diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,91345 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9998518006224373,
+  "global_step": 15180,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 12.3801,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 12.3929,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 12.3967,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 12.3743,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 12.3937,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 12.3454,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-07,
+      "loss": 12.3453,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 12.3661,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 12.3232,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 12.3285,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 12.3217,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 12.3083,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 12.2884,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4e-07,
+      "loss": 12.2589,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-07,
+      "loss": 12.2341,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 12.1871,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 12.2138,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08e-06,
+      "loss": 12.1842,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.14e-06,
+      "loss": 12.1688,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 12.1013,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.26e-06,
+      "loss": 12.0711,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 12.0639,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.38e-06,
+      "loss": 11.993,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.44e-06,
+      "loss": 11.9764,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 11.9265,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 11.8626,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.62e-06,
+      "loss": 11.8609,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.68e-06,
+      "loss": 11.8234,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 11.8372,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 11.697,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86e-06,
+      "loss": 11.68,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 11.5844,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 11.5602,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 11.4774,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 11.4952,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 11.4523,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 11.3282,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22e-06,
+      "loss": 11.3368,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 11.2236,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.34e-06,
+      "loss": 11.1318,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 11.0139,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-06,
+      "loss": 10.9061,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.52e-06,
+      "loss": 10.8178,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58e-06,
+      "loss": 10.718,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 10.5777,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7e-06,
+      "loss": 10.5334,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.76e-06,
+      "loss": 10.4455,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.82e-06,
+      "loss": 10.3529,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88e-06,
+      "loss": 10.2405,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 10.071,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 9.8866,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06e-06,
+      "loss": 9.8695,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 9.7723,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.18e-06,
+      "loss": 9.6178,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.24e-06,
+      "loss": 9.5235,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-06,
+      "loss": 9.5275,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.36e-06,
+      "loss": 9.5092,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 9.3425,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.48e-06,
+      "loss": 9.2305,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.54e-06,
+      "loss": 9.2398,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 9.1602,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.66e-06,
+      "loss": 9.1101,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.72e-06,
+      "loss": 9.1138,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 9.1466,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 9.117,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9e-06,
+      "loss": 8.9632,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96e-06,
+      "loss": 8.8877,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 8.7984,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 8.9423,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.14e-06,
+      "loss": 8.7744,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 8.8704,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.26e-06,
+      "loss": 8.7296,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.32e-06,
+      "loss": 8.7101,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 8.6498,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.44e-06,
+      "loss": 8.5908,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-06,
+      "loss": 8.7431,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.56e-06,
+      "loss": 8.7721,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.62e-06,
+      "loss": 8.592,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.68e-06,
+      "loss": 8.588,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.74e-06,
+      "loss": 8.5709,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 8.5152,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.86e-06,
+      "loss": 8.5545,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92e-06,
+      "loss": 8.4252,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 8.3584,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.04e-06,
+      "loss": 8.3339,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1e-06,
+      "loss": 8.2722,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.16e-06,
+      "loss": 8.2823,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.22e-06,
+      "loss": 8.3021,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 8.182,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.34e-06,
+      "loss": 8.1713,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4e-06,
+      "loss": 8.0977,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.46e-06,
+      "loss": 8.0096,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.52e-06,
+      "loss": 8.0779,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.58e-06,
+      "loss": 7.9567,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.64e-06,
+      "loss": 7.9677,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 8.0375,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.76e-06,
+      "loss": 7.8269,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.82e-06,
+      "loss": 8.0384,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 7.7336,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 7.708,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 7.6561,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 7.6553,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.12e-06,
+      "loss": 7.6969,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.18e-06,
+      "loss": 7.5265,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 7.5305,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3e-06,
+      "loss": 7.5742,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.36e-06,
+      "loss": 7.5435,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.42e-06,
+      "loss": 7.4033,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.48e-06,
+      "loss": 7.3895,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.54e-06,
+      "loss": 7.3846,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-06,
+      "loss": 7.329,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 7.0505,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.72e-06,
+      "loss": 7.1392,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.78e-06,
+      "loss": 7.181,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 7.3663,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 7.0387,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.96e-06,
+      "loss": 7.254,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 7.0285,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.08e-06,
+      "loss": 7.0737,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.14e-06,
+      "loss": 6.9673,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-06,
+      "loss": 7.0091,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.26e-06,
+      "loss": 6.9219,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.32e-06,
+      "loss": 6.8354,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 6.9131,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.44e-06,
+      "loss": 6.7805,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5e-06,
+      "loss": 6.7293,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 6.6881,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.62e-06,
+      "loss": 6.6949,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 6.7204,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.74e-06,
+      "loss": 6.5903,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.8e-06,
+      "loss": 6.5115,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 6.5847,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.92e-06,
+      "loss": 6.6192,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.98e-06,
+      "loss": 6.38,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 6.4788,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.1e-06,
+      "loss": 6.425,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 6.3648,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 6.4753,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.28e-06,
+      "loss": 6.3986,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 6.3726,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 6.3897,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 6.1938,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.52e-06,
+      "loss": 6.3174,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.58e-06,
+      "loss": 6.2911,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.64e-06,
+      "loss": 6.1742,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.7e-06,
+      "loss": 6.2254,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 6.0813,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.82e-06,
+      "loss": 5.9342,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.88e-06,
+      "loss": 6.237,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 6.1599,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-06,
+      "loss": 6.1491,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.06e-06,
+      "loss": 6.12,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.12e-06,
+      "loss": 6.007,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.18e-06,
+      "loss": 5.922,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.24e-06,
+      "loss": 6.0716,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.3e-06,
+      "loss": 5.9792,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.36e-06,
+      "loss": 5.9546,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.42e-06,
+      "loss": 5.8486,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.48e-06,
+      "loss": 5.8958,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.54e-06,
+      "loss": 5.851,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 5.7608,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.66e-06,
+      "loss": 5.7972,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.72e-06,
+      "loss": 5.7225,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 5.718,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.84e-06,
+      "loss": 6.0537,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9e-06,
+      "loss": 5.6478,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 5.63,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.002e-05,
+      "loss": 5.6676,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.008e-05,
+      "loss": 5.6907,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 5.622,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.02e-05,
+      "loss": 5.5274,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 5.5308,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.032e-05,
+      "loss": 5.4303,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 5.4158,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.044e-05,
+      "loss": 5.5195,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.05e-05,
+      "loss": 5.5684,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 5.5652,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.062e-05,
+      "loss": 5.5245,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.068e-05,
+      "loss": 5.1648,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.074e-05,
+      "loss": 5.3321,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.08e-05,
+      "loss": 5.2989,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.086e-05,
+      "loss": 5.3067,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.092e-05,
+      "loss": 5.3192,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.098e-05,
+      "loss": 5.3819,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.104e-05,
+      "loss": 5.3156,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.11e-05,
+      "loss": 5.205,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.116e-05,
+      "loss": 5.2229,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 5.4222,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.128e-05,
+      "loss": 5.2472,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.134e-05,
+      "loss": 5.2894,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 5.1497,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.146e-05,
+      "loss": 5.186,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.152e-05,
+      "loss": 5.0915,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 5.0453,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.164e-05,
+      "loss": 4.9805,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 5.1284,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 5.1363,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.182e-05,
+      "loss": 5.0012,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 4.9152,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 5.0248,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2e-05,
+      "loss": 5.224,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 5.0139,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 4.9067,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 4.9474,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.224e-05,
+      "loss": 4.9904,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 4.8646,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.236e-05,
+      "loss": 4.9858,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.242e-05,
+      "loss": 4.7796,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 4.9048,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.254e-05,
+      "loss": 4.7484,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.26e-05,
+      "loss": 4.8515,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 4.7031,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.272e-05,
+      "loss": 4.6517,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.278e-05,
+      "loss": 4.8576,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.284e-05,
+      "loss": 4.6228,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.29e-05,
+      "loss": 4.7242,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.296e-05,
+      "loss": 4.544,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.302e-05,
+      "loss": 4.756,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.308e-05,
+      "loss": 4.7433,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.314e-05,
+      "loss": 4.5988,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.32e-05,
+      "loss": 4.9302,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.326e-05,
+      "loss": 4.7219,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 4.5483,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.338e-05,
+      "loss": 4.7253,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.344e-05,
+      "loss": 4.5723,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 4.5342,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.356e-05,
+      "loss": 4.6296,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.362e-05,
+      "loss": 4.3451,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 4.4181,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.374e-05,
+      "loss": 4.4814,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.3577,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 4.332,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.392e-05,
+      "loss": 4.4446,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 4.2664,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 4.4636,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 4.4082,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.416e-05,
+      "loss": 4.4765,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.422e-05,
+      "loss": 4.3272,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.428e-05,
+      "loss": 4.4187,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.434e-05,
+      "loss": 4.1713,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.44e-05,
+      "loss": 4.347,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.446e-05,
+      "loss": 4.1727,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.452e-05,
+      "loss": 4.2819,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.458e-05,
+      "loss": 4.2059,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.464e-05,
+      "loss": 4.2926,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.47e-05,
+      "loss": 4.0421,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 4.2229,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.482e-05,
+      "loss": 4.0699,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.488e-05,
+      "loss": 4.2471,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 4.1093,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5e-05,
+      "loss": 4.0828,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.506e-05,
+      "loss": 4.158,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 4.0412,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.518e-05,
+      "loss": 4.1331,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.524e-05,
+      "loss": 4.0727,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.53e-05,
+      "loss": 4.1218,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 4.0166,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.542e-05,
+      "loss": 3.9105,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.548e-05,
+      "loss": 4.1143,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.554e-05,
+      "loss": 3.9257,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.56e-05,
+      "loss": 3.9827,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 3.8388,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 3.8569,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.578e-05,
+      "loss": 3.9764,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.584e-05,
+      "loss": 4.0168,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.59e-05,
+      "loss": 3.9387,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.596e-05,
+      "loss": 3.8402,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 3.7839,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 3.9186,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.614e-05,
+      "loss": 3.8354,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.62e-05,
+      "loss": 3.7232,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.626e-05,
+      "loss": 3.7987,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 3.6899,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 3.6989,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 3.7435,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.65e-05,
+      "loss": 3.7494,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.656e-05,
+      "loss": 3.7046,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 3.7632,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 3.8118,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 3.7233,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 3.7142,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.686e-05,
+      "loss": 3.7401,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 3.6723,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.698e-05,
+      "loss": 3.6506,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.704e-05,
+      "loss": 3.7487,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.71e-05,
+      "loss": 3.6867,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.716e-05,
+      "loss": 3.5443,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 3.6806,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.728e-05,
+      "loss": 3.5317,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.734e-05,
+      "loss": 3.6411,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.74e-05,
+      "loss": 3.5101,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.746e-05,
+      "loss": 3.3998,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 3.3966,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.758e-05,
+      "loss": 3.5462,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.764e-05,
+      "loss": 3.5081,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.77e-05,
+      "loss": 3.4589,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.776e-05,
+      "loss": 3.5029,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.782e-05,
+      "loss": 3.5852,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 3.5732,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.794e-05,
+      "loss": 3.4392,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8e-05,
+      "loss": 3.5542,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.806e-05,
+      "loss": 3.5326,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.812e-05,
+      "loss": 3.3307,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.818e-05,
+      "loss": 3.4199,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.824e-05,
+      "loss": 3.4354,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.83e-05,
+      "loss": 3.4544,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.836e-05,
+      "loss": 3.3532,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.842e-05,
+      "loss": 3.4311,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.848e-05,
+      "loss": 3.449,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.854e-05,
+      "loss": 3.2284,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.86e-05,
+      "loss": 3.3197,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.866e-05,
+      "loss": 3.3924,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.872e-05,
+      "loss": 3.3603,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.878e-05,
+      "loss": 3.3255,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.884e-05,
+      "loss": 3.294,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 3.2856,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.896e-05,
+      "loss": 3.297,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.902e-05,
+      "loss": 3.2013,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.908e-05,
+      "loss": 3.2123,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.914e-05,
+      "loss": 3.3766,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 3.2169,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 3.2414,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.932e-05,
+      "loss": 3.208,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.938e-05,
+      "loss": 3.2303,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.944e-05,
+      "loss": 3.3072,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.95e-05,
+      "loss": 3.1741,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 3.2902,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 3.0851,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.968e-05,
+      "loss": 3.3922,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974e-05,
+      "loss": 3.3095,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98e-05,
+      "loss": 3.195,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 3.1421,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 3.2243,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 3.3346,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.004e-05,
+      "loss": 3.1639,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.01e-05,
+      "loss": 3.3177,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.016e-05,
+      "loss": 3.1052,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 3.2687,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 2.9727,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 3.3701,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.04e-05,
+      "loss": 3.1012,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.046e-05,
+      "loss": 3.0635,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 3.0932,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 3.0162,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.064e-05,
+      "loss": 3.3317,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.07e-05,
+      "loss": 3.0678,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 3.0766,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.082e-05,
+      "loss": 3.0972,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.088e-05,
+      "loss": 3.034,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.094e-05,
+      "loss": 3.2712,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1e-05,
+      "loss": 3.0938,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 3.0031,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 3.05,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.118e-05,
+      "loss": 3.1495,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.124e-05,
+      "loss": 2.9805,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.13e-05,
+      "loss": 3.1091,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.136e-05,
+      "loss": 3.0721,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 3.0507,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.148e-05,
+      "loss": 2.9576,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.154e-05,
+      "loss": 3.1014,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.16e-05,
+      "loss": 3.1092,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.166e-05,
+      "loss": 3.1224,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.172e-05,
+      "loss": 2.8643,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.178e-05,
+      "loss": 3.0159,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.184e-05,
+      "loss": 2.9211,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19e-05,
+      "loss": 2.9402,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.196e-05,
+      "loss": 2.8248,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.202e-05,
+      "loss": 2.9189,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.208e-05,
+      "loss": 2.9373,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.214e-05,
+      "loss": 2.8908,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.22e-05,
+      "loss": 2.9209,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.226e-05,
+      "loss": 2.8966,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.232e-05,
+      "loss": 3.0599,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.238e-05,
+      "loss": 2.8957,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2440000000000002e-05,
+      "loss": 2.8859,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.25e-05,
+      "loss": 2.9529,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.256e-05,
+      "loss": 2.896,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.262e-05,
+      "loss": 2.9233,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.268e-05,
+      "loss": 2.8866,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.274e-05,
+      "loss": 3.1204,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 2.9606,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.286e-05,
+      "loss": 2.6931,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.292e-05,
+      "loss": 2.9361,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.298e-05,
+      "loss": 2.9372,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.304e-05,
+      "loss": 2.8246,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 2.9921,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 2.8385,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.322e-05,
+      "loss": 2.7818,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.328e-05,
+      "loss": 2.8416,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.334e-05,
+      "loss": 2.8757,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 2.8672,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 2.8684,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 2.8921,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.358e-05,
+      "loss": 2.9504,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.364e-05,
+      "loss": 2.8815,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.37e-05,
+      "loss": 2.9419,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 2.7901,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 2.7493,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 2.8212,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.394e-05,
+      "loss": 2.8882,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4e-05,
+      "loss": 2.7358,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 2.6398,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4120000000000003e-05,
+      "loss": 2.704,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 2.8057,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 2.9874,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.43e-05,
+      "loss": 2.8645,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 2.8209,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.442e-05,
+      "loss": 2.7177,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.448e-05,
+      "loss": 2.6773,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.454e-05,
+      "loss": 2.8735,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 2.7748,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4659999999999998e-05,
+      "loss": 2.7497,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.472e-05,
+      "loss": 2.7039,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.478e-05,
+      "loss": 2.7781,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.484e-05,
+      "loss": 2.7243,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.49e-05,
+      "loss": 2.6549,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4959999999999998e-05,
+      "loss": 2.6735,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.502e-05,
+      "loss": 2.7178,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.508e-05,
+      "loss": 2.5894,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.514e-05,
+      "loss": 2.7761,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.52e-05,
+      "loss": 2.7245,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.526e-05,
+      "loss": 2.7532,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 2.5883,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.538e-05,
+      "loss": 2.7041,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.544e-05,
+      "loss": 2.8014,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.55e-05,
+      "loss": 2.695,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.556e-05,
+      "loss": 2.853,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.562e-05,
+      "loss": 2.6956,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.568e-05,
+      "loss": 2.864,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.574e-05,
+      "loss": 2.7395,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.58e-05,
+      "loss": 2.7908,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.586e-05,
+      "loss": 2.6616,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.592e-05,
+      "loss": 2.6582,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 2.6051,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.604e-05,
+      "loss": 2.698,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.61e-05,
+      "loss": 2.7423,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.616e-05,
+      "loss": 2.5528,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.622e-05,
+      "loss": 2.7297,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.628e-05,
+      "loss": 2.5698,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 2.729,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.64e-05,
+      "loss": 2.6491,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.646e-05,
+      "loss": 2.7376,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.652e-05,
+      "loss": 2.5554,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.658e-05,
+      "loss": 2.6081,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 2.6791,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 2.6808,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.676e-05,
+      "loss": 2.6644,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.682e-05,
+      "loss": 2.6703,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.688e-05,
+      "loss": 2.668,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6940000000000003e-05,
+      "loss": 2.5844,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 2.6493,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 2.5271,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.712e-05,
+      "loss": 2.598,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.718e-05,
+      "loss": 2.6091,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.724e-05,
+      "loss": 2.6002,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 2.6429,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 2.4507,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 2.6189,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.748e-05,
+      "loss": 2.5751,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.754e-05,
+      "loss": 2.6647,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 2.5896,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 2.6107,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 2.6181,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.778e-05,
+      "loss": 2.6913,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.784e-05,
+      "loss": 2.6338,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.79e-05,
+      "loss": 2.6302,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 2.5553,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 2.63,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 2.3935,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 2.577,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 2.4727,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.826e-05,
+      "loss": 2.6581,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.832e-05,
+      "loss": 2.5075,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.838e-05,
+      "loss": 2.5514,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.844e-05,
+      "loss": 2.4789,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 2.5421,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.856e-05,
+      "loss": 2.5403,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.862e-05,
+      "loss": 2.5571,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.868e-05,
+      "loss": 2.4614,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.874e-05,
+      "loss": 2.5497,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.88e-05,
+      "loss": 2.5314,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8859999999999998e-05,
+      "loss": 2.6329,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.892e-05,
+      "loss": 2.6226,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.898e-05,
+      "loss": 2.5398,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.904e-05,
+      "loss": 2.513,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.91e-05,
+      "loss": 2.517,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.916e-05,
+      "loss": 2.5076,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.922e-05,
+      "loss": 2.6412,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.928e-05,
+      "loss": 2.4502,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.934e-05,
+      "loss": 2.6583,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.94e-05,
+      "loss": 2.4653,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.946e-05,
