8925000000000002e-05,
+      "loss": 1.4189,
+      "step": 1686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8912500000000005e-05,
+      "loss": 1.0537,
+      "step": 1687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 1.4492,
+      "step": 1688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.88875e-05,
+      "loss": 1.2305,
+      "step": 1689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8875e-05,
+      "loss": 1.3672,
+      "step": 1690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8862500000000004e-05,
+      "loss": 1.0762,
+      "step": 1691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.885e-05,
+      "loss": 1.4023,
+      "step": 1692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.88375e-05,
+      "loss": 1.0703,
+      "step": 1693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8825000000000003e-05,
+      "loss": 1.5312,
+      "step": 1694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8812500000000002e-05,
+      "loss": 1.248,
+      "step": 1695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.88e-05,
+      "loss": 1.5654,
+      "step": 1696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8787500000000002e-05,
+      "loss": 1.1475,
+      "step": 1697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8775e-05,
+      "loss": 1.4336,
+      "step": 1698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8762500000000005e-05,
+      "loss": 1.2275,
+      "step": 1699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 1.1504,
+      "step": 1700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.87375e-05,
+      "loss": 1.3906,
+      "step": 1701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8725e-05,
+      "loss": 1.1484,
+      "step": 1702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8712500000000003e-05,
+      "loss": 1.2041,
+      "step": 1703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.87e-05,
+      "loss": 1.3027,
+      "step": 1704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.86875e-05,
+      "loss": 1.0781,
+      "step": 1705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8675000000000002e-05,
+      "loss": 1.666,
+      "step": 1706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8662500000000002e-05,
+      "loss": 1.0518,
+      "step": 1707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.865e-05,
+      "loss": 1.2334,
+      "step": 1708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.86375e-05,
+      "loss": 1.3975,
+      "step": 1709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8625e-05,
+      "loss": 1.2051,
+      "step": 1710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8612500000000004e-05,
+      "loss": 1.1943,
+      "step": 1711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.86e-05,
+      "loss": 1.2129,
+      "step": 1712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.85875e-05,
+      "loss": 1.1182,
+      "step": 1713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8575000000000003e-05,
+      "loss": 1.1914,
+      "step": 1714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8562500000000003e-05,
+      "loss": 1.3281,
+      "step": 1715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.855e-05,
+      "loss": 1.0088,
+      "step": 1716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.85375e-05,
+      "loss": 1.3564,
+      "step": 1717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8525000000000002e-05,
+      "loss": 1.1055,
+      "step": 1718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8512500000000002e-05,
+      "loss": 1.7266,
+      "step": 1719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 1.1768,
+      "step": 1720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.84875e-05,
+      "loss": 1.4541,
+      "step": 1721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8475e-05,
+      "loss": 1.1484,
+      "step": 1722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8462500000000004e-05,
+      "loss": 1.0469,
+      "step": 1723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.845e-05,
+      "loss": 1.0107,
+      "step": 1724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.84375e-05,
+      "loss": 1.3311,
+      "step": 1725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8425000000000003e-05,
+      "loss": 1.5527,
+      "step": 1726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8412500000000003e-05,
+      "loss": 1.1211,
+      "step": 1727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.84e-05,
+      "loss": 1.4092,
+      "step": 1728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.83875e-05,
+      "loss": 1.4316,
+      "step": 1729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8375000000000002e-05,
+      "loss": 1.0146,
+      "step": 1730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.83625e-05,
+      "loss": 1.127,
+      "step": 1731
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8349999999999998e-05,
+      "loss": 1.665,
+      "step": 1732
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.83375e-05,
+      "loss": 1.0225,
+      "step": 1733
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8325e-05,
+      "loss": 1.5898,
+      "step": 1734
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8312500000000004e-05,
+      "loss": 1.4336,
+      "step": 1735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.83e-05,
+      "loss": 1.1475,
+      "step": 1736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.82875e-05,
+      "loss": 1.0869,
+      "step": 1737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8275000000000003e-05,
+      "loss": 1.3291,
+      "step": 1738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8262500000000003e-05,
+      "loss": 1.5645,
+      "step": 1739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.825e-05,
+      "loss": 1.1758,
+      "step": 1740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.82375e-05,
+      "loss": 1.2822,
+      "step": 1741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8225e-05,
+      "loss": 1.2871,
+      "step": 1742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.82125e-05,
+      "loss": 1.3008,
+      "step": 1743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 1.165,
+      "step": 1744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.81875e-05,
+      "loss": 1.1387,
+      "step": 1745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8175e-05,
+      "loss": 1.3135,
+      "step": 1746
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8162500000000004e-05,
+      "loss": 1.375,
+      "step": 1747
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.815e-05,
+      "loss": 1.3223,
+      "step": 1748
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.81375e-05,
+      "loss": 1.4072,
+      "step": 1749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 1.208,
+      "step": 1750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8112500000000002e-05,
+      "loss": 1.1689,
+      "step": 1751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8100000000000005e-05,
+      "loss": 1.25,
+      "step": 1752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.80875e-05,
+      "loss": 1.2148,
+      "step": 1753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8075e-05,
+      "loss": 1.4756,
+      "step": 1754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.80625e-05,
+      "loss": 1.2852,
+      "step": 1755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8050000000000004e-05,
+      "loss": 1.3145,
+      "step": 1756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.80375e-05,
+      "loss": 0.9722,
+      "step": 1757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8025e-05,
+      "loss": 1.4541,
+      "step": 1758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8012500000000003e-05,
+      "loss": 1.4609,
+      "step": 1759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.3779,
+      "step": 1760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.79875e-05,
+      "loss": 1.1064,
+      "step": 1761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7975000000000002e-05,
+      "loss": 1.2656,
+      "step": 1762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7962500000000002e-05,
+      "loss": 1.3828,
+      "step": 1763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7950000000000005e-05,
+      "loss": 1.2871,
+      "step": 1764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.79375e-05,
+      "loss": 1.3066,
+      "step": 1765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7925e-05,
+      "loss": 1.3145,
+      "step": 1766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7912500000000004e-05,
+      "loss": 1.3574,
+      "step": 1767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 1.3555,
+      "step": 1768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.78875e-05,
+      "loss": 1.4229,
+      "step": 1769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7875e-05,
+      "loss": 1.1484,
+      "step": 1770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7862500000000003e-05,
+      "loss": 1.0088,
+      "step": 1771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7850000000000003e-05,
+      "loss": 1.1104,
+      "step": 1772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.78375e-05,
+      "loss": 1.0088,
+      "step": 1773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7825000000000002e-05,
+      "loss": 1.3115,
+      "step": 1774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7812500000000002e-05,
+      "loss": 1.3906,
+      "step": 1775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.209,
+      "step": 1776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.77875e-05,
+      "loss": 1.2471,
+      "step": 1777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7775e-05,
+      "loss": 1.2646,
+      "step": 1778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7762500000000004e-05,
+      "loss": 1.2939,
+      "step": 1779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 1.2549,
+      "step": 1780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.77375e-05,
+      "loss": 1.4199,
+      "step": 1781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7725e-05,
+      "loss": 1.2988,
+      "step": 1782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7712500000000003e-05,
+      "loss": 1.3203,
+      "step": 1783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7700000000000002e-05,
+      "loss": 1.1621,
+      "step": 1784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.76875e-05,
+      "loss": 1.3037,
+      "step": 1785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7675000000000002e-05,
+      "loss": 1.3291,
+      "step": 1786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.76625e-05,
+      "loss": 1.4072,
+      "step": 1787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7650000000000005e-05,
+      "loss": 1.5225,
+      "step": 1788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.76375e-05,
+      "loss": 1.1602,
+      "step": 1789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7625e-05,
+      "loss": 1.3799,
+      "step": 1790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7612500000000004e-05,
+      "loss": 1.2041,
+      "step": 1791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 1.1855,
+      "step": 1792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.75875e-05,
+      "loss": 1.1699,
+      "step": 1793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7575e-05,
+      "loss": 1.2324,
+      "step": 1794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7562500000000002e-05,
+      "loss": 1.0869,
+      "step": 1795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7550000000000002e-05,
+      "loss": 1.0654,
+      "step": 1796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.75375e-05,
+      "loss": 1.1895,
+      "step": 1797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7525e-05,
+      "loss": 1.0576,
+      "step": 1798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.75125e-05,
+      "loss": 1.1475,
+      "step": 1799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 1.0566,
+      "step": 1800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.74875e-05,
+      "loss": 1.1445,
+      "step": 1801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7475e-05,
+      "loss": 1.4756,
+      "step": 1802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7462500000000003e-05,
+      "loss": 1.5527,
+      "step": 1803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7450000000000003e-05,
+      "loss": 1.2939,
+      "step": 1804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.74375e-05,
+      "loss": 1.3252,
+      "step": 1805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7425e-05,
+      "loss": 1.6338,
+      "step": 1806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7412500000000002e-05,
+      "loss": 1.2861,
+      "step": 1807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 0.9116,
+      "step": 1808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7387499999999998e-05,
+      "loss": 1.0986,
+      "step": 1809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7375e-05,
+      "loss": 1.1172,
+      "step": 1810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.73625e-05,
+      "loss": 1.2988,
+      "step": 1811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7350000000000004e-05,
+      "loss": 1.1338,
+      "step": 1812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.73375e-05,
+      "loss": 1.2451,
+      "step": 1813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7325e-05,
+      "loss": 1.1055,
+      "step": 1814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7312500000000003e-05,
+      "loss": 1.1484,
+      "step": 1815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 1.4248,
+      "step": 1816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.72875e-05,
+      "loss": 1.4502,
+      "step": 1817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7275e-05,
+      "loss": 1.2715,
+      "step": 1818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7262500000000002e-05,
+      "loss": 1.4385,
+      "step": 1819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.725e-05,
+      "loss": 1.4893,
+      "step": 1820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7237499999999998e-05,
+      "loss": 1.1621,
+      "step": 1821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7225e-05,
+      "loss": 1.3643,
+      "step": 1822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.72125e-05,
+      "loss": 1.3184,
+      "step": 1823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 1.2295,
+      "step": 1824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.71875e-05,
+      "loss": 1.165,
+      "step": 1825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7175e-05,
+      "loss": 1.1094,
+      "step": 1826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7162500000000003e-05,
+      "loss": 1.4385,
+      "step": 1827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7150000000000003e-05,
+      "loss": 1.4219,
+      "step": 1828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.71375e-05,
+      "loss": 1.4102,
+      "step": 1829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7125000000000002e-05,
+      "loss": 0.9448,
+      "step": 1830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7112500000000002e-05,
+      "loss": 1.2217,
+      "step": 1831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7100000000000005e-05,
+      "loss": 1.0977,
+      "step": 1832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7087499999999998e-05,
+      "loss": 1.1377,
+      "step": 1833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7075e-05,
+      "loss": 1.2373,
+      "step": 1834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.70625e-05,
+      "loss": 1.4795,
+      "step": 1835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7050000000000004e-05,
+      "loss": 1.127,
+      "step": 1836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.70375e-05,
+      "loss": 1.4404,
+      "step": 1837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7025e-05,
+      "loss": 1.374,
+      "step": 1838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7012500000000003e-05,
+      "loss": 1.5361,
+      "step": 1839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.3311,
+      "step": 1840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69875e-05,
+      "loss": 1.2881,
+      "step": 1841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6975000000000002e-05,
+      "loss": 1.0156,
+      "step": 1842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69625e-05,
+      "loss": 1.2139,
+      "step": 1843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6950000000000005e-05,
+      "loss": 0.9971,
+      "step": 1844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6937499999999997e-05,
+      "loss": 1.2119,
+      "step": 1845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6925e-05,
+      "loss": 1.334,
+      "step": 1846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69125e-05,
+      "loss": 1.4404,
+      "step": 1847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 1.2393,
+      "step": 1848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.68875e-05,
+      "loss": 1.3857,
+      "step": 1849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6875e-05,
+      "loss": 1.2021,
+      "step": 1850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6862500000000002e-05,
+      "loss": 1.0098,
+      "step": 1851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6850000000000002e-05,
+      "loss": 1.1631,
+      "step": 1852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.68375e-05,
+      "loss": 1.3164,
+      "step": 1853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6825e-05,
+      "loss": 1.1396,
+      "step": 1854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.68125e-05,
+      "loss": 1.7607,
+      "step": 1855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 1.4092,
+      "step": 1856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6787499999999997e-05,
+      "loss": 1.4297,
+      "step": 1857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6775e-05,
+      "loss": 1.2715,
+      "step": 1858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.67625e-05,
+      "loss": 1.3486,
+      "step": 1859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6750000000000003e-05,
+      "loss": 1.6211,
+      "step": 1860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.67375e-05,
+      "loss": 1.0928,
+      "step": 1861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6725e-05,
+      "loss": 1.1953,
+      "step": 1862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6712500000000002e-05,
+      "loss": 1.1455,
+      "step": 1863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 1.3398,
+      "step": 1864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6687499999999998e-05,
+      "loss": 1.0947,
+      "step": 1865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6675e-05,
+      "loss": 1.5049,
+      "step": 1866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.66625e-05,
+      "loss": 1.3828,
+      "step": 1867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6650000000000004e-05,
+      "loss": 1.3799,
+      "step": 1868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6637499999999997e-05,
+      "loss": 1.1943,
+      "step": 1869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6625e-05,
+      "loss": 1.4678,
+      "step": 1870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.66125e-05,
+      "loss": 0.9907,
+      "step": 1871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 1.418,
+      "step": 1872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.65875e-05,
+      "loss": 1.377,
+      "step": 1873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6575e-05,
+      "loss": 1.1602,
+      "step": 1874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 1.4834,
+      "step": 1875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.655e-05,
+      "loss": 1.2256,
+      "step": 1876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6537500000000005e-05,
+      "loss": 1.1309,
+      "step": 1877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6525e-05,
+      "loss": 1.2451,
+      "step": 1878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.65125e-05,
+      "loss": 1.5303,
+      "step": 1879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 1.3066,
+      "step": 1880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6487500000000003e-05,
+      "loss": 1.3008,
+      "step": 1881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6475e-05,
+      "loss": 1.3428,
+      "step": 1882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6462500000000003e-05,
+      "loss": 1.291,
+      "step": 1883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6450000000000003e-05,
+      "loss": 1.502,
+      "step": 1884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6437500000000002e-05,
+      "loss": 1.2881,
+      "step": 1885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6425e-05,
+      "loss": 1.3555,
+      "step": 1886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.64125e-05,
+      "loss": 1.0488,
+      "step": 1887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.64e-05,
+      "loss": 1.3896,
+      "step": 1888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6387500000000004e-05,
+      "loss": 1.3467,
+      "step": 1889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6375e-05,
+      "loss": 1.4736,
+      "step": 1890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.63625e-05,
+      "loss": 1.1826,
+      "step": 1891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6350000000000004e-05,
+      "loss": 1.1602,
+      "step": 1892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6337500000000003e-05,
+      "loss": 1.0684,
+      "step": 1893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6325e-05,
+      "loss": 1.0742,
+      "step": 1894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6312500000000003e-05,
+      "loss": 1.2998,
+      "step": 1895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 1.1514,
+      "step": 1896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6287500000000005e-05,
+      "loss": 0.9907,
+      "step": 1897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6275e-05,
+      "loss": 1.1709,
+      "step": 1898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.62625e-05,
+      "loss": 1.0703,
+      "step": 1899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.625e-05,
+      "loss": 1.1055,
+      "step": 1900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6237500000000004e-05,
+      "loss": 1.1816,
+      "step": 1901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6225e-05,
+      "loss": 1.376,
+      "step": 1902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.62125e-05,
+      "loss": 1.1943,
+      "step": 1903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 1.1445,
+      "step": 1904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6187500000000003e-05,
+      "loss": 1.2725,
+      "step": 1905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6175e-05,
+      "loss": 1.4775,
+      "step": 1906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6162500000000002e-05,
+      "loss": 1.0928,
+      "step": 1907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6150000000000002e-05,
+      "loss": 1.4082,
+      "step": 1908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6137500000000005e-05,
+      "loss": 1.1387,
+      "step": 1909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6124999999999998e-05,
+      "loss": 1.4648,
+      "step": 1910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.61125e-05,
+      "loss": 1.3438,
+      "step": 1911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.61e-05,
+      "loss": 1.1016,
+      "step": 1912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6087500000000004e-05,
+      "loss": 1.1289,
+      "step": 1913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6075e-05,
+      "loss": 1.0908,
+      "step": 1914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.60625e-05,
+      "loss": 1.2354,
+      "step": 1915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6050000000000003e-05,
+      "loss": 1.3955,
+      "step": 1916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6037500000000003e-05,
+      "loss": 1.2549,
+      "step": 1917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6025e-05,
+      "loss": 1.3115,
+      "step": 1918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6012500000000002e-05,
+      "loss": 1.4502,
+      "step": 1919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.2227,
+      "step": 1920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5987500000000005e-05,
+      "loss": 1.4863,
+      "step": 1921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5974999999999998e-05,
+      "loss": 1.3076,
+      "step": 1922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.59625e-05,
+      "loss": 1.0967,
+      "step": 1923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.595e-05,
+      "loss": 1.4668,
+      "step": 1924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5937500000000004e-05,
+      "loss": 1.3008,
+      "step": 1925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5925e-05,
+      "loss": 1.4375,
+      "step": 1926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.59125e-05,
+      "loss": 1.207,
+      "step": 1927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 1.1094,
+      "step": 1928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5887500000000002e-05,
+      "loss": 1.1865,
+      "step": 1929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5875e-05,
+      "loss": 1.3438,
+      "step": 1930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5862500000000002e-05,
+      "loss": 1.2051,
+      "step": 1931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.585e-05,
+      "loss": 1.2148,
+      "step": 1932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5837500000000005e-05,
+      "loss": 1.3379,
+      "step": 1933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5824999999999998e-05,
+      "loss": 1.251,
+      "step": 1934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.58125e-05,
+      "loss": 1.1357,
+      "step": 1935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.58e-05,
+      "loss": 1.3076,
+      "step": 1936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5787500000000003e-05,
+      "loss": 1.1133,
+      "step": 1937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5775e-05,
+      "loss": 1.1162,
+      "step": 1938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.57625e-05,
+      "loss": 1.3125,
+      "step": 1939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5750000000000002e-05,
+      "loss": 1.2471,
+      "step": 1940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5737500000000002e-05,
+      "loss": 0.9346,
+      "step": 1941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5725e-05,
+      "loss": 1.2158,
+      "step": 1942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.57125e-05,
+      "loss": 1.3389,
+      "step": 1943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.57e-05,
+      "loss": 1.3066,
+      "step": 1944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5687500000000004e-05,
+      "loss": 1.3994,
+      "step": 1945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5675e-05,
+      "loss": 1.0752,
+      "step": 1946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.56625e-05,
+      "loss": 1.4482,
+      "step": 1947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5650000000000003e-05,
+      "loss": 1.3711,
+      "step": 1948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5637500000000003e-05,
+      "loss": 1.334,
+      "step": 1949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5625e-05,
+      "loss": 1.2207,
+      "step": 1950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.56125e-05,
+      "loss": 1.2695,
+      "step": 1951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 1.3232,
+      "step": 1952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5587500000000002e-05,
+      "loss": 0.9185,
+      "step": 1953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5574999999999998e-05,
+      "loss": 1.4551,
+      "step": 1954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.55625e-05,
+      "loss": 1.0967,
+      "step": 1955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.555e-05,
+      "loss": 1.167,
+      "step": 1956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5537500000000004e-05,
+      "loss": 1.2549,
+      "step": 1957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5525e-05,
+      "loss": 1.5566,
+      "step": 1958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.55125e-05,
+      "loss": 1.1953,
+      "step": 1959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 1.4326,
+      "step": 1960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5487500000000003e-05,
+      "loss": 1.4883,
+      "step": 1961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5475e-05,
+      "loss": 1.1963,
+      "step": 1962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.54625e-05,
+      "loss": 1.3447,
+      "step": 1963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5450000000000002e-05,
+      "loss": 1.0713,
+      "step": 1964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.54375e-05,
+      "loss": 1.3262,
+      "step": 1965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5424999999999998e-05,
+      "loss": 1.3301,
+      "step": 1966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.54125e-05,
+      "loss": 1.4658,
+      "step": 1967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.54e-05,
+      "loss": 1.29,
+      "step": 1968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5387500000000004e-05,
+      "loss": 1.4902,
+      "step": 1969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5375e-05,
+      "loss": 1.4775,
+      "step": 1970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.53625e-05,
+      "loss": 0.9277,
+      "step": 1971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5350000000000003e-05,
+      "loss": 1.2246,
+      "step": 1972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5337500000000003e-05,