+      "loss": 2.6104,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 2.4697,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.958e-05,
+      "loss": 2.4799,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.964e-05,
+      "loss": 2.4511,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.97e-05,
+      "loss": 2.5321,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.976e-05,
+      "loss": 2.5027,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.982e-05,
+      "loss": 2.4613,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 2.5935,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.994e-05,
+      "loss": 2.5636,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.2399110794067383,
+      "eval_runtime": 728.7565,
+      "eval_samples_per_second": 5.695,
+      "eval_steps_per_second": 1.424,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3e-05,
+      "loss": 2.4248,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9997956403269756e-05,
+      "loss": 2.4602,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.999591280653951e-05,
+      "loss": 2.4505,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9993869209809264e-05,
+      "loss": 2.3054,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.999182561307902e-05,
+      "loss": 2.5762,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9989782016348776e-05,
+      "loss": 2.5597,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998773841961853e-05,
+      "loss": 2.5569,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9985694822888283e-05,
+      "loss": 2.5018,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998365122615804e-05,
+      "loss": 2.5209,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981607629427795e-05,
+      "loss": 2.5445,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979564032697547e-05,
+      "loss": 2.5205,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9977520435967303e-05,
+      "loss": 2.3011,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975476839237058e-05,
+      "loss": 2.4657,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973433242506814e-05,
+      "loss": 2.5306,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997138964577657e-05,
+      "loss": 2.482,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969346049046322e-05,
+      "loss": 2.5655,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967302452316077e-05,
+      "loss": 2.5447,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996525885558583e-05,
+      "loss": 2.3372,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963215258855585e-05,
+      "loss": 2.3873,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9961171662125344e-05,
+      "loss": 2.5628,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9959128065395096e-05,
+      "loss": 2.4877,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9957084468664852e-05,
+      "loss": 2.3684,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9955040871934604e-05,
+      "loss": 2.3295,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.995299727520436e-05,
+      "loss": 2.4358,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9950953678474116e-05,
+      "loss": 2.4743,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9948910081743868e-05,
+      "loss": 2.4866,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9946866485013623e-05,
+      "loss": 2.5405,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.994482288828338e-05,
+      "loss": 2.442,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9942779291553135e-05,
+      "loss": 2.5246,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.994073569482289e-05,
+      "loss": 2.4173,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9938692098092643e-05,
+      "loss": 2.2504,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9936648501362398e-05,
+      "loss": 2.4856,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9934604904632154e-05,
+      "loss": 2.529,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9932561307901906e-05,
+      "loss": 2.5071,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9930517711171665e-05,
+      "loss": 2.3689,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9928474114441417e-05,
+      "loss": 2.4892,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9926430517711173e-05,
+      "loss": 2.3611,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.992438692098093e-05,
+      "loss": 2.5086,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.992234332425068e-05,
+      "loss": 2.4777,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9920299727520437e-05,
+      "loss": 2.453,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.991825613079019e-05,
+      "loss": 2.4161,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9916212534059944e-05,
+      "loss": 2.4228,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9914168937329703e-05,
+      "loss": 2.4055,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9912125340599456e-05,
+      "loss": 2.4232,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.991008174386921e-05,
+      "loss": 2.3001,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9908038147138964e-05,
+      "loss": 2.3646,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.990599455040872e-05,
+      "loss": 2.4095,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9903950953678475e-05,
+      "loss": 2.6815,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9901907356948227e-05,
+      "loss": 2.408,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9899863760217986e-05,
+      "loss": 2.5317,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9897820163487742e-05,
+      "loss": 2.4629,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9895776566757494e-05,
+      "loss": 2.4236,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.989373297002725e-05,
+      "loss": 2.4135,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9891689373297002e-05,
+      "loss": 2.4303,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9889645776566757e-05,
+      "loss": 2.4199,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9887602179836513e-05,
+      "loss": 2.3967,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.988555858310627e-05,
+      "loss": 2.3507,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9883514986376024e-05,
+      "loss": 2.4983,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9881471389645777e-05,
+      "loss": 2.4641,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9879427792915532e-05,
+      "loss": 2.4929,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9877384196185288e-05,
+      "loss": 2.4224,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.987534059945504e-05,
+      "loss": 2.2867,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9873297002724796e-05,
+      "loss": 2.4719,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9871253405994548e-05,
+      "loss": 2.2612,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9869209809264307e-05,
+      "loss": 2.4678,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9867166212534063e-05,
+      "loss": 2.4929,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9865122615803815e-05,
+      "loss": 2.3401,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.986307901907357e-05,
+      "loss": 2.4002,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9861035422343326e-05,
+      "loss": 2.4996,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.985899182561308e-05,
+      "loss": 2.3747,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9856948228882834e-05,
+      "loss": 2.4703,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.985490463215259e-05,
+      "loss": 2.4612,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9852861035422345e-05,
+      "loss": 2.3502,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.98508174386921e-05,
+      "loss": 2.3452,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9848773841961853e-05,
+      "loss": 2.2388,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.984673024523161e-05,
+      "loss": 2.3653,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.984468664850136e-05,
+      "loss": 2.4652,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9842643051771117e-05,
+      "loss": 2.2776,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9840599455040872e-05,
+      "loss": 2.4145,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9838555858310628e-05,
+      "loss": 2.4154,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9836512261580384e-05,
+      "loss": 2.5065,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9834468664850136e-05,
+      "loss": 2.5041,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.983242506811989e-05,
+      "loss": 2.2813,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9830381471389647e-05,
+      "loss": 2.3214,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.98283378746594e-05,
+      "loss": 2.3951,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9826294277929155e-05,
+      "loss": 2.453,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9824250681198914e-05,
+      "loss": 2.4337,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9822207084468666e-05,
+      "loss": 2.4373,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9820163487738422e-05,
+      "loss": 2.4124,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9818119891008174e-05,
+      "loss": 2.5282,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.981607629427793e-05,
+      "loss": 2.3704,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9814032697547685e-05,
+      "loss": 2.4899,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9811989100817438e-05,
+      "loss": 2.3083,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9809945504087193e-05,
+      "loss": 2.2678,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.980790190735695e-05,
+      "loss": 2.3309,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9805858310626705e-05,
+      "loss": 2.3492,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.980381471389646e-05,
+      "loss": 2.3422,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9801771117166212e-05,
+      "loss": 2.3608,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9799727520435968e-05,
+      "loss": 2.3159,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.979768392370572e-05,
+      "loss": 2.4553,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9795640326975476e-05,
+      "loss": 2.3998,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9793596730245235e-05,
+      "loss": 2.3923,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9791553133514987e-05,
+      "loss": 2.2469,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9789509536784743e-05,
+      "loss": 2.4529,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.97874659400545e-05,
+      "loss": 2.2854,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.978542234332425e-05,
+      "loss": 2.4183,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9783378746594006e-05,
+      "loss": 2.4061,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.978133514986376e-05,
+      "loss": 2.3072,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9779291553133514e-05,
+      "loss": 2.4387,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9777247956403273e-05,
+      "loss": 2.2426,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9775204359673025e-05,
+      "loss": 2.43,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.977316076294278e-05,
+      "loss": 2.576,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9771117166212533e-05,
+      "loss": 2.3264,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.976907356948229e-05,
+      "loss": 2.3505,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9767029972752045e-05,
+      "loss": 2.6384,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9764986376021797e-05,
+      "loss": 2.4253,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9762942779291556e-05,
+      "loss": 2.5578,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9760899182561308e-05,
+      "loss": 2.4473,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9758855585831064e-05,
+      "loss": 2.2754,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.975681198910082e-05,
+      "loss": 2.3329,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.975476839237057e-05,
+      "loss": 2.2919,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9752724795640327e-05,
+      "loss": 2.4208,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9750681198910083e-05,
+      "loss": 2.2981,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9748637602179835e-05,
+      "loss": 2.3347,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9746594005449594e-05,
+      "loss": 2.3595,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9744550408719346e-05,
+      "loss": 2.4058,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9742506811989102e-05,
+      "loss": 2.2694,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9740463215258858e-05,
+      "loss": 2.3705,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.973841961852861e-05,
+      "loss": 2.2611,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9736376021798366e-05,
+      "loss": 2.3002,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9734332425068118e-05,
+      "loss": 2.4016,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9732288828337877e-05,
+      "loss": 2.4497,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9730245231607632e-05,
+      "loss": 2.4718,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9728201634877385e-05,
+      "loss": 2.3159,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.972615803814714e-05,
+      "loss": 2.305,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9724114441416893e-05,
+      "loss": 2.342,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9722070844686648e-05,
+      "loss": 2.2115,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9720027247956404e-05,
+      "loss": 2.4659,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.971798365122616e-05,
+      "loss": 2.3064,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9715940054495915e-05,
+      "loss": 2.4322,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.971389645776567e-05,
+      "loss": 2.2604,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9711852861035423e-05,
+      "loss": 2.3569,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.970980926430518e-05,
+      "loss": 2.3684,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.970776566757493e-05,
+      "loss": 2.4119,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9705722070844687e-05,
+      "loss": 2.3403,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9703678474114442e-05,
+      "loss": 2.3016,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9701634877384198e-05,
+      "loss": 2.1005,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9699591280653953e-05,
+      "loss": 2.2217,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9697547683923706e-05,
+      "loss": 2.3483,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.969550408719346e-05,
+      "loss": 2.2446,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9693460490463217e-05,
+      "loss": 2.2034,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.969141689373297e-05,
+      "loss": 2.3899,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9689373297002725e-05,
+      "loss": 2.3733,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.968732970027248e-05,
+      "loss": 2.3757,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9685286103542236e-05,
+      "loss": 2.2525,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.968324250681199e-05,
+      "loss": 2.3626,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9681198910081744e-05,
+      "loss": 2.3122,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.96791553133515e-05,
+      "loss": 2.406,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9677111716621255e-05,
+      "loss": 2.3299,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9675068119891007e-05,
+      "loss": 2.2191,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9673024523160763e-05,
+      "loss": 2.2551,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.967098092643052e-05,
+      "loss": 2.2483,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9668937329700274e-05,
+      "loss": 2.2706,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.966689373297003e-05,
+      "loss": 2.2552,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9664850136239782e-05,
+      "loss": 2.1593,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9662806539509538e-05,
+      "loss": 2.2505,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.966076294277929e-05,
+      "loss": 2.3625,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9658719346049046e-05,
+      "loss": 2.327,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9656675749318805e-05,
+      "loss": 2.2089,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9654632152588557e-05,
+      "loss": 2.3537,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9652588555858313e-05,
+      "loss": 2.2406,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9650544959128065e-05,
+      "loss": 2.2519,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.964850136239782e-05,
+      "loss": 2.2514,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9646457765667576e-05,
+      "loss": 2.2632,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.964441416893733e-05,
+      "loss": 2.4283,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9642370572207084e-05,
+      "loss": 2.1899,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.964032697547684e-05,
+      "loss": 2.1519,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9638283378746595e-05,
+      "loss": 2.4794,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.963623978201635e-05,
+      "loss": 2.3071,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9634196185286103e-05,
+      "loss": 2.296,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.963215258855586e-05,
+      "loss": 2.2783,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9630108991825614e-05,
+      "loss": 2.2544,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9628065395095367e-05,
+      "loss": 2.3645,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9626021798365126e-05,
+      "loss": 2.0835,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9623978201634878e-05,
+      "loss": 2.2284,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9621934604904634e-05,
+      "loss": 2.2298,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.961989100817439e-05,
+      "loss": 2.4187,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.961784741144414e-05,
+      "loss": 2.334,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9615803814713897e-05,
+      "loss": 2.2603,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.961376021798365e-05,
+      "loss": 2.2171,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9611716621253405e-05,
+      "loss": 2.1078,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9609673024523164e-05,
+      "loss": 2.3455,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9607629427792916e-05,
+      "loss": 2.317,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9605585831062672e-05,
+      "loss": 2.1024,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9603542234332427e-05,
+      "loss": 2.1551,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.960149863760218e-05,
+      "loss": 2.3243,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9599455040871935e-05,
+      "loss": 2.209,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9597411444141688e-05,
+      "loss": 2.2303,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9595367847411447e-05,
+      "loss": 2.3003,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9593324250681202e-05,
+      "loss": 2.2001,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9591280653950955e-05,
+      "loss": 2.2707,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.958923705722071e-05,
+      "loss": 2.1908,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9587193460490462e-05,
+      "loss": 2.22,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9585149863760218e-05,
+      "loss": 2.2466,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9583106267029974e-05,
+      "loss": 2.4015,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9581062670299726e-05,
+      "loss": 2.3484,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9579019073569485e-05,
+      "loss": 2.1012,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9576975476839237e-05,
+      "loss": 2.386,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9574931880108993e-05,
+      "loss": 2.384,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.957288828337875e-05,
+      "loss": 2.0524,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.95708446866485e-05,
+      "loss": 2.2267,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9568801089918256e-05,
+      "loss": 2.3581,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9566757493188012e-05,
+      "loss": 2.458,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9564713896457768e-05,
+      "loss": 2.1955,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9562670299727523e-05,
+      "loss": 2.3318,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9560626702997275e-05,
+      "loss": 2.1704,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.955858310626703e-05,
+      "loss": 2.2717,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9556539509536787e-05,
+      "loss": 2.115,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.955449591280654e-05,
+      "loss": 2.3032,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9552452316076295e-05,
+      "loss": 2.1911,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9550408719346047e-05,
+      "loss": 2.4192,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9548365122615806e-05,
+      "loss": 2.3311,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.954632152588556e-05,
+      "loss": 2.3407,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9544277929155314e-05,
+      "loss": 2.2481,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.954223433242507e-05,
+      "loss": 2.235,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.954019073569482e-05,
+      "loss": 2.2904,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9538147138964577e-05,
+      "loss": 2.1494,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9536103542234333e-05,
+      "loss": 2.2306,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.953405994550409e-05,
+      "loss": 2.2438,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9532016348773844e-05,
+      "loss": 2.2685,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9529972752043596e-05,
+      "loss": 2.2514,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9527929155313352e-05,
+      "loss": 2.2022,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9525885558583108e-05,
+      "loss": 2.339,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.952384196185286e-05,
+      "loss": 2.2136,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9521798365122616e-05,
+      "loss": 2.4144,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9519754768392375e-05,
+      "loss": 2.3438,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9517711171662127e-05,
+      "loss": 2.2269,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9515667574931882e-05,
+      "loss": 2.2288,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9513623978201635e-05,
+      "loss": 2.2474,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.951158038147139e-05,
+      "loss": 2.3348,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9509536784741146e-05,
+      "loss": 2.2044,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9507493188010898e-05,
+      "loss": 2.3238,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9505449591280654e-05,
+      "loss": 2.2495,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.950340599455041e-05,
+      "loss": 2.2476,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9501362397820165e-05,
+      "loss": 2.2178,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.949931880108992e-05,
+      "loss": 2.2318,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9497275204359673e-05,
+      "loss": 2.325,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.949523160762943e-05,
+      "loss": 2.2691,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.949318801089918e-05,
+      "loss": 2.2515,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9491144414168936e-05,
+      "loss": 2.2084,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9489100817438695e-05,
+      "loss": 2.2617,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9487057220708448e-05,
+      "loss": 2.1756,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9485013623978203e-05,
+      "loss": 2.1534,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.948297002724796e-05,
+      "loss": 2.168,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.948092643051771e-05,
+      "loss": 2.2658,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9478882833787467e-05,
+      "loss": 2.1672,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.947683923705722e-05,
+      "loss": 2.197,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9474795640326975e-05,
+      "loss": 2.1644,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9472752043596734e-05,
+      "loss": 2.1109,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9470708446866486e-05,
+      "loss": 2.2624,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.946866485013624e-05,
+      "loss": 2.2552,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9466621253405994e-05,
+      "loss": 2.2357,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.946457765667575e-05,
+      "loss": 2.1121,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9462534059945505e-05,
+      "loss": 2.1896,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9460490463215257e-05,
+      "loss": 2.3156,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9458446866485016e-05,
+      "loss": 2.1704,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.945640326975477e-05,
+      "loss": 2.1475,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9454359673024524e-05,
+      "loss": 2.228,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.945231607629428e-05,
+      "loss": 2.125,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9450272479564032e-05,
+      "loss": 2.0879,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9448228882833788e-05,
+      "loss": 2.2653,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9446185286103543e-05,
+      "loss": 2.1513,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9444141689373296e-05,
+      "loss": 2.2983,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9442098092643055e-05,
+      "loss": 2.142,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9440054495912807e-05,
+      "loss": 2.213,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9438010899182563e-05,
+      "loss": 2.2446,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9435967302452318e-05,
+      "loss": 2.2791,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.943392370572207e-05,
+      "loss": 2.166,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9431880108991826e-05,
+      "loss": 2.2346,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.942983651226158e-05,
+      "loss": 2.2042,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9427792915531337e-05,
+      "loss": 2.2225,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9425749318801093e-05,
+      "loss": 2.1479,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9423705722070845e-05,
+      "loss": 2.1984,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.94216621253406e-05,
+      "loss": 2.2266,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9419618528610353e-05,
+      "loss": 2.255,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.941757493188011e-05,
+      "loss": 2.236,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9415531335149864e-05,
+      "loss": 2.2285,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9413487738419617e-05,
+      "loss": 2.1949,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9411444141689376e-05,
+      "loss": 2.0705,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.940940054495913e-05,
+      "loss": 2.1927,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9407356948228884e-05,
+      "loss": 2.1382,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.940531335149864e-05,
+      "loss": 2.2774,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.940326975476839e-05,
+      "loss": 2.2573,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9401226158038147e-05,
+      "loss": 2.1348,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9399182561307903e-05,
+      "loss": 2.2696,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9397138964577658e-05,
+      "loss": 2.1033,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9395095367847414e-05,
+      "loss": 2.2463,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9393051771117166e-05,
+      "loss": 2.1571,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9391008174386922e-05,
+      "loss": 2.0776,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9388964577656677e-05,
+      "loss": 2.3216,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.938692098092643e-05,
+      "loss": 2.2064,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9384877384196185e-05,
+      "loss": 2.0976,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9382833787465938e-05,
+      "loss": 2.1209,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9380790190735697e-05,
+      "loss": 2.283,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9378746594005452e-05,
+      "loss": 2.1344,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9376702997275204e-05,
+      "loss": 2.2109,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.937465940054496e-05,
+      "loss": 2.1387,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9372615803814716e-05,
+      "loss": 2.301,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9370572207084468e-05,
+      "loss": 2.235,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9368528610354224e-05,
+      "loss": 2.276,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.936648501362398e-05,
+      "loss": 2.2436,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9364441416893735e-05,
+      "loss": 2.138,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.936239782016349e-05,
+      "loss": 2.2157,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9360354223433243e-05,
+      "loss": 2.2541,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9358310626703e-05,
+      "loss": 2.1956,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.935626702997275e-05,
+      "loss": 2.235,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9354223433242506e-05,
+      "loss": 2.1211,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9352179836512262e-05,
+      "loss": 2.1918,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9350136239782018e-05,
+      "loss": 2.0978,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9348092643051773e-05,
+      "loss": 2.2577,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9346049046321525e-05,
+      "loss": 2.1815,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.934400544959128e-05,
+      "loss": 2.1302,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9341961852861037e-05,
+      "loss": 2.2619,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.933991825613079e-05,
+      "loss": 2.1998,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9337874659400545e-05,
+      "loss": 2.1895,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9335831062670304e-05,
+      "loss": 2.1204,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9333787465940056e-05,
+      "loss": 2.3214,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.933174386920981e-05,
+      "loss": 2.3871,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9329700272479564e-05,
+      "loss": 2.1756,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.932765667574932e-05,
+      "loss": 2.2367,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9325613079019075e-05,
+      "loss": 2.1955,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9323569482288827e-05,
+      "loss": 2.0981,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9321525885558586e-05,
+      "loss": 2.1326,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.931948228882834e-05,
+      "loss": 2.2726,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9317438692098094e-05,
+      "loss": 2.1551,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.931539509536785e-05,
+      "loss": 2.1143,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9313351498637602e-05,
+      "loss": 2.2829,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9311307901907358e-05,
+      "loss": 2.1441,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.930926430517711e-05,
+      "loss": 2.1856,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9307220708446865e-05,
+      "loss": 2.24,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9305177111716624e-05,
+      "loss": 2.1284,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9303133514986377e-05,
+      "loss": 2.0891,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9301089918256132e-05,
+      "loss": 2.1058,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9299046321525888e-05,
+      "loss": 2.1995,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.929700272479564e-05,
+      "loss": 2.1243,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9294959128065396e-05,
+      "loss": 2.1366,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9292915531335148e-05,
+      "loss": 2.2043,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9290871934604907e-05,
+      "loss": 2.1231,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9288828337874663e-05,
+      "loss": 2.1791,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9286784741144415e-05,
+      "loss": 2.133,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.928474114441417e-05,
+      "loss": 2.0653,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9282697547683923e-05,
+      "loss": 2.2685,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.928065395095368e-05,
+      "loss": 2.29,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9278610354223434e-05,
+      "loss": 2.1199,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9276566757493186e-05,
+      "loss": 2.2682,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9274523160762945e-05,
+      "loss": 2.2187,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9272479564032698e-05,
+      "loss": 2.1355,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9270435967302453e-05,
+      "loss": 2.1748,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.926839237057221e-05,
+      "loss": 2.1981,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.926634877384196e-05,
+      "loss": 2.2377,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9264305177111717e-05,
+      "loss": 2.0802,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9262261580381472e-05,
+      "loss": 2.0072,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9260217983651228e-05,
+      "loss": 2.2269,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9258174386920984e-05,
+      "loss": 2.2811,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9256130790190736e-05,
+      "loss": 2.2222,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.925408719346049e-05,
+      "loss": 1.9857,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9252043596730247e-05,
+      "loss": 2.1337,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.925e-05,
+      "loss": 2.1588,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9247956403269755e-05,
+      "loss": 2.2817,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9245912806539507e-05,
+      "loss": 2.1523,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9243869209809266e-05,
+      "loss": 2.1486,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9241825613079022e-05,
+      "loss": 2.1363,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9239782016348774e-05,
+      "loss": 2.1211,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.923773841961853e-05,
+      "loss": 2.1066,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9235694822888282e-05,
+      "loss": 2.2123,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9233651226158038e-05,
+      "loss": 2.1783,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9231607629427793e-05,
+      "loss": 2.0734,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.922956403269755e-05,
+      "loss": 2.0662,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9227520435967305e-05,
+      "loss": 2.1487,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.922547683923706e-05,
+      "loss": 2.2447,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9223433242506813e-05,
+      "loss": 2.1261,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9221389645776568e-05,
+      "loss": 2.2266,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.921934604904632e-05,
+      "loss": 2.137,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9217302452316076e-05,