+      "loss": 1.2207,
+      "step": 1973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5325e-05,
+      "loss": 1.3555,
+      "step": 1974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.53125e-05,
+      "loss": 1.5615,
+      "step": 1975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 1.332,
+      "step": 1976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.52875e-05,
+      "loss": 1.2051,
+      "step": 1977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5274999999999998e-05,
+      "loss": 1.4277,
+      "step": 1978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.52625e-05,
+      "loss": 1.332,
+      "step": 1979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.525e-05,
+      "loss": 1.2695,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5237500000000004e-05,
+      "loss": 1.3916,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5225e-05,
+      "loss": 1.3027,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.52125e-05,
+      "loss": 1.2588,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 1.2578,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5187500000000002e-05,
+      "loss": 1.1699,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5175e-05,
+      "loss": 1.3955,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.51625e-05,
+      "loss": 1.3682,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.515e-05,
+      "loss": 1.1758,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.51375e-05,
+      "loss": 1.4287,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5124999999999997e-05,
+      "loss": 1.415,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.51125e-05,
+      "loss": 1.2275,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.51e-05,
+      "loss": 1.3594,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5087500000000003e-05,
+      "loss": 1.2637,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5075e-05,
+      "loss": 1.249,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.50625e-05,
+      "loss": 1.3301,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5050000000000002e-05,
+      "loss": 1.5488,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5037500000000002e-05,
+      "loss": 1.4238,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5025e-05,
+      "loss": 1.1572,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.50125e-05,
+      "loss": 1.4131,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1191,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.49875e-05,
+      "loss": 1.335,
+      "step": 2001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4975e-05,
+      "loss": 1.1143,
+      "step": 2002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4962500000000004e-05,
+      "loss": 1.2988,
+      "step": 2003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.495e-05,
+      "loss": 1.1709,
+      "step": 2004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4937500000000003e-05,
+      "loss": 1.0752,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4925000000000003e-05,
+      "loss": 1.3711,
+      "step": 2006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4912500000000002e-05,
+      "loss": 1.0137,
+      "step": 2007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 1.292,
+      "step": 2008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4887500000000002e-05,
+      "loss": 1.3887,
+      "step": 2009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4875e-05,
+      "loss": 1.1348,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.48625e-05,
+      "loss": 1.4004,
+      "step": 2011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.485e-05,
+      "loss": 1.2676,
+      "step": 2012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4837500000000004e-05,
+      "loss": 1.3096,
+      "step": 2013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4825e-05,
+      "loss": 1.123,
+      "step": 2014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4812500000000003e-05,
+      "loss": 1.5586,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.48e-05,
+      "loss": 1.2471,
+      "step": 2016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4787500000000003e-05,
+      "loss": 1.3037,
+      "step": 2017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4775000000000003e-05,
+      "loss": 1.2021,
+      "step": 2018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4762500000000002e-05,
+      "loss": 1.2676,
+      "step": 2019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4750000000000002e-05,
+      "loss": 1.3584,
+      "step": 2020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.47375e-05,
+      "loss": 1.2598,
+      "step": 2021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4725e-05,
+      "loss": 1.1895,
+      "step": 2022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.47125e-05,
+      "loss": 1.4346,
+      "step": 2023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.47e-05,
+      "loss": 1.3193,
+      "step": 2024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4687500000000004e-05,
+      "loss": 1.3574,
+      "step": 2025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4675e-05,
+      "loss": 1.2559,
+      "step": 2026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4662500000000003e-05,
+      "loss": 1.4121,
+      "step": 2027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.465e-05,
+      "loss": 1.2783,
+      "step": 2028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4637500000000003e-05,
+      "loss": 1.1621,
+      "step": 2029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4625000000000002e-05,
+      "loss": 1.2607,
+      "step": 2030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4612500000000002e-05,
+      "loss": 1.2939,
+      "step": 2031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.46e-05,
+      "loss": 1.0459,
+      "step": 2032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45875e-05,
+      "loss": 1.3184,
+      "step": 2033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4575e-05,
+      "loss": 1.2949,
+      "step": 2034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45625e-05,
+      "loss": 1.3584,
+      "step": 2035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.455e-05,
+      "loss": 1.4209,
+      "step": 2036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4537500000000004e-05,
+      "loss": 1.1982,
+      "step": 2037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4525e-05,
+      "loss": 1.3506,
+      "step": 2038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4512500000000003e-05,
+      "loss": 0.9731,
+      "step": 2039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45e-05,
+      "loss": 1.2197,
+      "step": 2040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4487500000000002e-05,
+      "loss": 1.5684,
+      "step": 2041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4475000000000002e-05,
+      "loss": 1.0664,
+      "step": 2042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44625e-05,
+      "loss": 1.3408,
+      "step": 2043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.445e-05,
+      "loss": 1.3613,
+      "step": 2044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44375e-05,
+      "loss": 1.4141,
+      "step": 2045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4425e-05,
+      "loss": 1.2383,
+      "step": 2046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44125e-05,
+      "loss": 1.0498,
+      "step": 2047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44e-05,
+      "loss": 0.9482,
+      "step": 2048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4387500000000003e-05,
+      "loss": 1.2168,
+      "step": 2049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4375e-05,
+      "loss": 1.1475,
+      "step": 2050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4362500000000003e-05,
+      "loss": 1.5293,
+      "step": 2051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.435e-05,
+      "loss": 1.4912,
+      "step": 2052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4337500000000002e-05,
+      "loss": 1.3691,
+      "step": 2053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4325000000000002e-05,
+      "loss": 1.2129,
+      "step": 2054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.43125e-05,
+      "loss": 1.4619,
+      "step": 2055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.43e-05,
+      "loss": 1.5059,
+      "step": 2056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.42875e-05,
+      "loss": 1.1699,
+      "step": 2057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4275e-05,
+      "loss": 1.3838,
+      "step": 2058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.42625e-05,
+      "loss": 1.084,
+      "step": 2059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.425e-05,
+      "loss": 1.3379,
+      "step": 2060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4237500000000003e-05,
+      "loss": 1.1064,
+      "step": 2061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4225e-05,
+      "loss": 1.1055,
+      "step": 2062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4212500000000002e-05,
+      "loss": 0.9912,
+      "step": 2063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 1.2412,
+      "step": 2064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4187500000000002e-05,
+      "loss": 1.498,
+      "step": 2065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4175e-05,
+      "loss": 1.3672,
+      "step": 2066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.41625e-05,
+      "loss": 1.2188,
+      "step": 2067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.415e-05,
+      "loss": 1.2715,
+      "step": 2068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.41375e-05,
+      "loss": 1.1143,
+      "step": 2069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4125e-05,
+      "loss": 1.4531,
+      "step": 2070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4112500000000003e-05,
+      "loss": 1.2451,
+      "step": 2071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.41e-05,
+      "loss": 1.1836,
+      "step": 2072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4087500000000003e-05,
+      "loss": 1.1064,
+      "step": 2073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4075e-05,
+      "loss": 1.2266,
+      "step": 2074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4062500000000002e-05,
+      "loss": 1.2695,
+      "step": 2075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4050000000000002e-05,
+      "loss": 1.4219,
+      "step": 2076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.40375e-05,
+      "loss": 0.9878,
+      "step": 2077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4025e-05,
+      "loss": 1.4014,
+      "step": 2078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.40125e-05,
+      "loss": 1.4648,
+      "step": 2079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4e-05,
+      "loss": 1.3525,
+      "step": 2080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.39875e-05,
+      "loss": 1.4082,
+      "step": 2081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3975e-05,
+      "loss": 1.3457,
+      "step": 2082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3962500000000003e-05,
+      "loss": 1.3984,
+      "step": 2083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.395e-05,
+      "loss": 1.1826,
+      "step": 2084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3937500000000002e-05,
+      "loss": 1.3965,
+      "step": 2085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3925e-05,
+      "loss": 1.0371,
+      "step": 2086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3912500000000002e-05,
+      "loss": 1.1904,
+      "step": 2087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.39e-05,
+      "loss": 1.4277,
+      "step": 2088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.38875e-05,
+      "loss": 1.2266,
+      "step": 2089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3875e-05,
+      "loss": 1.4385,
+      "step": 2090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.38625e-05,
+      "loss": 1.2559,
+      "step": 2091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.385e-05,
+      "loss": 1.3848,
+      "step": 2092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.38375e-05,
+      "loss": 1.4043,
+      "step": 2093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3825e-05,
+      "loss": 1.4707,
+      "step": 2094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3812500000000003e-05,
+      "loss": 1.0635,
+      "step": 2095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.38e-05,
+      "loss": 1.3975,
+      "step": 2096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3787500000000002e-05,
+      "loss": 1.3721,
+      "step": 2097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3775e-05,
+      "loss": 1.2939,
+      "step": 2098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.37625e-05,
+      "loss": 1.3701,
+      "step": 2099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.375e-05,
+      "loss": 1.3125,
+      "step": 2100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.37375e-05,
+      "loss": 1.1777,
+      "step": 2101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3725e-05,
+      "loss": 1.2246,
+      "step": 2102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.37125e-05,
+      "loss": 1.2656,
+      "step": 2103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.37e-05,
+      "loss": 1.2744,
+      "step": 2104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.36875e-05,
+      "loss": 1.3555,
+      "step": 2105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3675e-05,
+      "loss": 1.1816,
+      "step": 2106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3662500000000003e-05,
+      "loss": 1.2969,
+      "step": 2107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.365e-05,
+      "loss": 1.1133,
+      "step": 2108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3637500000000002e-05,
+      "loss": 1.2422,
+      "step": 2109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3624999999999998e-05,
+      "loss": 1.3008,
+      "step": 2110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.36125e-05,
+      "loss": 0.8477,
+      "step": 2111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.36e-05,
+      "loss": 0.9927,
+      "step": 2112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35875e-05,
+      "loss": 1.4229,
+      "step": 2113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3575e-05,
+      "loss": 1.375,
+      "step": 2114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35625e-05,
+      "loss": 1.2432,
+      "step": 2115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.355e-05,
+      "loss": 1.2812,
+      "step": 2116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35375e-05,
+      "loss": 1.377,
+      "step": 2117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3525e-05,
+      "loss": 1.0107,
+      "step": 2118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3512500000000002e-05,
+      "loss": 1.1455,
+      "step": 2119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35e-05,
+      "loss": 1.2744,
+      "step": 2120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3487500000000002e-05,
+      "loss": 1.1816,
+      "step": 2121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3475e-05,
+      "loss": 1.165,
+      "step": 2122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.34625e-05,
+      "loss": 1.3135,
+      "step": 2123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.345e-05,
+      "loss": 1.4453,
+      "step": 2124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.34375e-05,
+      "loss": 1.4355,
+      "step": 2125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3425000000000004e-05,
+      "loss": 1.2031,
+      "step": 2126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.34125e-05,
+      "loss": 1.5039,
+      "step": 2127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 1.5723,
+      "step": 2128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3387500000000003e-05,
+      "loss": 1.3408,
+      "step": 2129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3375000000000002e-05,
+      "loss": 1.1191,
+      "step": 2130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3362500000000002e-05,
+      "loss": 1.4404,
+      "step": 2131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3350000000000002e-05,
+      "loss": 1.2402,
+      "step": 2132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.33375e-05,
+      "loss": 1.2002,
+      "step": 2133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3325e-05,
+      "loss": 1.1875,
+      "step": 2134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.33125e-05,
+      "loss": 1.2314,
+      "step": 2135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 1.5215,
+      "step": 2136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.32875e-05,
+      "loss": 1.4609,
+      "step": 2137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3275000000000003e-05,
+      "loss": 1.293,
+      "step": 2138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.32625e-05,
+      "loss": 1.4707,
+      "step": 2139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 0.9834,
+      "step": 2140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3237500000000002e-05,
+      "loss": 1.3789,
+      "step": 2141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3225000000000002e-05,
+      "loss": 1.4531,
+      "step": 2142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3212500000000002e-05,
+      "loss": 1.2998,
+      "step": 2143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.32e-05,
+      "loss": 1.2197,
+      "step": 2144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31875e-05,
+      "loss": 1.2715,
+      "step": 2145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3175e-05,
+      "loss": 1.1953,
+      "step": 2146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31625e-05,
+      "loss": 1.3809,
+      "step": 2147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3150000000000004e-05,
+      "loss": 1.1035,
+      "step": 2148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31375e-05,
+      "loss": 1.4277,
+      "step": 2149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3125000000000003e-05,
+      "loss": 1.4814,
+      "step": 2150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31125e-05,
+      "loss": 1.4521,
+      "step": 2151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 1.334,
+      "step": 2152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3087500000000002e-05,
+      "loss": 1.4355,
+      "step": 2153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3075000000000002e-05,
+      "loss": 1.1729,
+      "step": 2154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.30625e-05,
+      "loss": 0.9692,
+      "step": 2155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.305e-05,
+      "loss": 1.3438,
+      "step": 2156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.30375e-05,
+      "loss": 1.4014,
+      "step": 2157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3025e-05,
+      "loss": 1.3027,
+      "step": 2158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.30125e-05,
+      "loss": 1.0254,
+      "step": 2159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.334,
+      "step": 2160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29875e-05,
+      "loss": 1.4697,
+      "step": 2161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2975000000000003e-05,
+      "loss": 1.3438,
+      "step": 2162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29625e-05,
+      "loss": 1.3887,
+      "step": 2163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2950000000000002e-05,
+      "loss": 1.1338,
+      "step": 2164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2937500000000002e-05,
+      "loss": 1.0771,
+      "step": 2165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2925e-05,
+      "loss": 1.2539,
+      "step": 2166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29125e-05,
+      "loss": 1.3672,
+      "step": 2167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29e-05,
+      "loss": 1.252,
+      "step": 2168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.28875e-05,
+      "loss": 1.2861,
+      "step": 2169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2875e-05,
+      "loss": 1.3135,
+      "step": 2170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.28625e-05,
+      "loss": 1.2305,
+      "step": 2171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2850000000000003e-05,
+      "loss": 1.0869,
+      "step": 2172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.28375e-05,
+      "loss": 1.2939,
+      "step": 2173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2825000000000003e-05,
+      "loss": 1.2949,
+      "step": 2174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.28125e-05,
+      "loss": 1.2051,
+      "step": 2175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 1.3086,
+      "step": 2176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.27875e-05,
+      "loss": 1.4951,
+      "step": 2177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2775e-05,
+      "loss": 1.5654,
+      "step": 2178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.27625e-05,
+      "loss": 1.0254,
+      "step": 2179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.275e-05,
+      "loss": 1.0752,
+      "step": 2180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.27375e-05,
+      "loss": 1.168,
+      "step": 2181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2725000000000003e-05,
+      "loss": 1.4385,
+      "step": 2182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.27125e-05,
+      "loss": 1.3652,
+      "step": 2183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 0.8047,
+      "step": 2184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.26875e-05,
+      "loss": 1.4619,
+      "step": 2185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2675000000000002e-05,
+      "loss": 1.2402,
+      "step": 2186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2662500000000002e-05,
+      "loss": 1.0957,
+      "step": 2187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.265e-05,
+      "loss": 1.332,
+      "step": 2188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.26375e-05,
+      "loss": 1.3633,
+      "step": 2189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2625e-05,
+      "loss": 1.5205,
+      "step": 2190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.26125e-05,
+      "loss": 1.2715,
+      "step": 2191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.26e-05,
+      "loss": 1.1504,
+      "step": 2192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25875e-05,
+      "loss": 1.3438,
+      "step": 2193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2575000000000003e-05,
+      "loss": 1.1064,
+      "step": 2194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25625e-05,
+      "loss": 1.2266,
+      "step": 2195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2550000000000003e-05,
+      "loss": 1.583,
+      "step": 2196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25375e-05,
+      "loss": 1.1289,
+      "step": 2197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2525000000000002e-05,
+      "loss": 1.4336,
+      "step": 2198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2512500000000002e-05,
+      "loss": 1.2617,
+      "step": 2199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25e-05,
+      "loss": 1.3906,
+      "step": 2200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.24875e-05,
+      "loss": 1.2256,
+      "step": 2201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2475e-05,
+      "loss": 1.6602,
+      "step": 2202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.24625e-05,
+      "loss": 1.2656,
+      "step": 2203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.245e-05,
+      "loss": 1.3096,
+      "step": 2204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.24375e-05,
+      "loss": 1.2725,
+      "step": 2205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2425000000000003e-05,
+      "loss": 1.2363,
+      "step": 2206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.24125e-05,
+      "loss": 1.5166,
+      "step": 2207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.3955,
+      "step": 2208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23875e-05,
+      "loss": 1.2285,
+      "step": 2209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2375000000000002e-05,
+      "loss": 1.4805,
+      "step": 2210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23625e-05,
+      "loss": 1.1426,
+      "step": 2211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.235e-05,
+      "loss": 1.3896,
+      "step": 2212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23375e-05,
+      "loss": 1.2412,
+      "step": 2213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2325e-05,
+      "loss": 1.3779,
+      "step": 2214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23125e-05,
+      "loss": 1.3301,
+      "step": 2215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23e-05,
+      "loss": 1.4541,
+      "step": 2216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.22875e-05,
+      "loss": 1.1084,
+      "step": 2217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2275000000000003e-05,
+      "loss": 1.0498,
+      "step": 2218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.22625e-05,
+      "loss": 1.1602,
+      "step": 2219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2250000000000002e-05,
+      "loss": 1.3223,
+      "step": 2220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.22375e-05,
+      "loss": 1.1924,
+      "step": 2221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2225e-05,
+      "loss": 1.085,
+      "step": 2222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.22125e-05,
+      "loss": 1.2305,
+      "step": 2223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.22e-05,
+      "loss": 1.2529,
+      "step": 2224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21875e-05,
+      "loss": 1.1963,
+      "step": 2225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2175e-05,
+      "loss": 1.2852,
+      "step": 2226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21625e-05,
+      "loss": 1.2373,
+      "step": 2227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.215e-05,
+      "loss": 1.2334,
+      "step": 2228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21375e-05,
+      "loss": 1.4102,
+      "step": 2229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2125000000000002e-05,
+      "loss": 1.1709,
+      "step": 2230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21125e-05,
+      "loss": 1.3252,
+      "step": 2231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 1.6309,
+      "step": 2232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2087499999999998e-05,
+      "loss": 1.1357,
+      "step": 2233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2075e-05,
+      "loss": 1.2998,
+      "step": 2234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.20625e-05,
+      "loss": 1.1992,
+      "step": 2235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.205e-05,
+      "loss": 1.3428,
+      "step": 2236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.20375e-05,
+      "loss": 1.1963,
+      "step": 2237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2025e-05,
+      "loss": 1.2979,
+      "step": 2238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.20125e-05,
+      "loss": 1.2217,
+      "step": 2239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.2129,
+      "step": 2240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.19875e-05,
+      "loss": 1.1846,
+      "step": 2241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1975000000000002e-05,
+      "loss": 1.3291,
+      "step": 2242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.19625e-05,
+      "loss": 1.2168,
+      "step": 2243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.195e-05,
+      "loss": 1.3369,
+      "step": 2244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.19375e-05,
+      "loss": 1.3506,
+      "step": 2245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1925e-05,
+      "loss": 1.082,
+      "step": 2246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.19125e-05,
+      "loss": 1.0391,
+      "step": 2247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.19e-05,
+      "loss": 1.2129,
+      "step": 2248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.18875e-05,
+      "loss": 1.5215,
+      "step": 2249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1875e-05,
+      "loss": 1.2168,
+      "step": 2250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1862500000000003e-05,
+      "loss": 1.1504,
+      "step": 2251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1850000000000003e-05,
+      "loss": 1.209,
+      "step": 2252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1837500000000002e-05,
+      "loss": 1.5088,
+      "step": 2253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1825000000000002e-05,
+      "loss": 1.2783,
+      "step": 2254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.18125e-05,
+      "loss": 1.3408,
+      "step": 2255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.18e-05,
+      "loss": 1.2725,
+      "step": 2256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.17875e-05,
+      "loss": 1.1934,
+      "step": 2257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1775e-05,
+      "loss": 1.1436,
+      "step": 2258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1762500000000004e-05,
+      "loss": 1.1143,
+      "step": 2259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.175e-05,
+      "loss": 1.1562,
+      "step": 2260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1737500000000003e-05,
+      "loss": 1.2178,
+      "step": 2261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1725e-05,
+      "loss": 1.1914,
+      "step": 2262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1712500000000003e-05,
+      "loss": 1.2598,
+      "step": 2263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 1.2568,
+      "step": 2264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1687500000000002e-05,