+      "loss": 2.2017,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.921525885558583e-05,
+      "loss": 2.3049,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9213215258855587e-05,
+      "loss": 2.304,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9211171662125343e-05,
+      "loss": 2.2432,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9209128065395095e-05,
+      "loss": 2.1802,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.920708446866485e-05,
+      "loss": 2.0916,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9205040871934606e-05,
+      "loss": 2.1687,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.920299727520436e-05,
+      "loss": 2.2279,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9200953678474114e-05,
+      "loss": 2.2406,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.919891008174387e-05,
+      "loss": 2.1819,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9196866485013626e-05,
+      "loss": 2.2131,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.919482288828338e-05,
+      "loss": 2.108,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9192779291553133e-05,
+      "loss": 2.0236,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.919073569482289e-05,
+      "loss": 2.2204,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9188692098092645e-05,
+      "loss": 2.053,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9186648501362397e-05,
+      "loss": 2.0777,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9184604904632153e-05,
+      "loss": 2.292,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9182561307901908e-05,
+      "loss": 2.0786,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9180517711171664e-05,
+      "loss": 2.1407,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.917847411444142e-05,
+      "loss": 2.1652,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9176430517711172e-05,
+      "loss": 2.2019,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9174386920980927e-05,
+      "loss": 2.1764,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.917234332425068e-05,
+      "loss": 2.1496,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9170299727520435e-05,
+      "loss": 2.1957,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9168256130790194e-05,
+      "loss": 2.2293,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9166212534059947e-05,
+      "loss": 2.1563,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9164168937329702e-05,
+      "loss": 2.1739,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9162125340599454e-05,
+      "loss": 2.065,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.916008174386921e-05,
+      "loss": 2.2057,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9158038147138966e-05,
+      "loss": 2.3068,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9155994550408718e-05,
+      "loss": 2.1414,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9153950953678477e-05,
+      "loss": 2.1581,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.915190735694823e-05,
+      "loss": 2.1245,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9149863760217985e-05,
+      "loss": 2.2266,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.914782016348774e-05,
+      "loss": 2.184,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9145776566757493e-05,
+      "loss": 2.0066,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.914373297002725e-05,
+      "loss": 2.1845,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9141689373297004e-05,
+      "loss": 2.1962,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9139645776566756e-05,
+      "loss": 2.2877,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9137602179836515e-05,
+      "loss": 2.2505,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9135558583106267e-05,
+      "loss": 2.4288,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9133514986376023e-05,
+      "loss": 2.1,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.913147138964578e-05,
+      "loss": 2.1338,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.912942779291553e-05,
+      "loss": 2.0953,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9127384196185287e-05,
+      "loss": 2.093,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.912534059945504e-05,
+      "loss": 2.0987,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9123297002724798e-05,
+      "loss": 2.2341,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9121253405994554e-05,
+      "loss": 2.0651,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9119209809264306e-05,
+      "loss": 2.2126,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.911716621253406e-05,
+      "loss": 2.1409,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9115122615803814e-05,
+      "loss": 2.1445,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.911307901907357e-05,
+      "loss": 2.119,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9111035422343325e-05,
+      "loss": 2.2083,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9108991825613077e-05,
+      "loss": 2.1359,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9106948228882836e-05,
+      "loss": 2.0471,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9104904632152592e-05,
+      "loss": 2.1169,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9102861035422344e-05,
+      "loss": 2.1817,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.91008174386921e-05,
+      "loss": 2.1208,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9098773841961852e-05,
+      "loss": 2.1945,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9096730245231608e-05,
+      "loss": 2.2511,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9094686648501363e-05,
+      "loss": 2.1576,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.909264305177112e-05,
+      "loss": 2.1802,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9090599455040874e-05,
+      "loss": 2.1543,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9088555858310627e-05,
+      "loss": 2.1088,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9086512261580382e-05,
+      "loss": 2.1978,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9084468664850138e-05,
+      "loss": 2.0814,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.908242506811989e-05,
+      "loss": 2.2989,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9080381471389646e-05,
+      "loss": 2.1781,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9078337874659398e-05,
+      "loss": 2.0672,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9076294277929157e-05,
+      "loss": 2.1865,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9074250681198913e-05,
+      "loss": 2.2305,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9072207084468665e-05,
+      "loss": 2.3189,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.907016348773842e-05,
+      "loss": 2.0923,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9068119891008176e-05,
+      "loss": 2.0924,
+      "step": 957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.906607629427793e-05,
+      "loss": 2.1646,
+      "step": 958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9064032697547684e-05,
+      "loss": 2.0978,
+      "step": 959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.906198910081744e-05,
+      "loss": 2.1462,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9059945504087195e-05,
+      "loss": 2.0645,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.905790190735695e-05,
+      "loss": 2.2537,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9055858310626703e-05,
+      "loss": 2.3624,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.905381471389646e-05,
+      "loss": 2.1492,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.905177111716621e-05,
+      "loss": 2.1467,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9049727520435967e-05,
+      "loss": 2.2528,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9047683923705722e-05,
+      "loss": 1.9763,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9045640326975478e-05,
+      "loss": 2.0146,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9043596730245234e-05,
+      "loss": 2.1393,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9041553133514986e-05,
+      "loss": 2.1411,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.903950953678474e-05,
+      "loss": 2.1563,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9037465940054497e-05,
+      "loss": 2.1585,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.903542234332425e-05,
+      "loss": 2.1299,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9033378746594005e-05,
+      "loss": 2.0934,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9031335149863764e-05,
+      "loss": 2.1353,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9029291553133516e-05,
+      "loss": 2.0733,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9027247956403272e-05,
+      "loss": 2.0818,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9025204359673024e-05,
+      "loss": 2.1462,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.902316076294278e-05,
+      "loss": 2.1365,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9021117166212535e-05,
+      "loss": 2.0998,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9019073569482288e-05,
+      "loss": 2.1242,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9017029972752043e-05,
+      "loss": 2.1666,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.90149863760218e-05,
+      "loss": 2.0299,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9012942779291555e-05,
+      "loss": 2.1146,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.901089918256131e-05,
+      "loss": 2.081,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9008855585831062e-05,
+      "loss": 2.1564,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9006811989100818e-05,
+      "loss": 2.0854,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.900476839237057e-05,
+      "loss": 2.0577,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9002724795640326e-05,
+      "loss": 2.0693,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9000681198910085e-05,
+      "loss": 2.0763,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8998637602179837e-05,
+      "loss": 2.0598,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8996594005449593e-05,
+      "loss": 2.1548,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.899455040871935e-05,
+      "loss": 2.182,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.89925068119891e-05,
+      "loss": 2.1207,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8990463215258856e-05,
+      "loss": 2.1741,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.898841961852861e-05,
+      "loss": 2.0478,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8986376021798364e-05,
+      "loss": 2.0311,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8984332425068123e-05,
+      "loss": 2.0589,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8982288828337876e-05,
+      "loss": 2.1818,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.898024523160763e-05,
+      "loss": 2.0426,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.8841488361358643,
+      "eval_runtime": 723.6028,
+      "eval_samples_per_second": 5.735,
+      "eval_steps_per_second": 1.434,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8978201634877383e-05,
+      "loss": 2.1217,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.897615803814714e-05,
+      "loss": 2.2176,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8974114441416895e-05,
+      "loss": 1.982,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8972070844686647e-05,
+      "loss": 2.2898,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8970027247956406e-05,
+      "loss": 2.0935,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8967983651226158e-05,
+      "loss": 2.1029,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8965940054495914e-05,
+      "loss": 2.0967,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.896389645776567e-05,
+      "loss": 2.1395,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8961852861035422e-05,
+      "loss": 2.1916,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8959809264305177e-05,
+      "loss": 2.1169,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8957765667574933e-05,
+      "loss": 2.0299,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.895572207084469e-05,
+      "loss": 2.1039,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8953678474114444e-05,
+      "loss": 2.1993,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8951634877384196e-05,
+      "loss": 2.0438,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8949591280653952e-05,
+      "loss": 2.0622,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8947547683923708e-05,
+      "loss": 2.1165,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.894550408719346e-05,
+      "loss": 2.0655,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8943460490463216e-05,
+      "loss": 2.1138,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8941416893732968e-05,
+      "loss": 1.9803,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8939373297002727e-05,
+      "loss": 2.0118,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8937329700272483e-05,
+      "loss": 2.1344,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8935286103542235e-05,
+      "loss": 2.0848,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.893324250681199e-05,
+      "loss": 2.0962,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8931198910081743e-05,
+      "loss": 2.0252,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8929155313351498e-05,
+      "loss": 2.0949,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8927111716621254e-05,
+      "loss": 2.0595,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.892506811989101e-05,
+      "loss": 1.7636,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8923024523160765e-05,
+      "loss": 2.1131,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.892098092643052e-05,
+      "loss": 2.1672,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8918937329700273e-05,
+      "loss": 2.0983,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.891689373297003e-05,
+      "loss": 2.1643,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.891485013623978e-05,
+      "loss": 2.0449,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8912806539509537e-05,
+      "loss": 2.2065,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8910762942779292e-05,
+      "loss": 2.0943,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8908719346049048e-05,
+      "loss": 2.2054,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8906675749318803e-05,
+      "loss": 1.9968,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8904632152588556e-05,
+      "loss": 2.0615,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.890258855585831e-05,
+      "loss": 1.9347,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8900544959128067e-05,
+      "loss": 2.1957,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.889850136239782e-05,
+      "loss": 2.1301,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8896457765667575e-05,
+      "loss": 2.1605,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.889441416893733e-05,
+      "loss": 2.0913,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8892370572207086e-05,
+      "loss": 1.9808,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8890326975476842e-05,
+      "loss": 2.0008,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8888283378746594e-05,
+      "loss": 2.0502,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.888623978201635e-05,
+      "loss": 2.1785,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8884196185286105e-05,
+      "loss": 2.0267,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8882152588555857e-05,
+      "loss": 2.1141,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8880108991825613e-05,
+      "loss": 2.1603,
+      "step": 1049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.887806539509537e-05,
+      "loss": 2.1968,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8876021798365124e-05,
+      "loss": 1.975,
+      "step": 1051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.887397820163488e-05,
+      "loss": 2.0832,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8871934604904632e-05,
+      "loss": 2.0483,
+      "step": 1053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8869891008174388e-05,
+      "loss": 2.0269,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.886784741144414e-05,
+      "loss": 2.1593,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8865803814713896e-05,
+      "loss": 2.1202,
+      "step": 1056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8863760217983655e-05,
+      "loss": 2.1137,
+      "step": 1057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8861716621253407e-05,
+      "loss": 2.1401,
+      "step": 1058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8859673024523163e-05,
+      "loss": 2.3027,
+      "step": 1059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8857629427792915e-05,
+      "loss": 2.0619,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.885558583106267e-05,
+      "loss": 2.1357,
+      "step": 1061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8853542234332426e-05,
+      "loss": 2.0256,
+      "step": 1062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.885149863760218e-05,
+      "loss": 2.0156,
+      "step": 1063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8849455040871934e-05,
+      "loss": 2.0119,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8847411444141693e-05,
+      "loss": 2.075,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8845367847411445e-05,
+      "loss": 2.1243,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.88433242506812e-05,
+      "loss": 2.158,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8841280653950953e-05,
+      "loss": 2.1637,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.883923705722071e-05,
+      "loss": 2.031,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8837193460490464e-05,
+      "loss": 1.9607,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8835149863760217e-05,
+      "loss": 2.2258,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8833106267029976e-05,
+      "loss": 2.1091,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8831062670299728e-05,
+      "loss": 2.2534,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8829019073569484e-05,
+      "loss": 2.0797,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.882697547683924e-05,
+      "loss": 2.0128,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.882493188010899e-05,
+      "loss": 2.0428,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8822888283378747e-05,
+      "loss": 2.0595,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.88208446866485e-05,
+      "loss": 2.0816,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8818801089918255e-05,
+      "loss": 2.0431,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8816757493188014e-05,
+      "loss": 2.1595,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8814713896457766e-05,
+      "loss": 2.1377,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8812670299727522e-05,
+      "loss": 1.9792,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8810626702997278e-05,
+      "loss": 2.0934,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.880858310626703e-05,
+      "loss": 2.0259,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8806539509536785e-05,
+      "loss": 2.1009,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8804495912806538e-05,
+      "loss": 2.0275,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8802452316076297e-05,
+      "loss": 2.0012,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8800408719346052e-05,
+      "loss": 2.1138,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8798365122615805e-05,
+      "loss": 2.1026,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.879632152588556e-05,
+      "loss": 2.087,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8794277929155312e-05,
+      "loss": 2.1828,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8792234332425068e-05,
+      "loss": 2.1751,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8790190735694824e-05,
+      "loss": 2.1362,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8788147138964576e-05,
+      "loss": 2.0958,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8786103542234335e-05,
+      "loss": 2.0255,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8784059945504087e-05,
+      "loss": 2.1172,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8782016348773843e-05,
+      "loss": 2.1133,
+      "step": 1097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.87799727520436e-05,
+      "loss": 2.3357,
+      "step": 1098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.877792915531335e-05,
+      "loss": 2.0244,
+      "step": 1099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8775885558583106e-05,
+      "loss": 2.1841,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8773841961852862e-05,
+      "loss": 2.075,
+      "step": 1101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8771798365122618e-05,
+      "loss": 2.0283,
+      "step": 1102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8769754768392373e-05,
+      "loss": 2.0335,
+      "step": 1103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8767711171662125e-05,
+      "loss": 2.0819,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.876566757493188e-05,
+      "loss": 2.0125,
+      "step": 1105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8763623978201637e-05,
+      "loss": 2.113,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.876158038147139e-05,
+      "loss": 2.1268,
+      "step": 1107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8759536784741145e-05,
+      "loss": 1.9827,
+      "step": 1108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.87574931880109e-05,
+      "loss": 2.1545,
+      "step": 1109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8755449591280656e-05,
+      "loss": 2.0518,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.875340599455041e-05,
+      "loss": 2.0754,
+      "step": 1111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8751362397820164e-05,
+      "loss": 2.025,
+      "step": 1112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.874931880108992e-05,
+      "loss": 1.9802,
+      "step": 1113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.874727520435967e-05,
+      "loss": 2.056,
+      "step": 1114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8745231607629427e-05,
+      "loss": 2.0438,
+      "step": 1115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8743188010899183e-05,
+      "loss": 2.1055,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.874114441416894e-05,
+      "loss": 2.0597,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8739100817438694e-05,
+      "loss": 1.9877,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8737057220708446e-05,
+      "loss": 2.0769,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8735013623978202e-05,
+      "loss": 1.989,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8732970027247958e-05,
+      "loss": 2.0867,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.873092643051771e-05,
+      "loss": 1.9485,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8728882833787466e-05,
+      "loss": 2.0059,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8726839237057225e-05,
+      "loss": 2.0502,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8724795640326977e-05,
+      "loss": 1.9426,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8722752043596732e-05,
+      "loss": 2.0807,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8720708446866485e-05,
+      "loss": 2.1255,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.871866485013624e-05,
+      "loss": 2.0505,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8716621253405996e-05,
+      "loss": 2.1498,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8714577656675748e-05,
+      "loss": 2.0854,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8712534059945504e-05,
+      "loss": 2.0545,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.871049046321526e-05,
+      "loss": 2.0726,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8708446866485015e-05,
+      "loss": 2.1048,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.870640326975477e-05,
+      "loss": 2.079,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8704359673024523e-05,
+      "loss": 2.058,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.870231607629428e-05,
+      "loss": 2.1202,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.870027247956403e-05,
+      "loss": 2.0534,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8698228882833787e-05,
+      "loss": 2.0015,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8696185286103546e-05,
+      "loss": 2.0549,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8694141689373298e-05,
+      "loss": 2.083,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8692098092643053e-05,
+      "loss": 2.1094,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.869005449591281e-05,
+      "loss": 2.0465,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.868801089918256e-05,
+      "loss": 2.0043,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8685967302452317e-05,
+      "loss": 2.1632,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.868392370572207e-05,
+      "loss": 2.1125,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8681880108991825e-05,
+      "loss": 2.0903,
+      "step": 1146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8679836512261584e-05,
+      "loss": 2.0642,
+      "step": 1147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8677792915531336e-05,
+      "loss": 2.0893,
+      "step": 1148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8675749318801092e-05,
+      "loss": 2.1718,
+      "step": 1149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8673705722070844e-05,
+      "loss": 1.9276,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.86716621253406e-05,
+      "loss": 1.9387,
+      "step": 1151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8669618528610355e-05,
+      "loss": 1.925,
+      "step": 1152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8667574931880107e-05,
+      "loss": 1.9449,
+      "step": 1153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8665531335149866e-05,
+      "loss": 1.9053,
+      "step": 1154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.866348773841962e-05,
+      "loss": 2.0746,
+      "step": 1155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8661444141689374e-05,
+      "loss": 1.9758,
+      "step": 1156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.865940054495913e-05,
+      "loss": 2.2293,
+      "step": 1157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8657356948228882e-05,
+      "loss": 2.1213,
+      "step": 1158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8655313351498638e-05,
+      "loss": 2.0115,
+      "step": 1159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8653269754768393e-05,
+      "loss": 1.9814,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8651226158038146e-05,
+      "loss": 2.006,
+      "step": 1161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8649182561307905e-05,
+      "loss": 2.1013,
+      "step": 1162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8647138964577657e-05,
+      "loss": 2.0791,
+      "step": 1163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8645095367847413e-05,
+      "loss": 1.9204,
+      "step": 1164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8643051771117168e-05,
+      "loss": 1.9788,
+      "step": 1165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.864100817438692e-05,
+      "loss": 2.0619,
+      "step": 1166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8638964577656676e-05,
+      "loss": 1.9947,
+      "step": 1167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.863692098092643e-05,
+      "loss": 2.0422,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8634877384196187e-05,
+      "loss": 2.0853,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8632833787465943e-05,
+      "loss": 2.3041,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8630790190735695e-05,
+      "loss": 1.9753,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.862874659400545e-05,
+      "loss": 1.9503,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8626702997275203e-05,
+      "loss": 2.1707,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.862465940054496e-05,
+      "loss": 2.152,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8622615803814714e-05,
+      "loss": 2.0868,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8620572207084467e-05,
+      "loss": 2.1644,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8618528610354226e-05,
+      "loss": 2.0984,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.861648501362398e-05,
+      "loss": 1.995,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8614441416893734e-05,
+      "loss": 1.9546,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.861239782016349e-05,
+      "loss": 2.0081,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.861035422343324e-05,
+      "loss": 2.1274,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8608310626702997e-05,
+      "loss": 2.0774,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8606267029972753e-05,
+      "loss": 1.9856,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.860422343324251e-05,
+      "loss": 2.157,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8602179836512264e-05,
+      "loss": 1.9722,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8600136239782016e-05,
+      "loss": 2.0919,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8598092643051772e-05,
+      "loss": 2.0251,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8596049046321527e-05,
+      "loss": 1.8813,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.859400544959128e-05,
+      "loss": 1.9438,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8591961852861035e-05,
+      "loss": 1.9778,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.858991825613079e-05,
+      "loss": 2.0334,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8587874659400547e-05,
+      "loss": 2.1092,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8585831062670302e-05,
+      "loss": 2.0554,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8583787465940055e-05,
+      "loss": 2.0991,
+      "step": 1194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.858174386920981e-05,
+      "loss": 2.0729,
+      "step": 1195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8579700272479566e-05,
+      "loss": 1.9219,
+      "step": 1196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8577656675749318e-05,
+      "loss": 2.1256,
+      "step": 1197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8575613079019074e-05,
+      "loss": 2.0284,
+      "step": 1198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.857356948228883e-05,
+      "loss": 1.9397,
+      "step": 1199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8571525885558585e-05,
+      "loss": 2.2779,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.856948228882834e-05,
+      "loss": 1.8958,
+      "step": 1201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8567438692098093e-05,
+      "loss": 1.8909,
+      "step": 1202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.856539509536785e-05,
+      "loss": 2.0035,
+      "step": 1203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.85633514986376e-05,
+      "loss": 2.1069,
+      "step": 1204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8561307901907356e-05,
+      "loss": 2.2476,
+      "step": 1205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8559264305177115e-05,
+      "loss": 1.9751,
+      "step": 1206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8557220708446868e-05,
+      "loss": 2.1143,
+      "step": 1207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8555177111716623e-05,
+      "loss": 2.0242,
+      "step": 1208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8553133514986375e-05,
+      "loss": 2.0721,
+      "step": 1209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.855108991825613e-05,
+      "loss": 2.0784,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8549046321525887e-05,
+      "loss": 2.173,
+      "step": 1211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.854700272479564e-05,
+      "loss": 2.1584,
+      "step": 1212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8544959128065395e-05,
+      "loss": 1.907,
+      "step": 1213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8542915531335154e-05,
+      "loss": 1.9943,
+      "step": 1214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8540871934604906e-05,
+      "loss": 2.0769,
+      "step": 1215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.853882833787466e-05,
+      "loss": 2.1291,
+      "step": 1216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8536784741144414e-05,
+      "loss": 2.2674,
+      "step": 1217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.853474114441417e-05,
+      "loss": 1.9787,
+      "step": 1218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8532697547683925e-05,
+      "loss": 2.2067,
+      "step": 1219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8530653950953677e-05,
+      "loss": 2.0721,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8528610354223436e-05,
+      "loss": 1.9209,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.852656675749319e-05,
+      "loss": 2.0706,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8524523160762944e-05,
+      "loss": 2.0295,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.85224795640327e-05,
+      "loss": 2.0767,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8520435967302452e-05,
+      "loss": 2.1795,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8518392370572208e-05,
+      "loss": 2.0226,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.851634877384196e-05,
+      "loss": 2.0727,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8514305177111716e-05,
+      "loss": 2.03,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8512261580381475e-05,
+      "loss": 1.9812,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8510217983651227e-05,
+      "loss": 2.1117,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8508174386920982e-05,
+      "loss": 1.8877,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8506130790190738e-05,
+      "loss": 2.0262,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.850408719346049e-05,
+      "loss": 2.1465,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8502043596730246e-05,
+      "loss": 2.0005,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 2.0188,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8497956403269757e-05,
+      "loss": 2.0338,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8495912806539513e-05,
+      "loss": 2.0647,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8493869209809265e-05,
+      "loss": 2.1052,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.849182561307902e-05,
+      "loss": 2.0632,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8489782016348773e-05,
+      "loss": 1.9777,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.848773841961853e-05,
+      "loss": 2.0048,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8485694822888284e-05,
+      "loss": 1.9885,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8483651226158036e-05,
+      "loss": 1.9291,
+      "step": 1243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8481607629427795e-05,
+      "loss": 2.0725,
+      "step": 1244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8479564032697548e-05,
+      "loss": 2.023,
+      "step": 1245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8477520435967303e-05,
+      "loss": 2.1349,
+      "step": 1246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.847547683923706e-05,
+      "loss": 1.9621,
+      "step": 1247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.847343324250681e-05,
+      "loss": 2.0892,
+      "step": 1248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8471389645776567e-05,
+      "loss": 2.0333,
+      "step": 1249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8469346049046323e-05,
+      "loss": 2.1451,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8467302452316078e-05,
+      "loss": 2.0219,
+      "step": 1251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8465258855585834e-05,
+      "loss": 2.0426,
+      "step": 1252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8463215258855586e-05,
+      "loss": 2.1088,
+      "step": 1253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.846117166212534e-05,
+      "loss": 1.9643,
+      "step": 1254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8459128065395097e-05,
+      "loss": 2.0354,
+      "step": 1255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.845708446866485e-05,
+      "loss": 2.1895,
+      "step": 1256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8455040871934605e-05,
+      "loss": 1.9441,