+      "loss": 1.3105,
+      "step": 2265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1675e-05,
+      "loss": 1.041,
+      "step": 2266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.16625e-05,
+      "loss": 1.3359,
+      "step": 2267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.165e-05,
+      "loss": 1.082,
+      "step": 2268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.16375e-05,
+      "loss": 1.2695,
+      "step": 2269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1625e-05,
+      "loss": 1.3877,
+      "step": 2270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1612500000000004e-05,
+      "loss": 1.1465,
+      "step": 2271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.16e-05,
+      "loss": 0.9707,
+      "step": 2272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1587500000000003e-05,
+      "loss": 1.3311,
+      "step": 2273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1575e-05,
+      "loss": 1.2881,
+      "step": 2274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1562500000000002e-05,
+      "loss": 1.0654,
+      "step": 2275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1550000000000002e-05,
+      "loss": 1.3447,
+      "step": 2276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1537500000000002e-05,
+      "loss": 1.1738,
+      "step": 2277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1525e-05,
+      "loss": 1.2363,
+      "step": 2278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.15125e-05,
+      "loss": 1.2383,
+      "step": 2279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.15e-05,
+      "loss": 1.2783,
+      "step": 2280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.14875e-05,
+      "loss": 1.4482,
+      "step": 2281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1475e-05,
+      "loss": 1.4658,
+      "step": 2282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1462500000000003e-05,
+      "loss": 1.4248,
+      "step": 2283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.145e-05,
+      "loss": 1.0098,
+      "step": 2284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1437500000000003e-05,
+      "loss": 1.0508,
+      "step": 2285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1425e-05,
+      "loss": 1.0918,
+      "step": 2286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1412500000000002e-05,
+      "loss": 1.2627,
+      "step": 2287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 1.2314,
+      "step": 2288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.13875e-05,
+      "loss": 1.1992,
+      "step": 2289
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1375e-05,
+      "loss": 1.3418,
+      "step": 2290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.13625e-05,
+      "loss": 1.1572,
+      "step": 2291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.135e-05,
+      "loss": 1.1787,
+      "step": 2292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.13375e-05,
+      "loss": 1.334,
+      "step": 2293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1325e-05,
+      "loss": 1.3418,
+      "step": 2294
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1312500000000003e-05,
+      "loss": 1.1211,
+      "step": 2295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.13e-05,
+      "loss": 0.9102,
+      "step": 2296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1287500000000002e-05,
+      "loss": 1.1484,
+      "step": 2297
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1275000000000002e-05,
+      "loss": 1.1875,
+      "step": 2298
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1262500000000002e-05,
+      "loss": 1.1787,
+      "step": 2299
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.125e-05,
+      "loss": 1.4072,
+      "step": 2300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.12375e-05,
+      "loss": 1.0557,
+      "step": 2301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1225e-05,
+      "loss": 1.3857,
+      "step": 2302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.12125e-05,
+      "loss": 1.2656,
+      "step": 2303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.12e-05,
+      "loss": 1.2412,
+      "step": 2304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1187500000000003e-05,
+      "loss": 1.1084,
+      "step": 2305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1175e-05,
+      "loss": 1.1455,
+      "step": 2306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1162500000000003e-05,
+      "loss": 1.2383,
+      "step": 2307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.115e-05,
+      "loss": 1.4717,
+      "step": 2308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1137500000000002e-05,
+      "loss": 1.1162,
+      "step": 2309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1125000000000002e-05,
+      "loss": 1.1221,
+      "step": 2310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.11125e-05,
+      "loss": 1.3701,
+      "step": 2311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.11e-05,
+      "loss": 1.3252,
+      "step": 2312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.10875e-05,
+      "loss": 1.1084,
+      "step": 2313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1075e-05,
+      "loss": 1.2197,
+      "step": 2314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.10625e-05,
+      "loss": 1.373,
+      "step": 2315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.105e-05,
+      "loss": 1.1318,
+      "step": 2316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1037500000000003e-05,
+      "loss": 1.2422,
+      "step": 2317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1025e-05,
+      "loss": 1.4746,
+      "step": 2318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1012500000000003e-05,
+      "loss": 1.1885,
+      "step": 2319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1e-05,
+      "loss": 1.4688,
+      "step": 2320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0987500000000002e-05,
+      "loss": 1.1094,
+      "step": 2321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0975e-05,
+      "loss": 1.2607,
+      "step": 2322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.09625e-05,
+      "loss": 1.418,
+      "step": 2323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.095e-05,
+      "loss": 1.0957,
+      "step": 2324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.09375e-05,
+      "loss": 1.1631,
+      "step": 2325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0925e-05,
+      "loss": 1.3574,
+      "step": 2326
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.09125e-05,
+      "loss": 1.1738,
+      "step": 2327
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.09e-05,
+      "loss": 1.2324,
+      "step": 2328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0887500000000003e-05,
+      "loss": 1.4404,
+      "step": 2329
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0875e-05,
+      "loss": 1.1592,
+      "step": 2330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0862500000000002e-05,
+      "loss": 1.1143,
+      "step": 2331
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.085e-05,
+      "loss": 1.3066,
+      "step": 2332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.08375e-05,
+      "loss": 1.1973,
+      "step": 2333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0825e-05,
+      "loss": 1.2549,
+      "step": 2334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.08125e-05,
+      "loss": 1.5088,
+      "step": 2335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.08e-05,
+      "loss": 1.2979,
+      "step": 2336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.07875e-05,
+      "loss": 1.2168,
+      "step": 2337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0775e-05,
+      "loss": 1.0957,
+      "step": 2338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.07625e-05,
+      "loss": 1.4443,
+      "step": 2339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.075e-05,
+      "loss": 1.4111,
+      "step": 2340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0737500000000003e-05,
+      "loss": 1.4014,
+      "step": 2341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0725e-05,
+      "loss": 1.1963,
+      "step": 2342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0712500000000002e-05,
+      "loss": 1.3486,
+      "step": 2343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.07e-05,
+      "loss": 0.9595,
+      "step": 2344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.06875e-05,
+      "loss": 1.1973,
+      "step": 2345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0675e-05,
+      "loss": 1.2607,
+      "step": 2346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.06625e-05,
+      "loss": 1.3193,
+      "step": 2347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.065e-05,
+      "loss": 1.2422,
+      "step": 2348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.06375e-05,
+      "loss": 1.3799,
+      "step": 2349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0625e-05,
+      "loss": 1.4746,
+      "step": 2350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.06125e-05,
+      "loss": 1.458,
+      "step": 2351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.06e-05,
+      "loss": 1.3809,
+      "step": 2352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0587500000000002e-05,
+      "loss": 1.2734,
+      "step": 2353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0575e-05,
+      "loss": 1.2549,
+      "step": 2354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0562500000000002e-05,
+      "loss": 1.2803,
+      "step": 2355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.055e-05,
+      "loss": 1.2158,
+      "step": 2356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.05375e-05,
+      "loss": 1.3301,
+      "step": 2357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0525e-05,
+      "loss": 1.3447,
+      "step": 2358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.05125e-05,
+      "loss": 1.1396,
+      "step": 2359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.05e-05,
+      "loss": 1.3467,
+      "step": 2360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.04875e-05,
+      "loss": 1.1982,
+      "step": 2361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0475e-05,
+      "loss": 1.3408,
+      "step": 2362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0462500000000003e-05,
+      "loss": 1.4893,
+      "step": 2363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.045e-05,
+      "loss": 1.3203,
+      "step": 2364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0437500000000002e-05,
+      "loss": 1.3867,
+      "step": 2365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0425e-05,
+      "loss": 1.3975,
+      "step": 2366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.04125e-05,
+      "loss": 1.4141,
+      "step": 2367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.04e-05,
+      "loss": 1.2041,
+      "step": 2368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.03875e-05,
+      "loss": 0.9854,
+      "step": 2369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0375e-05,
+      "loss": 1.0107,
+      "step": 2370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.03625e-05,
+      "loss": 1.207,
+      "step": 2371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.035e-05,
+      "loss": 1.4277,
+      "step": 2372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.03375e-05,
+      "loss": 1.3408,
+      "step": 2373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0325e-05,
+      "loss": 1.1426,
+      "step": 2374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 1.0,
+      "step": 2375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 1.4531,
+      "step": 2376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0287500000000002e-05,
+      "loss": 1.1504,
+      "step": 2377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0275e-05,
+      "loss": 1.1797,
+      "step": 2378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.02625e-05,
+      "loss": 1.2598,
+      "step": 2379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.025e-05,
+      "loss": 1.1299,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.02375e-05,
+      "loss": 1.4648,
+      "step": 2381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0225000000000004e-05,
+      "loss": 1.2305,
+      "step": 2382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.02125e-05,
+      "loss": 1.1836,
+      "step": 2383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 1.0625,
+      "step": 2384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.01875e-05,
+      "loss": 1.335,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0175000000000003e-05,
+      "loss": 1.4707,
+      "step": 2386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0162500000000002e-05,
+      "loss": 1.4824,
+      "step": 2387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0150000000000002e-05,
+      "loss": 1.1787,
+      "step": 2388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.01375e-05,
+      "loss": 1.3506,
+      "step": 2389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0125e-05,
+      "loss": 1.2822,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.01125e-05,
+      "loss": 1.2197,
+      "step": 2391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.01e-05,
+      "loss": 1.4795,
+      "step": 2392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.00875e-05,
+      "loss": 1.3018,
+      "step": 2393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0075000000000003e-05,
+      "loss": 1.3174,
+      "step": 2394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.00625e-05,
+      "loss": 1.3271,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0050000000000003e-05,
+      "loss": 1.1943,
+      "step": 2396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.00375e-05,
+      "loss": 1.4609,
+      "step": 2397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0025000000000002e-05,
+      "loss": 1.3047,
+      "step": 2398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0012500000000002e-05,
+      "loss": 1.1084,
+      "step": 2399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 1.3174,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99875e-05,
+      "loss": 1.5342,
+      "step": 2401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975e-05,
+      "loss": 1.1357,
+      "step": 2402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99625e-05,
+      "loss": 1.0986,
+      "step": 2403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.995e-05,
+      "loss": 1.1455,
+      "step": 2404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99375e-05,
+      "loss": 1.3799,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9925000000000003e-05,
+      "loss": 1.5479,
+      "step": 2406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99125e-05,
+      "loss": 1.3418,
+      "step": 2407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 1.1631,
+      "step": 2408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.98875e-05,
+      "loss": 1.3594,
+      "step": 2409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9875000000000002e-05,
+      "loss": 1.1719,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.98625e-05,
+      "loss": 1.4121,
+      "step": 2411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.985e-05,
+      "loss": 0.9971,
+      "step": 2412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.98375e-05,
+      "loss": 1.2988,
+      "step": 2413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9825e-05,
+      "loss": 1.3223,
+      "step": 2414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.98125e-05,
+      "loss": 1.3613,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.4385,
+      "step": 2416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.97875e-05,
+      "loss": 1.2549,
+      "step": 2417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9775000000000003e-05,
+      "loss": 1.3047,
+      "step": 2418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.97625e-05,
+      "loss": 1.1279,
+      "step": 2419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 1.3184,
+      "step": 2420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9737500000000002e-05,
+      "loss": 1.1973,
+      "step": 2421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9725000000000002e-05,
+      "loss": 1.5977,
+      "step": 2422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.97125e-05,
+      "loss": 1.2314,
+      "step": 2423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.97e-05,
+      "loss": 1.124,
+      "step": 2424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.96875e-05,
+      "loss": 1.1055,
+      "step": 2425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9675e-05,
+      "loss": 1.2275,
+      "step": 2426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.96625e-05,
+      "loss": 1.3877,
+      "step": 2427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9650000000000003e-05,
+      "loss": 1.2959,
+      "step": 2428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.96375e-05,
+      "loss": 1.3379,
+      "step": 2429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9625000000000003e-05,
+      "loss": 1.1816,
+      "step": 2430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.96125e-05,
+      "loss": 1.5234,
+      "step": 2431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.4746,
+      "step": 2432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9587500000000002e-05,
+      "loss": 1.1377,
+      "step": 2433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9575e-05,
+      "loss": 1.4033,
+      "step": 2434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95625e-05,
+      "loss": 1.209,
+      "step": 2435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955e-05,
+      "loss": 1.2676,
+      "step": 2436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95375e-05,
+      "loss": 1.3086,
+      "step": 2437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9525e-05,
+      "loss": 0.9058,
+      "step": 2438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95125e-05,
+      "loss": 1.4512,
+      "step": 2439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 1.1289,
+      "step": 2440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.94875e-05,
+      "loss": 1.4707,
+      "step": 2441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9475000000000002e-05,
+      "loss": 0.96,
+      "step": 2442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.94625e-05,
+      "loss": 0.9707,
+      "step": 2443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9450000000000002e-05,
+      "loss": 1.0244,
+      "step": 2444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.94375e-05,
+      "loss": 1.2021,
+      "step": 2445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9425e-05,
+      "loss": 1.3584,
+      "step": 2446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.94125e-05,
+      "loss": 1.1572,
+      "step": 2447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.94e-05,
+      "loss": 1.3682,
+      "step": 2448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.93875e-05,
+      "loss": 1.0264,
+      "step": 2449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9375e-05,
+      "loss": 1.3584,
+      "step": 2450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.93625e-05,
+      "loss": 1.3701,
+      "step": 2451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9350000000000003e-05,
+      "loss": 1.1689,
+      "step": 2452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.93375e-05,
+      "loss": 1.2236,
+      "step": 2453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9325000000000002e-05,
+      "loss": 1.4873,
+      "step": 2454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.93125e-05,
+      "loss": 1.3184,
+      "step": 2455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.93e-05,
+      "loss": 1.1895,
+      "step": 2456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.92875e-05,
+      "loss": 0.9536,
+      "step": 2457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9275e-05,
+      "loss": 1.5,
+      "step": 2458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.92625e-05,
+      "loss": 1.5225,
+      "step": 2459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.925e-05,
+      "loss": 1.3447,
+      "step": 2460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.92375e-05,
+      "loss": 1.084,
+      "step": 2461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9225e-05,
+      "loss": 1.1758,
+      "step": 2462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.92125e-05,
+      "loss": 1.04,
+      "step": 2463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.3086,
+      "step": 2464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.91875e-05,
+      "loss": 1.0498,
+      "step": 2465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9175000000000002e-05,
+      "loss": 1.3496,
+      "step": 2466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9162499999999998e-05,
+      "loss": 1.1826,
+      "step": 2467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.915e-05,
+      "loss": 1.1201,
+      "step": 2468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.91375e-05,
+      "loss": 1.2412,
+      "step": 2469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9125e-05,
+      "loss": 1.1631,
+      "step": 2470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.91125e-05,
+      "loss": 1.2002,
+      "step": 2471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.91e-05,
+      "loss": 1.1152,
+      "step": 2472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90875e-05,
+      "loss": 1.3232,
+      "step": 2473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9075000000000003e-05,
+      "loss": 1.3584,
+      "step": 2474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90625e-05,
+      "loss": 1.3242,
+      "step": 2475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9050000000000002e-05,
+      "loss": 1.3311,
+      "step": 2476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90375e-05,
+      "loss": 1.4473,
+      "step": 2477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9025e-05,
+      "loss": 1.2129,
+      "step": 2478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90125e-05,
+      "loss": 1.1484,
+      "step": 2479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9e-05,
+      "loss": 1.084,
+      "step": 2480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.89875e-05,
+      "loss": 1.2217,
+      "step": 2481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8975e-05,
+      "loss": 1.1406,
+      "step": 2482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.89625e-05,
+      "loss": 0.9233,
+      "step": 2483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.895e-05,
+      "loss": 1.0879,
+      "step": 2484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.89375e-05,
+      "loss": 1.209,
+      "step": 2485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8925000000000003e-05,
+      "loss": 1.3379,
+      "step": 2486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.89125e-05,
+      "loss": 1.293,
+      "step": 2487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 1.2061,
+      "step": 2488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88875e-05,
+      "loss": 1.1367,
+      "step": 2489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8875e-05,
+      "loss": 1.4229,
+      "step": 2490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88625e-05,
+      "loss": 1.2969,
+      "step": 2491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885e-05,
+      "loss": 1.2451,
+      "step": 2492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88375e-05,
+      "loss": 1.2373,
+      "step": 2493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8825e-05,
+      "loss": 1.1797,
+      "step": 2494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88125e-05,
+      "loss": 0.9692,
+      "step": 2495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88e-05,
+      "loss": 1.2881,
+      "step": 2496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.87875e-05,
+      "loss": 1.5078,
+      "step": 2497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8775000000000002e-05,
+      "loss": 1.1973,
+      "step": 2498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.87625e-05,
+      "loss": 1.2559,
+      "step": 2499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.2881,
+      "step": 2500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.87375e-05,
+      "loss": 1.3584,
+      "step": 2501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8725e-05,
+      "loss": 1.5469,
+      "step": 2502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.87125e-05,
+      "loss": 1.4395,
+      "step": 2503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.87e-05,
+      "loss": 1.1162,
+      "step": 2504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8687500000000004e-05,
+      "loss": 1.3428,
+      "step": 2505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8675e-05,
+      "loss": 1.1992,
+      "step": 2506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8662500000000003e-05,
+      "loss": 1.3174,
+      "step": 2507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.865e-05,
+      "loss": 1.3916,
+      "step": 2508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8637500000000002e-05,
+      "loss": 1.5684,
+      "step": 2509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8625000000000002e-05,
+      "loss": 1.2539,
+      "step": 2510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8612500000000002e-05,
+      "loss": 1.3242,
+      "step": 2511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.86e-05,
+      "loss": 1.4014,
+      "step": 2512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.85875e-05,
+      "loss": 1.3008,
+      "step": 2513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8575e-05,
+      "loss": 1.3213,
+      "step": 2514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.85625e-05,
+      "loss": 1.3799,
+      "step": 2515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.855e-05,
+      "loss": 1.5264,
+      "step": 2516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8537500000000003e-05,
+      "loss": 1.332,
+      "step": 2517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8525e-05,
+      "loss": 1.3408,
+      "step": 2518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8512500000000003e-05,
+      "loss": 1.3154,
+      "step": 2519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.85e-05,
+      "loss": 1.4883,
+      "step": 2520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8487500000000002e-05,
+      "loss": 0.9668,
+      "step": 2521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8475000000000002e-05,
+      "loss": 1.1094,
+      "step": 2522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.84625e-05,
+      "loss": 1.1055,
+      "step": 2523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.845e-05,
+      "loss": 1.3369,
+      "step": 2524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.84375e-05,
+      "loss": 1.4043,
+      "step": 2525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8425e-05,
+      "loss": 1.4717,
+      "step": 2526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.84125e-05,
+      "loss": 1.1113,
+      "step": 2527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.84e-05,
+      "loss": 1.0977,
+      "step": 2528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8387500000000003e-05,
+      "loss": 1.3281,
+      "step": 2529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375e-05,
+      "loss": 1.377,
+      "step": 2530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362500000000002e-05,
+      "loss": 1.209,
+      "step": 2531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350000000000002e-05,
+      "loss": 1.2227,
+      "step": 2532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337500000000002e-05,
+      "loss": 1.1982,
+      "step": 2533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325e-05,
+      "loss": 1.126,
+      "step": 2534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83125e-05,
+      "loss": 1.3359,
+      "step": 2535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83e-05,
+      "loss": 1.2207,
+      "step": 2536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.82875e-05,
+      "loss": 1.0479,
+      "step": 2537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8275e-05,
+      "loss": 1.1377,
+      "step": 2538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8262500000000003e-05,
+      "loss": 1.2715,
+      "step": 2539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825e-05,
+      "loss": 1.1748,
+      "step": 2540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8237500000000003e-05,
+      "loss": 1.3994,
+      "step": 2541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8225e-05,
+      "loss": 1.6729,
+      "step": 2542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8212500000000002e-05,
+      "loss": 1.3145,
+      "step": 2543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 1.1104,
+      "step": 2544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.81875e-05,
+      "loss": 1.4678,
+      "step": 2545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8175e-05,
+      "loss": 1.3916,
+      "step": 2546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.81625e-05,
+      "loss": 1.2383,
+      "step": 2547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.815e-05,
+      "loss": 1.3887,
+      "step": 2548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.81375e-05,
+      "loss": 1.3662,
+      "step": 2549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8125e-05,
+      "loss": 1.2314,
+      "step": 2550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8112500000000003e-05,
+      "loss": 1.3213,
+      "step": 2551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.81e-05,
+      "loss": 1.2471,
+      "step": 2552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8087500000000003e-05,
+      "loss": 1.3682,
+      "step": 2553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8075e-05,
+      "loss": 1.1855,
+      "step": 2554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8062500000000002e-05,