+      "step": 1257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8452997275204357e-05,
+      "loss": 2.0857,
+      "step": 1258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8450953678474116e-05,
+      "loss": 2.1392,
+      "step": 1259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8448910081743872e-05,
+      "loss": 2.0594,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8446866485013624e-05,
+      "loss": 2.0045,
+      "step": 1261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.844482288828338e-05,
+      "loss": 1.9632,
+      "step": 1262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8442779291553132e-05,
+      "loss": 2.0308,
+      "step": 1263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8440735694822888e-05,
+      "loss": 2.042,
+      "step": 1264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8438692098092643e-05,
+      "loss": 2.0551,
+      "step": 1265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.84366485013624e-05,
+      "loss": 1.9958,
+      "step": 1266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8434604904632155e-05,
+      "loss": 1.9965,
+      "step": 1267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.843256130790191e-05,
+      "loss": 2.0472,
+      "step": 1268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8430517711171663e-05,
+      "loss": 2.0511,
+      "step": 1269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8428474114441418e-05,
+      "loss": 2.0717,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.842643051771117e-05,
+      "loss": 2.1638,
+      "step": 1271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8424386920980926e-05,
+      "loss": 2.0922,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8422343324250682e-05,
+      "loss": 2.0948,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8420299727520437e-05,
+      "loss": 2.1072,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8418256130790193e-05,
+      "loss": 2.2364,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8416212534059945e-05,
+      "loss": 1.9418,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.84141689373297e-05,
+      "loss": 1.9482,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8412125340599457e-05,
+      "loss": 2.0112,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.841008174386921e-05,
+      "loss": 2.0901,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8408038147138964e-05,
+      "loss": 2.0215,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.840599455040872e-05,
+      "loss": 2.0013,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8403950953678476e-05,
+      "loss": 2.0446,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.840190735694823e-05,
+      "loss": 2.0048,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8399863760217984e-05,
+      "loss": 2.1482,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.839782016348774e-05,
+      "loss": 1.9353,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8395776566757495e-05,
+      "loss": 2.0952,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8393732970027247e-05,
+      "loss": 2.0845,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8391689373297006e-05,
+      "loss": 2.0138,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8389645776566758e-05,
+      "loss": 1.8655,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8387602179836514e-05,
+      "loss": 2.0634,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.838555858310627e-05,
+      "loss": 1.9325,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8383514986376022e-05,
+      "loss": 2.0194,
+      "step": 1292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8381471389645777e-05,
+      "loss": 2.1384,
+      "step": 1293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.837942779291553e-05,
+      "loss": 2.0249,
+      "step": 1294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8377384196185285e-05,
+      "loss": 1.9726,
+      "step": 1295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8375340599455044e-05,
+      "loss": 1.9177,
+      "step": 1296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8373297002724797e-05,
+      "loss": 2.0126,
+      "step": 1297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8371253405994552e-05,
+      "loss": 2.0882,
+      "step": 1298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8369209809264304e-05,
+      "loss": 1.9821,
+      "step": 1299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.836716621253406e-05,
+      "loss": 1.9855,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8365122615803816e-05,
+      "loss": 2.0114,
+      "step": 1301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8363079019073568e-05,
+      "loss": 2.1174,
+      "step": 1302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8361035422343327e-05,
+      "loss": 2.0753,
+      "step": 1303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.835899182561308e-05,
+      "loss": 2.0657,
+      "step": 1304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8356948228882835e-05,
+      "loss": 2.0554,
+      "step": 1305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.835490463215259e-05,
+      "loss": 1.9825,
+      "step": 1306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8352861035422343e-05,
+      "loss": 2.0166,
+      "step": 1307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.83508174386921e-05,
+      "loss": 1.9753,
+      "step": 1308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8348773841961854e-05,
+      "loss": 2.0128,
+      "step": 1309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8346730245231606e-05,
+      "loss": 2.0695,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8344686648501365e-05,
+      "loss": 2.0989,
+      "step": 1311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8342643051771118e-05,
+      "loss": 2.0314,
+      "step": 1312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8340599455040873e-05,
+      "loss": 2.0491,
+      "step": 1313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.833855585831063e-05,
+      "loss": 2.0255,
+      "step": 1314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.833651226158038e-05,
+      "loss": 2.0702,
+      "step": 1315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8334468664850137e-05,
+      "loss": 2.091,
+      "step": 1316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.833242506811989e-05,
+      "loss": 1.9721,
+      "step": 1317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8330381471389648e-05,
+      "loss": 2.0011,
+      "step": 1318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8328337874659404e-05,
+      "loss": 2.1116,
+      "step": 1319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8326294277929156e-05,
+      "loss": 2.0061,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.832425068119891e-05,
+      "loss": 2.1219,
+      "step": 1321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8322207084468664e-05,
+      "loss": 2.0204,
+      "step": 1322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.832016348773842e-05,
+      "loss": 1.9047,
+      "step": 1323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8318119891008175e-05,
+      "loss": 2.049,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8316076294277927e-05,
+      "loss": 2.0477,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8314032697547686e-05,
+      "loss": 1.9868,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8311989100817442e-05,
+      "loss": 1.9874,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8309945504087194e-05,
+      "loss": 2.0452,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.830790190735695e-05,
+      "loss": 1.9005,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8305858310626702e-05,
+      "loss": 1.9904,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8303814713896458e-05,
+      "loss": 1.901,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8301771117166213e-05,
+      "loss": 1.9495,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.829972752043597e-05,
+      "loss": 2.0596,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8297683923705725e-05,
+      "loss": 1.9597,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8295640326975477e-05,
+      "loss": 2.0721,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8293596730245232e-05,
+      "loss": 2.0253,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8291553133514988e-05,
+      "loss": 2.0391,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.828950953678474e-05,
+      "loss": 1.9586,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8287465940054496e-05,
+      "loss": 2.0071,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8285422343324248e-05,
+      "loss": 1.9542,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8283378746594007e-05,
+      "loss": 1.9817,
+      "step": 1341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8281335149863763e-05,
+      "loss": 2.0102,
+      "step": 1342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8279291553133515e-05,
+      "loss": 2.217,
+      "step": 1343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.827724795640327e-05,
+      "loss": 2.247,
+      "step": 1344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8275204359673026e-05,
+      "loss": 2.0337,
+      "step": 1345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.827316076294278e-05,
+      "loss": 2.0476,
+      "step": 1346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8271117166212534e-05,
+      "loss": 2.0624,
+      "step": 1347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.826907356948229e-05,
+      "loss": 1.9927,
+      "step": 1348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8267029972752045e-05,
+      "loss": 2.085,
+      "step": 1349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.82649863760218e-05,
+      "loss": 1.942,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8262942779291553e-05,
+      "loss": 2.0407,
+      "step": 1351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.826089918256131e-05,
+      "loss": 1.9392,
+      "step": 1352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.825885558583106e-05,
+      "loss": 1.9525,
+      "step": 1353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8256811989100817e-05,
+      "loss": 2.0107,
+      "step": 1354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8254768392370572e-05,
+      "loss": 2.1014,
+      "step": 1355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8252724795640328e-05,
+      "loss": 2.0825,
+      "step": 1356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8250681198910084e-05,
+      "loss": 1.9548,
+      "step": 1357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8248637602179836e-05,
+      "loss": 2.0102,
+      "step": 1358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.824659400544959e-05,
+      "loss": 2.2352,
+      "step": 1359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8244550408719347e-05,
+      "loss": 2.0705,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.82425068119891e-05,
+      "loss": 2.1815,
+      "step": 1361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8240463215258855e-05,
+      "loss": 2.0303,
+      "step": 1362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8238419618528614e-05,
+      "loss": 2.17,
+      "step": 1363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8236376021798366e-05,
+      "loss": 2.0178,
+      "step": 1364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8234332425068122e-05,
+      "loss": 1.9205,
+      "step": 1365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8232288828337874e-05,
+      "loss": 1.9559,
+      "step": 1366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.823024523160763e-05,
+      "loss": 2.1937,
+      "step": 1367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8228201634877386e-05,
+      "loss": 2.05,
+      "step": 1368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8226158038147138e-05,
+      "loss": 1.7555,
+      "step": 1369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8224114441416893e-05,
+      "loss": 1.9866,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.822207084468665e-05,
+      "loss": 2.0122,
+      "step": 1371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8220027247956405e-05,
+      "loss": 1.9416,
+      "step": 1372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.821798365122616e-05,
+      "loss": 2.0903,
+      "step": 1373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8215940054495913e-05,
+      "loss": 1.95,
+      "step": 1374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8213896457765668e-05,
+      "loss": 2.0757,
+      "step": 1375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.821185286103542e-05,
+      "loss": 2.0425,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8209809264305176e-05,
+      "loss": 2.2424,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8207765667574935e-05,
+      "loss": 2.0129,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8205722070844687e-05,
+      "loss": 2.0885,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8203678474114443e-05,
+      "loss": 2.0119,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.82016348773842e-05,
+      "loss": 2.0486,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.819959128065395e-05,
+      "loss": 2.017,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8197547683923706e-05,
+      "loss": 1.961,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.819550408719346e-05,
+      "loss": 1.9294,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8193460490463218e-05,
+      "loss": 1.9872,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8191416893732973e-05,
+      "loss": 2.1444,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8189373297002726e-05,
+      "loss": 2.1076,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.818732970027248e-05,
+      "loss": 1.9262,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8185286103542233e-05,
+      "loss": 2.047,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.818324250681199e-05,
+      "loss": 2.0565,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8181198910081745e-05,
+      "loss": 2.1998,
+      "step": 1391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8179155313351497e-05,
+      "loss": 1.9191,
+      "step": 1392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8177111716621256e-05,
+      "loss": 1.9666,
+      "step": 1393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8175068119891008e-05,
+      "loss": 2.0641,
+      "step": 1394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8173024523160764e-05,
+      "loss": 2.1095,
+      "step": 1395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.817098092643052e-05,
+      "loss": 2.0855,
+      "step": 1396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8168937329700272e-05,
+      "loss": 1.9961,
+      "step": 1397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8166893732970027e-05,
+      "loss": 2.0155,
+      "step": 1398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8164850136239783e-05,
+      "loss": 2.1024,
+      "step": 1399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.816280653950954e-05,
+      "loss": 2.0522,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8160762942779294e-05,
+      "loss": 1.8714,
+      "step": 1401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8158719346049047e-05,
+      "loss": 2.0283,
+      "step": 1402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8156675749318802e-05,
+      "loss": 2.1183,
+      "step": 1403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8154632152588558e-05,
+      "loss": 2.0766,
+      "step": 1404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.815258855585831e-05,
+      "loss": 1.9406,
+      "step": 1405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8150544959128066e-05,
+      "loss": 1.9816,
+      "step": 1406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8148501362397818e-05,
+      "loss": 2.129,
+      "step": 1407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8146457765667577e-05,
+      "loss": 2.0449,
+      "step": 1408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8144414168937333e-05,
+      "loss": 1.9703,
+      "step": 1409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8142370572207085e-05,
+      "loss": 1.9588,
+      "step": 1410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.814032697547684e-05,
+      "loss": 1.8364,
+      "step": 1411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8138283378746593e-05,
+      "loss": 2.0339,
+      "step": 1412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.813623978201635e-05,
+      "loss": 1.987,
+      "step": 1413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8134196185286104e-05,
+      "loss": 1.9976,
+      "step": 1414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.813215258855586e-05,
+      "loss": 2.0105,
+      "step": 1415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8130108991825615e-05,
+      "loss": 2.1046,
+      "step": 1416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.812806539509537e-05,
+      "loss": 1.9618,
+      "step": 1417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8126021798365123e-05,
+      "loss": 2.0674,
+      "step": 1418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.812397820163488e-05,
+      "loss": 1.9791,
+      "step": 1419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.812193460490463e-05,
+      "loss": 1.9459,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8119891008174387e-05,
+      "loss": 2.0655,
+      "step": 1421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8117847411444142e-05,
+      "loss": 1.9089,
+      "step": 1422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8115803814713898e-05,
+      "loss": 1.9737,
+      "step": 1423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8113760217983654e-05,
+      "loss": 2.0943,
+      "step": 1424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8111716621253406e-05,
+      "loss": 1.9502,
+      "step": 1425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.810967302452316e-05,
+      "loss": 1.9448,
+      "step": 1426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8107629427792917e-05,
+      "loss": 2.0328,
+      "step": 1427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.810558583106267e-05,
+      "loss": 1.934,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8103542234332425e-05,
+      "loss": 2.0533,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.810149863760218e-05,
+      "loss": 2.0119,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8099455040871936e-05,
+      "loss": 1.8824,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8097411444141692e-05,
+      "loss": 1.9739,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8095367847411444e-05,
+      "loss": 1.866,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.80933242506812e-05,
+      "loss": 1.8839,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8091280653950955e-05,
+      "loss": 1.9534,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8089237057220708e-05,
+      "loss": 2.015,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8087193460490463e-05,
+      "loss": 1.938,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.808514986376022e-05,
+      "loss": 1.9916,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8083106267029974e-05,
+      "loss": 1.9141,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.808106267029973e-05,
+      "loss": 2.0976,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8079019073569482e-05,
+      "loss": 2.0064,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8076975476839238e-05,
+      "loss": 2.0165,
+      "step": 1442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.807493188010899e-05,
+      "loss": 1.8537,
+      "step": 1443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8072888283378746e-05,
+      "loss": 2.1097,
+      "step": 1444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8070844686648505e-05,
+      "loss": 2.0446,
+      "step": 1445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8068801089918257e-05,
+      "loss": 1.8704,
+      "step": 1446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8066757493188013e-05,
+      "loss": 1.9725,
+      "step": 1447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8064713896457765e-05,
+      "loss": 1.9289,
+      "step": 1448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.806267029972752e-05,
+      "loss": 2.0081,
+      "step": 1449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8060626702997276e-05,
+      "loss": 2.1283,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.805858310626703e-05,
+      "loss": 1.951,
+      "step": 1451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8056539509536784e-05,
+      "loss": 1.9561,
+      "step": 1452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8054495912806543e-05,
+      "loss": 1.9237,
+      "step": 1453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8052452316076295e-05,
+      "loss": 1.9139,
+      "step": 1454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.805040871934605e-05,
+      "loss": 1.9936,
+      "step": 1455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8048365122615803e-05,
+      "loss": 2.007,
+      "step": 1456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.804632152588556e-05,
+      "loss": 1.9874,
+      "step": 1457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8044277929155315e-05,
+      "loss": 2.0343,
+      "step": 1458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8042234332425067e-05,
+      "loss": 2.0021,
+      "step": 1459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8040190735694826e-05,
+      "loss": 1.9154,
+      "step": 1460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8038147138964578e-05,
+      "loss": 2.1512,
+      "step": 1461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8036103542234334e-05,
+      "loss": 2.0672,
+      "step": 1462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.803405994550409e-05,
+      "loss": 1.9521,
+      "step": 1463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.803201634877384e-05,
+      "loss": 1.9872,
+      "step": 1464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8029972752043597e-05,
+      "loss": 1.9962,
+      "step": 1465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.802792915531335e-05,
+      "loss": 1.9854,
+      "step": 1466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.802588555858311e-05,
+      "loss": 2.0059,
+      "step": 1467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8023841961852864e-05,
+      "loss": 2.0674,
+      "step": 1468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8021798365122616e-05,
+      "loss": 1.9726,
+      "step": 1469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8019754768392372e-05,
+      "loss": 2.0607,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8017711171662128e-05,
+      "loss": 2.0303,
+      "step": 1471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.801566757493188e-05,
+      "loss": 1.9574,
+      "step": 1472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8013623978201635e-05,
+      "loss": 2.1477,
+      "step": 1473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8011580381471388e-05,
+      "loss": 1.8651,
+      "step": 1474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8009536784741147e-05,
+      "loss": 2.0136,
+      "step": 1475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8007493188010902e-05,
+      "loss": 1.9708,
+      "step": 1476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8005449591280655e-05,
+      "loss": 1.9276,
+      "step": 1477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.800340599455041e-05,
+      "loss": 1.8435,
+      "step": 1478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8001362397820162e-05,
+      "loss": 2.0862,
+      "step": 1479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7999318801089918e-05,
+      "loss": 1.951,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7997275204359674e-05,
+      "loss": 2.0574,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.799523160762943e-05,
+      "loss": 1.9646,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7993188010899185e-05,
+      "loss": 1.9634,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7991144414168937e-05,
+      "loss": 1.9778,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7989100817438693e-05,
+      "loss": 2.0179,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.798705722070845e-05,
+      "loss": 2.056,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.79850136239782e-05,
+      "loss": 1.966,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7982970027247956e-05,
+      "loss": 2.032,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7980926430517712e-05,
+      "loss": 2.1211,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7978882833787468e-05,
+      "loss": 2.1128,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7976839237057223e-05,
+      "loss": 1.9083,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7974795640326976e-05,
+      "loss": 1.9709,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.797275204359673e-05,
+      "loss": 1.8692,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7970708446866487e-05,
+      "loss": 2.0489,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.796866485013624e-05,
+      "loss": 2.0638,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7966621253405995e-05,
+      "loss": 1.9718,
+      "step": 1496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.796457765667575e-05,
+      "loss": 2.167,
+      "step": 1497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7962534059945506e-05,
+      "loss": 1.9225,
+      "step": 1498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.796049046321526e-05,
+      "loss": 1.9614,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7958446866485014e-05,
+      "loss": 1.9653,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.7807098627090454,
+      "eval_runtime": 723.2961,
+      "eval_samples_per_second": 5.738,
+      "eval_steps_per_second": 1.435,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.795640326975477e-05,
+      "loss": 2.1714,
+      "step": 1501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7954359673024522e-05,
+      "loss": 1.8559,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7952316076294277e-05,
+      "loss": 2.0495,
+      "step": 1503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7950272479564033e-05,
+      "loss": 1.8735,
+      "step": 1504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.794822888283379e-05,
+      "loss": 2.01,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7946185286103544e-05,
+      "loss": 2.1058,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7944141689373296e-05,
+      "loss": 2.0036,
+      "step": 1507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7942098092643052e-05,
+      "loss": 2.0506,
+      "step": 1508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7940054495912808e-05,
+      "loss": 2.0476,
+      "step": 1509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.793801089918256e-05,
+      "loss": 2.1181,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7935967302452316e-05,
+      "loss": 2.0837,
+      "step": 1511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7933923705722075e-05,
+      "loss": 1.8698,
+      "step": 1512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7931880108991827e-05,
+      "loss": 2.0448,
+      "step": 1513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7929836512261583e-05,
+      "loss": 2.0125,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7927792915531335e-05,
+      "loss": 2.1108,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.792574931880109e-05,
+      "loss": 1.9507,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7923705722070846e-05,
+      "loss": 1.9451,
+      "step": 1517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7921662125340598e-05,
+      "loss": 2.0215,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7919618528610354e-05,
+      "loss": 2.003,
+      "step": 1519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.791757493188011e-05,
+      "loss": 1.9873,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7915531335149865e-05,
+      "loss": 1.8443,
+      "step": 1521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.791348773841962e-05,
+      "loss": 2.2012,
+      "step": 1522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7911444141689373e-05,
+      "loss": 1.9615,
+      "step": 1523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.790940054495913e-05,
+      "loss": 1.979,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.790735694822888e-05,
+      "loss": 1.9855,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7905313351498637e-05,
+      "loss": 1.9405,
+      "step": 1526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7903269754768396e-05,
+      "loss": 1.9802,
+      "step": 1527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7901226158038148e-05,
+      "loss": 1.9384,
+      "step": 1528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7899182561307903e-05,
+      "loss": 1.9668,
+      "step": 1529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.789713896457766e-05,
+      "loss": 1.9207,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.789509536784741e-05,
+      "loss": 1.9693,
+      "step": 1531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7893051771117167e-05,
+      "loss": 1.9576,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.789100817438692e-05,
+      "loss": 1.9554,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7888964577656675e-05,
+      "loss": 2.0719,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7886920980926434e-05,
+      "loss": 1.9884,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7884877384196186e-05,
+      "loss": 2.0172,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7882833787465942e-05,
+      "loss": 2.0603,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7880790190735694e-05,
+      "loss": 1.9458,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.787874659400545e-05,
+      "loss": 1.934,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7876702997275205e-05,
+      "loss": 1.9673,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7874659400544958e-05,
+      "loss": 1.8949,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7872615803814717e-05,
+      "loss": 1.99,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.787057220708447e-05,
+      "loss": 1.9766,
+      "step": 1543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7868528610354224e-05,
+      "loss": 2.0183,
+      "step": 1544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.786648501362398e-05,
+      "loss": 1.7916,
+      "step": 1545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7864441416893732e-05,
+      "loss": 1.9361,
+      "step": 1546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7862397820163488e-05,
+      "loss": 1.8125,
+      "step": 1547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7860354223433244e-05,
+      "loss": 1.9474,
+      "step": 1548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7858310626702996e-05,
+      "loss": 2.0556,
+      "step": 1549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7856267029972755e-05,
+      "loss": 1.9199,
+      "step": 1550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7854223433242507e-05,
+      "loss": 1.9267,
+      "step": 1551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7852179836512263e-05,
+      "loss": 1.7936,
+      "step": 1552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.785013623978202e-05,
+      "loss": 2.1166,
+      "step": 1553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.784809264305177e-05,
+      "loss": 1.9569,
+      "step": 1554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7846049046321526e-05,
+      "loss": 1.9071,
+      "step": 1555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.784400544959128e-05,
+      "loss": 1.9762,
+      "step": 1556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7841961852861037e-05,
+      "loss": 1.8938,
+      "step": 1557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7839918256130793e-05,
+      "loss": 2.0184,
+      "step": 1558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7837874659400545e-05,
+      "loss": 1.9127,
+      "step": 1559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.78358310626703e-05,
+      "loss": 1.9019,
+      "step": 1560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7833787465940053e-05,
+      "loss": 1.8811,
+      "step": 1561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.783174386920981e-05,
+      "loss": 1.907,
+      "step": 1562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7829700272479564e-05,
+      "loss": 1.9083,
+      "step": 1563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.782765667574932e-05,
+      "loss": 1.9797,
+      "step": 1564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7825613079019076e-05,
+      "loss": 1.9581,
+      "step": 1565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.782356948228883e-05,
+      "loss": 2.0608,
+      "step": 1566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7821525885558584e-05,
+      "loss": 1.9202,
+      "step": 1567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.781948228882834e-05,
+      "loss": 1.7897,
+      "step": 1568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.781743869209809e-05,
+      "loss": 1.9714,
+      "step": 1569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7815395095367847e-05,
+      "loss": 1.8891,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7813351498637603e-05,
+      "loss": 1.9951,
+      "step": 1571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.781130790190736e-05,
+      "loss": 2.0477,
+      "step": 1572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7809264305177114e-05,
+      "loss": 1.9641,
+      "step": 1573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7807220708446866e-05,
+      "loss": 2.0495,
+      "step": 1574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7805177111716622e-05,
+      "loss": 1.9963,
+      "step": 1575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7803133514986378e-05,
+      "loss": 2.0674,
+      "step": 1576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.780108991825613e-05,
+      "loss": 1.9737,
+      "step": 1577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7799046321525885e-05,
+      "loss": 1.9743,
+      "step": 1578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.779700272479564e-05,
+      "loss": 2.0658,
+      "step": 1579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7794959128065397e-05,
+      "loss": 2.0308,
+      "step": 1580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7792915531335152e-05,
+      "loss": 2.0985,
+      "step": 1581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7790871934604905e-05,
+      "loss": 2.0641,
+      "step": 1582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.778882833787466e-05,
+      "loss": 2.1278,
+      "step": 1583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7786784741144416e-05,
+      "loss": 1.9166,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7784741144414168e-05,
+      "loss": 1.9982,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7782697547683924e-05,
+      "loss": 2.0802,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.778065395095368e-05,
+      "loss": 1.9131,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7778610354223435e-05,
+      "loss": 1.8803,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.777656675749319e-05,
+      "loss": 1.83,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7774523160762943e-05,
+      "loss": 1.9815,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.77724795640327e-05,
+      "loss": 1.9683,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.777043596730245e-05,
+      "loss": 1.9756,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7768392370572206e-05,
+      "loss": 1.8578,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7766348773841965e-05,
+      "loss": 2.03,
+      "step": 1594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7764305177111718e-05,
+      "loss": 1.9831,
+      "step": 1595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7762261580381473e-05,
+      "loss": 1.9778,
+      "step": 1596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7760217983651226e-05,
+      "loss": 1.9772,
+      "step": 1597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.775817438692098e-05,
+      "loss": 1.9733,
+      "step": 1598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7756130790190737e-05,
+      "loss": 1.9707,
+      "step": 1599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.775408719346049e-05,
+      "loss": 1.9322,
+      "step": 1600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7752043596730245e-05,
+      "loss": 1.8767,
+      "step": 1601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 1.917,
+      "step": 1602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7747956403269756e-05,
+      "loss": 2.0381,
+      "step": 1603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.774591280653951e-05,
+      "loss": 1.9284,
+      "step": 1604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7743869209809264e-05,
+      "loss": 2.0511,
+      "step": 1605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.774182561307902e-05,
+      "loss": 1.918,
+      "step": 1606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7739782016348775e-05,
+      "loss": 1.8763,
+      "step": 1607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7737738419618527e-05,
+      "loss": 2.0057,
+      "step": 1608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7735694822888286e-05,
+      "loss": 2.0088,
+      "step": 1609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.773365122615804e-05,
+      "loss": 1.9097,
+      "step": 1610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7731607629427794e-05,
+      "loss": 1.9072,
+      "step": 1611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.772956403269755e-05,
+      "loss": 2.0516,
+      "step": 1612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7727520435967302e-05,
+      "loss": 2.0169,
+      "step": 1613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7725476839237058e-05,
+      "loss": 2.0035,
+      "step": 1614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.772343324250681e-05,
+      "loss": 1.9571,
+      "step": 1615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7721389645776566e-05,
+      "loss": 1.98,
+      "step": 1616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7719346049046325e-05,
+      "loss": 2.2177,
+      "step": 1617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7717302452316077e-05,
+      "loss": 1.9528,
+      "step": 1618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7715258855585832e-05,
+      "loss": 2.0449,
+      "step": 1619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7713215258855588e-05,
+      "loss": 1.9684,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.771117166212534e-05,
+      "loss": 1.9241,
+      "step": 1621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7709128065395096e-05,
+      "loss": 1.9007,
+      "step": 1622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7707084468664848e-05,
+      "loss": 2.0488,
+      "step": 1623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7705040871934607e-05,
+      "loss": 1.912,
+      "step": 1624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7702997275204363e-05,
+      "loss": 1.9093,
+      "step": 1625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7700953678474115e-05,
+      "loss": 1.9948,
+      "step": 1626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.769891008174387e-05,
+      "loss": 1.8719,
+      "step": 1627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7696866485013623e-05,
+      "loss": 1.9047,
+      "step": 1628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.769482288828338e-05,
+      "loss": 1.9734,
+      "step": 1629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7692779291553134e-05,
+      "loss": 2.0197,
+      "step": 1630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7690735694822887e-05,
+      "loss": 2.0528,
+      "step": 1631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7688692098092646e-05,
+      "loss": 1.8721,
+      "step": 1632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7686648501362398e-05,
+      "loss": 2.1041,
+      "step": 1633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7684604904632153e-05,
+      "loss": 1.9415,
+      "step": 1634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.768256130790191e-05,
+      "loss": 1.9372,
+      "step": 1635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.768051771117166e-05,
+      "loss": 2.0657,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7678474114441417e-05,
+      "loss": 2.0597,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7676430517711173e-05,
+      "loss": 1.9951,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7674386920980928e-05,
+      "loss": 2.0591,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7672343324250684e-05,
+      "loss": 1.9495,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7670299727520436e-05,
+      "loss": 1.9733,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7668256130790192e-05,
+      "loss": 2.0512,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.7666212534059947e-05,
+      "loss": 2.031,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.76641689373297e-05,
+      "loss": 1.9051,
+      "step": 1644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7662125340599455e-05,
+      "loss": 1.9399,
+      "step": 1645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7660081743869207e-05,
+      "loss": 1.8961,
+      "step": 1646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7658038147138966e-05,
+      "loss": 1.9803,
+      "step": 1647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7655994550408722e-05,
+      "loss": 2.0087,
+      "step": 1648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7653950953678474e-05,
+      "loss": 1.9236,
+      "step": 1649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.765190735694823e-05,
+      "loss": 2.1224,
+      "step": 1650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7649863760217982e-05,
+      "loss": 1.9893,
+      "step": 1651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7647820163487738e-05,
+      "loss": 1.9095,
+      "step": 1652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7645776566757494e-05,
+      "loss": 1.9479,
+      "step": 1653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.764373297002725e-05,
+      "loss": 2.0224,
+      "step": 1654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7641689373297005e-05,
+      "loss": 2.0525,
+      "step": 1655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.763964577656676e-05,
+      "loss": 1.8049,
+      "step": 1656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7637602179836513e-05,
+      "loss": 1.8835,
+      "step": 1657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7635558583106268e-05,
+      "loss": 1.9295,
+      "step": 1658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.763351498637602e-05,
+      "loss": 2.0172,
+      "step": 1659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7631471389645776e-05,
+      "loss": 1.9525,
+      "step": 1660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7629427792915535e-05,
+      "loss": 2.0098,
+      "step": 1661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7627384196185287e-05,
+      "loss": 2.078,
+      "step": 1662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7625340599455043e-05,
+      "loss": 1.885,
+      "step": 1663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7623297002724795e-05,
+      "loss": 1.8045,
+      "step": 1664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.762125340599455e-05,
+      "loss": 1.8797,
+      "step": 1665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7619209809264307e-05,
+      "loss": 1.9503,
+      "step": 1666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.761716621253406e-05,
+      "loss": 2.0171,
+      "step": 1667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7615122615803814e-05,
+      "loss": 1.933,
+      "step": 1668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.761307901907357e-05,
+      "loss": 2.1329,
+      "step": 1669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7611035422343326e-05,
+      "loss": 2.006,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.760899182561308e-05,
+      "loss": 2.0241,
+      "step": 1671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7606948228882834e-05,
+      "loss": 1.889,
+      "step": 1672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.760490463215259e-05,
+      "loss": 1.9461,
+      "step": 1673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7602861035422345e-05,
+      "loss": 1.942,
+      "step": 1674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7600817438692097e-05,
+      "loss": 1.9012,
+      "step": 1675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7598773841961856e-05,
+      "loss": 1.9326,
+      "step": 1676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.759673024523161e-05,
+      "loss": 2.0366,
+      "step": 1677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7594686648501364e-05,
+      "loss": 1.9999,
+      "step": 1678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.759264305177112e-05,
+      "loss": 1.875,
+      "step": 1679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7590599455040872e-05,
+      "loss": 1.9443,
+      "step": 1680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7588555858310627e-05,
+      "loss": 2.047,
+      "step": 1681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.758651226158038e-05,
+      "loss": 1.9346,
+      "step": 1682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7584468664850135e-05,
+      "loss": 1.8894,
+      "step": 1683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7582425068119894e-05,
+      "loss": 1.9365,
+      "step": 1684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7580381471389647e-05,
+      "loss": 1.7715,
+      "step": 1685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7578337874659402e-05,
+      "loss": 2.0212,
+      "step": 1686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7576294277929155e-05,
+      "loss": 2.0829,
+      "step": 1687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.757425068119891e-05,
+      "loss": 1.9641,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7572207084468666e-05,
+      "loss": 1.9153,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7570163487738418e-05,
+      "loss": 1.7853,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7568119891008177e-05,
+      "loss": 1.9631,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.756607629427793e-05,
+      "loss": 2.0215,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7564032697547685e-05,
+      "loss": 1.9234,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.756198910081744e-05,
+      "loss": 2.0037,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7559945504087193e-05,
+      "loss": 1.9084,
+      "step": 1695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.755790190735695e-05,
+      "loss": 2.123,
+      "step": 1696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7555858310626704e-05,
+      "loss": 1.9397,
+      "step": 1697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7553814713896456e-05,
+      "loss": 2.0095,
+      "step": 1698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7551771117166215e-05,
+      "loss": 1.8616,
+      "step": 1699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7549727520435968e-05,
+      "loss": 1.9641,
+      "step": 1700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7547683923705723e-05,
+      "loss": 2.0677,
+      "step": 1701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.754564032697548e-05,
+      "loss": 1.8893,
+      "step": 1702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.754359673024523e-05,
+      "loss": 2.0575,
+      "step": 1703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7541553133514987e-05,
+      "loss": 1.9524,
+      "step": 1704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.753950953678474e-05,
+      "loss": 1.9396,
+      "step": 1705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7537465940054498e-05,
+      "loss": 1.9468,
+      "step": 1706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7535422343324254e-05,
+      "loss": 1.9033,
+      "step": 1707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7533378746594006e-05,
+      "loss": 1.8053,
+      "step": 1708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.753133514986376e-05,
+      "loss": 1.9704,
+      "step": 1709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7529291553133514e-05,
+      "loss": 1.8918,
+      "step": 1710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.752724795640327e-05,
+      "loss": 2.0357,
+      "step": 1711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7525204359673025e-05,
+      "loss": 1.9626,
+      "step": 1712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7523160762942777e-05,
+      "loss": 1.899,
+      "step": 1713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7521117166212536e-05,
+      "loss": 2.0686,
+      "step": 1714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7519073569482292e-05,
+      "loss": 2.0911,
+      "step": 1715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7517029972752044e-05,
+      "loss": 1.8802,
+      "step": 1716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.75149863760218e-05,
+      "loss": 2.0656,
+      "step": 1717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7512942779291552e-05,
+      "loss": 2.0207,
+      "step": 1718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7510899182561308e-05,
+      "loss": 1.8201,
+      "step": 1719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7508855585831063e-05,
+      "loss": 1.9117,
+      "step": 1720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.750681198910082e-05,
+      "loss": 2.0582,
+      "step": 1721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7504768392370575e-05,
+      "loss": 1.8135,
+      "step": 1722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7502724795640327e-05,
+      "loss": 1.9632,
+      "step": 1723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7500681198910082e-05,
+      "loss": 1.9214,
+      "step": 1724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7498637602179838e-05,
+      "loss": 1.845,
+      "step": 1725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.749659400544959e-05,
+      "loss": 1.9048,
+      "step": 1726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7494550408719346e-05,
+      "loss": 1.9463,
+      "step": 1727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7492506811989098e-05,
+      "loss": 1.9432,
+      "step": 1728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7490463215258857e-05,
+      "loss": 1.941,
+      "step": 1729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7488419618528613e-05,
+      "loss": 2.1292,
+      "step": 1730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7486376021798365e-05,
+      "loss": 1.9645,
+      "step": 1731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.748433242506812e-05,
+      "loss": 1.9568,
+      "step": 1732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7482288828337876e-05,
+      "loss": 2.0355,
+      "step": 1733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.748024523160763e-05,
+      "loss": 1.9531,
+      "step": 1734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7478201634877384e-05,
+      "loss": 2.0188,
+      "step": 1735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.747615803814714e-05,
+      "loss": 1.873,
+      "step": 1736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7474114441416895e-05,
+      "loss": 1.9224,
+      "step": 1737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.747207084468665e-05,
+      "loss": 1.8823,
+      "step": 1738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7470027247956403e-05,
+      "loss": 1.9188,
+      "step": 1739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.746798365122616e-05,
+      "loss": 2.0133,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.746594005449591e-05,
+      "loss": 1.9328,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7463896457765667e-05,
+      "loss": 1.9259,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7461852861035426e-05,
+      "loss": 1.9503,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7459809264305178e-05,
+      "loss": 1.8999,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7457765667574934e-05,
+      "loss": 1.9307,
+      "step": 1745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7455722070844686e-05,
+      "loss": 2.0276,
+      "step": 1746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.745367847411444e-05,
+      "loss": 1.9216,
+      "step": 1747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7451634877384197e-05,
+      "loss": 1.7932,
+      "step": 1748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.744959128065395e-05,
+      "loss": 1.9881,
+      "step": 1749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7447547683923705e-05,
+      "loss": 1.9915,
+      "step": 1750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7445504087193464e-05,
+      "loss": 1.8792,
+      "step": 1751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7443460490463216e-05,
+      "loss": 1.9946,
+      "step": 1752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7441416893732972e-05,
+      "loss": 1.7708,
+      "step": 1753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7439373297002724e-05,
+      "loss": 2.107,
+      "step": 1754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.743732970027248e-05,
+      "loss": 1.9189,
+      "step": 1755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7435286103542236e-05,
+      "loss": 1.9799,
+      "step": 1756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7433242506811988e-05,
+      "loss": 2.0022,
+      "step": 1757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7431198910081747e-05,
+      "loss": 1.9892,
+      "step": 1758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.74291553133515e-05,
+      "loss": 2.0174,
+      "step": 1759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7427111716621255e-05,
+      "loss": 1.8811,
+      "step": 1760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.742506811989101e-05,
+      "loss": 2.0172,
+      "step": 1761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7423024523160763e-05,
+      "loss": 1.9431,
+      "step": 1762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7420980926430518e-05,
+      "loss": 1.8263,
+      "step": 1763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.741893732970027e-05,
+      "loss": 1.8605,
+      "step": 1764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7416893732970026e-05,
+      "loss": 1.9179,
+      "step": 1765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7414850136239785e-05,
+      "loss": 2.1166,
+      "step": 1766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7412806539509537e-05,
+      "loss": 2.0779,
+      "step": 1767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7410762942779293e-05,
+      "loss": 1.8374,
+      "step": 1768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.740871934604905e-05,
+      "loss": 2.0167,
+      "step": 1769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.74066757493188e-05,
+      "loss": 2.0626,
+      "step": 1770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7404632152588557e-05,
+      "loss": 1.9327,
+      "step": 1771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.740258855585831e-05,
+      "loss": 1.9882,
+      "step": 1772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7400544959128068e-05,
+      "loss": 1.8724,
+      "step": 1773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7398501362397823e-05,
+      "loss": 2.0493,
+      "step": 1774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7396457765667576e-05,
+      "loss": 1.9911,
+      "step": 1775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.739441416893733e-05,
+      "loss": 1.9319,
+      "step": 1776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7392370572207084e-05,
+      "loss": 1.8659,
+      "step": 1777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.739032697547684e-05,
+      "loss": 1.9708,
+      "step": 1778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7388283378746595e-05,
+      "loss": 2.043,
+      "step": 1779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7386239782016347e-05,
+      "loss": 1.9473,
+      "step": 1780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7384196185286106e-05,
+      "loss": 1.9012,
+      "step": 1781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.738215258855586e-05,
+      "loss": 2.0768,
+      "step": 1782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7380108991825614e-05,
+      "loss": 2.0741,
+      "step": 1783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.737806539509537e-05,
+      "loss": 1.946,
+      "step": 1784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7376021798365122e-05,
+      "loss": 1.9781,
+      "step": 1785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7373978201634877e-05,
+      "loss": 2.0388,
+      "step": 1786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7371934604904633e-05,
+      "loss": 1.7587,
+      "step": 1787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.736989100817439e-05,
+      "loss": 1.9346,
+      "step": 1788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7367847411444144e-05,
+      "loss": 1.9844,
+      "step": 1789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7365803814713897e-05,
+      "loss": 2.001,
+      "step": 1790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7363760217983652e-05,
+      "loss": 2.1662,
+      "step": 1791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7361716621253408e-05,
+      "loss": 1.9166,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.735967302452316e-05,
+      "loss": 1.9527,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7357629427792916e-05,
+      "loss": 1.948,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7355585831062668e-05,
+      "loss": 1.8248,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7353542234332427e-05,
+      "loss": 2.0491,
+      "step": 1796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7351498637602183e-05,
+      "loss": 1.97,
+      "step": 1797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7349455040871935e-05,
+      "loss": 1.9996,
+      "step": 1798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.734741144414169e-05,
+      "loss": 1.8854,
+      "step": 1799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7345367847411443e-05,
+      "loss": 2.0183,
+      "step": 1800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.73433242506812e-05,
+      "loss": 1.8531,
+      "step": 1801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7341280653950954e-05,
+      "loss": 1.895,
+      "step": 1802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.733923705722071e-05,
+      "loss": 1.9819,
+      "step": 1803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7337193460490465e-05,
+      "loss": 2.2423,
+      "step": 1804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.733514986376022e-05,
+      "loss": 1.9981,
+      "step": 1805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7333106267029973e-05,
+      "loss": 1.8961,
+      "step": 1806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.733106267029973e-05,
+      "loss": 1.8931,
+      "step": 1807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.732901907356948e-05,
+      "loss": 1.857,
+      "step": 1808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7326975476839237e-05,
+      "loss": 1.9183,
+      "step": 1809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7324931880108992e-05,
+      "loss": 1.8568,
+      "step": 1810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7322888283378748e-05,
+      "loss": 2.0226,
+      "step": 1811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7320844686648504e-05,
+      "loss": 1.8688,
+      "step": 1812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7318801089918256e-05,
+      "loss": 2.0503,
+      "step": 1813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.731675749318801e-05,
+      "loss": 1.9882,
+      "step": 1814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7314713896457767e-05,
+      "loss": 1.86,
+      "step": 1815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.731267029972752e-05,
+      "loss": 2.0318,
+      "step": 1816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7310626702997275e-05,
+      "loss": 1.9705,
+      "step": 1817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.730858310626703e-05,
+      "loss": 1.9882,
+      "step": 1818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7306539509536786e-05,
+      "loss": 1.7852,
+      "step": 1819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7304495912806542e-05,
+      "loss": 1.9781,
+      "step": 1820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7302452316076294e-05,
+      "loss": 1.8021,
+      "step": 1821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.730040871934605e-05,
+      "loss": 1.9574,
+      "step": 1822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7298365122615805e-05,
+      "loss": 1.9223,
+      "step": 1823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7296321525885558e-05,
+      "loss": 2.0167,
+      "step": 1824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7294277929155313e-05,
+      "loss": 2.0328,
+      "step": 1825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.729223433242507e-05,
+      "loss": 1.9271,
+      "step": 1826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7290190735694825e-05,
+      "loss": 2.0307,
+      "step": 1827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.728814713896458e-05,
+      "loss": 1.9615,
+      "step": 1828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7286103542234332e-05,
+      "loss": 1.9535,
+      "step": 1829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7284059945504088e-05,
+      "loss": 1.9382,
+      "step": 1830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.728201634877384e-05,
+      "loss": 2.0648,
+      "step": 1831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7279972752043596e-05,
+      "loss": 1.9576,
+      "step": 1832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7277929155313355e-05,
+      "loss": 1.8282,
+      "step": 1833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7275885558583107e-05,
+      "loss": 1.9415,
+      "step": 1834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7273841961852863e-05,
+      "loss": 1.8957,
+      "step": 1835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7271798365122615e-05,
+      "loss": 1.8927,
+      "step": 1836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.726975476839237e-05,
+      "loss": 2.0034,
+      "step": 1837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7267711171662126e-05,
+      "loss": 2.0373,
+      "step": 1838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.726566757493188e-05,
+      "loss": 1.9389,
+      "step": 1839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7263623978201638e-05,
+      "loss": 2.1016,
+      "step": 1840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7261580381471393e-05,
+      "loss": 1.8945,
+      "step": 1841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7259536784741145e-05,
+      "loss": 1.8161,
+      "step": 1842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.72574931880109e-05,
+      "loss": 1.9182,
+      "step": 1843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7255449591280653e-05,
+      "loss": 1.9995,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.725340599455041e-05,
+      "loss": 1.9828,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7251362397820165e-05,
+      "loss": 2.0222,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.7249318801089917e-05,
+      "loss": 1.9356,
+      "step": 1847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7247275204359676e-05,
+      "loss": 1.9438,
+      "step": 1848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7245231607629428e-05,
+      "loss": 1.7861,
+      "step": 1849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7243188010899184e-05,
+      "loss": 1.9435,
+      "step": 1850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.724114441416894e-05,
+      "loss": 2.0353,
+      "step": 1851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.723910081743869e-05,
+      "loss": 1.887,
+      "step": 1852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7237057220708447e-05,
+      "loss": 1.9499,
+      "step": 1853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.72350136239782e-05,
+      "loss": 2.0417,
+      "step": 1854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.723297002724796e-05,
+      "loss": 1.8717,
+      "step": 1855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7230926430517714e-05,
+      "loss": 1.9052,
+      "step": 1856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7228882833787466e-05,
+      "loss": 1.9713,
+      "step": 1857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7226839237057222e-05,
+      "loss": 1.9202,
+      "step": 1858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7224795640326978e-05,
+      "loss": 1.944,
+      "step": 1859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.722275204359673e-05,
+      "loss": 1.9269,
+      "step": 1860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7220708446866486e-05,
+      "loss": 2.0297,
+      "step": 1861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7218664850136238e-05,
+      "loss": 2.0791,
+      "step": 1862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7216621253405997e-05,
+      "loss": 1.9785,
+      "step": 1863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7214577656675752e-05,
+      "loss": 1.9604,
+      "step": 1864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7212534059945505e-05,
+      "loss": 1.9934,
+      "step": 1865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.721049046321526e-05,
+      "loss": 1.8593,
+      "step": 1866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7208446866485013e-05,
+      "loss": 1.8929,
+      "step": 1867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7206403269754768e-05,
+      "loss": 1.8308,
+      "step": 1868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7204359673024524e-05,
+      "loss": 1.9766,
+      "step": 1869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.720231607629428e-05,
+      "loss": 1.7726,
+      "step": 1870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7200272479564035e-05,
+      "loss": 1.929,
+      "step": 1871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7198228882833787e-05,
+      "loss": 1.9997,
+      "step": 1872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7196185286103543e-05,
+      "loss": 1.873,
+      "step": 1873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.71941416893733e-05,
+      "loss": 1.9355,
+      "step": 1874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.719209809264305e-05,
+      "loss": 1.8403,
+      "step": 1875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7190054495912806e-05,
+      "loss": 2.0345,
+      "step": 1876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7188010899182562e-05,
+      "loss": 2.0971,
+      "step": 1877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7185967302452318e-05,
+      "loss": 1.8697,
+      "step": 1878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7183923705722073e-05,
+      "loss": 1.8196,
+      "step": 1879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7181880108991826e-05,
+      "loss": 2.1044,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.717983651226158e-05,
+      "loss": 2.0153,
+      "step": 1881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7177792915531337e-05,
+      "loss": 1.9361,
+      "step": 1882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.717574931880109e-05,
+      "loss": 1.923,
+      "step": 1883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7173705722070845e-05,
+      "loss": 1.9371,
+      "step": 1884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.71716621253406e-05,
+      "loss": 1.8271,
+      "step": 1885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7169618528610356e-05,
+      "loss": 1.8892,
+      "step": 1886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.716757493188011e-05,
+      "loss": 1.8548,
+      "step": 1887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7165531335149864e-05,
+      "loss": 1.8878,
+      "step": 1888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.716348773841962e-05,
+      "loss": 1.9696,
+      "step": 1889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7161444141689372e-05,
+      "loss": 1.7956,
+      "step": 1890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7159400544959127e-05,
+      "loss": 1.9153,
+      "step": 1891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7157356948228883e-05,
+      "loss": 1.9545,
+      "step": 1892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.715531335149864e-05,
+      "loss": 1.9615,
+      "step": 1893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7153269754768394e-05,
+      "loss": 1.9725,
+      "step": 1894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7151226158038147e-05,
+      "loss": 2.0073,
+      "step": 1895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7149182561307902e-05,
+      "loss": 1.9462,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.7147138964577658e-05,
+      "loss": 1.8622,
+      "step": 1897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.714509536784741e-05,
+      "loss": 1.9316,
+      "step": 1898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7143051771117166e-05,
+      "loss": 1.9202,
+      "step": 1899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7141008174386925e-05,
+      "loss": 1.9277,
+      "step": 1900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7138964577656677e-05,
+      "loss": 1.9521,
+      "step": 1901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7136920980926433e-05,
+      "loss": 1.892,
+      "step": 1902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7134877384196185e-05,
+      "loss": 2.0034,
+      "step": 1903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.713283378746594e-05,
+      "loss": 1.9659,
+      "step": 1904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7130790190735696e-05,
+      "loss": 1.8676,
+      "step": 1905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.712874659400545e-05,
+      "loss": 1.9484,
+      "step": 1906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7126702997275204e-05,
+      "loss": 1.9469,
+      "step": 1907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.712465940054496e-05,
+      "loss": 1.8806,
+      "step": 1908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7122615803814715e-05,
+      "loss": 2.01,
+      "step": 1909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.712057220708447e-05,
+      "loss": 2.0599,
+      "step": 1910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7118528610354223e-05,
+      "loss": 1.8208,
+      "step": 1911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.711648501362398e-05,
+      "loss": 1.8425,
+      "step": 1912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.711444141689373e-05,
+      "loss": 2.0212,
+      "step": 1913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7112397820163487e-05,
+      "loss": 1.9842,
+      "step": 1914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7110354223433246e-05,
+      "loss": 1.8862,
+      "step": 1915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7108310626702998e-05,
+      "loss": 2.0509,
+      "step": 1916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7106267029972754e-05,
+      "loss": 1.916,
+      "step": 1917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.710422343324251e-05,
+      "loss": 1.9547,
+      "step": 1918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.710217983651226e-05,
+      "loss": 1.908,
+      "step": 1919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7100136239782017e-05,
+      "loss": 1.8462,
+      "step": 1920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.709809264305177e-05,
+      "loss": 1.8827,
+      "step": 1921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7096049046321525e-05,
+      "loss": 1.9807,
+      "step": 1922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7094005449591284e-05,
+      "loss": 1.8521,
+      "step": 1923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7091961852861036e-05,
+      "loss": 1.9481,
+      "step": 1924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7089918256130792e-05,
+      "loss": 1.7447,
+      "step": 1925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7087874659400544e-05,
+      "loss": 1.9497,
+      "step": 1926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.70858310626703e-05,
+      "loss": 1.9898,
+      "step": 1927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7083787465940055e-05,
+      "loss": 1.8618,
+      "step": 1928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7081743869209808e-05,
+      "loss": 1.8287,
+      "step": 1929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7079700272479567e-05,
+      "loss": 1.9118,
+      "step": 1930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.707765667574932e-05,
+      "loss": 2.0357,
+      "step": 1931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7075613079019074e-05,
+      "loss": 2.0582,
+      "step": 1932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.707356948228883e-05,
+      "loss": 1.9729,