+      "loss": 1.3809,
+      "step": 2555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.805e-05,
+      "loss": 1.5498,
+      "step": 2556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.80375e-05,
+      "loss": 1.2324,
+      "step": 2557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8025e-05,
+      "loss": 1.2559,
+      "step": 2558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.80125e-05,
+      "loss": 1.2324,
+      "step": 2559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8e-05,
+      "loss": 1.2129,
+      "step": 2560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79875e-05,
+      "loss": 1.0293,
+      "step": 2561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7975e-05,
+      "loss": 1.0811,
+      "step": 2562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7962500000000003e-05,
+      "loss": 1.627,
+      "step": 2563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.795e-05,
+      "loss": 1.3027,
+      "step": 2564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7937500000000002e-05,
+      "loss": 1.1025,
+      "step": 2565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7925e-05,
+      "loss": 1.3574,
+      "step": 2566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7912500000000002e-05,
+      "loss": 1.4414,
+      "step": 2567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79e-05,
+      "loss": 1.1104,
+      "step": 2568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.78875e-05,
+      "loss": 1.4092,
+      "step": 2569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7875e-05,
+      "loss": 1.2314,
+      "step": 2570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.78625e-05,
+      "loss": 1.083,
+      "step": 2571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.785e-05,
+      "loss": 1.1494,
+      "step": 2572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.78375e-05,
+      "loss": 1.3535,
+      "step": 2573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7825e-05,
+      "loss": 1.2744,
+      "step": 2574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7812500000000003e-05,
+      "loss": 1.0273,
+      "step": 2575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.78e-05,
+      "loss": 1.46,
+      "step": 2576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7787500000000002e-05,
+      "loss": 1.1709,
+      "step": 2577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7775e-05,
+      "loss": 1.2939,
+      "step": 2578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.77625e-05,
+      "loss": 1.4736,
+      "step": 2579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.775e-05,
+      "loss": 1.4648,
+      "step": 2580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.77375e-05,
+      "loss": 1.3789,
+      "step": 2581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7725e-05,
+      "loss": 1.1592,
+      "step": 2582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.77125e-05,
+      "loss": 1.0508,
+      "step": 2583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.77e-05,
+      "loss": 0.9014,
+      "step": 2584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.76875e-05,
+      "loss": 1.2695,
+      "step": 2585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7675e-05,
+      "loss": 1.1357,
+      "step": 2586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7662500000000002e-05,
+      "loss": 1.2812,
+      "step": 2587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.765e-05,
+      "loss": 0.9966,
+      "step": 2588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7637500000000002e-05,
+      "loss": 1.0898,
+      "step": 2589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7625e-05,
+      "loss": 1.04,
+      "step": 2590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.76125e-05,
+      "loss": 1.4512,
+      "step": 2591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.76e-05,
+      "loss": 1.5039,
+      "step": 2592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.75875e-05,
+      "loss": 1.2422,
+      "step": 2593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7575e-05,
+      "loss": 1.4307,
+      "step": 2594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.75625e-05,
+      "loss": 1.2217,
+      "step": 2595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.755e-05,
+      "loss": 1.3955,
+      "step": 2596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7537500000000003e-05,
+      "loss": 1.2939,
+      "step": 2597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7525e-05,
+      "loss": 1.085,
+      "step": 2598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7512500000000002e-05,
+      "loss": 1.3174,
+      "step": 2599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.75e-05,
+      "loss": 1.3203,
+      "step": 2600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.74875e-05,
+      "loss": 1.2441,
+      "step": 2601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7475e-05,
+      "loss": 1.4746,
+      "step": 2602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.74625e-05,
+      "loss": 1.0117,
+      "step": 2603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.745e-05,
+      "loss": 1.2852,
+      "step": 2604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.74375e-05,
+      "loss": 1.4551,
+      "step": 2605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7425e-05,
+      "loss": 1.3408,
+      "step": 2606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.74125e-05,
+      "loss": 1.1279,
+      "step": 2607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.74e-05,
+      "loss": 1.2188,
+      "step": 2608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7387500000000003e-05,
+      "loss": 1.1465,
+      "step": 2609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7375e-05,
+      "loss": 1.2549,
+      "step": 2610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7362500000000002e-05,
+      "loss": 1.2744,
+      "step": 2611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7349999999999998e-05,
+      "loss": 1.2344,
+      "step": 2612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.73375e-05,
+      "loss": 0.9639,
+      "step": 2613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7325e-05,
+      "loss": 1.3447,
+      "step": 2614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.73125e-05,
+      "loss": 1.3799,
+      "step": 2615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.73e-05,
+      "loss": 1.2783,
+      "step": 2616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.72875e-05,
+      "loss": 1.3975,
+      "step": 2617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7275e-05,
+      "loss": 1.376,
+      "step": 2618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.72625e-05,
+      "loss": 1.2764,
+      "step": 2619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.725e-05,
+      "loss": 1.5625,
+      "step": 2620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7237500000000002e-05,
+      "loss": 1.3027,
+      "step": 2621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7225e-05,
+      "loss": 1.4463,
+      "step": 2622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.72125e-05,
+      "loss": 1.252,
+      "step": 2623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 1.2666,
+      "step": 2624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.9683,
+      "step": 2625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7175e-05,
+      "loss": 1.1855,
+      "step": 2626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.71625e-05,
+      "loss": 1.4229,
+      "step": 2627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7150000000000004e-05,
+      "loss": 1.1416,
+      "step": 2628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.71375e-05,
+      "loss": 0.8525,
+      "step": 2629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7125000000000003e-05,
+      "loss": 1.4766,
+      "step": 2630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.71125e-05,
+      "loss": 1.04,
+      "step": 2631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 0.9473,
+      "step": 2632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7087500000000002e-05,
+      "loss": 1.2451,
+      "step": 2633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7075e-05,
+      "loss": 1.2178,
+      "step": 2634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.70625e-05,
+      "loss": 1.3428,
+      "step": 2635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.705e-05,
+      "loss": 1.3789,
+      "step": 2636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.70375e-05,
+      "loss": 1.2871,
+      "step": 2637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7025e-05,
+      "loss": 1.4492,
+      "step": 2638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.70125e-05,
+      "loss": 1.2549,
+      "step": 2639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.376,
+      "step": 2640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.69875e-05,
+      "loss": 1.2783,
+      "step": 2641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6975000000000003e-05,
+      "loss": 1.0684,
+      "step": 2642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.69625e-05,
+      "loss": 1.0029,
+      "step": 2643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6950000000000002e-05,
+      "loss": 1.373,
+      "step": 2644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6937500000000002e-05,
+      "loss": 1.3818,
+      "step": 2645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6925e-05,
+      "loss": 1.418,
+      "step": 2646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.69125e-05,
+      "loss": 1.4111,
+      "step": 2647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.69e-05,
+      "loss": 1.2764,
+      "step": 2648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.68875e-05,
+      "loss": 1.2559,
+      "step": 2649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6875000000000004e-05,
+      "loss": 1.4434,
+      "step": 2650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.68625e-05,
+      "loss": 1.1602,
+      "step": 2651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6850000000000003e-05,
+      "loss": 1.1914,
+      "step": 2652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.68375e-05,
+      "loss": 1.3691,
+      "step": 2653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6825000000000002e-05,
+      "loss": 1.2881,
+      "step": 2654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6812500000000002e-05,
+      "loss": 1.3428,
+      "step": 2655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.2432,
+      "step": 2656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.67875e-05,
+      "loss": 1.2471,
+      "step": 2657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6775e-05,
+      "loss": 1.1924,
+      "step": 2658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.67625e-05,
+      "loss": 1.5146,
+      "step": 2659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.675e-05,
+      "loss": 1.4053,
+      "step": 2660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.67375e-05,
+      "loss": 1.3975,
+      "step": 2661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6725000000000003e-05,
+      "loss": 1.3369,
+      "step": 2662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.67125e-05,
+      "loss": 0.9771,
+      "step": 2663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 1.0342,
+      "step": 2664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66875e-05,
+      "loss": 1.0586,
+      "step": 2665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6675000000000002e-05,
+      "loss": 1.3848,
+      "step": 2666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6662500000000002e-05,
+      "loss": 1.1855,
+      "step": 2667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.665e-05,
+      "loss": 1.3486,
+      "step": 2668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66375e-05,
+      "loss": 1.1338,
+      "step": 2669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6625e-05,
+      "loss": 1.3213,
+      "step": 2670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66125e-05,
+      "loss": 1.1846,
+      "step": 2671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66e-05,
+      "loss": 1.1172,
+      "step": 2672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.65875e-05,
+      "loss": 1.2656,
+      "step": 2673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6575000000000003e-05,
+      "loss": 1.0684,
+      "step": 2674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.65625e-05,
+      "loss": 1.0498,
+      "step": 2675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6550000000000002e-05,
+      "loss": 1.2939,
+      "step": 2676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.65375e-05,
+      "loss": 1.4268,
+      "step": 2677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6525000000000002e-05,
+      "loss": 1.2793,
+      "step": 2678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.65125e-05,
+      "loss": 1.3213,
+      "step": 2679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.65e-05,
+      "loss": 1.2969,
+      "step": 2680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.64875e-05,
+      "loss": 1.1602,
+      "step": 2681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6475e-05,
+      "loss": 1.4346,
+      "step": 2682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.64625e-05,
+      "loss": 1.2549,
+      "step": 2683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.645e-05,
+      "loss": 1.3037,
+      "step": 2684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.64375e-05,
+      "loss": 1.2803,
+      "step": 2685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6425000000000003e-05,
+      "loss": 1.2354,
+      "step": 2686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.64125e-05,
+      "loss": 1.2344,
+      "step": 2687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 1.0938,
+      "step": 2688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.63875e-05,
+      "loss": 1.3613,
+      "step": 2689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6375e-05,
+      "loss": 1.1406,
+      "step": 2690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.63625e-05,
+      "loss": 1.1699,
+      "step": 2691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.635e-05,
+      "loss": 1.4326,
+      "step": 2692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.63375e-05,
+      "loss": 1.3906,
+      "step": 2693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6325e-05,
+      "loss": 1.3164,
+      "step": 2694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.63125e-05,
+      "loss": 1.0957,
+      "step": 2695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.63e-05,
+      "loss": 1.3398,
+      "step": 2696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.62875e-05,
+      "loss": 1.3164,
+      "step": 2697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6275000000000003e-05,
+      "loss": 1.3027,
+      "step": 2698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.62625e-05,
+      "loss": 1.0557,
+      "step": 2699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 1.0996,
+      "step": 2700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6237499999999998e-05,
+      "loss": 1.1592,
+      "step": 2701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6225e-05,
+      "loss": 1.293,
+      "step": 2702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.62125e-05,
+      "loss": 1.21,
+      "step": 2703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.62e-05,
+      "loss": 1.5342,
+      "step": 2704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.61875e-05,
+      "loss": 1.1377,
+      "step": 2705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6175e-05,
+      "loss": 1.1641,
+      "step": 2706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.61625e-05,
+      "loss": 1.2939,
+      "step": 2707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6150000000000003e-05,
+      "loss": 1.3379,
+      "step": 2708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.61375e-05,
+      "loss": 1.0361,
+      "step": 2709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6125000000000002e-05,
+      "loss": 1.166,
+      "step": 2710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.61125e-05,
+      "loss": 1.3994,
+      "step": 2711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 1.2568,
+      "step": 2712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.60875e-05,
+      "loss": 1.2656,
+      "step": 2713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6075e-05,
+      "loss": 1.3311,
+      "step": 2714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.60625e-05,
+      "loss": 1.5566,
+      "step": 2715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.605e-05,
+      "loss": 1.084,
+      "step": 2716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.60375e-05,
+      "loss": 1.5391,
+      "step": 2717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6025e-05,
+      "loss": 1.54,
+      "step": 2718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.60125e-05,
+      "loss": 1.5117,
+      "step": 2719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.3477,
+      "step": 2720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59875e-05,
+      "loss": 1.2305,
+      "step": 2721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5975000000000002e-05,
+      "loss": 1.126,
+      "step": 2722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59625e-05,
+      "loss": 1.4355,
+      "step": 2723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.595e-05,
+      "loss": 1.4414,
+      "step": 2724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59375e-05,
+      "loss": 1.4004,
+      "step": 2725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5925e-05,
+      "loss": 1.0625,
+      "step": 2726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59125e-05,
+      "loss": 1.3486,
+      "step": 2727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59e-05,
+      "loss": 1.166,
+      "step": 2728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.58875e-05,
+      "loss": 1.2578,
+      "step": 2729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5875e-05,
+      "loss": 1.0752,
+      "step": 2730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.58625e-05,
+      "loss": 1.4844,
+      "step": 2731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5850000000000002e-05,
+      "loss": 1.1553,
+      "step": 2732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.58375e-05,
+      "loss": 1.6562,
+      "step": 2733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5825000000000002e-05,
+      "loss": 1.042,
+      "step": 2734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5812499999999998e-05,
+      "loss": 1.2852,
+      "step": 2735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.58e-05,
+      "loss": 1.2168,
+      "step": 2736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.57875e-05,
+      "loss": 1.2178,
+      "step": 2737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5775e-05,
+      "loss": 1.335,
+      "step": 2738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.57625e-05,
+      "loss": 1.2188,
+      "step": 2739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.575e-05,
+      "loss": 1.4346,
+      "step": 2740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.57375e-05,
+      "loss": 1.332,
+      "step": 2741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5725e-05,
+      "loss": 1.3359,
+      "step": 2742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.57125e-05,
+      "loss": 1.2041,
+      "step": 2743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 1.2783,
+      "step": 2744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.56875e-05,
+      "loss": 1.4648,
+      "step": 2745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5675e-05,
+      "loss": 1.0791,
+      "step": 2746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5662499999999998e-05,
+      "loss": 1.2539,
+      "step": 2747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.565e-05,
+      "loss": 1.0674,
+      "step": 2748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.56375e-05,
+      "loss": 1.5889,
+      "step": 2749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5625e-05,
+      "loss": 1.3857,
+      "step": 2750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5612500000000003e-05,
+      "loss": 1.4014,
+      "step": 2751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.56e-05,
+      "loss": 1.2598,
+      "step": 2752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5587500000000003e-05,
+      "loss": 1.3066,
+      "step": 2753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5575e-05,
+      "loss": 1.3857,
+      "step": 2754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5562500000000002e-05,
+      "loss": 1.0508,
+      "step": 2755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5550000000000002e-05,
+      "loss": 1.3291,
+      "step": 2756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.55375e-05,
+      "loss": 1.1943,
+      "step": 2757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5525e-05,
+      "loss": 1.2959,
+      "step": 2758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.55125e-05,
+      "loss": 1.1592,
+      "step": 2759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.55e-05,
+      "loss": 1.0283,
+      "step": 2760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.54875e-05,
+      "loss": 1.5117,
+      "step": 2761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5475e-05,
+      "loss": 0.9194,
+      "step": 2762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5462500000000003e-05,
+      "loss": 1.2939,
+      "step": 2763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.545e-05,
+      "loss": 1.4258,
+      "step": 2764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5437500000000003e-05,
+      "loss": 0.9976,
+      "step": 2765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5425000000000002e-05,
+      "loss": 1.2607,
+      "step": 2766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5412500000000002e-05,
+      "loss": 1.1709,
+      "step": 2767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.54e-05,
+      "loss": 1.4092,
+      "step": 2768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.53875e-05,
+      "loss": 1.1465,
+      "step": 2769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5375e-05,
+      "loss": 1.0117,
+      "step": 2770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.53625e-05,
+      "loss": 1.2881,
+      "step": 2771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.535e-05,
+      "loss": 1.1816,
+      "step": 2772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5337500000000004e-05,
+      "loss": 1.2842,
+      "step": 2773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5325e-05,
+      "loss": 0.9487,
+      "step": 2774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5312500000000003e-05,
+      "loss": 1.1729,
+      "step": 2775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.53e-05,
+      "loss": 1.2793,
+      "step": 2776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5287500000000002e-05,
+      "loss": 1.3584,
+      "step": 2777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5275000000000002e-05,
+      "loss": 1.4229,
+      "step": 2778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.52625e-05,
+      "loss": 1.4521,
+      "step": 2779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.525e-05,
+      "loss": 1.3076,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5237500000000001e-05,
+      "loss": 1.0596,
+      "step": 2781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5225e-05,
+      "loss": 1.2471,
+      "step": 2782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5212500000000002e-05,
+      "loss": 1.2119,
+      "step": 2783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.52e-05,
+      "loss": 1.502,
+      "step": 2784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5187500000000002e-05,
+      "loss": 1.2998,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5175e-05,
+      "loss": 1.0625,
+      "step": 2786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5162500000000001e-05,
+      "loss": 1.2432,
+      "step": 2787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.515e-05,
+      "loss": 1.1807,
+      "step": 2788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5137500000000002e-05,
+      "loss": 1.3545,
+      "step": 2789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5125e-05,
+      "loss": 1.2734,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5112500000000001e-05,
+      "loss": 1.334,
+      "step": 2791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.51e-05,
+      "loss": 1.3066,
+      "step": 2792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.50875e-05,
+      "loss": 1.4473,
+      "step": 2793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5075e-05,
+      "loss": 1.0371,
+      "step": 2794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5062500000000002e-05,
+      "loss": 1.0957,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.505e-05,
+      "loss": 1.1992,
+      "step": 2796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5037500000000001e-05,
+      "loss": 1.4238,
+      "step": 2797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5025000000000001e-05,
+      "loss": 1.3086,
+      "step": 2798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5012500000000002e-05,
+      "loss": 1.3076,
+      "step": 2799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5e-05,
+      "loss": 1.2314,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4987500000000002e-05,
+      "loss": 1.1211,
+      "step": 2801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4975e-05,
+      "loss": 1.6826,
+      "step": 2802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4962500000000001e-05,
+      "loss": 1.2773,
+      "step": 2803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4950000000000001e-05,
+      "loss": 1.2451,
+      "step": 2804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4937500000000002e-05,
+      "loss": 1.2715,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4925e-05,
+      "loss": 1.3184,
+      "step": 2806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4912500000000002e-05,
+      "loss": 1.2305,
+      "step": 2807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.49e-05,
+      "loss": 1.2168,
+      "step": 2808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4887500000000001e-05,
+      "loss": 1.0957,
+      "step": 2809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4875e-05,
+      "loss": 1.2754,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4862500000000002e-05,
+      "loss": 1.0674,
+      "step": 2811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.485e-05,
+      "loss": 1.2754,
+      "step": 2812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4837500000000002e-05,
+      "loss": 1.3535,
+      "step": 2813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4825e-05,
+      "loss": 0.9873,
+      "step": 2814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4812500000000001e-05,
+      "loss": 1.0244,
+      "step": 2815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.48e-05,
+      "loss": 1.2725,
+      "step": 2816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4787500000000002e-05,
+      "loss": 1.1113,
+      "step": 2817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4775e-05,
+      "loss": 1.4053,
+      "step": 2818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4762500000000001e-05,
+      "loss": 1.3174,
+      "step": 2819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.475e-05,
+      "loss": 1.4434,
+      "step": 2820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.47375e-05,
+      "loss": 1.3037,
+      "step": 2821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4725e-05,
+      "loss": 1.1309,
+      "step": 2822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4712500000000002e-05,
+      "loss": 1.0859,
+      "step": 2823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.47e-05,
+      "loss": 0.9502,
+      "step": 2824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4687500000000001e-05,
+      "loss": 1.1543,
+      "step": 2825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4675e-05,
+      "loss": 1.0742,
+      "step": 2826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.46625e-05,
+      "loss": 1.2754,
+      "step": 2827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.465e-05,
+      "loss": 1.1123,
+      "step": 2828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4637500000000002e-05,
+      "loss": 1.2461,
+      "step": 2829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4625e-05,
+      "loss": 1.3008,
+      "step": 2830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4612500000000001e-05,
+      "loss": 1.2861,
+      "step": 2831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 1.4375,
+      "step": 2832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.45875e-05,
+      "loss": 1.3359,
+      "step": 2833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4575e-05,
+      "loss": 1.1191,
+      "step": 2834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4562500000000002e-05,
+      "loss": 1.335,
+      "step": 2835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.455e-05,
+      "loss": 1.4248,
+      "step": 2836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4537500000000001e-05,
+      "loss": 1.1533,
+      "step": 2837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4524999999999999e-05,
+      "loss": 1.209,
+      "step": 2838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.45125e-05,
+      "loss": 1.4268,
+      "step": 2839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.45e-05,
+      "loss": 1.3564,
+      "step": 2840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4487500000000001e-05,
+      "loss": 1.3574,
+      "step": 2841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4475e-05,
+      "loss": 1.0664,
+      "step": 2842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4462500000000001e-05,
+      "loss": 1.3867,
+      "step": 2843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4449999999999999e-05,
+      "loss": 1.1904,
+      "step": 2844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.44375e-05,
+      "loss": 1.0234,
+      "step": 2845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4425e-05,
+      "loss": 1.5762,
+      "step": 2846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4412500000000001e-05,
+      "loss": 1.2568,
+      "step": 2847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.44e-05,
+      "loss": 1.4619,
+      "step": 2848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.43875e-05,
+      "loss": 1.3867,
+      "step": 2849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4374999999999999e-05,
+      "loss": 1.1816,
+      "step": 2850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.43625e-05,
+      "loss": 1.3691,
+      "step": 2851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.435e-05,
+      "loss": 1.1885,
+      "step": 2852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4337500000000001e-05,
+      "loss": 1.4307,
+      "step": 2853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4325e-05,
+      "loss": 1.3037,
+      "step": 2854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.43125e-05,
+      "loss": 1.2783,
+      "step": 2855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.43e-05,
+      "loss": 0.9351,
+      "step": 2856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4287500000000002e-05,
+      "loss": 1.3291,
+      "step": 2857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4275e-05,
+      "loss": 1.0977,
+      "step": 2858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4262500000000001e-05,
+      "loss": 1.3594,
+      "step": 2859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 1.334,
+      "step": 2860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.42375e-05,
+      "loss": 1.1484,
+      "step": 2861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4225e-05,
+      "loss": 1.3486,
+      "step": 2862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4212500000000002e-05,
+      "loss": 1.2783,
+      "step": 2863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.42e-05,
+      "loss": 1.4092,
+      "step": 2864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4187500000000001e-05,
+      "loss": 1.2959,
+      "step": 2865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4174999999999999e-05,
+      "loss": 1.2275,
+      "step": 2866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.41625e-05,
+      "loss": 1.1787,
+      "step": 2867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.415e-05,
+      "loss": 1.3604,
+      "step": 2868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4137500000000001e-05,
+      "loss": 1.1328,
+      "step": 2869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4125e-05,
+      "loss": 1.2197,
+      "step": 2870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.41125e-05,
+      "loss": 1.4736,
+      "step": 2871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 1.4248,
+      "step": 2872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.40875e-05,
+      "loss": 1.2012,
+      "step": 2873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4075e-05,
+      "loss": 1.1211,
+      "step": 2874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 1.5928,
+      "step": 2875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4050000000000003e-05,
+      "loss": 1.1934,
+      "step": 2876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.40375e-05,
+      "loss": 1.4346,
+      "step": 2877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4025000000000002e-05,
+      "loss": 1.3877,
+      "step": 2878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.40125e-05,
+      "loss": 1.1631,
+      "step": 2879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.1191,
+      "step": 2880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3987500000000001e-05,
+      "loss": 1.4521,
+      "step": 2881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3975000000000003e-05,
+      "loss": 0.9575,
+      "step": 2882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.39625e-05,
+      "loss": 1.2373,
+      "step": 2883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3950000000000002e-05,
+      "loss": 1.2881,
+      "step": 2884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.39375e-05,
+      "loss": 1.3037,
+      "step": 2885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3925000000000001e-05,
+      "loss": 1.335,
+      "step": 2886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3912500000000001e-05,
+      "loss": 1.0195,
+      "step": 2887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 1.2988,
+      "step": 2888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.38875e-05,
+      "loss": 1.3545,
+      "step": 2889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3875000000000002e-05,
+      "loss": 1.3076,
+      "step": 2890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.38625e-05,
+      "loss": 1.0957,
+      "step": 2891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3850000000000001e-05,
+      "loss": 1.293,
+      "step": 2892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3837500000000001e-05,
+      "loss": 1.4844,
+      "step": 2893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3825000000000002e-05,
+      "loss": 1.0361,
+      "step": 2894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.38125e-05,
+      "loss": 1.4082,
+      "step": 2895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.3018,
+      "step": 2896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.37875e-05,
+      "loss": 1.4678,
+      "step": 2897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3775000000000001e-05,
+      "loss": 1.4092,
+      "step": 2898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.37625e-05,
+      "loss": 1.0898,
+      "step": 2899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 1.1992,
+      "step": 2900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.37375e-05,
+      "loss": 1.166,
+      "step": 2901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3725000000000002e-05,
+      "loss": 1.2031,
+      "step": 2902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.37125e-05,
+      "loss": 1.2686,
+      "step": 2903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 1.5146,
+      "step": 2904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.36875e-05,
+      "loss": 1.2236,
+      "step": 2905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3675000000000002e-05,
+      "loss": 1.1494,
+      "step": 2906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.36625e-05,
+      "loss": 1.0908,
+      "step": 2907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3650000000000001e-05,
+      "loss": 1.1699,
+      "step": 2908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.36375e-05,
+      "loss": 1.083,
+      "step": 2909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3625e-05,
+      "loss": 1.3184,
+      "step": 2910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.36125e-05,
+      "loss": 1.2969,
+      "step": 2911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.2344,
+      "step": 2912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.35875e-05,
+      "loss": 1.1885,
+      "step": 2913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3575000000000001e-05,
+      "loss": 1.3047,
+      "step": 2914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3562500000000001e-05,
+      "loss": 1.2539,
+      "step": 2915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3550000000000002e-05,
+      "loss": 1.1641,
+      "step": 2916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.35375e-05,
+      "loss": 1.2568,
+      "step": 2917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3525000000000002e-05,
+      "loss": 1.5527,
+      "step": 2918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.35125e-05,
+      "loss": 1.2129,
+      "step": 2919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.4844,
+      "step": 2920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3487500000000001e-05,
+      "loss": 1.2236,
+      "step": 2921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3475000000000002e-05,
+      "loss": 1.4092,
+      "step": 2922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.34625e-05,
+      "loss": 1.6309,
+      "step": 2923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3450000000000002e-05,
+      "loss": 1.582,
+      "step": 2924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.34375e-05,
+      "loss": 1.3311,
+      "step": 2925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3425000000000001e-05,
+      "loss": 1.1973,
+      "step": 2926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.34125e-05,
+      "loss": 1.2852,
+      "step": 2927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 1.1094,
+      "step": 2928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.33875e-05,
+      "loss": 1.377,
+      "step": 2929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3375000000000002e-05,
+      "loss": 1.2441,
+      "step": 2930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.33625e-05,
+      "loss": 1.2451,
+      "step": 2931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3350000000000001e-05,
+      "loss": 1.3799,
+      "step": 2932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.33375e-05,
+      "loss": 1.1758,
+      "step": 2933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3325000000000002e-05,
+      "loss": 1.2666,
+      "step": 2934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.33125e-05,
+      "loss": 1.2168,
+      "step": 2935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 1.1885,
+      "step": 2936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.32875e-05,
+      "loss": 1.4004,
+      "step": 2937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3275e-05,
+      "loss": 1.3477,
+      "step": 2938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.32625e-05,
+      "loss": 1.2021,
+      "step": 2939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3250000000000002e-05,
+      "loss": 1.1299,
+      "step": 2940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.32375e-05,
+      "loss": 1.4961,
+      "step": 2941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3225000000000001e-05,
+      "loss": 1.2734,
+      "step": 2942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.32125e-05,
+      "loss": 1.1074,
+      "step": 2943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.32e-05,
+      "loss": 1.2861,
+      "step": 2944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.31875e-05,
+      "loss": 1.0645,
+      "step": 2945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3175000000000002e-05,
+      "loss": 1.2305,
+      "step": 2946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.31625e-05,
+      "loss": 1.2969,
+      "step": 2947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3150000000000001e-05,
+      "loss": 1.2539,
+      "step": 2948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.31375e-05,
+      "loss": 1.2637,
+      "step": 2949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3125e-05,
+      "loss": 1.5391,
+      "step": 2950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.31125e-05,
+      "loss": 1.5381,
+      "step": 2951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 1.165,
+      "step": 2952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.30875e-05,
+      "loss": 1.3955,
+      "step": 2953
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3075000000000001e-05,
+      "loss": 1.2246,
+      "step": 2954
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3062499999999999e-05,
+      "loss": 1.5342,
+      "step": 2955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.305e-05,
+      "loss": 0.9648,
+      "step": 2956
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.30375e-05,
+      "loss": 1.3525,
+      "step": 2957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3025000000000002e-05,
+      "loss": 1.2207,
+      "step": 2958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.30125e-05,
+      "loss": 1.2715,
+      "step": 2959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.4844,
+      "step": 2960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2987499999999999e-05,
+      "loss": 1.3125,
+      "step": 2961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2975e-05,
+      "loss": 1.6523,
+      "step": 2962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.29625e-05,
+      "loss": 1.1211,
+      "step": 2963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2950000000000001e-05,
+      "loss": 1.2129,
+      "step": 2964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.29375e-05,
+      "loss": 1.0605,
+      "step": 2965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2925e-05,
+      "loss": 1.5625,
+      "step": 2966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2912499999999999e-05,
+      "loss": 1.3389,
+      "step": 2967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.29e-05,
+      "loss": 1.3789,
+      "step": 2968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.28875e-05,
+      "loss": 1.5391,
+      "step": 2969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2875000000000001e-05,
+      "loss": 1.4902,
+      "step": 2970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.28625e-05,
+      "loss": 1.5723,
+      "step": 2971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.285e-05,
+      "loss": 1.373,
+      "step": 2972
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.28375e-05,
+      "loss": 1.1855,
+      "step": 2973
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2825000000000002e-05,
+      "loss": 1.3643,
+      "step": 2974
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.28125e-05,
+      "loss": 1.2246,
+      "step": 2975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.2656,
+      "step": 2976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2787499999999999e-05,
+      "loss": 1.4385,
+      "step": 2977
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2775e-05,
+      "loss": 1.0742,
+      "step": 2978
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.27625e-05,
+      "loss": 1.3799,
+      "step": 2979
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2750000000000002e-05,
+      "loss": 1.3887,
+      "step": 2980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.27375e-05,
+      "loss": 1.1602,
+      "step": 2981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2725000000000001e-05,
+      "loss": 1.2559,
+      "step": 2982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2712499999999999e-05,
+      "loss": 1.3125,
+      "step": 2983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.27e-05,
+      "loss": 1.1641,
+      "step": 2984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.26875e-05,
+      "loss": 1.1182,
+      "step": 2985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2675000000000001e-05,
+      "loss": 1.2002,
+      "step": 2986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.26625e-05,
+      "loss": 1.1748,
+      "step": 2987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2650000000000001e-05,
+      "loss": 1.1445,
+      "step": 2988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2637499999999999e-05,
+      "loss": 1.1299,
+      "step": 2989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2625e-05,
+      "loss": 1.5352,
+      "step": 2990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.26125e-05,
+      "loss": 1.373,
+      "step": 2991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 1.2217,
+      "step": 2992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.25875e-05,
+      "loss": 1.1807,
+      "step": 2993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2575e-05,
+      "loss": 1.3643,
+      "step": 2994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2562499999999999e-05,
+      "loss": 1.1348,
+      "step": 2995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.255e-05,
+      "loss": 1.4756,
+      "step": 2996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.25375e-05,
+      "loss": 1.1143,
+      "step": 2997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2525000000000001e-05,
+      "loss": 0.9243,
+      "step": 2998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.25125e-05,
+      "loss": 1.1064,
+      "step": 2999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.25e-05,
+      "loss": 1.3105,
+      "step": 3000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.24875e-05,
+      "loss": 1.3359,
+      "step": 3001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2475e-05,
+      "loss": 1.3711,
+      "step": 3002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2462500000000001e-05,
+      "loss": 1.1846,
+      "step": 3003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2450000000000001e-05,
+      "loss": 1.292,
+      "step": 3004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.24375e-05,
+      "loss": 1.3047,
+      "step": 3005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2425e-05,
+      "loss": 1.1865,
+      "step": 3006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.24125e-05,
+      "loss": 1.3604,
+      "step": 3007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.24e-05,
+      "loss": 1.2373,
+      "step": 3008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2387500000000001e-05,
+      "loss": 1.0596,
+      "step": 3009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2375000000000001e-05,
+      "loss": 1.2529,
+      "step": 3010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.23625e-05,
+      "loss": 1.4561,
+      "step": 3011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.235e-05,
+      "loss": 1.3691,
+      "step": 3012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.23375e-05,
+      "loss": 1.1533,
+      "step": 3013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2325e-05,
+      "loss": 1.3545,
+      "step": 3014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2312500000000001e-05,
+      "loss": 1.1777,
+      "step": 3015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.23e-05,
+      "loss": 1.1123,
+      "step": 3016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.22875e-05,
+      "loss": 1.3955,
+      "step": 3017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2275e-05,
+      "loss": 1.3418,
+      "step": 3018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.22625e-05,
+      "loss": 1.25,
+      "step": 3019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.225e-05,
+      "loss": 1.293,
+      "step": 3020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2237500000000001e-05,
+      "loss": 1.1816,
+      "step": 3021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2225e-05,
+      "loss": 1.1816,
+      "step": 3022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.22125e-05,
+      "loss": 1.083,
+      "step": 3023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.22e-05,
+      "loss": 0.9658,
+      "step": 3024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.21875e-05,
+      "loss": 1.2246,
+      "step": 3025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2175e-05,
+      "loss": 1.0996,
+      "step": 3026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2162500000000001e-05,
+      "loss": 1.3232,
+      "step": 3027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.215e-05,
+      "loss": 1.0898,
+      "step": 3028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.21375e-05,
+      "loss": 1.4414,
+      "step": 3029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2125e-05,
+      "loss": 1.4961,
+      "step": 3030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.21125e-05,
+      "loss": 1.3242,
+      "step": 3031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 1.1396,
+      "step": 3032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.20875e-05,
+      "loss": 1.2441,
+      "step": 3033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2075e-05,
+      "loss": 1.2783,
+      "step": 3034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.20625e-05,
+      "loss": 1.3125,
+      "step": 3035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.205e-05,
+      "loss": 1.1113,
+      "step": 3036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.20375e-05,
+      "loss": 1.3066,
+      "step": 3037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2025000000000001e-05,
+      "loss": 1.376,
+      "step": 3038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.20125e-05,
+      "loss": 1.1748,
+      "step": 3039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3252,
+      "step": 3040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.19875e-05,
+      "loss": 1.1064,
+      "step": 3041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1975e-05,
+      "loss": 1.1836,
+      "step": 3042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.19625e-05,
+      "loss": 1.1758,
+      "step": 3043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.195e-05,
+      "loss": 1.625,
+      "step": 3044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.19375e-05,
+      "loss": 1.2383,
+      "step": 3045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1925e-05,
+      "loss": 1.4824,
+      "step": 3046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.19125e-05,
+      "loss": 1.2002,
+      "step": 3047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.19e-05,
+      "loss": 1.4395,
+      "step": 3048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.18875e-05,
+      "loss": 1.4551,
+      "step": 3049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1875e-05,
+      "loss": 1.1514,
+      "step": 3050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.18625e-05,
+      "loss": 1.4785,
+      "step": 3051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.185e-05,
+      "loss": 1.7188,
+      "step": 3052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.18375e-05,
+      "loss": 1.3457,
+      "step": 3053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1825e-05,
+      "loss": 1.2637,
+      "step": 3054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1812499999999999e-05,
+      "loss": 1.1494,
+      "step": 3055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.18e-05,
+      "loss": 1.2168,
+      "step": 3056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.17875e-05,
+      "loss": 1.252,
+      "step": 3057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1775e-05,
+      "loss": 1.1689,
+      "step": 3058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.17625e-05,
+      "loss": 1.1201,
+      "step": 3059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.175e-05,
+      "loss": 1.4229,
+      "step": 3060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.17375e-05,
+      "loss": 1.4268,
+      "step": 3061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1725e-05,
+      "loss": 1.1885,
+      "step": 3062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1712500000000002e-05,
+      "loss": 1.2168,
+      "step": 3063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 1.2627,
+      "step": 3064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1687500000000001e-05,
+      "loss": 1.1553,
+      "step": 3065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1675000000000001e-05,
+      "loss": 1.2129,
+      "step": 3066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.16625e-05,
+      "loss": 1.1436,
+      "step": 3067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1650000000000002e-05,
+      "loss": 1.2266,
+      "step": 3068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1637500000000002e-05,
+      "loss": 1.3232,
+      "step": 3069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1625000000000001e-05,
+      "loss": 1.3135,
+      "step": 3070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1612500000000001e-05,
+      "loss": 1.0869,
+      "step": 3071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.16e-05,
+      "loss": 1.2812,
+      "step": 3072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.15875e-05,
+      "loss": 1.4961,
+      "step": 3073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1575000000000002e-05,
+      "loss": 1.3271,
+      "step": 3074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1562500000000002e-05,
+      "loss": 1.1328,
+      "step": 3075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1550000000000001e-05,
+      "loss": 1.2188,
+      "step": 3076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1537500000000001e-05,
+      "loss": 1.2949,
+      "step": 3077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1525e-05,
+      "loss": 1.4902,
+      "step": 3078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.15125e-05,
+      "loss": 1.3203,
+      "step": 3079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 1.2803,
+      "step": 3080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1487500000000001e-05,
+      "loss": 1.2852,
+      "step": 3081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1475000000000001e-05,
+      "loss": 1.6074,
+      "step": 3082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.14625e-05,
+      "loss": 1.084,
+      "step": 3083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.145e-05,
+      "loss": 1.3271,
+      "step": 3084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.14375e-05,
+      "loss": 1.2129,
+      "step": 3085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1425000000000002e-05,
+      "loss": 1.0869,
+      "step": 3086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1412500000000001e-05,
+      "loss": 1.2021,
+      "step": 3087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.3311,
+      "step": 3088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.13875e-05,
+      "loss": 1.2832,
+      "step": 3089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1375e-05,
+      "loss": 1.2666,
+      "step": 3090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1362500000000002e-05,
+      "loss": 1.291,
+      "step": 3091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1350000000000001e-05,
+      "loss": 1.0791,
+      "step": 3092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1337500000000001e-05,
+      "loss": 1.2031,
+      "step": 3093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1325e-05,
+      "loss": 1.4844,
+      "step": 3094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.13125e-05,
+      "loss": 1.1797,
+      "step": 3095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.13e-05,
+      "loss": 1.3389,
+      "step": 3096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1287500000000002e-05,
+      "loss": 1.1006,
+      "step": 3097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1275000000000001e-05,
+      "loss": 1.3984,
+      "step": 3098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1262500000000001e-05,
+      "loss": 1.1475,
+      "step": 3099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.125e-05,
+      "loss": 1.2031,
+      "step": 3100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.12375e-05,
+      "loss": 1.1328,
+      "step": 3101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1225e-05,
+      "loss": 1.3311,
+      "step": 3102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1212500000000001e-05,
+      "loss": 1.3477,
+      "step": 3103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.2695,
+      "step": 3104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1187500000000001e-05,
+      "loss": 1.0879,
+      "step": 3105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1175e-05,
+      "loss": 1.1084,
+      "step": 3106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.11625e-05,
+      "loss": 1.1885,
+      "step": 3107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.115e-05,
+      "loss": 1.2988,
+      "step": 3108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1137500000000001e-05,
+      "loss": 1.1533,
+      "step": 3109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1125000000000001e-05,
+      "loss": 1.5928,
+      "step": 3110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.11125e-05,
+      "loss": 1.3613,
+      "step": 3111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.11e-05,
+      "loss": 1.21,
+      "step": 3112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.10875e-05,
+      "loss": 1.1533,
+      "step": 3113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1075e-05,
+      "loss": 1.4014,
+      "step": 3114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1062500000000001e-05,
+      "loss": 1.1924,
+      "step": 3115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1050000000000001e-05,
+      "loss": 1.0977,
+      "step": 3116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.10375e-05,
+      "loss": 1.0918,
+      "step": 3117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1025e-05,
+      "loss": 1.1309,
+      "step": 3118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.10125e-05,
+      "loss": 1.3047,
+      "step": 3119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.2969,
+      "step": 3120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0987500000000001e-05,
+      "loss": 1.2236,
+      "step": 3121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0975e-05,
+      "loss": 1.457,
+      "step": 3122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.09625e-05,
+      "loss": 1.0762,
+      "step": 3123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.095e-05,
+      "loss": 1.3574,
+      "step": 3124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.09375e-05,
+      "loss": 1.2129,
+      "step": 3125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0925000000000001e-05,
+      "loss": 1.2676,
+      "step": 3126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0912500000000001e-05,
+      "loss": 1.334,
+      "step": 3127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.09e-05,
+      "loss": 1.3135,
+      "step": 3128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.08875e-05,
+      "loss": 0.9771,
+      "step": 3129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0875e-05,
+      "loss": 1.4482,