+      "step": 1933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7071525885558582e-05,
+      "loss": 1.8912,
+      "step": 1934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7069482288828338e-05,
+      "loss": 1.9169,
+      "step": 1935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7067438692098094e-05,
+      "loss": 1.7676,
+      "step": 1936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.706539509536785e-05,
+      "loss": 1.9129,
+      "step": 1937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7063351498637605e-05,
+      "loss": 1.9163,
+      "step": 1938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7061307901907357e-05,
+      "loss": 1.9025,
+      "step": 1939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7059264305177113e-05,
+      "loss": 1.8675,
+      "step": 1940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.705722070844687e-05,
+      "loss": 1.9911,
+      "step": 1941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.705517711171662e-05,
+      "loss": 2.0466,
+      "step": 1942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7053133514986376e-05,
+      "loss": 1.8536,
+      "step": 1943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.705108991825613e-05,
+      "loss": 1.7948,
+      "step": 1944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7049046321525888e-05,
+      "loss": 1.9573,
+      "step": 1945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7047002724795643e-05,
+      "loss": 1.8782,
+      "step": 1946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7044959128065395e-05,
+      "loss": 1.985,
+      "step": 1947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.704291553133515e-05,
+      "loss": 1.8695,
+      "step": 1948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7040871934604903e-05,
+      "loss": 1.9192,
+      "step": 1949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.703882833787466e-05,
+      "loss": 1.9706,
+      "step": 1950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7036784741144415e-05,
+      "loss": 1.8219,
+      "step": 1951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.703474114441417e-05,
+      "loss": 1.9216,
+      "step": 1952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7032697547683926e-05,
+      "loss": 2.1252,
+      "step": 1953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.703065395095368e-05,
+      "loss": 1.946,
+      "step": 1954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7028610354223434e-05,
+      "loss": 1.9975,
+      "step": 1955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.702656675749319e-05,
+      "loss": 1.8636,
+      "step": 1956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.702452316076294e-05,
+      "loss": 1.9864,
+      "step": 1957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7022479564032697e-05,
+      "loss": 1.9459,
+      "step": 1958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7020435967302453e-05,
+      "loss": 2.0421,
+      "step": 1959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.701839237057221e-05,
+      "loss": 2.0148,
+      "step": 1960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7016348773841964e-05,
+      "loss": 1.837,
+      "step": 1961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7014305177111716e-05,
+      "loss": 1.9642,
+      "step": 1962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7012261580381472e-05,
+      "loss": 1.8588,
+      "step": 1963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7010217983651228e-05,
+      "loss": 1.9597,
+      "step": 1964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.700817438692098e-05,
+      "loss": 1.8939,
+      "step": 1965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7006130790190735e-05,
+      "loss": 1.9833,
+      "step": 1966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.700408719346049e-05,
+      "loss": 1.9795,
+      "step": 1967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7002043596730247e-05,
+      "loss": 1.8141,
+      "step": 1968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 2.0065,
+      "step": 1969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6997956403269755e-05,
+      "loss": 1.8777,
+      "step": 1970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.699591280653951e-05,
+      "loss": 1.9577,
+      "step": 1971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6993869209809266e-05,
+      "loss": 2.008,
+      "step": 1972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6991825613079018e-05,
+      "loss": 2.0703,
+      "step": 1973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6989782016348774e-05,
+      "loss": 1.8962,
+      "step": 1974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.698773841961853e-05,
+      "loss": 1.9401,
+      "step": 1975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6985694822888285e-05,
+      "loss": 1.841,
+      "step": 1976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.698365122615804e-05,
+      "loss": 1.8977,
+      "step": 1977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6981607629427793e-05,
+      "loss": 2.0994,
+      "step": 1978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.697956403269755e-05,
+      "loss": 1.9391,
+      "step": 1979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.69775204359673e-05,
+      "loss": 2.0248,
+      "step": 1980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6975476839237056e-05,
+      "loss": 1.8382,
+      "step": 1981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6973433242506815e-05,
+      "loss": 1.9249,
+      "step": 1982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6971389645776568e-05,
+      "loss": 1.9387,
+      "step": 1983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6969346049046323e-05,
+      "loss": 1.8981,
+      "step": 1984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6967302452316076e-05,
+      "loss": 1.8682,
+      "step": 1985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.696525885558583e-05,
+      "loss": 1.8111,
+      "step": 1986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6963215258855587e-05,
+      "loss": 1.9205,
+      "step": 1987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.696117166212534e-05,
+      "loss": 1.915,
+      "step": 1988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6959128065395095e-05,
+      "loss": 1.9121,
+      "step": 1989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6957084468664854e-05,
+      "loss": 1.9728,
+      "step": 1990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6955040871934606e-05,
+      "loss": 1.9113,
+      "step": 1991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.695299727520436e-05,
+      "loss": 1.9928,
+      "step": 1992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6950953678474114e-05,
+      "loss": 1.8253,
+      "step": 1993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.694891008174387e-05,
+      "loss": 1.942,
+      "step": 1994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6946866485013625e-05,
+      "loss": 1.9559,
+      "step": 1995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6944822888283377e-05,
+      "loss": 1.7674,
+      "step": 1996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6942779291553136e-05,
+      "loss": 2.0095,
+      "step": 1997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.694073569482289e-05,
+      "loss": 1.8254,
+      "step": 1998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6938692098092644e-05,
+      "loss": 1.8794,
+      "step": 1999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.69366485013624e-05,
+      "loss": 1.9422,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.7206156253814697,
+      "eval_runtime": 722.8136,
+      "eval_samples_per_second": 5.741,
+      "eval_steps_per_second": 1.436,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6934604904632152e-05,
+      "loss": 1.9528,
+      "step": 2001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6932561307901908e-05,
+      "loss": 1.8663,
+      "step": 2002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.693051771117166e-05,
+      "loss": 1.8714,
+      "step": 2003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6928474114441416e-05,
+      "loss": 1.9399,
+      "step": 2004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6926430517711175e-05,
+      "loss": 2.0547,
+      "step": 2005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6924386920980927e-05,
+      "loss": 1.976,
+      "step": 2006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6922343324250683e-05,
+      "loss": 1.892,
+      "step": 2007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6920299727520438e-05,
+      "loss": 1.9277,
+      "step": 2008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.691825613079019e-05,
+      "loss": 1.9398,
+      "step": 2009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6916212534059946e-05,
+      "loss": 1.9503,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6914168937329698e-05,
+      "loss": 1.9426,
+      "step": 2011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6912125340599457e-05,
+      "loss": 1.9292,
+      "step": 2012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6910081743869213e-05,
+      "loss": 2.0577,
+      "step": 2013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6908038147138965e-05,
+      "loss": 1.9305,
+      "step": 2014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.690599455040872e-05,
+      "loss": 1.899,
+      "step": 2015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6903950953678473e-05,
+      "loss": 2.0274,
+      "step": 2016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.690190735694823e-05,
+      "loss": 1.9625,
+      "step": 2017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6899863760217984e-05,
+      "loss": 1.8415,
+      "step": 2018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.689782016348774e-05,
+      "loss": 1.9015,
+      "step": 2019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6895776566757496e-05,
+      "loss": 1.7343,
+      "step": 2020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6893732970027248e-05,
+      "loss": 1.8621,
+      "step": 2021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6891689373297003e-05,
+      "loss": 1.899,
+      "step": 2022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.688964577656676e-05,
+      "loss": 1.9104,
+      "step": 2023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.688760217983651e-05,
+      "loss": 1.9122,
+      "step": 2024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6885558583106267e-05,
+      "loss": 1.8684,
+      "step": 2025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6883514986376023e-05,
+      "loss": 1.9817,
+      "step": 2026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6881471389645778e-05,
+      "loss": 1.9104,
+      "step": 2027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6879427792915534e-05,
+      "loss": 1.9496,
+      "step": 2028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6877384196185286e-05,
+      "loss": 2.0356,
+      "step": 2029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6875340599455042e-05,
+      "loss": 1.9037,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6873297002724797e-05,
+      "loss": 2.0414,
+      "step": 2031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.687125340599455e-05,
+      "loss": 1.9661,
+      "step": 2032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6869209809264305e-05,
+      "loss": 1.9949,
+      "step": 2033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.686716621253406e-05,
+      "loss": 1.9212,
+      "step": 2034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6865122615803817e-05,
+      "loss": 1.9125,
+      "step": 2035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6863079019073572e-05,
+      "loss": 1.9206,
+      "step": 2036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6861035422343324e-05,
+      "loss": 1.8517,
+      "step": 2037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.685899182561308e-05,
+      "loss": 1.7546,
+      "step": 2038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6856948228882832e-05,
+      "loss": 2.0497,
+      "step": 2039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6854904632152588e-05,
+      "loss": 1.9379,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6852861035422344e-05,
+      "loss": 1.9526,
+      "step": 2041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.68508174386921e-05,
+      "loss": 1.9509,
+      "step": 2042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6848773841961855e-05,
+      "loss": 1.7614,
+      "step": 2043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.684673024523161e-05,
+      "loss": 1.9339,
+      "step": 2044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6844686648501363e-05,
+      "loss": 1.852,
+      "step": 2045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.684264305177112e-05,
+      "loss": 1.973,
+      "step": 2046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.684059945504087e-05,
+      "loss": 1.9584,
+      "step": 2047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6838555858310626e-05,
+      "loss": 1.8114,
+      "step": 2048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6836512261580385e-05,
+      "loss": 2.0937,
+      "step": 2049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6834468664850137e-05,
+      "loss": 1.83,
+      "step": 2050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6832425068119893e-05,
+      "loss": 1.9544,
+      "step": 2051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6830381471389645e-05,
+      "loss": 1.9273,
+      "step": 2052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.68283378746594e-05,
+      "loss": 1.9749,
+      "step": 2053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6826294277929157e-05,
+      "loss": 1.8686,
+      "step": 2054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.682425068119891e-05,
+      "loss": 1.9034,
+      "step": 2055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6822207084468664e-05,
+      "loss": 1.8439,
+      "step": 2056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.682016348773842e-05,
+      "loss": 1.7831,
+      "step": 2057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6818119891008176e-05,
+      "loss": 1.8386,
+      "step": 2058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.681607629427793e-05,
+      "loss": 1.7533,
+      "step": 2059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6814032697547684e-05,
+      "loss": 1.9584,
+      "step": 2060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.681198910081744e-05,
+      "loss": 1.9381,
+      "step": 2061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6809945504087195e-05,
+      "loss": 1.8929,
+      "step": 2062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6807901907356947e-05,
+      "loss": 1.8474,
+      "step": 2063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6805858310626706e-05,
+      "loss": 1.8706,
+      "step": 2064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.680381471389646e-05,
+      "loss": 1.918,
+      "step": 2065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6801771117166214e-05,
+      "loss": 1.8646,
+      "step": 2066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.679972752043597e-05,
+      "loss": 1.9563,
+      "step": 2067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6797683923705722e-05,
+      "loss": 1.9789,
+      "step": 2068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6795640326975478e-05,
+      "loss": 1.8542,
+      "step": 2069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.679359673024523e-05,
+      "loss": 1.8515,
+      "step": 2070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6791553133514985e-05,
+      "loss": 1.9518,
+      "step": 2071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6789509536784744e-05,
+      "loss": 1.9448,
+      "step": 2072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6787465940054497e-05,
+      "loss": 1.8338,
+      "step": 2073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6785422343324252e-05,
+      "loss": 1.8413,
+      "step": 2074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6783378746594005e-05,
+      "loss": 1.8926,
+      "step": 2075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.678133514986376e-05,
+      "loss": 2.0089,
+      "step": 2076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6779291553133516e-05,
+      "loss": 1.8435,
+      "step": 2077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6777247956403268e-05,
+      "loss": 1.912,
+      "step": 2078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6775204359673027e-05,
+      "loss": 1.8968,
+      "step": 2079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.677316076294278e-05,
+      "loss": 1.9679,
+      "step": 2080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6771117166212535e-05,
+      "loss": 1.9738,
+      "step": 2081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.676907356948229e-05,
+      "loss": 1.9876,
+      "step": 2082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6767029972752043e-05,
+      "loss": 1.9327,
+      "step": 2083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.67649863760218e-05,
+      "loss": 1.9599,
+      "step": 2084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6762942779291554e-05,
+      "loss": 1.9526,
+      "step": 2085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6760899182561306e-05,
+      "loss": 1.9362,
+      "step": 2086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6758855585831065e-05,
+      "loss": 1.8768,
+      "step": 2087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6756811989100818e-05,
+      "loss": 1.8974,
+      "step": 2088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6754768392370573e-05,
+      "loss": 2.0681,
+      "step": 2089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.675272479564033e-05,
+      "loss": 1.8981,
+      "step": 2090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.675068119891008e-05,
+      "loss": 1.9877,
+      "step": 2091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6748637602179837e-05,
+      "loss": 1.9129,
+      "step": 2092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.674659400544959e-05,
+      "loss": 1.9432,
+      "step": 2093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6744550408719348e-05,
+      "loss": 1.9558,
+      "step": 2094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6742506811989104e-05,
+      "loss": 1.993,
+      "step": 2095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6740463215258856e-05,
+      "loss": 1.9532,
+      "step": 2096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.673841961852861e-05,
+      "loss": 1.8956,
+      "step": 2097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6736376021798364e-05,
+      "loss": 1.8534,
+      "step": 2098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.673433242506812e-05,
+      "loss": 1.9475,
+      "step": 2099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6732288828337875e-05,
+      "loss": 1.8482,
+      "step": 2100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6730245231607627e-05,
+      "loss": 1.9355,
+      "step": 2101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6728201634877386e-05,
+      "loss": 1.7951,
+      "step": 2102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6726158038147142e-05,
+      "loss": 1.8653,
+      "step": 2103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6724114441416894e-05,
+      "loss": 1.8849,
+      "step": 2104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.672207084468665e-05,
+      "loss": 1.9908,
+      "step": 2105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6720027247956402e-05,
+      "loss": 1.9467,
+      "step": 2106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6717983651226158e-05,
+      "loss": 1.8248,
+      "step": 2107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6715940054495913e-05,
+      "loss": 1.9002,
+      "step": 2108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.671389645776567e-05,
+      "loss": 1.8518,
+      "step": 2109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6711852861035425e-05,
+      "loss": 1.8291,
+      "step": 2110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6709809264305177e-05,
+      "loss": 1.9109,
+      "step": 2111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6707765667574932e-05,
+      "loss": 1.8939,
+      "step": 2112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6705722070844688e-05,
+      "loss": 1.838,
+      "step": 2113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.670367847411444e-05,
+      "loss": 2.0517,
+      "step": 2114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6701634877384196e-05,
+      "loss": 1.9119,
+      "step": 2115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.669959128065395e-05,
+      "loss": 1.9041,
+      "step": 2116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6697547683923707e-05,
+      "loss": 1.8585,
+      "step": 2117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6695504087193463e-05,
+      "loss": 1.8267,
+      "step": 2118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6693460490463215e-05,
+      "loss": 1.888,
+      "step": 2119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.669141689373297e-05,
+      "loss": 1.9162,
+      "step": 2120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6689373297002726e-05,
+      "loss": 1.9187,
+      "step": 2121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.668732970027248e-05,
+      "loss": 1.8813,
+      "step": 2122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6685286103542234e-05,
+      "loss": 1.738,
+      "step": 2123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.668324250681199e-05,
+      "loss": 1.7886,
+      "step": 2124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6681198910081746e-05,
+      "loss": 1.8464,
+      "step": 2125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.66791553133515e-05,
+      "loss": 1.8805,
+      "step": 2126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6677111716621253e-05,
+      "loss": 1.8946,
+      "step": 2127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.667506811989101e-05,
+      "loss": 1.8367,
+      "step": 2128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.667302452316076e-05,
+      "loss": 1.8376,
+      "step": 2129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6670980926430517e-05,
+      "loss": 2.0008,
+      "step": 2130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6668937329700276e-05,
+      "loss": 1.7813,
+      "step": 2131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6666893732970028e-05,
+      "loss": 1.8083,
+      "step": 2132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6664850136239784e-05,
+      "loss": 1.9076,
+      "step": 2133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6662806539509536e-05,
+      "loss": 1.8415,
+      "step": 2134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6660762942779292e-05,
+      "loss": 1.8245,
+      "step": 2135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6658719346049047e-05,
+      "loss": 1.9373,
+      "step": 2136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.66566757493188e-05,
+      "loss": 1.9988,
+      "step": 2137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6654632152588555e-05,
+      "loss": 2.0322,
+      "step": 2138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6652588555858314e-05,
+      "loss": 2.0565,
+      "step": 2139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6650544959128066e-05,
+      "loss": 1.9038,
+      "step": 2140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6648501362397822e-05,
+      "loss": 2.0221,
+      "step": 2141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6646457765667574e-05,
+      "loss": 1.8893,
+      "step": 2142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.664441416893733e-05,
+      "loss": 1.9064,
+      "step": 2143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6642370572207086e-05,
+      "loss": 1.7561,
+      "step": 2144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6640326975476838e-05,
+      "loss": 1.8754,
+      "step": 2145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6638283378746597e-05,
+      "loss": 1.9164,
+      "step": 2146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.663623978201635e-05,
+      "loss": 1.8899,
+      "step": 2147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6634196185286105e-05,
+      "loss": 1.8828,
+      "step": 2148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.663215258855586e-05,
+      "loss": 1.9626,
+      "step": 2149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6630108991825613e-05,
+      "loss": 1.8562,
+      "step": 2150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6628065395095368e-05,
+      "loss": 1.9413,
+      "step": 2151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.662602179836512e-05,
+      "loss": 2.069,
+      "step": 2152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6623978201634876e-05,
+      "loss": 1.8532,
+      "step": 2153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6621934604904635e-05,
+      "loss": 1.9358,
+      "step": 2154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6619891008174387e-05,
+      "loss": 1.9538,
+      "step": 2155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6617847411444143e-05,
+      "loss": 1.9466,
+      "step": 2156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.66158038147139e-05,
+      "loss": 1.8593,
+      "step": 2157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.661376021798365e-05,
+      "loss": 1.9008,
+      "step": 2158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6611716621253407e-05,
+      "loss": 1.954,
+      "step": 2159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.660967302452316e-05,
+      "loss": 1.8544,
+      "step": 2160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6607629427792918e-05,
+      "loss": 2.0094,
+      "step": 2161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6605585831062673e-05,
+      "loss": 1.7665,
+      "step": 2162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6603542234332426e-05,
+      "loss": 1.9209,
+      "step": 2163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.660149863760218e-05,
+      "loss": 2.0007,
+      "step": 2164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6599455040871934e-05,
+      "loss": 1.7537,
+      "step": 2165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.659741144414169e-05,
+      "loss": 1.9086,
+      "step": 2166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6595367847411445e-05,
+      "loss": 1.933,
+      "step": 2167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6593324250681197e-05,
+      "loss": 1.9817,
+      "step": 2168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6591280653950956e-05,
+      "loss": 2.0577,
+      "step": 2169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.658923705722071e-05,
+      "loss": 1.9799,
+      "step": 2170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6587193460490464e-05,
+      "loss": 1.8779,
+      "step": 2171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.658514986376022e-05,
+      "loss": 1.9729,
+      "step": 2172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6583106267029972e-05,
+      "loss": 1.9364,
+      "step": 2173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6581062670299728e-05,
+      "loss": 1.9303,
+      "step": 2174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6579019073569483e-05,
+      "loss": 1.9434,
+      "step": 2175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.657697547683924e-05,
+      "loss": 1.8429,
+      "step": 2176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6574931880108994e-05,
+      "loss": 1.7931,
+      "step": 2177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6572888283378747e-05,
+      "loss": 1.8515,
+      "step": 2178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6570844686648502e-05,
+      "loss": 1.8602,
+      "step": 2179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6568801089918258e-05,
+      "loss": 2.0656,
+      "step": 2180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.656675749318801e-05,
+      "loss": 1.8471,
+      "step": 2181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6564713896457766e-05,
+      "loss": 1.8098,
+      "step": 2182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6562670299727518e-05,
+      "loss": 1.8784,
+      "step": 2183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6560626702997277e-05,
+      "loss": 1.9403,
+      "step": 2184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6558583106267033e-05,
+      "loss": 1.9824,
+      "step": 2185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6556539509536785e-05,
+      "loss": 1.9147,
+      "step": 2186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.655449591280654e-05,
+      "loss": 1.895,
+      "step": 2187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6552452316076293e-05,
+      "loss": 1.7985,
+      "step": 2188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.655040871934605e-05,
+      "loss": 1.8525,
+      "step": 2189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6548365122615804e-05,
+      "loss": 1.9523,
+      "step": 2190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.654632152588556e-05,
+      "loss": 2.0004,
+      "step": 2191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6544277929155315e-05,
+      "loss": 1.8509,
+      "step": 2192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.654223433242507e-05,
+      "loss": 1.9427,
+      "step": 2193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6540190735694823e-05,
+      "loss": 1.9328,
+      "step": 2194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.653814713896458e-05,
+      "loss": 2.0243,
+      "step": 2195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.653610354223433e-05,
+      "loss": 1.9072,
+      "step": 2196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6534059945504087e-05,
+      "loss": 1.8826,
+      "step": 2197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6532016348773842e-05,
+      "loss": 2.0076,
+      "step": 2198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6529972752043598e-05,
+      "loss": 1.9087,
+      "step": 2199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6527929155313354e-05,
+      "loss": 2.0179,
+      "step": 2200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6525885558583106e-05,
+      "loss": 1.9642,
+      "step": 2201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.652384196185286e-05,
+      "loss": 1.8709,
+      "step": 2202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6521798365122617e-05,
+      "loss": 1.8183,
+      "step": 2203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.651975476839237e-05,
+      "loss": 1.9233,
+      "step": 2204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6517711171662125e-05,
+      "loss": 1.8901,
+      "step": 2205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.651566757493188e-05,
+      "loss": 1.8494,
+      "step": 2206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6513623978201636e-05,
+      "loss": 1.9497,
+      "step": 2207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6511580381471392e-05,
+      "loss": 1.9043,
+      "step": 2208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6509536784741144e-05,
+      "loss": 1.8972,
+      "step": 2209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.65074931880109e-05,
+      "loss": 1.8815,
+      "step": 2210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6505449591280655e-05,
+      "loss": 1.8815,
+      "step": 2211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6503405994550408e-05,
+      "loss": 1.9505,
+      "step": 2212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6501362397820167e-05,
+      "loss": 1.8783,
+      "step": 2213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.649931880108992e-05,
+      "loss": 1.8398,
+      "step": 2214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6497275204359675e-05,
+      "loss": 2.0582,
+      "step": 2215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.649523160762943e-05,
+      "loss": 1.9125,
+      "step": 2216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6493188010899182e-05,
+      "loss": 1.9293,
+      "step": 2217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6491144414168938e-05,
+      "loss": 1.9547,
+      "step": 2218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.648910081743869e-05,
+      "loss": 1.9701,
+      "step": 2219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6487057220708446e-05,
+      "loss": 1.8372,
+      "step": 2220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6485013623978205e-05,
+      "loss": 1.9324,
+      "step": 2221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6482970027247957e-05,
+      "loss": 1.8932,
+      "step": 2222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6480926430517713e-05,
+      "loss": 1.9164,
+      "step": 2223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6478882833787465e-05,
+      "loss": 1.8556,
+      "step": 2224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.647683923705722e-05,
+      "loss": 1.9718,
+      "step": 2225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6474795640326976e-05,
+      "loss": 1.9379,
+      "step": 2226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.647275204359673e-05,
+      "loss": 1.9073,
+      "step": 2227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6470708446866488e-05,
+      "loss": 1.903,
+      "step": 2228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6468664850136243e-05,
+      "loss": 1.722,
+      "step": 2229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6466621253405996e-05,
+      "loss": 1.898,
+      "step": 2230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.646457765667575e-05,
+      "loss": 1.863,
+      "step": 2231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6462534059945503e-05,
+      "loss": 1.8471,
+      "step": 2232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.646049046321526e-05,
+      "loss": 1.9869,
+      "step": 2233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6458446866485015e-05,
+      "loss": 1.8745,
+      "step": 2234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6456403269754767e-05,
+      "loss": 1.7946,
+      "step": 2235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6454359673024526e-05,
+      "loss": 1.8451,
+      "step": 2236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6452316076294278e-05,
+      "loss": 1.9746,
+      "step": 2237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6450272479564034e-05,
+      "loss": 1.8806,
+      "step": 2238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.644822888283379e-05,
+      "loss": 2.1698,
+      "step": 2239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.644618528610354e-05,
+      "loss": 1.9639,
+      "step": 2240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6444141689373297e-05,
+      "loss": 1.9395,
+      "step": 2241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.644209809264305e-05,
+      "loss": 1.949,
+      "step": 2242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.644005449591281e-05,
+      "loss": 1.8376,
+      "step": 2243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6438010899182564e-05,
+      "loss": 1.9353,
+      "step": 2244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6435967302452316e-05,
+      "loss": 1.965,
+      "step": 2245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6433923705722072e-05,
+      "loss": 1.9242,
+      "step": 2246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6431880108991828e-05,
+      "loss": 1.9328,
+      "step": 2247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.642983651226158e-05,
+      "loss": 1.8933,
+      "step": 2248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6427792915531336e-05,
+      "loss": 1.9342,
+      "step": 2249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6425749318801088e-05,
+      "loss": 1.9439,
+      "step": 2250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6423705722070847e-05,
+      "loss": 1.8112,
+      "step": 2251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.6421662125340602e-05,
+      "loss": 2.0072,
+      "step": 2252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6419618528610355e-05,
+      "loss": 1.8895,
+      "step": 2253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.641757493188011e-05,
+      "loss": 2.0026,
+      "step": 2254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6415531335149863e-05,
+      "loss": 1.9227,
+      "step": 2255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6413487738419618e-05,
+      "loss": 1.9529,
+      "step": 2256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6411444141689374e-05,
+      "loss": 1.837,