+      "step": 3130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.08625e-05,
+      "loss": 1.4971,
+      "step": 3131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0850000000000001e-05,
+      "loss": 1.3037,
+      "step": 3132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.08375e-05,
+      "loss": 1.2236,
+      "step": 3133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0825e-05,
+      "loss": 1.4941,
+      "step": 3134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.08125e-05,
+      "loss": 1.0752,
+      "step": 3135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.08e-05,
+      "loss": 1.5146,
+      "step": 3136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.07875e-05,
+      "loss": 1.166,
+      "step": 3137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0775000000000001e-05,
+      "loss": 1.1904,
+      "step": 3138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.07625e-05,
+      "loss": 1.4277,
+      "step": 3139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.075e-05,
+      "loss": 1.416,
+      "step": 3140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.07375e-05,
+      "loss": 1.2393,
+      "step": 3141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0725e-05,
+      "loss": 1.1748,
+      "step": 3142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.07125e-05,
+      "loss": 1.0801,
+      "step": 3143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 1.1748,
+      "step": 3144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.06875e-05,
+      "loss": 1.2607,
+      "step": 3145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0675e-05,
+      "loss": 1.2236,
+      "step": 3146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.06625e-05,
+      "loss": 1.1641,
+      "step": 3147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.065e-05,
+      "loss": 1.1777,
+      "step": 3148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0637500000000001e-05,
+      "loss": 1.1348,
+      "step": 3149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0625e-05,
+      "loss": 1.2959,
+      "step": 3150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.06125e-05,
+      "loss": 1.1777,
+      "step": 3151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.06e-05,
+      "loss": 1.5029,
+      "step": 3152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.05875e-05,
+      "loss": 1.4004,
+      "step": 3153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0575e-05,
+      "loss": 1.2061,
+      "step": 3154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0562500000000001e-05,
+      "loss": 1.4463,
+      "step": 3155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.055e-05,
+      "loss": 1.1367,
+      "step": 3156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.05375e-05,
+      "loss": 1.3828,
+      "step": 3157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0525e-05,
+      "loss": 1.2705,
+      "step": 3158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.05125e-05,
+      "loss": 1.2812,
+      "step": 3159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.05e-05,
+      "loss": 1.2744,
+      "step": 3160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.04875e-05,
+      "loss": 1.2822,
+      "step": 3161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0475e-05,
+      "loss": 1.3506,
+      "step": 3162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.04625e-05,
+      "loss": 1.1641,
+      "step": 3163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.045e-05,
+      "loss": 1.4688,
+      "step": 3164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.04375e-05,
+      "loss": 1.1973,
+      "step": 3165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0425e-05,
+      "loss": 1.3135,
+      "step": 3166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.04125e-05,
+      "loss": 1.4414,
+      "step": 3167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.04e-05,
+      "loss": 1.3506,
+      "step": 3168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03875e-05,
+      "loss": 1.2822,
+      "step": 3169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0375e-05,
+      "loss": 0.9712,
+      "step": 3170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03625e-05,
+      "loss": 1.3096,
+      "step": 3171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.035e-05,
+      "loss": 1.1367,
+      "step": 3172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03375e-05,
+      "loss": 1.4941,
+      "step": 3173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0325e-05,
+      "loss": 1.2461,
+      "step": 3174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03125e-05,
+      "loss": 1.1084,
+      "step": 3175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03e-05,
+      "loss": 1.168,
+      "step": 3176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.02875e-05,
+      "loss": 1.123,
+      "step": 3177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0275e-05,
+      "loss": 0.9756,
+      "step": 3178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.02625e-05,
+      "loss": 0.9487,
+      "step": 3179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025e-05,
+      "loss": 1.3984,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.02375e-05,
+      "loss": 1.3936,
+      "step": 3181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0225e-05,
+      "loss": 1.2158,
+      "step": 3182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.02125e-05,
+      "loss": 1.2627,
+      "step": 3183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.02e-05,
+      "loss": 1.1553,
+      "step": 3184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01875e-05,
+      "loss": 1.248,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0175e-05,
+      "loss": 1.248,
+      "step": 3186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01625e-05,
+      "loss": 1.5762,
+      "step": 3187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0150000000000001e-05,
+      "loss": 1.3662,
+      "step": 3188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01375e-05,
+      "loss": 1.5156,
+      "step": 3189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0125e-05,
+      "loss": 1.2139,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0112500000000002e-05,
+      "loss": 1.292,
+      "step": 3191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 1.0898,
+      "step": 3192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0087500000000001e-05,
+      "loss": 1.2119,
+      "step": 3193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0075000000000001e-05,
+      "loss": 1.3115,
+      "step": 3194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.00625e-05,
+      "loss": 1.2559,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.005e-05,
+      "loss": 1.0654,
+      "step": 3196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0037500000000002e-05,
+      "loss": 1.1064,
+      "step": 3197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0025000000000001e-05,
+      "loss": 1.292,
+      "step": 3198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0012500000000001e-05,
+      "loss": 1.1357,
+      "step": 3199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 1.1104,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.9875e-06,
+      "loss": 1.2051,
+      "step": 3201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.975e-06,
+      "loss": 1.2832,
+      "step": 3202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.962500000000002e-06,
+      "loss": 1.1582,
+      "step": 3203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.950000000000001e-06,
+      "loss": 1.6836,
+      "step": 3204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.937500000000001e-06,
+      "loss": 1.5127,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.925e-06,
+      "loss": 1.5986,
+      "step": 3206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.9125e-06,
+      "loss": 1.334,
+      "step": 3207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 1.1104,
+      "step": 3208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.887500000000001e-06,
+      "loss": 1.3613,
+      "step": 3209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.875000000000001e-06,
+      "loss": 1.2568,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.862500000000001e-06,
+      "loss": 1.1484,
+      "step": 3211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.85e-06,
+      "loss": 1.4297,
+      "step": 3212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.8375e-06,
+      "loss": 0.9863,
+      "step": 3213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.825000000000002e-06,
+      "loss": 1.4248,
+      "step": 3214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.812500000000001e-06,
+      "loss": 1.2383,
+      "step": 3215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.4102,
+      "step": 3216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.7875e-06,
+      "loss": 1.4717,
+      "step": 3217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.775e-06,
+      "loss": 1.2432,
+      "step": 3218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.7625e-06,
+      "loss": 1.0293,
+      "step": 3219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.750000000000002e-06,
+      "loss": 1.1807,
+      "step": 3220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.737500000000001e-06,
+      "loss": 1.3418,
+      "step": 3221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.725000000000001e-06,
+      "loss": 1.3613,
+      "step": 3222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.7125e-06,
+      "loss": 1.0361,
+      "step": 3223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.7e-06,
+      "loss": 1.1973,
+      "step": 3224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.6875e-06,
+      "loss": 1.0645,
+      "step": 3225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.675000000000001e-06,
+      "loss": 1.0527,
+      "step": 3226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.662500000000001e-06,
+      "loss": 1.3477,
+      "step": 3227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.65e-06,
+      "loss": 1.5391,
+      "step": 3228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.6375e-06,
+      "loss": 1.5918,
+      "step": 3229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.625e-06,
+      "loss": 1.2041,
+      "step": 3230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.6125e-06,
+      "loss": 1.1025,
+      "step": 3231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.2441,
+      "step": 3232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.587500000000001e-06,
+      "loss": 1.2207,
+      "step": 3233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.575e-06,
+      "loss": 1.1836,
+      "step": 3234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.5625e-06,
+      "loss": 1.1055,
+      "step": 3235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.55e-06,
+      "loss": 1.2168,
+      "step": 3236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.537500000000001e-06,
+      "loss": 1.1445,
+      "step": 3237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.525000000000001e-06,
+      "loss": 1.3359,
+      "step": 3238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.5125e-06,
+      "loss": 1.0342,
+      "step": 3239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.5e-06,
+      "loss": 1.2803,
+      "step": 3240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.4875e-06,
+      "loss": 1.3682,
+      "step": 3241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.475e-06,
+      "loss": 1.0957,
+      "step": 3242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.462500000000001e-06,
+      "loss": 1.3857,
+      "step": 3243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 1.3662,
+      "step": 3244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.4375e-06,
+      "loss": 1.1006,
+      "step": 3245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.425e-06,
+      "loss": 1.2969,
+      "step": 3246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.4125e-06,
+      "loss": 1.1182,
+      "step": 3247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.4e-06,
+      "loss": 1.2656,
+      "step": 3248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.387500000000001e-06,
+      "loss": 1.168,
+      "step": 3249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 1.459,
+      "step": 3250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.3625e-06,
+      "loss": 1.084,
+      "step": 3251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.35e-06,
+      "loss": 1.0742,
+      "step": 3252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.3375e-06,
+      "loss": 1.3467,
+      "step": 3253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.325e-06,
+      "loss": 1.3037,
+      "step": 3254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.312500000000001e-06,
+      "loss": 1.1855,
+      "step": 3255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.3e-06,
+      "loss": 1.251,
+      "step": 3256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.2875e-06,
+      "loss": 1.4053,
+      "step": 3257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.275e-06,
+      "loss": 1.3193,
+      "step": 3258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.2625e-06,
+      "loss": 1.335,
+      "step": 3259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.25e-06,
+      "loss": 1.0283,
+      "step": 3260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.237500000000001e-06,
+      "loss": 1.0371,
+      "step": 3261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.225e-06,
+      "loss": 1.3174,
+      "step": 3262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.2125e-06,
+      "loss": 1.1943,
+      "step": 3263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.2e-06,
+      "loss": 1.4453,
+      "step": 3264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.1875e-06,
+      "loss": 1.623,
+      "step": 3265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.175000000000001e-06,
+      "loss": 1.0898,
+      "step": 3266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.1625e-06,
+      "loss": 1.5635,
+      "step": 3267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.15e-06,
+      "loss": 1.457,
+      "step": 3268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.1375e-06,
+      "loss": 1.1914,
+      "step": 3269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.125e-06,
+      "loss": 1.3545,
+      "step": 3270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.1125e-06,
+      "loss": 1.2012,
+      "step": 3271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 1.3848,
+      "step": 3272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.0875e-06,
+      "loss": 1.1299,
+      "step": 3273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.075e-06,
+      "loss": 1.3516,
+      "step": 3274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.0625e-06,
+      "loss": 1.4268,
+      "step": 3275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.05e-06,
+      "loss": 1.2197,
+      "step": 3276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.0375e-06,
+      "loss": 1.1162,
+      "step": 3277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.025e-06,
+      "loss": 1.2783,
+      "step": 3278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.0125e-06,
+      "loss": 1.0527,
+      "step": 3279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9e-06,
+      "loss": 1.377,
+      "step": 3280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.9875e-06,
+      "loss": 1.1807,
+      "step": 3281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.975e-06,
+      "loss": 1.1611,
+      "step": 3282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.9625e-06,
+      "loss": 1.374,
+      "step": 3283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.95e-06,
+      "loss": 1.2637,
+      "step": 3284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.9375e-06,
+      "loss": 1.2119,
+      "step": 3285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.925e-06,
+      "loss": 1.4727,
+      "step": 3286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.9125e-06,
+      "loss": 1.3291,
+      "step": 3287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.9e-06,
+      "loss": 1.4902,
+      "step": 3288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.8875e-06,
+      "loss": 1.0723,
+      "step": 3289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.875e-06,
+      "loss": 1.0811,
+      "step": 3290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.8625e-06,
+      "loss": 1.4609,
+      "step": 3291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.85e-06,
+      "loss": 1.3564,
+      "step": 3292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.8375e-06,
+      "loss": 1.2158,
+      "step": 3293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.825e-06,
+      "loss": 1.3125,
+      "step": 3294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.8125e-06,
+      "loss": 1.2129,
+      "step": 3295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.8e-06,
+      "loss": 1.4307,
+      "step": 3296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.7875e-06,
+      "loss": 1.3447,
+      "step": 3297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.775e-06,
+      "loss": 1.3047,
+      "step": 3298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.7625e-06,
+      "loss": 1.4453,
+      "step": 3299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.75e-06,
+      "loss": 1.3975,
+      "step": 3300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.7375e-06,
+      "loss": 1.3896,
+      "step": 3301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.725e-06,
+      "loss": 0.9629,
+      "step": 3302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.7125e-06,
+      "loss": 1.3047,
+      "step": 3303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.7e-06,
+      "loss": 1.3115,
+      "step": 3304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.6875e-06,
+      "loss": 1.4082,
+      "step": 3305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.674999999999999e-06,
+      "loss": 1.3789,
+      "step": 3306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.6625e-06,
+      "loss": 1.293,
+      "step": 3307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.65e-06,
+      "loss": 1.1406,
+      "step": 3308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.6375e-06,
+      "loss": 1.3486,
+      "step": 3309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.625e-06,
+      "loss": 1.0781,
+      "step": 3310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.6125e-06,
+      "loss": 1.335,
+      "step": 3311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 1.2305,
+      "step": 3312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.5875e-06,
+      "loss": 1.3008,
+      "step": 3313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.575000000000002e-06,
+      "loss": 0.9849,
+      "step": 3314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.562500000000001e-06,
+      "loss": 1.2754,
+      "step": 3315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.550000000000001e-06,
+      "loss": 1.4033,
+      "step": 3316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.5375e-06,
+      "loss": 1.4102,
+      "step": 3317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.525e-06,
+      "loss": 1.2832,
+      "step": 3318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.5125e-06,
+      "loss": 1.1074,
+      "step": 3319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 1.1523,
+      "step": 3320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.487500000000001e-06,
+      "loss": 1.3467,
+      "step": 3321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.475000000000001e-06,
+      "loss": 1.5645,
+      "step": 3322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.4625e-06,
+      "loss": 1.2119,
+      "step": 3323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.45e-06,
+      "loss": 1.4551,
+      "step": 3324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.437500000000002e-06,
+      "loss": 0.9956,
+      "step": 3325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.425000000000001e-06,
+      "loss": 1.5303,
+      "step": 3326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.412500000000001e-06,
+      "loss": 1.1475,
+      "step": 3327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.2041,
+      "step": 3328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.3875e-06,
+      "loss": 1.4121,
+      "step": 3329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.375e-06,
+      "loss": 1.6191,
+      "step": 3330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.362500000000002e-06,
+      "loss": 1.5566,
+      "step": 3331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.350000000000001e-06,
+      "loss": 1.2959,
+      "step": 3332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.337500000000001e-06,
+      "loss": 1.124,
+      "step": 3333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.325e-06,
+      "loss": 1.3086,
+      "step": 3334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.3125e-06,
+      "loss": 1.2539,
+      "step": 3335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.3e-06,
+      "loss": 1.1631,
+      "step": 3336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.287500000000002e-06,
+      "loss": 1.3066,
+      "step": 3337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.275000000000001e-06,
+      "loss": 1.418,
+      "step": 3338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.262500000000001e-06,
+      "loss": 1.2734,
+      "step": 3339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.25e-06,
+      "loss": 1.2256,
+      "step": 3340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.2375e-06,
+      "loss": 1.0762,
+      "step": 3341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.225e-06,
+      "loss": 1.2852,
+      "step": 3342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.212500000000001e-06,
+      "loss": 1.4775,
+      "step": 3343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 1.3613,
+      "step": 3344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.1875e-06,
+      "loss": 1.0957,
+      "step": 3345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.175e-06,
+      "loss": 1.3457,
+      "step": 3346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.1625e-06,
+      "loss": 1.042,
+      "step": 3347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.15e-06,
+      "loss": 1.2656,
+      "step": 3348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.137500000000001e-06,
+      "loss": 1.1396,
+      "step": 3349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.8096,
+      "step": 3350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.1125e-06,
+      "loss": 1.3281,
+      "step": 3351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.1e-06,
+      "loss": 1.2168,
+      "step": 3352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.0875e-06,
+      "loss": 1.542,
+      "step": 3353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.075000000000001e-06,
+      "loss": 1.2607,
+      "step": 3354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.062500000000001e-06,
+      "loss": 1.2979,
+      "step": 3355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.050000000000001e-06,
+      "loss": 1.5488,
+      "step": 3356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.0375e-06,
+      "loss": 1.3115,
+      "step": 3357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.025e-06,
+      "loss": 1.2441,
+      "step": 3358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.0125e-06,
+      "loss": 1.4482,
+      "step": 3359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0137,
+      "step": 3360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.987500000000001e-06,
+      "loss": 1.4336,
+      "step": 3361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.975e-06,
+      "loss": 1.1592,
+      "step": 3362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.9625e-06,
+      "loss": 1.3535,
+      "step": 3363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.95e-06,
+      "loss": 1.4355,
+      "step": 3364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.9375e-06,
+      "loss": 1.1484,
+      "step": 3365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.925000000000001e-06,
+      "loss": 1.2236,
+      "step": 3366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.912500000000001e-06,
+      "loss": 1.3018,
+      "step": 3367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.9e-06,
+      "loss": 1.4717,
+      "step": 3368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.8875e-06,
+      "loss": 1.3193,
+      "step": 3369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.875e-06,
+      "loss": 1.2266,
+      "step": 3370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.8625e-06,
+      "loss": 1.3115,
+      "step": 3371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.850000000000001e-06,
+      "loss": 1.2188,
+      "step": 3372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.8375e-06,
+      "loss": 1.293,
+      "step": 3373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.825e-06,
+      "loss": 1.4971,
+      "step": 3374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.8125e-06,
+      "loss": 1.3379,
+      "step": 3375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.8e-06,
+      "loss": 1.2051,
+      "step": 3376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.7875e-06,
+      "loss": 1.2686,
+      "step": 3377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.775000000000001e-06,
+      "loss": 1.2402,
+      "step": 3378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.7625e-06,
+      "loss": 1.167,
+      "step": 3379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.75e-06,
+      "loss": 1.5,
+      "step": 3380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.7375e-06,
+      "loss": 1.2666,
+      "step": 3381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.725e-06,
+      "loss": 1.2812,
+      "step": 3382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.712500000000001e-06,
+      "loss": 1.2568,
+      "step": 3383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.7e-06,
+      "loss": 1.3955,
+      "step": 3384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6875e-06,
+      "loss": 1.1641,
+      "step": 3385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.675e-06,
+      "loss": 1.5625,
+      "step": 3386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6625e-06,
+      "loss": 1.3398,
+      "step": 3387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.65e-06,
+      "loss": 1.2695,
+      "step": 3388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.637500000000001e-06,
+      "loss": 0.9883,
+      "step": 3389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.625e-06,
+      "loss": 1.166,
+      "step": 3390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6125e-06,
+      "loss": 1.1094,
+      "step": 3391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6e-06,
+      "loss": 1.2754,
+      "step": 3392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5875e-06,
+      "loss": 1.375,
+      "step": 3393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.575e-06,
+      "loss": 1.3945,
+      "step": 3394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5625e-06,
+      "loss": 1.1582,
+      "step": 3395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.55e-06,
+      "loss": 1.0264,
+      "step": 3396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5375e-06,
+      "loss": 0.9795,
+      "step": 3397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.525e-06,
+      "loss": 1.0439,
+      "step": 3398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5125000000000005e-06,
+      "loss": 1.1152,
+      "step": 3399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5e-06,
+      "loss": 1.083,
+      "step": 3400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4875e-06,
+      "loss": 1.3682,
+      "step": 3401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4750000000000004e-06,
+      "loss": 1.2461,
+      "step": 3402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4625e-06,
+      "loss": 1.2236,
+      "step": 3403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.45e-06,
+      "loss": 1.1172,
+      "step": 3404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4375e-06,
+      "loss": 1.291,
+      "step": 3405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.425e-06,
+      "loss": 1.1201,
+      "step": 3406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4125e-06,
+      "loss": 1.3477,
+      "step": 3407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4e-06,
+      "loss": 1.1504,
+      "step": 3408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.3875e-06,
+      "loss": 1.0684,
+      "step": 3409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.375e-06,
+      "loss": 1.4463,
+      "step": 3410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.3625e-06,
+      "loss": 1.2617,
+      "step": 3411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.35e-06,
+      "loss": 1.2295,
+      "step": 3412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.3375e-06,
+      "loss": 1.2559,
+      "step": 3413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.325e-06,
+      "loss": 1.0967,
+      "step": 3414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.3125e-06,
+      "loss": 1.083,
+      "step": 3415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.2999999999999996e-06,
+      "loss": 1.1377,
+      "step": 3416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.2875e-06,
+      "loss": 1.3926,
+      "step": 3417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.275e-06,
+      "loss": 1.1826,
+      "step": 3418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.2624999999999995e-06,
+      "loss": 1.1914,
+      "step": 3419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.25e-06,
+      "loss": 1.042,
+      "step": 3420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.2375e-06,
+      "loss": 1.3994,
+      "step": 3421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.2249999999999994e-06,
+      "loss": 1.1104,
+      "step": 3422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.2125e-06,
+      "loss": 1.3008,
+      "step": 3423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.2e-06,
+      "loss": 1.0693,
+      "step": 3424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.187499999999999e-06,
+      "loss": 0.98,
+      "step": 3425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.175e-06,
+      "loss": 1.2803,
+      "step": 3426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1625e-06,
+      "loss": 1.6572,
+      "step": 3427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.15e-06,
+      "loss": 1.0557,
+      "step": 3428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1375e-06,
+      "loss": 1.501,
+      "step": 3429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1249999999999995e-06,
+      "loss": 1.1084,
+      "step": 3430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1125e-06,
+      "loss": 1.1562,
+      "step": 3431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1e-06,
+      "loss": 1.4346,
+      "step": 3432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0874999999999995e-06,
+      "loss": 1.0518,
+      "step": 3433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.075e-06,
+      "loss": 1.418,
+      "step": 3434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0625e-06,
+      "loss": 1.334,
+      "step": 3435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.049999999999999e-06,
+      "loss": 1.3701,
+      "step": 3436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0375e-06,
+      "loss": 0.9424,
+      "step": 3437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.025000000000001e-06,
+      "loss": 1.1279,
+      "step": 3438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.012500000000001e-06,
+      "loss": 1.1562,
+      "step": 3439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.2842,
+      "step": 3440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.987500000000001e-06,
+      "loss": 1.5479,
+      "step": 3441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.975000000000001e-06,
+      "loss": 1.1758,
+      "step": 3442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.962500000000001e-06,
+      "loss": 1.4004,
+      "step": 3443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.950000000000001e-06,
+      "loss": 1.3408,
+      "step": 3444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.937500000000001e-06,
+      "loss": 1.1738,
+      "step": 3445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.925000000000001e-06,
+      "loss": 1.0889,
+      "step": 3446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.912500000000001e-06,
+      "loss": 1.082,
+      "step": 3447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 1.2578,
+      "step": 3448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8875000000000005e-06,
+      "loss": 0.9844,
+      "step": 3449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.875000000000001e-06,
+      "loss": 1.3359,
+      "step": 3450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.862500000000001e-06,
+      "loss": 1.0459,
+      "step": 3451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8500000000000005e-06,
+      "loss": 1.0049,
+      "step": 3452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.837500000000001e-06,
+      "loss": 1.166,
+      "step": 3453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.825000000000001e-06,
+      "loss": 1.3037,
+      "step": 3454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8125e-06,
+      "loss": 1.1992,
+      "step": 3455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.1465,
+      "step": 3456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.787500000000001e-06,
+      "loss": 1.0537,
+      "step": 3457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.775000000000001e-06,
+      "loss": 0.9404,
+      "step": 3458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.762500000000001e-06,
+      "loss": 1.3896,
+      "step": 3459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.6318,
+      "step": 3460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.737500000000001e-06,
+      "loss": 1.1758,
+      "step": 3461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.725000000000001e-06,
+      "loss": 1.2949,
+      "step": 3462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.7125000000000005e-06,
+      "loss": 0.9399,
+      "step": 3463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 1.2344,
+      "step": 3464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.687500000000001e-06,
+      "loss": 1.4502,
+      "step": 3465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6750000000000005e-06,
+      "loss": 1.0117,
+      "step": 3466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.662500000000001e-06,
+      "loss": 1.2734,
+      "step": 3467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.650000000000001e-06,
+      "loss": 1.2842,
+      "step": 3468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6375e-06,
+      "loss": 1.3789,
+      "step": 3469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.625000000000001e-06,
+      "loss": 1.0518,
+      "step": 3470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.612500000000001e-06,
+      "loss": 1.1484,
+      "step": 3471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6e-06,
+      "loss": 1.4131,
+      "step": 3472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.587500000000001e-06,
+      "loss": 1.1455,
+      "step": 3473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5750000000000006e-06,
+      "loss": 1.3887,
+      "step": 3474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5625e-06,
+      "loss": 1.3887,
+      "step": 3475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.550000000000001e-06,
+      "loss": 1.4346,
+      "step": 3476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5375000000000005e-06,
+      "loss": 1.2041,
+      "step": 3477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.525e-06,
+      "loss": 1.123,
+      "step": 3478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.512500000000001e-06,
+      "loss": 1.0752,
+      "step": 3479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.3252,
+      "step": 3480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4875e-06,
+      "loss": 1.3906,
+      "step": 3481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.475000000000001e-06,
+      "loss": 1.4395,
+      "step": 3482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4625e-06,
+      "loss": 1.2412,
+      "step": 3483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.45e-06,
+      "loss": 1.0957,
+      "step": 3484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.437500000000001e-06,
+      "loss": 1.1494,
+      "step": 3485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.425e-06,
+      "loss": 1.291,
+      "step": 3486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.412500000000001e-06,
+      "loss": 1.415,
+      "step": 3487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.1748,
+      "step": 3488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3875e-06,
+      "loss": 1.2393,
+      "step": 3489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.375000000000001e-06,
+      "loss": 1.125,
+      "step": 3490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3625000000000005e-06,
+      "loss": 1.54,
+      "step": 3491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.35e-06,
+      "loss": 1.3652,
+      "step": 3492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.337500000000001e-06,
+      "loss": 1.3867,
+      "step": 3493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3250000000000004e-06,
+      "loss": 1.0469,
+      "step": 3494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3125e-06,
+      "loss": 1.0156,
+      "step": 3495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 1.4131,
+      "step": 3496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.2875e-06,
+      "loss": 1.2227,
+      "step": 3497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.275e-06,
+      "loss": 1.0869,
+      "step": 3498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.262500000000001e-06,
+      "loss": 1.1807,
+      "step": 3499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.25e-06,
+      "loss": 1.5889,
+      "step": 3500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.2375e-06,
+      "loss": 1.2578,
+      "step": 3501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.2250000000000005e-06,
+      "loss": 1.4951,
+      "step": 3502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.2125e-06,
+      "loss": 1.1748,
+      "step": 3503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.2e-06,
+      "loss": 1.0703,
+      "step": 3504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1875000000000005e-06,
+      "loss": 1.3037,
+      "step": 3505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.175e-06,
+      "loss": 1.4541,
+      "step": 3506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1625e-06,
+      "loss": 1.1621,
+      "step": 3507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.15e-06,
+      "loss": 1.0889,
+      "step": 3508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1375e-06,
+      "loss": 1.4902,
+      "step": 3509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.125e-06,
+      "loss": 1.2754,
+      "step": 3510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1125e-06,
+      "loss": 1.3799,
+      "step": 3511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1e-06,
+      "loss": 1.3516,
+      "step": 3512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0875e-06,
+      "loss": 1.3945,
+      "step": 3513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.075e-06,
+      "loss": 1.0537,
+      "step": 3514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0625e-06,
+      "loss": 1.2754,
+      "step": 3515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0500000000000005e-06,
+      "loss": 1.3916,
+      "step": 3516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0375e-06,
+      "loss": 1.0947,
+      "step": 3517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.025e-06,
+      "loss": 1.1016,
+      "step": 3518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0125000000000005e-06,
+      "loss": 1.2158,
+      "step": 3519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6e-06,
+      "loss": 1.127,
+      "step": 3520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9875e-06,
+      "loss": 1.3896,
+      "step": 3521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.975e-06,
+      "loss": 1.2168,
+      "step": 3522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9625e-06,
+      "loss": 1.3408,
+      "step": 3523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.95e-06,
+      "loss": 0.9072,
+      "step": 3524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9375e-06,
+      "loss": 1.5195,
+      "step": 3525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.925e-06,
+      "loss": 1.1943,
+      "step": 3526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9125e-06,
+      "loss": 1.0215,
+      "step": 3527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9e-06,
+      "loss": 1.4834,
+      "step": 3528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8875e-06,
+      "loss": 1.1406,
+      "step": 3529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.875e-06,
+      "loss": 1.3516,
+      "step": 3530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8625e-06,
+      "loss": 1.3105,
+      "step": 3531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.850000000000001e-06,
+      "loss": 1.2402,
+      "step": 3532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8375000000000004e-06,
+      "loss": 1.1543,
+      "step": 3533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.825000000000001e-06,
+      "loss": 0.9839,
+      "step": 3534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.812500000000001e-06,
+      "loss": 1.2139,
+      "step": 3535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8e-06,
+      "loss": 1.1396,
+      "step": 3536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.787500000000001e-06,
+      "loss": 1.1172,
+      "step": 3537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.775000000000001e-06,
+      "loss": 1.4229,
+      "step": 3538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.7625e-06,
+      "loss": 1.2334,
+      "step": 3539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.750000000000001e-06,
+      "loss": 1.5059,
+      "step": 3540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7375000000000005e-06,
+      "loss": 1.1934,
+      "step": 3541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.725e-06,
+      "loss": 1.375,
+      "step": 3542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.712500000000001e-06,
+      "loss": 1.2129,
+      "step": 3543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 1.3711,
+      "step": 3544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6875e-06,
+      "loss": 1.2803,
+      "step": 3545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.675000000000001e-06,
+      "loss": 1.2422,
+      "step": 3546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6625e-06,
+      "loss": 0.9521,
+      "step": 3547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.65e-06,
+      "loss": 1.2012,
+      "step": 3548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.637500000000001e-06,
+      "loss": 1.209,
+      "step": 3549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.625e-06,
+      "loss": 1.4121,
+      "step": 3550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6125e-06,
+      "loss": 1.21,
+      "step": 3551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.2646,
+      "step": 3552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5875e-06,
+      "loss": 1.3311,
+      "step": 3553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.575e-06,
+      "loss": 1.335,
+      "step": 3554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5625000000000005e-06,
+      "loss": 1.3447,
+      "step": 3555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.55e-06,
+      "loss": 1.291,
+      "step": 3556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5375e-06,
+      "loss": 1.0898,
+      "step": 3557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5250000000000005e-06,
+      "loss": 1.4551,
+      "step": 3558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5125e-06,
+      "loss": 1.1074,
+      "step": 3559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.415,
+      "step": 3560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4875e-06,
+      "loss": 1.0146,
+      "step": 3561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.475e-06,
+      "loss": 1.2119,
+      "step": 3562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.462500000000001e-06,
+      "loss": 1.2129,
+      "step": 3563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.45e-06,
+      "loss": 1.3115,
+      "step": 3564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4375e-06,
+      "loss": 1.2422,
+      "step": 3565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4250000000000006e-06,
+      "loss": 1.3242,
+      "step": 3566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4125e-06,
+      "loss": 1.5059,
+      "step": 3567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4e-06,
+      "loss": 1.334,
+      "step": 3568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3875000000000005e-06,
+      "loss": 1.2148,
+      "step": 3569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.375e-06,
+      "loss": 1.3691,
+      "step": 3570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3625e-06,
+      "loss": 1.0391,
+      "step": 3571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3500000000000004e-06,
+      "loss": 1.458,
+      "step": 3572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3375e-06,
+      "loss": 1.1406,
+      "step": 3573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.325e-06,
+      "loss": 1.1094,
+      "step": 3574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3125e-06,
+      "loss": 1.2998,
+      "step": 3575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3e-06,
+      "loss": 1.2578,
+      "step": 3576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2875e-06,
+      "loss": 0.9634,
+      "step": 3577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.275e-06,
+      "loss": 1.3867,
+      "step": 3578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2625e-06,
+      "loss": 1.3252,
+      "step": 3579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.25e-06,
+      "loss": 1.459,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2375e-06,
+      "loss": 1.3906,
+      "step": 3581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.225e-06,
+      "loss": 1.2188,
+      "step": 3582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2125e-06,
+      "loss": 1.2959,
+      "step": 3583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2e-06,
+      "loss": 1.2939,
+      "step": 3584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1875e-06,
+      "loss": 1.1025,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175e-06,
+      "loss": 1.6758,
+      "step": 3586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1625e-06,
+      "loss": 1.166,
+      "step": 3587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15e-06,
+      "loss": 1.4492,
+      "step": 3588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1375e-06,
+      "loss": 1.4814,
+      "step": 3589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.125e-06,
+      "loss": 1.4434,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1125e-06,
+      "loss": 1.3525,
+      "step": 3591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1e-06,
+      "loss": 1.4141,
+      "step": 3592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0875e-06,
+      "loss": 1.4727,
+      "step": 3593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0750000000000005e-06,
+      "loss": 1.3594,
+      "step": 3594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0625e-06,
+      "loss": 1.1621,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.050000000000001e-06,
+      "loss": 1.1807,
+      "step": 3596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0375000000000005e-06,
+      "loss": 1.499,
+      "step": 3597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025e-06,
+      "loss": 1.4443,
+      "step": 3598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.012500000000001e-06,
+      "loss": 1.0576,
+      "step": 3599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 1.0615,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9875e-06,
+      "loss": 1.3652,
+      "step": 3601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975000000000001e-06,
+      "loss": 1.627,
+      "step": 3602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9625e-06,
+      "loss": 1.2988,
+      "step": 3603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.950000000000001e-06,
+      "loss": 1.4795,
+      "step": 3604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.937500000000001e-06,
+      "loss": 1.1924,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.925e-06,
+      "loss": 1.1357,
+      "step": 3606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.912500000000001e-06,
+      "loss": 1.4434,
+      "step": 3607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 1.4385,
+      "step": 3608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8875e-06,
+      "loss": 1.4482,
+      "step": 3609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875000000000001e-06,
+      "loss": 1.5273,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8625000000000005e-06,
+      "loss": 1.1436,
+      "step": 3611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85e-06,
+      "loss": 1.2217,
+      "step": 3612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.837500000000001e-06,
+      "loss": 1.0723,
+      "step": 3613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.825e-06,
+      "loss": 1.0303,
+      "step": 3614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8125e-06,
+      "loss": 1.4717,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.3262,
+      "step": 3616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7875e-06,
+      "loss": 1.3789,
+      "step": 3617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.775e-06,
+      "loss": 1.0605,
+      "step": 3618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7625000000000006e-06,
+      "loss": 1.1797,
+      "step": 3619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.75e-06,
+      "loss": 1.2969,
+      "step": 3620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7375e-06,
+      "loss": 1.4746,
+      "step": 3621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7250000000000005e-06,
+      "loss": 1.3252,
+      "step": 3622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7125e-06,
+      "loss": 1.0088,
+      "step": 3623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7e-06,
+      "loss": 1.2188,
+      "step": 3624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 1.2363,
+      "step": 3625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.675e-06,
+      "loss": 1.0342,
+      "step": 3626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6625e-06,
+      "loss": 1.1035,
+      "step": 3627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.65e-06,
+      "loss": 1.251,
+      "step": 3628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6375e-06,
+      "loss": 1.1289,
+      "step": 3629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.625e-06,
+      "loss": 1.293,
+      "step": 3630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6125e-06,
+      "loss": 1.2354,
+      "step": 3631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6e-06,
+      "loss": 1.2705,
+      "step": 3632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5875000000000005e-06,
+      "loss": 1.4092,
+      "step": 3633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.575e-06,
+      "loss": 1.3965,
+      "step": 3634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5625e-06,
+      "loss": 1.4434,
+      "step": 3635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 1.1875,
+      "step": 3636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5375e-06,
+      "loss": 1.4795,
+      "step": 3637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.525e-06,
+      "loss": 1.5312,
+      "step": 3638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5125e-06,
+      "loss": 1.415,
+      "step": 3639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5e-06,
+      "loss": 1.3281,
+      "step": 3640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4875e-06,
+      "loss": 1.0889,
+      "step": 3641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.475e-06,
+      "loss": 1.084,
+      "step": 3642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4625e-06,
+      "loss": 1.3848,
+      "step": 3643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.45e-06,
+      "loss": 1.6123,
+      "step": 3644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4375e-06,
+      "loss": 1.0713,
+      "step": 3645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.425e-06,
+      "loss": 1.4453,
+      "step": 3646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4125e-06,
+      "loss": 1.1543,
+      "step": 3647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4e-06,
+      "loss": 1.5527,
+      "step": 3648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3875e-06,
+      "loss": 1.335,
+      "step": 3649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.375e-06,
+      "loss": 1.2832,
+      "step": 3650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3625e-06,
+      "loss": 1.3115,
+      "step": 3651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.35e-06,
+      "loss": 1.2881,
+      "step": 3652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3374999999999996e-06,
+      "loss": 1.3477,
+      "step": 3653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.325e-06,
+      "loss": 1.3516,
+      "step": 3654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3125e-06,
+      "loss": 1.2383,
+      "step": 3655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2999999999999995e-06,
+      "loss": 0.9585,
+      "step": 3656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.287500000000001e-06,
+      "loss": 1.3252,
+      "step": 3657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2750000000000006e-06,
+      "loss": 1.2998,
+      "step": 3658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2625e-06,
+      "loss": 1.123,
+      "step": 3659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.250000000000001e-06,
+      "loss": 1.2217,
+      "step": 3660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2375000000000005e-06,
+      "loss": 1.1582,
+      "step": 3661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.225e-06,
+      "loss": 1.417,
+      "step": 3662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.212500000000001e-06,
+      "loss": 1.1943,
+      "step": 3663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.2568,
+      "step": 3664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1875e-06,
+      "loss": 1.2891,
+      "step": 3665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.175000000000001e-06,
+      "loss": 1.2021,
+      "step": 3666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1625e-06,
+      "loss": 1.0068,
+      "step": 3667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.15e-06,
+      "loss": 1.1895,
+      "step": 3668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.137500000000001e-06,
+      "loss": 1.2676,
+      "step": 3669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.125e-06,
+      "loss": 1.1904,
+      "step": 3670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1125e-06,
+      "loss": 1.2764,
+      "step": 3671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1000000000000006e-06,
+      "loss": 1.1543,
+      "step": 3672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0875e-06,
+      "loss": 1.4619,
+      "step": 3673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.075e-06,
+      "loss": 1.2109,
+      "step": 3674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0625000000000005e-06,
+      "loss": 1.2363,
+      "step": 3675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.05e-06,
+      "loss": 1.2891,
+      "step": 3676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.037500000000001e-06,
+      "loss": 1.3672,
+      "step": 3677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0250000000000004e-06,
+      "loss": 1.4678,
+      "step": 3678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0125e-06,
+      "loss": 1.4199,
+      "step": 3679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.0488,
+      "step": 3680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9875e-06,
+      "loss": 1.2842,
+      "step": 3681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.975e-06,
+      "loss": 1.3535,
+      "step": 3682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.962500000000001e-06,
+      "loss": 1.0635,
+      "step": 3683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.95e-06,
+      "loss": 1.3438,
+      "step": 3684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9375e-06,
+      "loss": 1.1777,
+      "step": 3685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9250000000000005e-06,
+      "loss": 1.1201,
+      "step": 3686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9125e-06,
+      "loss": 1.1494,
+      "step": 3687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9e-06,
+      "loss": 1.0889,
+      "step": 3688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8875000000000005e-06,
+      "loss": 1.3672,
+      "step": 3689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.875e-06,
+      "loss": 1.3799,
+      "step": 3690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8625e-06,
+      "loss": 1.3711,
+      "step": 3691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.85e-06,
+      "loss": 1.2285,
+      "step": 3692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8375e-06,
+      "loss": 1.209,
+      "step": 3693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.825e-06,
+      "loss": 1.1533,
+      "step": 3694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8125e-06,
+      "loss": 1.1709,
+      "step": 3695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8e-06,
+      "loss": 1.1348,
+      "step": 3696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7875e-06,
+      "loss": 1.2549,
+      "step": 3697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.775e-06,
+      "loss": 1.0488,
+      "step": 3698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7625e-06,
+      "loss": 1.2246,
+      "step": 3699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.75e-06,
+      "loss": 1.1865,
+      "step": 3700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7375000000000002e-06,
+      "loss": 1.1621,
+      "step": 3701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.725e-06,
+      "loss": 1.332,
+      "step": 3702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7125e-06,
+      "loss": 1.04,
+      "step": 3703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7e-06,
+      "loss": 1.3096,
+      "step": 3704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6875e-06,
+      "loss": 1.2959,
+      "step": 3705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.675e-06,
+      "loss": 1.3584,
+      "step": 3706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6625e-06,
+      "loss": 1.2441,
+      "step": 3707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6499999999999998e-06,
+      "loss": 1.1328,
+      "step": 3708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6375e-06,
+      "loss": 1.4717,
+      "step": 3709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.625e-06,
+      "loss": 1.3613,
+      "step": 3710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6124999999999997e-06,
+      "loss": 1.2188,
+      "step": 3711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6e-06,
+      "loss": 1.1768,
+      "step": 3712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5875e-06,