+      "step": 2257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.640940054495913e-05,
+      "loss": 1.989,
+      "step": 2258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6407356948228885e-05,
+      "loss": 1.932,
+      "step": 2259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6405313351498637e-05,
+      "loss": 1.9278,
+      "step": 2260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6403269754768393e-05,
+      "loss": 1.9131,
+      "step": 2261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.640122615803815e-05,
+      "loss": 1.8608,
+      "step": 2262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.63991825613079e-05,
+      "loss": 2.0937,
+      "step": 2263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6397138964577657e-05,
+      "loss": 1.9812,
+      "step": 2264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6395095367847412e-05,
+      "loss": 1.9157,
+      "step": 2265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6393051771117168e-05,
+      "loss": 1.942,
+      "step": 2266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6391008174386923e-05,
+      "loss": 1.9121,
+      "step": 2267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6388964577656676e-05,
+      "loss": 1.9152,
+      "step": 2268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.638692098092643e-05,
+      "loss": 1.7612,
+      "step": 2269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6384877384196187e-05,
+      "loss": 1.8336,
+      "step": 2270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.638283378746594e-05,
+      "loss": 1.9398,
+      "step": 2271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6380790190735695e-05,
+      "loss": 1.9344,
+      "step": 2272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.637874659400545e-05,
+      "loss": 2.1251,
+      "step": 2273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6376702997275206e-05,
+      "loss": 1.8504,
+      "step": 2274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6374659400544962e-05,
+      "loss": 1.7908,
+      "step": 2275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6372615803814714e-05,
+      "loss": 1.8327,
+      "step": 2276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.637057220708447e-05,
+      "loss": 1.7763,
+      "step": 2277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6368528610354222e-05,
+      "loss": 1.839,
+      "step": 2278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6366485013623977e-05,
+      "loss": 1.8347,
+      "step": 2279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6364441416893733e-05,
+      "loss": 1.771,
+      "step": 2280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.636239782016349e-05,
+      "loss": 1.8701,
+      "step": 2281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6360354223433244e-05,
+      "loss": 1.8655,
+      "step": 2282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6358310626702997e-05,
+      "loss": 1.9063,
+      "step": 2283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6356267029972752e-05,
+      "loss": 1.799,
+      "step": 2284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6354223433242508e-05,
+      "loss": 1.9087,
+      "step": 2285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.635217983651226e-05,
+      "loss": 1.8131,
+      "step": 2286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6350136239782016e-05,
+      "loss": 1.9693,
+      "step": 2287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6348092643051775e-05,
+      "loss": 1.873,
+      "step": 2288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6346049046321527e-05,
+      "loss": 1.8486,
+      "step": 2289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6344005449591283e-05,
+      "loss": 1.8387,
+      "step": 2290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6341961852861035e-05,
+      "loss": 1.8858,
+      "step": 2291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.633991825613079e-05,
+      "loss": 1.9109,
+      "step": 2292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6337874659400546e-05,
+      "loss": 1.9165,
+      "step": 2293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.63358310626703e-05,
+      "loss": 1.9509,
+      "step": 2294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6333787465940057e-05,
+      "loss": 1.8454,
+      "step": 2295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.633174386920981e-05,
+      "loss": 1.9674,
+      "step": 2296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6329700272479565e-05,
+      "loss": 1.8758,
+      "step": 2297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.632765667574932e-05,
+      "loss": 1.8299,
+      "step": 2298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6325613079019073e-05,
+      "loss": 1.9176,
+      "step": 2299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.632356948228883e-05,
+      "loss": 1.8413,
+      "step": 2300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6321525885558584e-05,
+      "loss": 1.9154,
+      "step": 2301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6319482288828337e-05,
+      "loss": 2.0646,
+      "step": 2302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6317438692098096e-05,
+      "loss": 1.984,
+      "step": 2303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6315395095367848e-05,
+      "loss": 1.9287,
+      "step": 2304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6313351498637604e-05,
+      "loss": 1.88,
+      "step": 2305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.631130790190736e-05,
+      "loss": 1.919,
+      "step": 2306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.630926430517711e-05,
+      "loss": 1.8056,
+      "step": 2307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6307220708446867e-05,
+      "loss": 1.8833,
+      "step": 2308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.630517711171662e-05,
+      "loss": 1.9228,
+      "step": 2309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.630313351498638e-05,
+      "loss": 1.9347,
+      "step": 2310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6301089918256134e-05,
+      "loss": 1.9383,
+      "step": 2311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6299046321525886e-05,
+      "loss": 2.0705,
+      "step": 2312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6297002724795642e-05,
+      "loss": 1.7926,
+      "step": 2313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6294959128065394e-05,
+      "loss": 1.9544,
+      "step": 2314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.629291553133515e-05,
+      "loss": 2.0074,
+      "step": 2315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6290871934604905e-05,
+      "loss": 1.832,
+      "step": 2316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6288828337874658e-05,
+      "loss": 1.9617,
+      "step": 2317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6286784741144417e-05,
+      "loss": 1.9361,
+      "step": 2318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.628474114441417e-05,
+      "loss": 1.9976,
+      "step": 2319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6282697547683925e-05,
+      "loss": 1.8165,
+      "step": 2320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.628065395095368e-05,
+      "loss": 1.8214,
+      "step": 2321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6278610354223432e-05,
+      "loss": 1.7523,
+      "step": 2322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6276566757493188e-05,
+      "loss": 1.9062,
+      "step": 2323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6274523160762944e-05,
+      "loss": 1.8762,
+      "step": 2324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.62724795640327e-05,
+      "loss": 1.9461,
+      "step": 2325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6270435967302455e-05,
+      "loss": 1.8432,
+      "step": 2326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6268392370572207e-05,
+      "loss": 1.729,
+      "step": 2327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6266348773841963e-05,
+      "loss": 2.0468,
+      "step": 2328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.626430517711172e-05,
+      "loss": 1.893,
+      "step": 2329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.626226158038147e-05,
+      "loss": 1.9034,
+      "step": 2330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6260217983651226e-05,
+      "loss": 1.8751,
+      "step": 2331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.625817438692098e-05,
+      "loss": 1.7147,
+      "step": 2332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6256130790190738e-05,
+      "loss": 1.7268,
+      "step": 2333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6254087193460493e-05,
+      "loss": 1.796,
+      "step": 2334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6252043596730245e-05,
+      "loss": 1.8983,
+      "step": 2335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.625e-05,
+      "loss": 1.8304,
+      "step": 2336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6247956403269753e-05,
+      "loss": 1.865,
+      "step": 2337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.624591280653951e-05,
+      "loss": 1.8809,
+      "step": 2338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6243869209809265e-05,
+      "loss": 1.9428,
+      "step": 2339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.624182561307902e-05,
+      "loss": 1.9806,
+      "step": 2340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6239782016348776e-05,
+      "loss": 1.9409,
+      "step": 2341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.623773841961853e-05,
+      "loss": 1.902,
+      "step": 2342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6235694822888284e-05,
+      "loss": 1.8423,
+      "step": 2343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.623365122615804e-05,
+      "loss": 2.0263,
+      "step": 2344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.623160762942779e-05,
+      "loss": 1.9012,
+      "step": 2345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6229564032697547e-05,
+      "loss": 1.8417,
+      "step": 2346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6227520435967303e-05,
+      "loss": 1.8669,
+      "step": 2347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.622547683923706e-05,
+      "loss": 2.0041,
+      "step": 2348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6223433242506814e-05,
+      "loss": 1.8006,
+      "step": 2349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6221389645776566e-05,
+      "loss": 1.8197,
+      "step": 2350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6219346049046322e-05,
+      "loss": 1.9969,
+      "step": 2351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6217302452316078e-05,
+      "loss": 1.8006,
+      "step": 2352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.621525885558583e-05,
+      "loss": 2.079,
+      "step": 2353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6213215258855586e-05,
+      "loss": 1.7949,
+      "step": 2354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.621117166212534e-05,
+      "loss": 1.9527,
+      "step": 2355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6209128065395097e-05,
+      "loss": 1.8222,
+      "step": 2356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6207084468664852e-05,
+      "loss": 1.8381,
+      "step": 2357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6205040871934605e-05,
+      "loss": 1.7632,
+      "step": 2358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.620299727520436e-05,
+      "loss": 1.8852,
+      "step": 2359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6200953678474116e-05,
+      "loss": 1.9809,
+      "step": 2360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6198910081743868e-05,
+      "loss": 1.9015,
+      "step": 2361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6196866485013624e-05,
+      "loss": 1.999,
+      "step": 2362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.619482288828338e-05,
+      "loss": 1.904,
+      "step": 2363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6192779291553135e-05,
+      "loss": 1.9221,
+      "step": 2364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.619073569482289e-05,
+      "loss": 1.9176,
+      "step": 2365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6188692098092643e-05,
+      "loss": 1.8928,
+      "step": 2366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.61866485013624e-05,
+      "loss": 2.1157,
+      "step": 2367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.618460490463215e-05,
+      "loss": 1.7773,
+      "step": 2368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6182561307901906e-05,
+      "loss": 1.9043,
+      "step": 2369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6180517711171665e-05,
+      "loss": 1.8159,
+      "step": 2370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6178474114441418e-05,
+      "loss": 1.8489,
+      "step": 2371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6176430517711173e-05,
+      "loss": 2.0177,
+      "step": 2372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6174386920980926e-05,
+      "loss": 1.88,
+      "step": 2373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.617234332425068e-05,
+      "loss": 1.87,
+      "step": 2374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6170299727520437e-05,
+      "loss": 1.8244,
+      "step": 2375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.616825613079019e-05,
+      "loss": 1.9345,
+      "step": 2376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6166212534059945e-05,
+      "loss": 1.8882,
+      "step": 2377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6164168937329704e-05,
+      "loss": 1.9249,
+      "step": 2378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6162125340599456e-05,
+      "loss": 1.9242,
+      "step": 2379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.616008174386921e-05,
+      "loss": 1.9526,
+      "step": 2380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6158038147138964e-05,
+      "loss": 2.1128,
+      "step": 2381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.615599455040872e-05,
+      "loss": 1.9749,
+      "step": 2382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6153950953678475e-05,
+      "loss": 1.8161,
+      "step": 2383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6151907356948227e-05,
+      "loss": 1.9052,
+      "step": 2384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6149863760217986e-05,
+      "loss": 1.9189,
+      "step": 2385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.614782016348774e-05,
+      "loss": 1.8431,
+      "step": 2386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6145776566757494e-05,
+      "loss": 1.8106,
+      "step": 2387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.614373297002725e-05,
+      "loss": 1.7803,
+      "step": 2388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6141689373297002e-05,
+      "loss": 1.8294,
+      "step": 2389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6139645776566758e-05,
+      "loss": 1.8857,
+      "step": 2390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.613760217983651e-05,
+      "loss": 1.7991,
+      "step": 2391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.613555858310627e-05,
+      "loss": 1.8865,
+      "step": 2392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6133514986376025e-05,
+      "loss": 1.8943,
+      "step": 2393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6131471389645777e-05,
+      "loss": 1.8902,
+      "step": 2394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6129427792915533e-05,
+      "loss": 2.0102,
+      "step": 2395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6127384196185288e-05,
+      "loss": 1.7755,
+      "step": 2396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.612534059945504e-05,
+      "loss": 1.9072,
+      "step": 2397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6123297002724796e-05,
+      "loss": 1.8318,
+      "step": 2398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.612125340599455e-05,
+      "loss": 1.9581,
+      "step": 2399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6119209809264307e-05,
+      "loss": 1.8377,
+      "step": 2400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6117166212534063e-05,
+      "loss": 1.8449,
+      "step": 2401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6115122615803815e-05,
+      "loss": 2.0545,
+      "step": 2402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.611307901907357e-05,
+      "loss": 1.8388,
+      "step": 2403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6111035422343323e-05,
+      "loss": 2.0021,
+      "step": 2404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.610899182561308e-05,
+      "loss": 1.8947,
+      "step": 2405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6106948228882834e-05,
+      "loss": 1.8385,
+      "step": 2406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.610490463215259e-05,
+      "loss": 1.7292,
+      "step": 2407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6102861035422346e-05,
+      "loss": 1.8457,
+      "step": 2408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6100817438692098e-05,
+      "loss": 2.1354,
+      "step": 2409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6098773841961854e-05,
+      "loss": 1.8839,
+      "step": 2410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.609673024523161e-05,
+      "loss": 1.8601,
+      "step": 2411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.609468664850136e-05,
+      "loss": 1.8488,
+      "step": 2412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6092643051771117e-05,
+      "loss": 1.883,
+      "step": 2413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6090599455040873e-05,
+      "loss": 1.7837,
+      "step": 2414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.608855585831063e-05,
+      "loss": 1.9513,
+      "step": 2415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6086512261580384e-05,
+      "loss": 2.0024,
+      "step": 2416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6084468664850136e-05,
+      "loss": 1.8673,
+      "step": 2417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6082425068119892e-05,
+      "loss": 1.9036,
+      "step": 2418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6080381471389647e-05,
+      "loss": 1.8414,
+      "step": 2419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.60783378746594e-05,
+      "loss": 1.854,
+      "step": 2420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6076294277929155e-05,
+      "loss": 1.9856,
+      "step": 2421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.607425068119891e-05,
+      "loss": 1.9612,
+      "step": 2422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6072207084468667e-05,
+      "loss": 2.0705,
+      "step": 2423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6070163487738422e-05,
+      "loss": 1.8613,
+      "step": 2424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6068119891008174e-05,
+      "loss": 1.9696,
+      "step": 2425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.606607629427793e-05,
+      "loss": 2.0191,
+      "step": 2426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6064032697547682e-05,
+      "loss": 1.8876,
+      "step": 2427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6061989100817438e-05,
+      "loss": 2.0352,
+      "step": 2428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6059945504087194e-05,
+      "loss": 1.8811,
+      "step": 2429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.605790190735695e-05,
+      "loss": 1.833,
+      "step": 2430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6055858310626705e-05,
+      "loss": 1.8189,
+      "step": 2431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.605381471389646e-05,
+      "loss": 1.8977,
+      "step": 2432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6051771117166213e-05,
+      "loss": 1.9348,
+      "step": 2433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.604972752043597e-05,
+      "loss": 1.9366,
+      "step": 2434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.604768392370572e-05,
+      "loss": 1.8141,
+      "step": 2435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6045640326975476e-05,
+      "loss": 1.917,
+      "step": 2436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6043596730245235e-05,
+      "loss": 1.9642,
+      "step": 2437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6041553133514988e-05,
+      "loss": 1.9216,
+      "step": 2438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6039509536784743e-05,
+      "loss": 1.8766,
+      "step": 2439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6037465940054495e-05,
+      "loss": 1.9437,
+      "step": 2440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.603542234332425e-05,
+      "loss": 1.8013,
+      "step": 2441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6033378746594007e-05,
+      "loss": 1.9665,
+      "step": 2442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.603133514986376e-05,
+      "loss": 1.9224,
+      "step": 2443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6029291553133515e-05,
+      "loss": 2.0846,
+      "step": 2444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.602724795640327e-05,
+      "loss": 1.8752,
+      "step": 2445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6025204359673026e-05,
+      "loss": 1.9392,
+      "step": 2446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.602316076294278e-05,
+      "loss": 1.8152,
+      "step": 2447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6021117166212534e-05,
+      "loss": 1.9177,
+      "step": 2448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.601907356948229e-05,
+      "loss": 2.0555,
+      "step": 2449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6017029972752045e-05,
+      "loss": 1.8404,
+      "step": 2450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6014986376021797e-05,
+      "loss": 2.0146,
+      "step": 2451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6012942779291556e-05,
+      "loss": 1.8505,
+      "step": 2452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.601089918256131e-05,
+      "loss": 1.8178,
+      "step": 2453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6008855585831064e-05,
+      "loss": 1.8087,
+      "step": 2454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.600681198910082e-05,
+      "loss": 1.9129,
+      "step": 2455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6004768392370572e-05,
+      "loss": 1.8223,
+      "step": 2456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6002724795640328e-05,
+      "loss": 2.029,
+      "step": 2457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.600068119891008e-05,
+      "loss": 1.915,
+      "step": 2458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5998637602179835e-05,
+      "loss": 1.997,
+      "step": 2459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5996594005449595e-05,
+      "loss": 1.8182,
+      "step": 2460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5994550408719347e-05,
+      "loss": 1.8932,
+      "step": 2461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5992506811989102e-05,
+      "loss": 1.7912,
+      "step": 2462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5990463215258855e-05,
+      "loss": 1.9223,
+      "step": 2463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.598841961852861e-05,
+      "loss": 1.8461,
+      "step": 2464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5986376021798366e-05,
+      "loss": 1.8565,
+      "step": 2465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5984332425068118e-05,
+      "loss": 1.8947,
+      "step": 2466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5982288828337877e-05,
+      "loss": 1.8707,
+      "step": 2467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.598024523160763e-05,
+      "loss": 1.9655,
+      "step": 2468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5978201634877385e-05,
+      "loss": 1.8354,
+      "step": 2469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.597615803814714e-05,
+      "loss": 1.9054,
+      "step": 2470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5974114441416893e-05,
+      "loss": 1.9514,
+      "step": 2471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.597207084468665e-05,
+      "loss": 1.8081,
+      "step": 2472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5970027247956404e-05,
+      "loss": 1.8882,
+      "step": 2473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.596798365122616e-05,
+      "loss": 1.8025,
+      "step": 2474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5965940054495915e-05,
+      "loss": 1.8783,
+      "step": 2475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5963896457765668e-05,
+      "loss": 1.8745,
+      "step": 2476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5961852861035423e-05,
+      "loss": 1.8969,
+      "step": 2477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.595980926430518e-05,
+      "loss": 1.8956,
+      "step": 2478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.595776566757493e-05,
+      "loss": 1.8095,
+      "step": 2479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5955722070844687e-05,
+      "loss": 1.7965,
+      "step": 2480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.595367847411444e-05,
+      "loss": 1.8594,
+      "step": 2481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5951634877384198e-05,
+      "loss": 1.7829,
+      "step": 2482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5949591280653954e-05,
+      "loss": 1.8084,
+      "step": 2483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5947547683923706e-05,
+      "loss": 1.8272,
+      "step": 2484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.594550408719346e-05,
+      "loss": 1.8026,
+      "step": 2485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5943460490463217e-05,
+      "loss": 1.9027,
+      "step": 2486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.594141689373297e-05,
+      "loss": 1.9931,
+      "step": 2487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5939373297002725e-05,
+      "loss": 1.8221,
+      "step": 2488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.593732970027248e-05,
+      "loss": 1.6734,
+      "step": 2489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5935286103542236e-05,
+      "loss": 1.9689,
+      "step": 2490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5933242506811992e-05,
+      "loss": 2.0002,
+      "step": 2491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5931198910081744e-05,
+      "loss": 1.8018,
+      "step": 2492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.59291553133515e-05,
+      "loss": 1.9242,
+      "step": 2493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5927111716621252e-05,
+      "loss": 1.8535,
+      "step": 2494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5925068119891008e-05,
+      "loss": 1.9556,
+      "step": 2495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5923024523160763e-05,
+      "loss": 1.8875,
+      "step": 2496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.592098092643052e-05,
+      "loss": 1.8753,
+      "step": 2497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5918937329700275e-05,
+      "loss": 1.848,
+      "step": 2498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5916893732970027e-05,
+      "loss": 1.8722,
+      "step": 2499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5914850136239783e-05,
+      "loss": 1.9323,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 1.67998468875885,
+      "eval_runtime": 727.0892,
+      "eval_samples_per_second": 5.708,
+      "eval_steps_per_second": 1.428,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5912806539509538e-05,
+      "loss": 1.9617,
+      "step": 2501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.591076294277929e-05,
+      "loss": 1.9429,
+      "step": 2502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5908719346049046e-05,
+      "loss": 1.8485,
+      "step": 2503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.59066757493188e-05,
+      "loss": 1.9645,
+      "step": 2504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5904632152588557e-05,
+      "loss": 1.8327,
+      "step": 2505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5902588555858313e-05,
+      "loss": 1.8604,
+      "step": 2506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5900544959128065e-05,
+      "loss": 1.8216,
+      "step": 2507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.589850136239782e-05,
+      "loss": 1.9416,
+      "step": 2508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5896457765667576e-05,
+      "loss": 1.9173,
+      "step": 2509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.589441416893733e-05,
+      "loss": 2.0128,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5892370572207084e-05,
+      "loss": 1.8907,
+      "step": 2511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.589032697547684e-05,
+      "loss": 1.797,
+      "step": 2512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5888283378746596e-05,
+      "loss": 1.8204,
+      "step": 2513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.588623978201635e-05,
+      "loss": 1.9039,
+      "step": 2514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5884196185286103e-05,
+      "loss": 1.9108,
+      "step": 2515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.588215258855586e-05,
+      "loss": 1.8312,
+      "step": 2516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.588010899182561e-05,
+      "loss": 1.9786,
+      "step": 2517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5878065395095367e-05,
+      "loss": 1.8373,
+      "step": 2518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5876021798365126e-05,
+      "loss": 1.9169,
+      "step": 2519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5873978201634878e-05,
+      "loss": 1.8342,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5871934604904634e-05,
+      "loss": 1.751,
+      "step": 2521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5869891008174386e-05,
+      "loss": 1.8303,
+      "step": 2522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5867847411444142e-05,
+      "loss": 1.8876,
+      "step": 2523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5865803814713897e-05,
+      "loss": 1.8616,
+      "step": 2524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.586376021798365e-05,
+      "loss": 1.7551,
+      "step": 2525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5861716621253405e-05,
+      "loss": 1.8928,
+      "step": 2526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5859673024523164e-05,
+      "loss": 1.9565,
+      "step": 2527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5857629427792917e-05,
+      "loss": 1.795,
+      "step": 2528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5855585831062672e-05,
+      "loss": 1.8889,
+      "step": 2529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5853542234332424e-05,
+      "loss": 1.9337,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.585149863760218e-05,
+      "loss": 1.9111,
+      "step": 2531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5849455040871936e-05,
+      "loss": 1.9543,
+      "step": 2532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5847411444141688e-05,
+      "loss": 1.8911,
+      "step": 2533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5845367847411447e-05,
+      "loss": 1.828,
+      "step": 2534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.58433242506812e-05,
+      "loss": 1.8578,
+      "step": 2535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5841280653950955e-05,
+      "loss": 1.9373,
+      "step": 2536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.583923705722071e-05,
+      "loss": 2.1078,
+      "step": 2537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5837193460490463e-05,
+      "loss": 1.8524,
+      "step": 2538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.583514986376022e-05,
+      "loss": 1.9492,
+      "step": 2539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.583310626702997e-05,
+      "loss": 2.0599,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5831062670299726e-05,
+      "loss": 1.8398,
+      "step": 2541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5829019073569485e-05,
+      "loss": 1.8854,
+      "step": 2542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5826975476839237e-05,
+      "loss": 1.9396,
+      "step": 2543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5824931880108993e-05,
+      "loss": 2.0216,
+      "step": 2544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.582288828337875e-05,
+      "loss": 1.9619,
+      "step": 2545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.58208446866485e-05,
+      "loss": 1.8609,
+      "step": 2546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5818801089918257e-05,
+      "loss": 1.8734,
+      "step": 2547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.581675749318801e-05,
+      "loss": 1.815,
+      "step": 2548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5814713896457768e-05,
+      "loss": 1.895,
+      "step": 2549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5812670299727524e-05,
+      "loss": 1.8204,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5810626702997276e-05,
+      "loss": 1.8519,
+      "step": 2551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.580858310626703e-05,
+      "loss": 1.8455,
+      "step": 2552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5806539509536784e-05,
+      "loss": 1.8752,
+      "step": 2553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.580449591280654e-05,
+      "loss": 1.7654,
+      "step": 2554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5802452316076295e-05,
+      "loss": 1.8338,
+      "step": 2555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5800408719346047e-05,
+      "loss": 1.9469,
+      "step": 2556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5798365122615806e-05,
+      "loss": 1.9592,
+      "step": 2557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.579632152588556e-05,
+      "loss": 2.0329,
+      "step": 2558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5794277929155314e-05,
+      "loss": 1.7939,
+      "step": 2559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.579223433242507e-05,
+      "loss": 1.8753,
+      "step": 2560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5790190735694822e-05,
+      "loss": 1.8682,
+      "step": 2561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5788147138964578e-05,
+      "loss": 1.9335,
+      "step": 2562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5786103542234333e-05,
+      "loss": 1.9965,
+      "step": 2563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.578405994550409e-05,
+      "loss": 1.7719,
+      "step": 2564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5782016348773844e-05,
+      "loss": 1.9333,
+      "step": 2565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5779972752043597e-05,
+      "loss": 1.7876,
+      "step": 2566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5777929155313352e-05,
+      "loss": 1.8784,
+      "step": 2567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5775885558583108e-05,
+      "loss": 1.8823,
+      "step": 2568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.577384196185286e-05,
+      "loss": 2.0199,
+      "step": 2569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5771798365122616e-05,
+      "loss": 1.7865,
+      "step": 2570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.576975476839237e-05,
+      "loss": 1.8188,
+      "step": 2571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5767711171662127e-05,
+      "loss": 1.8756,
+      "step": 2572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5765667574931883e-05,
+      "loss": 1.8228,
+      "step": 2573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5763623978201635e-05,
+      "loss": 1.9266,
+      "step": 2574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.576158038147139e-05,
+      "loss": 1.8833,
+      "step": 2575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5759536784741143e-05,
+      "loss": 2.0074,
+      "step": 2576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.57574931880109e-05,
+      "loss": 1.8382,
+      "step": 2577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5755449591280654e-05,
+      "loss": 1.9262,
+      "step": 2578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.575340599455041e-05,
+      "loss": 1.8414,
+      "step": 2579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5751362397820165e-05,
+      "loss": 1.9222,
+      "step": 2580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.574931880108992e-05,
+      "loss": 1.8679,
+      "step": 2581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5747275204359673e-05,