+      "loss": 1.124,
+      "step": 3713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.575e-06,
+      "loss": 1.2852,
+      "step": 3714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5624999999999998e-06,
+      "loss": 1.292,
+      "step": 3715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.55e-06,
+      "loss": 1.2891,
+      "step": 3716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5375e-06,
+      "loss": 1.2646,
+      "step": 3717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5249999999999997e-06,
+      "loss": 1.2197,
+      "step": 3718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5125000000000007e-06,
+      "loss": 1.46,
+      "step": 3719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 1.5762,
+      "step": 3720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4875000000000005e-06,
+      "loss": 1.4287,
+      "step": 3721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4750000000000006e-06,
+      "loss": 1.2949,
+      "step": 3722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4625000000000003e-06,
+      "loss": 1.0547,
+      "step": 3723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4500000000000004e-06,
+      "loss": 1.3066,
+      "step": 3724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4375000000000005e-06,
+      "loss": 1.2559,
+      "step": 3725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4250000000000002e-06,
+      "loss": 1.2959,
+      "step": 3726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4125000000000004e-06,
+      "loss": 1.3369,
+      "step": 3727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.2051,
+      "step": 3728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3875000000000006e-06,
+      "loss": 1.5889,
+      "step": 3729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3750000000000003e-06,
+      "loss": 1.2676,
+      "step": 3730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3625000000000004e-06,
+      "loss": 1.4492,
+      "step": 3731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 1.2988,
+      "step": 3732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3375000000000002e-06,
+      "loss": 1.1553,
+      "step": 3733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3250000000000004e-06,
+      "loss": 1.3574,
+      "step": 3734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3125000000000005e-06,
+      "loss": 1.0449,
+      "step": 3735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3e-06,
+      "loss": 1.4512,
+      "step": 3736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2875000000000003e-06,
+      "loss": 1.1143,
+      "step": 3737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2750000000000004e-06,
+      "loss": 1.2656,
+      "step": 3738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2625e-06,
+      "loss": 1.2695,
+      "step": 3739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 1.3389,
+      "step": 3740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2375000000000003e-06,
+      "loss": 1.4863,
+      "step": 3741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.225e-06,
+      "loss": 1.2119,
+      "step": 3742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2125e-06,
+      "loss": 1.043,
+      "step": 3743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.2002,
+      "step": 3744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1875000000000004e-06,
+      "loss": 1.2334,
+      "step": 3745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.175e-06,
+      "loss": 1.2334,
+      "step": 3746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1625000000000002e-06,
+      "loss": 1.1602,
+      "step": 3747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 1.2988,
+      "step": 3748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1375e-06,
+      "loss": 1.2383,
+      "step": 3749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.125e-06,
+      "loss": 1.3154,
+      "step": 3750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1125000000000003e-06,
+      "loss": 1.2764,
+      "step": 3751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1e-06,
+      "loss": 1.1846,
+      "step": 3752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0875e-06,
+      "loss": 1.3994,
+      "step": 3753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.075e-06,
+      "loss": 1.252,
+      "step": 3754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0625e-06,
+      "loss": 1.168,
+      "step": 3755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.05e-06,
+      "loss": 1.1104,
+      "step": 3756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0375e-06,
+      "loss": 1.291,
+      "step": 3757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0250000000000003e-06,
+      "loss": 1.4141,
+      "step": 3758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0125e-06,
+      "loss": 1.3926,
+      "step": 3759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3e-06,
+      "loss": 1.373,
+      "step": 3760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9875e-06,
+      "loss": 1.1992,
+      "step": 3761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.975e-06,
+      "loss": 1.4443,
+      "step": 3762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9625e-06,
+      "loss": 1.2168,
+      "step": 3763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.95e-06,
+      "loss": 1.4863,
+      "step": 3764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9375e-06,
+      "loss": 1.2666,
+      "step": 3765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9250000000000004e-06,
+      "loss": 1.2559,
+      "step": 3766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9125000000000005e-06,
+      "loss": 1.3779,
+      "step": 3767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9e-06,
+      "loss": 1.3389,
+      "step": 3768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8875000000000003e-06,
+      "loss": 1.1641,
+      "step": 3769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8750000000000004e-06,
+      "loss": 1.5107,
+      "step": 3770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8625e-06,
+      "loss": 1.6729,
+      "step": 3771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8500000000000002e-06,
+      "loss": 1.4609,
+      "step": 3772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8375000000000004e-06,
+      "loss": 1.4033,
+      "step": 3773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.825e-06,
+      "loss": 1.0898,
+      "step": 3774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8125e-06,
+      "loss": 1.1787,
+      "step": 3775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.3369,
+      "step": 3776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7875e-06,
+      "loss": 1.3555,
+      "step": 3777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.775e-06,
+      "loss": 1.3682,
+      "step": 3778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7625000000000002e-06,
+      "loss": 1.4492,
+      "step": 3779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 1.3662,
+      "step": 3780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7375e-06,
+      "loss": 1.4746,
+      "step": 3781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.725e-06,
+      "loss": 1.4502,
+      "step": 3782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7125000000000003e-06,
+      "loss": 1.084,
+      "step": 3783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7e-06,
+      "loss": 1.3223,
+      "step": 3784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6875e-06,
+      "loss": 1.2441,
+      "step": 3785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6750000000000002e-06,
+      "loss": 1.2471,
+      "step": 3786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6625e-06,
+      "loss": 1.1885,
+      "step": 3787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.65e-06,
+      "loss": 1.0996,
+      "step": 3788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6375e-06,
+      "loss": 1.4619,
+      "step": 3789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.625e-06,
+      "loss": 1.4902,
+      "step": 3790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6125e-06,
+      "loss": 1.3311,
+      "step": 3791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6e-06,
+      "loss": 1.2539,
+      "step": 3792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5875e-06,
+      "loss": 1.207,
+      "step": 3793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.575e-06,
+      "loss": 1.2402,
+      "step": 3794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5625e-06,
+      "loss": 1.2656,
+      "step": 3795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.55e-06,
+      "loss": 1.3477,
+      "step": 3796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5375000000000003e-06,
+      "loss": 1.3359,
+      "step": 3797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5250000000000004e-06,
+      "loss": 1.5078,
+      "step": 3798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5125e-06,
+      "loss": 1.2354,
+      "step": 3799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5e-06,
+      "loss": 1.3281,
+      "step": 3800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4875000000000003e-06,
+      "loss": 1.3818,
+      "step": 3801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4750000000000004e-06,
+      "loss": 1.3828,
+      "step": 3802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4625e-06,
+      "loss": 1.0039,
+      "step": 3803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.9927,
+      "step": 3804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4375000000000004e-06,
+      "loss": 1.0088,
+      "step": 3805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.425e-06,
+      "loss": 1.2783,
+      "step": 3806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4125e-06,
+      "loss": 0.9819,
+      "step": 3807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.3262,
+      "step": 3808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3875e-06,
+      "loss": 1.3252,
+      "step": 3809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.375e-06,
+      "loss": 1.0264,
+      "step": 3810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3625000000000003e-06,
+      "loss": 1.0664,
+      "step": 3811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.35e-06,
+      "loss": 1.0996,
+      "step": 3812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3375e-06,
+      "loss": 1.1348,
+      "step": 3813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.325e-06,
+      "loss": 1.3252,
+      "step": 3814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3125e-06,
+      "loss": 1.2949,
+      "step": 3815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3e-06,
+      "loss": 1.3857,
+      "step": 3816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2875e-06,
+      "loss": 1.3184,
+      "step": 3817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2750000000000002e-06,
+      "loss": 1.293,
+      "step": 3818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2625e-06,
+      "loss": 1.4854,
+      "step": 3819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.25e-06,
+      "loss": 1.0645,
+      "step": 3820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2375e-06,
+      "loss": 1.2246,
+      "step": 3821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.225e-06,
+      "loss": 1.418,
+      "step": 3822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2125e-06,
+      "loss": 1.3613,
+      "step": 3823
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2e-06,
+      "loss": 1.4756,
+      "step": 3824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1875e-06,
+      "loss": 1.2256,
+      "step": 3825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.175e-06,
+      "loss": 1.1406,
+      "step": 3826
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1625e-06,
+      "loss": 1.2148,
+      "step": 3827
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1499999999999997e-06,
+      "loss": 1.3506,
+      "step": 3828
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1375000000000003e-06,
+      "loss": 1.2285,
+      "step": 3829
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1250000000000004e-06,
+      "loss": 1.2881,
+      "step": 3830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1125e-06,
+      "loss": 1.4102,
+      "step": 3831
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.3613,
+      "step": 3832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0875000000000003e-06,
+      "loss": 1.417,
+      "step": 3833
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.075e-06,
+      "loss": 1.2607,
+      "step": 3834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0625e-06,
+      "loss": 1.1934,
+      "step": 3835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0500000000000003e-06,
+      "loss": 1.3535,
+      "step": 3836
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0375e-06,
+      "loss": 1.2998,
+      "step": 3837
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.025e-06,
+      "loss": 1.3965,
+      "step": 3838
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0125000000000002e-06,
+      "loss": 1.0449,
+      "step": 3839
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.2246,
+      "step": 3840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9875e-06,
+      "loss": 1.0557,
+      "step": 3841
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.975e-06,
+      "loss": 1.4668,
+      "step": 3842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9625000000000003e-06,
+      "loss": 1.1699,
+      "step": 3843
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.95e-06,
+      "loss": 1.3848,
+      "step": 3844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9375e-06,
+      "loss": 1.333,
+      "step": 3845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.925e-06,
+      "loss": 1.2119,
+      "step": 3846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9125e-06,
+      "loss": 1.3311,
+      "step": 3847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9e-06,
+      "loss": 1.3037,
+      "step": 3848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8875e-06,
+      "loss": 1.333,
+      "step": 3849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.875e-06,
+      "loss": 1.1738,
+      "step": 3850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8625e-06,
+      "loss": 1.1816,
+      "step": 3851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.85e-06,
+      "loss": 1.1045,
+      "step": 3852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8375e-06,
+      "loss": 1.0771,
+      "step": 3853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8249999999999999e-06,
+      "loss": 1.3105,
+      "step": 3854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8125e-06,
+      "loss": 1.123,
+      "step": 3855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8e-06,
+      "loss": 1.0811,
+      "step": 3856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7875e-06,
+      "loss": 1.3984,
+      "step": 3857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.775e-06,
+      "loss": 1.2549,
+      "step": 3858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7624999999999999e-06,
+      "loss": 1.1787,
+      "step": 3859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7500000000000002e-06,
+      "loss": 1.21,
+      "step": 3860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7375000000000003e-06,
+      "loss": 1.5391,
+      "step": 3861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7250000000000002e-06,
+      "loss": 1.3799,
+      "step": 3862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7125000000000001e-06,
+      "loss": 1.1201,
+      "step": 3863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.1396,
+      "step": 3864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6875000000000001e-06,
+      "loss": 1.2822,
+      "step": 3865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6750000000000003e-06,
+      "loss": 1.1436,
+      "step": 3866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6625000000000002e-06,
+      "loss": 1.1953,
+      "step": 3867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.65e-06,
+      "loss": 1.2012,
+      "step": 3868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6375000000000002e-06,
+      "loss": 1.1992,
+      "step": 3869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6250000000000001e-06,
+      "loss": 1.2686,
+      "step": 3870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6125e-06,
+      "loss": 1.2441,
+      "step": 3871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.2842,
+      "step": 3872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5875e-06,
+      "loss": 1.4326,
+      "step": 3873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5750000000000002e-06,
+      "loss": 1.248,
+      "step": 3874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5625e-06,
+      "loss": 0.9229,
+      "step": 3875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.55e-06,
+      "loss": 1.3838,
+      "step": 3876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5375e-06,
+      "loss": 1.418,
+      "step": 3877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.525e-06,
+      "loss": 1.2959,
+      "step": 3878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5125000000000001e-06,
+      "loss": 1.0225,
+      "step": 3879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5e-06,
+      "loss": 1.2207,
+      "step": 3880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4875e-06,
+      "loss": 1.3359,
+      "step": 3881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.475e-06,
+      "loss": 1.2246,
+      "step": 3882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4625000000000002e-06,
+      "loss": 1.1445,
+      "step": 3883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.45e-06,
+      "loss": 1.3145,
+      "step": 3884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4375000000000002e-06,
+      "loss": 1.3447,
+      "step": 3885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4250000000000001e-06,
+      "loss": 1.4814,
+      "step": 3886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4125e-06,
+      "loss": 1.1943,
+      "step": 3887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.2383,
+      "step": 3888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3875e-06,
+      "loss": 1.6289,
+      "step": 3889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3750000000000002e-06,
+      "loss": 1.3096,
+      "step": 3890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3625e-06,
+      "loss": 1.2725,
+      "step": 3891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.35e-06,
+      "loss": 1.0146,
+      "step": 3892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3375000000000001e-06,
+      "loss": 1.248,
+      "step": 3893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.325e-06,
+      "loss": 1.248,
+      "step": 3894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3125e-06,
+      "loss": 1.2939,
+      "step": 3895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3e-06,
+      "loss": 1.2373,
+      "step": 3896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2875e-06,
+      "loss": 1.6025,
+      "step": 3897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.275e-06,
+      "loss": 1.2646,
+      "step": 3898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2625000000000002e-06,
+      "loss": 1.0723,
+      "step": 3899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.25e-06,
+      "loss": 1.4121,
+      "step": 3900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2375000000000002e-06,
+      "loss": 1.5176,
+      "step": 3901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2250000000000001e-06,
+      "loss": 1.5586,
+      "step": 3902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2125e-06,
+      "loss": 1.418,
+      "step": 3903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.4131,
+      "step": 3904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1875e-06,
+      "loss": 1.0273,
+      "step": 3905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.175e-06,
+      "loss": 1.1084,
+      "step": 3906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1625e-06,
+      "loss": 1.1436,
+      "step": 3907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.15e-06,
+      "loss": 0.8188,
+      "step": 3908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1375000000000001e-06,
+      "loss": 1.0283,
+      "step": 3909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.125e-06,
+      "loss": 1.2119,
+      "step": 3910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1125e-06,
+      "loss": 1.3271,
+      "step": 3911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1e-06,
+      "loss": 1.2979,
+      "step": 3912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0875e-06,
+      "loss": 1.3213,
+      "step": 3913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0749999999999999e-06,
+      "loss": 1.5889,
+      "step": 3914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0625000000000002e-06,
+      "loss": 1.0137,
+      "step": 3915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 1.2217,
+      "step": 3916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0375e-06,
+      "loss": 1.5049,
+      "step": 3917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0250000000000001e-06,
+      "loss": 1.2471,
+      "step": 3918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0125e-06,
+      "loss": 1.334,
+      "step": 3919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.2021,
+      "step": 3920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.875e-07,
+      "loss": 1.3145,
+      "step": 3921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.75e-07,
+      "loss": 1.2441,
+      "step": 3922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.625e-07,
+      "loss": 1.3389,
+      "step": 3923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.5e-07,
+      "loss": 1.3398,
+      "step": 3924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.375e-07,
+      "loss": 1.2734,
+      "step": 3925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.25e-07,
+      "loss": 1.3096,
+      "step": 3926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.124999999999999e-07,
+      "loss": 1.167,
+      "step": 3927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9e-07,
+      "loss": 1.25,
+      "step": 3928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.875e-07,
+      "loss": 1.1689,
+      "step": 3929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.750000000000001e-07,
+      "loss": 1.127,
+      "step": 3930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.625000000000001e-07,
+      "loss": 1.3291,
+      "step": 3931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 1.5127,
+      "step": 3932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.375000000000001e-07,
+      "loss": 1.2793,
+      "step": 3933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.25e-07,
+      "loss": 1.2256,
+      "step": 3934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.125000000000001e-07,
+      "loss": 0.9551,
+      "step": 3935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.2109,
+      "step": 3936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.875000000000001e-07,
+      "loss": 1.29,
+      "step": 3937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.75e-07,
+      "loss": 1.0986,
+      "step": 3938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.625e-07,
+      "loss": 1.2354,
+      "step": 3939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.5e-07,
+      "loss": 1.3447,
+      "step": 3940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.375e-07,
+      "loss": 1.0684,
+      "step": 3941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.25e-07,
+      "loss": 1.3711,
+      "step": 3942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.125000000000001e-07,
+      "loss": 1.1572,
+      "step": 3943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 1.1396,
+      "step": 3944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.875000000000001e-07,
+      "loss": 1.2822,
+      "step": 3945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.75e-07,
+      "loss": 1.1562,
+      "step": 3946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.625e-07,
+      "loss": 1.2549,
+      "step": 3947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5e-07,
+      "loss": 1.1582,
+      "step": 3948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.375e-07,
+      "loss": 1.3613,
+      "step": 3949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.25e-07,
+      "loss": 1.1768,
+      "step": 3950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.125000000000001e-07,
+      "loss": 1.3281,
+      "step": 3951
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.1309,
+      "step": 3952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.875e-07,
+      "loss": 1.2266,
+      "step": 3953
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.75e-07,
+      "loss": 1.4609,
+      "step": 3954
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.625e-07,
+      "loss": 1.291,
+      "step": 3955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5e-07,
+      "loss": 1.4795,
+      "step": 3956
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.374999999999999e-07,
+      "loss": 1.1787,
+      "step": 3957
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.250000000000001e-07,
+      "loss": 1.4355,
+      "step": 3958
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.125000000000001e-07,
+      "loss": 1.2871,
+      "step": 3959
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.2705,
+      "step": 3960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.875e-07,
+      "loss": 1.1621,
+      "step": 3961
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.75e-07,
+      "loss": 1.0898,
+      "step": 3962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.625e-07,
+      "loss": 1.3193,
+      "step": 3963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5e-07,
+      "loss": 1.2773,
+      "step": 3964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3750000000000005e-07,
+      "loss": 1.3633,
+      "step": 3965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2500000000000006e-07,
+      "loss": 1.3447,
+      "step": 3966
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.125e-07,
+      "loss": 1.583,
+      "step": 3967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.3125,
+      "step": 3968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.875e-07,
+      "loss": 1.252,
+      "step": 3969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.75e-07,
+      "loss": 1.3906,
+      "step": 3970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.625e-07,
+      "loss": 1.1475,
+      "step": 3971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 1.1514,
+      "step": 3972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.375e-07,
+      "loss": 1.1504,
+      "step": 3973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.25e-07,
+      "loss": 1.0762,
+      "step": 3974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.125e-07,
+      "loss": 1.2246,
+      "step": 3975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.3662,
+      "step": 3976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.875e-07,
+      "loss": 1.2119,
+      "step": 3977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.75e-07,
+      "loss": 1.1064,
+      "step": 3978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6250000000000003e-07,
+      "loss": 1.1475,
+      "step": 3979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 1.0459,
+      "step": 3980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.375e-07,
+      "loss": 1.3818,
+      "step": 3981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.25e-07,
+      "loss": 1.6777,
+      "step": 3982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1250000000000003e-07,
+      "loss": 1.3271,
+      "step": 3983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.2764,
+      "step": 3984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.875e-07,
+      "loss": 1.1172,
+      "step": 3985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7500000000000002e-07,
+      "loss": 1.1182,
+      "step": 3986
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.625e-07,
+      "loss": 1.3701,
+      "step": 3987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 1.5996,
+      "step": 3988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.375e-07,
+      "loss": 1.3115,
+      "step": 3989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2500000000000002e-07,
+      "loss": 1.1465,
+      "step": 3990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.125e-07,
+      "loss": 1.4287,
+      "step": 3991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 1.5225,
+      "step": 3992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.750000000000001e-08,
+      "loss": 1.165,
+      "step": 3993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.500000000000001e-08,
+      "loss": 1.2617,
+      "step": 3994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.250000000000001e-08,
+      "loss": 0.9238,
+      "step": 3995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 1.2441,
+      "step": 3996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7500000000000005e-08,
+      "loss": 1.749,
+      "step": 3997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5000000000000002e-08,
+      "loss": 1.0469,
+      "step": 3998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2500000000000001e-08,
+      "loss": 1.2432,
+      "step": 3999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.2295,
+      "step": 4000
+    },
+    {
+      "epoch": 1.0,
+      "eval_bleu-4": 0.21864021599726502,
+      "eval_rouge-1": 59.13236399999999,
+      "eval_rouge-2": 29.168236,
+      "eval_rouge-l": 40.408967999999994,
+      "eval_runtime": 198.3621,
+      "eval_samples_per_second": 0.252,
+      "eval_steps_per_second": 0.126,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 4000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 4000,
+  "total_flos": 3.3328801620526694e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}