+      "loss": 1.9469,
+      "step": 2582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.574523160762943e-05,
+      "loss": 1.884,
+      "step": 2583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.574318801089918e-05,
+      "loss": 1.7859,
+      "step": 2584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5741144414168937e-05,
+      "loss": 1.8368,
+      "step": 2585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5739100817438696e-05,
+      "loss": 1.7922,
+      "step": 2586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5737057220708448e-05,
+      "loss": 1.9704,
+      "step": 2587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5735013623978204e-05,
+      "loss": 1.7271,
+      "step": 2588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5732970027247956e-05,
+      "loss": 1.8995,
+      "step": 2589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.573092643051771e-05,
+      "loss": 1.8007,
+      "step": 2590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5728882833787467e-05,
+      "loss": 1.8523,
+      "step": 2591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.572683923705722e-05,
+      "loss": 1.7843,
+      "step": 2592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5724795640326975e-05,
+      "loss": 1.8443,
+      "step": 2593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.572275204359673e-05,
+      "loss": 1.9286,
+      "step": 2594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5720708446866486e-05,
+      "loss": 1.8814,
+      "step": 2595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5718664850136242e-05,
+      "loss": 1.875,
+      "step": 2596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5716621253405994e-05,
+      "loss": 1.9885,
+      "step": 2597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.571457765667575e-05,
+      "loss": 1.8351,
+      "step": 2598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5712534059945505e-05,
+      "loss": 1.9225,
+      "step": 2599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5710490463215258e-05,
+      "loss": 1.7978,
+      "step": 2600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5708446866485017e-05,
+      "loss": 1.7768,
+      "step": 2601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.570640326975477e-05,
+      "loss": 1.9426,
+      "step": 2602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5704359673024525e-05,
+      "loss": 1.8829,
+      "step": 2603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.570231607629428e-05,
+      "loss": 1.835,
+      "step": 2604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5700272479564032e-05,
+      "loss": 1.8856,
+      "step": 2605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5698228882833788e-05,
+      "loss": 1.9082,
+      "step": 2606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.569618528610354e-05,
+      "loss": 1.865,
+      "step": 2607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5694141689373296e-05,
+      "loss": 1.7992,
+      "step": 2608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5692098092643055e-05,
+      "loss": 1.8052,
+      "step": 2609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5690054495912807e-05,
+      "loss": 1.8203,
+      "step": 2610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5688010899182563e-05,
+      "loss": 1.8342,
+      "step": 2611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5685967302452315e-05,
+      "loss": 2.0265,
+      "step": 2612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.568392370572207e-05,
+      "loss": 1.9486,
+      "step": 2613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5681880108991826e-05,
+      "loss": 1.8235,
+      "step": 2614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.567983651226158e-05,
+      "loss": 1.8432,
+      "step": 2615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5677792915531338e-05,
+      "loss": 1.9815,
+      "step": 2616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5675749318801093e-05,
+      "loss": 1.8289,
+      "step": 2617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5673705722070846e-05,
+      "loss": 1.9065,
+      "step": 2618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.56716621253406e-05,
+      "loss": 1.9085,
+      "step": 2619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5669618528610353e-05,
+      "loss": 1.9984,
+      "step": 2620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.566757493188011e-05,
+      "loss": 1.8924,
+      "step": 2621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5665531335149865e-05,
+      "loss": 1.7157,
+      "step": 2622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5663487738419617e-05,
+      "loss": 1.7433,
+      "step": 2623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5661444141689376e-05,
+      "loss": 1.9964,
+      "step": 2624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5659400544959128e-05,
+      "loss": 1.8809,
+      "step": 2625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5657356948228884e-05,
+      "loss": 1.7833,
+      "step": 2626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.565531335149864e-05,
+      "loss": 1.8311,
+      "step": 2627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5653269754768392e-05,
+      "loss": 1.8828,
+      "step": 2628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5651226158038147e-05,
+      "loss": 1.8403,
+      "step": 2629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.56491825613079e-05,
+      "loss": 1.8136,
+      "step": 2630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.564713896457766e-05,
+      "loss": 1.9386,
+      "step": 2631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5645095367847414e-05,
+      "loss": 1.8677,
+      "step": 2632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5643051771117166e-05,
+      "loss": 1.8093,
+      "step": 2633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5641008174386922e-05,
+      "loss": 1.9178,
+      "step": 2634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5638964577656678e-05,
+      "loss": 1.801,
+      "step": 2635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.563692098092643e-05,
+      "loss": 1.914,
+      "step": 2636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5634877384196186e-05,
+      "loss": 1.8512,
+      "step": 2637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5632833787465938e-05,
+      "loss": 1.9387,
+      "step": 2638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5630790190735697e-05,
+      "loss": 1.8269,
+      "step": 2639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5628746594005453e-05,
+      "loss": 1.8072,
+      "step": 2640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5626702997275205e-05,
+      "loss": 1.9635,
+      "step": 2641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.562465940054496e-05,
+      "loss": 1.9282,
+      "step": 2642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5622615803814713e-05,
+      "loss": 1.8151,
+      "step": 2643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5620572207084468e-05,
+      "loss": 1.7964,
+      "step": 2644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5618528610354224e-05,
+      "loss": 1.789,
+      "step": 2645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.561648501362398e-05,
+      "loss": 1.8568,
+      "step": 2646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5614441416893735e-05,
+      "loss": 1.9144,
+      "step": 2647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5612397820163487e-05,
+      "loss": 1.907,
+      "step": 2648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5610354223433243e-05,
+      "loss": 1.8037,
+      "step": 2649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5608310626703e-05,
+      "loss": 1.7893,
+      "step": 2650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.560626702997275e-05,
+      "loss": 1.9002,
+      "step": 2651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5604223433242507e-05,
+      "loss": 1.9755,
+      "step": 2652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5602179836512262e-05,
+      "loss": 1.9634,
+      "step": 2653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5600136239782018e-05,
+      "loss": 1.9874,
+      "step": 2654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5598092643051773e-05,
+      "loss": 1.8107,
+      "step": 2655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5596049046321526e-05,
+      "loss": 1.8535,
+      "step": 2656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.559400544959128e-05,
+      "loss": 1.7373,
+      "step": 2657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5591961852861037e-05,
+      "loss": 1.9944,
+      "step": 2658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.558991825613079e-05,
+      "loss": 1.866,
+      "step": 2659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5587874659400545e-05,
+      "loss": 1.7492,
+      "step": 2660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.55858310626703e-05,
+      "loss": 1.888,
+      "step": 2661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5583787465940056e-05,
+      "loss": 1.8237,
+      "step": 2662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5581743869209812e-05,
+      "loss": 2.053,
+      "step": 2663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5579700272479564e-05,
+      "loss": 1.8766,
+      "step": 2664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.557765667574932e-05,
+      "loss": 1.7394,
+      "step": 2665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5575613079019072e-05,
+      "loss": 1.8374,
+      "step": 2666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5573569482288828e-05,
+      "loss": 1.8455,
+      "step": 2667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5571525885558587e-05,
+      "loss": 1.9512,
+      "step": 2668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.556948228882834e-05,
+      "loss": 1.7419,
+      "step": 2669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5567438692098094e-05,
+      "loss": 1.8374,
+      "step": 2670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.556539509536785e-05,
+      "loss": 1.9094,
+      "step": 2671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5563351498637602e-05,
+      "loss": 1.8922,
+      "step": 2672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5561307901907358e-05,
+      "loss": 1.8705,
+      "step": 2673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.555926430517711e-05,
+      "loss": 1.8552,
+      "step": 2674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5557220708446866e-05,
+      "loss": 1.7992,
+      "step": 2675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5555177111716625e-05,
+      "loss": 1.7203,
+      "step": 2676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5553133514986377e-05,
+      "loss": 1.8095,
+      "step": 2677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5551089918256133e-05,
+      "loss": 1.7923,
+      "step": 2678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5549046321525885e-05,
+      "loss": 1.8568,
+      "step": 2679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.554700272479564e-05,
+      "loss": 1.9189,
+      "step": 2680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5544959128065396e-05,
+      "loss": 1.9479,
+      "step": 2681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.554291553133515e-05,
+      "loss": 1.7222,
+      "step": 2682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5540871934604907e-05,
+      "loss": 1.8643,
+      "step": 2683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.553882833787466e-05,
+      "loss": 1.8799,
+      "step": 2684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5536784741144415e-05,
+      "loss": 1.8174,
+      "step": 2685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.553474114441417e-05,
+      "loss": 1.8019,
+      "step": 2686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5532697547683923e-05,
+      "loss": 1.8358,
+      "step": 2687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.553065395095368e-05,
+      "loss": 1.7846,
+      "step": 2688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5528610354223434e-05,
+      "loss": 1.9264,
+      "step": 2689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5526566757493187e-05,
+      "loss": 1.7946,
+      "step": 2690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5524523160762946e-05,
+      "loss": 1.8304,
+      "step": 2691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5522479564032698e-05,
+      "loss": 1.8388,
+      "step": 2692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5520435967302454e-05,
+      "loss": 1.7586,
+      "step": 2693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.551839237057221e-05,
+      "loss": 1.9512,
+      "step": 2694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.551634877384196e-05,
+      "loss": 1.8102,
+      "step": 2695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5514305177111717e-05,
+      "loss": 1.7761,
+      "step": 2696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.551226158038147e-05,
+      "loss": 1.8866,
+      "step": 2697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.551021798365123e-05,
+      "loss": 1.8434,
+      "step": 2698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5508174386920984e-05,
+      "loss": 1.7977,
+      "step": 2699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5506130790190736e-05,
+      "loss": 1.9713,
+      "step": 2700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5504087193460492e-05,
+      "loss": 1.9291,
+      "step": 2701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5502043596730244e-05,
+      "loss": 1.8023,
+      "step": 2702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.55e-05,
+      "loss": 1.8192,
+      "step": 2703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5497956403269755e-05,
+      "loss": 1.8587,
+      "step": 2704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5495912806539508e-05,
+      "loss": 1.9707,
+      "step": 2705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5493869209809267e-05,
+      "loss": 1.8104,
+      "step": 2706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.549182561307902e-05,
+      "loss": 1.8565,
+      "step": 2707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5489782016348775e-05,
+      "loss": 1.8226,
+      "step": 2708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.548773841961853e-05,
+      "loss": 1.8146,
+      "step": 2709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5485694822888282e-05,
+      "loss": 1.8899,
+      "step": 2710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5483651226158038e-05,
+      "loss": 1.7932,
+      "step": 2711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5481607629427794e-05,
+      "loss": 1.92,
+      "step": 2712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.547956403269755e-05,
+      "loss": 1.7164,
+      "step": 2713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5477520435967305e-05,
+      "loss": 1.9217,
+      "step": 2714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5475476839237057e-05,
+      "loss": 1.7339,
+      "step": 2715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5473433242506813e-05,
+      "loss": 1.8828,
+      "step": 2716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.547138964577657e-05,
+      "loss": 1.774,
+      "step": 2717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.546934604904632e-05,
+      "loss": 2.0393,
+      "step": 2718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5467302452316076e-05,
+      "loss": 1.8917,
+      "step": 2719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.546525885558583e-05,
+      "loss": 1.8303,
+      "step": 2720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5463215258855588e-05,
+      "loss": 1.7527,
+      "step": 2721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5461171662125343e-05,
+      "loss": 1.8969,
+      "step": 2722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5459128065395096e-05,
+      "loss": 1.7888,
+      "step": 2723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.545708446866485e-05,
+      "loss": 1.8802,
+      "step": 2724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5455040871934603e-05,
+      "loss": 1.7655,
+      "step": 2725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.545299727520436e-05,
+      "loss": 1.8056,
+      "step": 2726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5450953678474115e-05,
+      "loss": 1.7165,
+      "step": 2727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.544891008174387e-05,
+      "loss": 1.901,
+      "step": 2728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5446866485013626e-05,
+      "loss": 1.897,
+      "step": 2729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.544482288828338e-05,
+      "loss": 1.8546,
+      "step": 2730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5442779291553134e-05,
+      "loss": 1.9678,
+      "step": 2731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.544073569482289e-05,
+      "loss": 1.9316,
+      "step": 2732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.543869209809264e-05,
+      "loss": 1.7658,
+      "step": 2733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5436648501362397e-05,
+      "loss": 1.8823,
+      "step": 2734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5434604904632153e-05,
+      "loss": 1.894,
+      "step": 2735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.543256130790191e-05,
+      "loss": 2.0216,
+      "step": 2736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5430517711171664e-05,
+      "loss": 1.9342,
+      "step": 2737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5428474114441416e-05,
+      "loss": 2.0624,
+      "step": 2738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5426430517711172e-05,
+      "loss": 1.9613,
+      "step": 2739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5424386920980928e-05,
+      "loss": 1.8697,
+      "step": 2740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.542234332425068e-05,
+      "loss": 1.7935,
+      "step": 2741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5420299727520436e-05,
+      "loss": 1.8452,
+      "step": 2742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.541825613079019e-05,
+      "loss": 2.0176,
+      "step": 2743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5416212534059947e-05,
+      "loss": 1.9111,
+      "step": 2744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5414168937329702e-05,
+      "loss": 1.8149,
+      "step": 2745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5412125340599455e-05,
+      "loss": 1.7707,
+      "step": 2746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.541008174386921e-05,
+      "loss": 1.9152,
+      "step": 2747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5408038147138966e-05,
+      "loss": 1.8494,
+      "step": 2748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5405994550408718e-05,
+      "loss": 1.9074,
+      "step": 2749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5403950953678477e-05,
+      "loss": 1.8651,
+      "step": 2750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.540190735694823e-05,
+      "loss": 1.6887,
+      "step": 2751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5399863760217985e-05,
+      "loss": 1.8662,
+      "step": 2752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.539782016348774e-05,
+      "loss": 1.9147,
+      "step": 2753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5395776566757493e-05,
+      "loss": 1.9499,
+      "step": 2754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.539373297002725e-05,
+      "loss": 1.7528,
+      "step": 2755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5391689373297e-05,
+      "loss": 2.0174,
+      "step": 2756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5389645776566757e-05,
+      "loss": 1.8336,
+      "step": 2757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5387602179836516e-05,
+      "loss": 1.8155,
+      "step": 2758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5385558583106268e-05,
+      "loss": 1.8794,
+      "step": 2759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5383514986376023e-05,
+      "loss": 1.9686,
+      "step": 2760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5381471389645776e-05,
+      "loss": 1.9769,
+      "step": 2761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.537942779291553e-05,
+      "loss": 1.8073,
+      "step": 2762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5377384196185287e-05,
+      "loss": 1.897,
+      "step": 2763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.537534059945504e-05,
+      "loss": 1.805,
+      "step": 2764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5373297002724798e-05,
+      "loss": 1.7828,
+      "step": 2765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5371253405994554e-05,
+      "loss": 1.9121,
+      "step": 2766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5369209809264306e-05,
+      "loss": 1.7957,
+      "step": 2767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5367166212534062e-05,
+      "loss": 1.9243,
+      "step": 2768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5365122615803814e-05,
+      "loss": 1.8264,
+      "step": 2769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.536307901907357e-05,
+      "loss": 1.8099,
+      "step": 2770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5361035422343325e-05,
+      "loss": 1.9764,
+      "step": 2771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5358991825613077e-05,
+      "loss": 1.8802,
+      "step": 2772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5356948228882836e-05,
+      "loss": 1.9195,
+      "step": 2773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.535490463215259e-05,
+      "loss": 1.9557,
+      "step": 2774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5352861035422344e-05,
+      "loss": 1.8373,
+      "step": 2775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.53508174386921e-05,
+      "loss": 1.9007,
+      "step": 2776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5348773841961852e-05,
+      "loss": 1.8283,
+      "step": 2777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5346730245231608e-05,
+      "loss": 1.9547,
+      "step": 2778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.534468664850136e-05,
+      "loss": 1.8534,
+      "step": 2779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.534264305177112e-05,
+      "loss": 1.7652,
+      "step": 2780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5340599455040875e-05,
+      "loss": 1.8978,
+      "step": 2781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5338555858310627e-05,
+      "loss": 1.7201,
+      "step": 2782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5336512261580383e-05,
+      "loss": 1.8171,
+      "step": 2783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5334468664850138e-05,
+      "loss": 1.8048,
+      "step": 2784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.533242506811989e-05,
+      "loss": 1.8894,
+      "step": 2785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5330381471389646e-05,
+      "loss": 1.8045,
+      "step": 2786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.53283378746594e-05,
+      "loss": 1.7854,
+      "step": 2787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5326294277929157e-05,
+      "loss": 1.8001,
+      "step": 2788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5324250681198913e-05,
+      "loss": 1.8796,
+      "step": 2789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5322207084468665e-05,
+      "loss": 1.7601,
+      "step": 2790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.532016348773842e-05,
+      "loss": 1.8545,
+      "step": 2791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5318119891008173e-05,
+      "loss": 1.9663,
+      "step": 2792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.531607629427793e-05,
+      "loss": 1.8709,
+      "step": 2793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5314032697547684e-05,
+      "loss": 1.8595,
+      "step": 2794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.531198910081744e-05,
+      "loss": 1.9132,
+      "step": 2795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5309945504087196e-05,
+      "loss": 1.7995,
+      "step": 2796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5307901907356948e-05,
+      "loss": 1.9608,
+      "step": 2797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5305858310626704e-05,
+      "loss": 1.8484,
+      "step": 2798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.530381471389646e-05,
+      "loss": 2.0133,
+      "step": 2799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.530177111716621e-05,
+      "loss": 2.0159,
+      "step": 2800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5299727520435967e-05,
+      "loss": 1.9311,
+      "step": 2801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5297683923705723e-05,
+      "loss": 1.8775,
+      "step": 2802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.529564032697548e-05,
+      "loss": 1.8226,
+      "step": 2803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5293596730245234e-05,
+      "loss": 1.9035,
+      "step": 2804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5291553133514986e-05,
+      "loss": 1.7868,
+      "step": 2805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5289509536784742e-05,
+      "loss": 1.9028,
+      "step": 2806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5287465940054498e-05,
+      "loss": 1.8637,
+      "step": 2807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.528542234332425e-05,
+      "loss": 1.8788,
+      "step": 2808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5283378746594005e-05,
+      "loss": 1.7687,
+      "step": 2809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.528133514986376e-05,
+      "loss": 1.8413,
+      "step": 2810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5279291553133517e-05,
+      "loss": 1.6779,
+      "step": 2811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5277247956403272e-05,
+      "loss": 1.7444,
+      "step": 2812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5275204359673025e-05,
+      "loss": 1.7654,
+      "step": 2813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.527316076294278e-05,
+      "loss": 1.9372,
+      "step": 2814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5271117166212532e-05,
+      "loss": 1.8411,
+      "step": 2815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5269073569482288e-05,
+      "loss": 1.9654,
+      "step": 2816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5267029972752044e-05,
+      "loss": 1.8641,
+      "step": 2817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.52649863760218e-05,
+      "loss": 1.7706,
+      "step": 2818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5262942779291555e-05,
+      "loss": 1.8748,
+      "step": 2819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.526089918256131e-05,
+      "loss": 1.9043,
+      "step": 2820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5258855585831063e-05,
+      "loss": 1.9018,
+      "step": 2821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.525681198910082e-05,
+      "loss": 1.7878,
+      "step": 2822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.525476839237057e-05,
+      "loss": 1.9033,
+      "step": 2823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5252724795640326e-05,
+      "loss": 1.9332,
+      "step": 2824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5250681198910085e-05,
+      "loss": 1.7764,
+      "step": 2825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5248637602179838e-05,
+      "loss": 1.8972,
+      "step": 2826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5246594005449593e-05,
+      "loss": 1.8213,
+      "step": 2827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5244550408719345e-05,
+      "loss": 1.9018,
+      "step": 2828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.52425068119891e-05,
+      "loss": 1.9076,
+      "step": 2829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5240463215258857e-05,
+      "loss": 1.9152,
+      "step": 2830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.523841961852861e-05,
+      "loss": 1.8901,
+      "step": 2831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5236376021798365e-05,
+      "loss": 1.8599,
+      "step": 2832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.523433242506812e-05,
+      "loss": 1.768,
+      "step": 2833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5232288828337876e-05,
+      "loss": 1.8087,
+      "step": 2834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.523024523160763e-05,
+      "loss": 1.6986,
+      "step": 2835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5228201634877384e-05,
+      "loss": 1.8795,
+      "step": 2836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.522615803814714e-05,
+      "loss": 1.7759,
+      "step": 2837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5224114441416895e-05,
+      "loss": 1.8521,
+      "step": 2838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5222070844686647e-05,
+      "loss": 1.9218,
+      "step": 2839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5220027247956406e-05,
+      "loss": 1.8683,
+      "step": 2840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.521798365122616e-05,
+      "loss": 1.8369,
+      "step": 2841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5215940054495914e-05,
+      "loss": 1.8857,
+      "step": 2842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.521389645776567e-05,
+      "loss": 1.9086,
+      "step": 2843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5211852861035422e-05,
+      "loss": 1.8144,
+      "step": 2844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5209809264305178e-05,
+      "loss": 1.8504,
+      "step": 2845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.520776566757493e-05,
+      "loss": 1.867,
+      "step": 2846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.520572207084469e-05,
+      "loss": 1.7581,
+      "step": 2847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5203678474114445e-05,
+      "loss": 1.8705,
+      "step": 2848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5201634877384197e-05,
+      "loss": 1.8366,
+      "step": 2849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5199591280653952e-05,
+      "loss": 1.9686,
+      "step": 2850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5197547683923705e-05,
+      "loss": 1.8536,
+      "step": 2851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.519550408719346e-05,
+      "loss": 2.0031,
+      "step": 2852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5193460490463216e-05,
+      "loss": 1.8925,
+      "step": 2853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5191416893732968e-05,
+      "loss": 1.8676,
+      "step": 2854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5189373297002727e-05,
+      "loss": 1.8729,
+      "step": 2855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5187329700272483e-05,
+      "loss": 1.9466,
+      "step": 2856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5185286103542235e-05,
+      "loss": 1.8617,
+      "step": 2857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.518324250681199e-05,
+      "loss": 1.901,
+      "step": 2858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5181198910081743e-05,
+      "loss": 2.0434,
+      "step": 2859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.51791553133515e-05,
+      "loss": 1.8213,
+      "step": 2860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5177111716621254e-05,
+      "loss": 1.8913,
+      "step": 2861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.517506811989101e-05,
+      "loss": 1.8209,
+      "step": 2862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5173024523160766e-05,
+      "loss": 1.7233,
+      "step": 2863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5170980926430518e-05,
+      "loss": 1.8288,
+      "step": 2864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5168937329700273e-05,
+      "loss": 1.9103,
+      "step": 2865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516689373297003e-05,
+      "loss": 1.8278,
+      "step": 2866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516485013623978e-05,
+      "loss": 1.8144,
+      "step": 2867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5162806539509537e-05,
+      "loss": 1.7896,
+      "step": 2868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516076294277929e-05,
+      "loss": 1.8288,
+      "step": 2869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5158719346049048e-05,
+      "loss": 1.818,
+      "step": 2870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5156675749318804e-05,
+      "loss": 1.8299,
+      "step": 2871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5154632152588556e-05,
+      "loss": 1.8676,
+      "step": 2872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.515258855585831e-05,
+      "loss": 1.9178,
+      "step": 2873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5150544959128067e-05,
+      "loss": 1.889,
+      "step": 2874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.514850136239782e-05,
+      "loss": 1.7749,
+      "step": 2875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5146457765667575e-05,
+      "loss": 1.7831,
+      "step": 2876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.514441416893733e-05,
+      "loss": 1.9321,
+      "step": 2877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5142370572207086e-05,
+      "loss": 1.6993,
+      "step": 2878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5140326975476842e-05,
+      "loss": 1.7612,
+      "step": 2879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5138283378746594e-05,
+      "loss": 1.9802,
+      "step": 2880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.513623978201635e-05,
+      "loss": 1.8096,
+      "step": 2881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5134196185286102e-05,
+      "loss": 1.7925,
+      "step": 2882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5132152588555858e-05,
+      "loss": 1.8629,
+      "step": 2883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5130108991825613e-05,
+      "loss": 1.8092,
+      "step": 2884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.512806539509537e-05,
+      "loss": 1.8792,
+      "step": 2885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5126021798365125e-05,
+      "loss": 1.7581,
+      "step": 2886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5123978201634877e-05,
+      "loss": 1.8096,
+      "step": 2887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5121934604904633e-05,
+      "loss": 1.7968,
+      "step": 2888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5119891008174388e-05,
+      "loss": 1.893,
+      "step": 2889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.511784741144414e-05,
+      "loss": 1.8541,
+      "step": 2890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5115803814713896e-05,
+      "loss": 1.9517,
+      "step": 2891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5113760217983652e-05,
+      "loss": 1.8648,
+      "step": 2892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5111716621253407e-05,
+      "loss": 2.0107,
+      "step": 2893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5109673024523163e-05,
+      "loss": 1.8076,
+      "step": 2894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5107629427792915e-05,
+      "loss": 1.8038,
+      "step": 2895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.510558583106267e-05,
+      "loss": 1.8311,
+      "step": 2896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5103542234332427e-05,
+      "loss": 1.8168,
+      "step": 2897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.510149863760218e-05,
+      "loss": 1.6839,
+      "step": 2898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5099455040871934e-05,
+      "loss": 1.7981,
+      "step": 2899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.509741144414169e-05,
+      "loss": 1.9046,
+      "step": 2900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5095367847411446e-05,
+      "loss": 1.8817,
+      "step": 2901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.50933242506812e-05,
+      "loss": 1.7743,
+      "step": 2902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5091280653950954e-05,
+      "loss": 1.9157,
+      "step": 2903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.508923705722071e-05,
+      "loss": 1.7742,
+      "step": 2904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.508719346049046e-05,
+      "loss": 1.9105,
+      "step": 2905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5085149863760217e-05,
+      "loss": 1.8217,
+      "step": 2906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5083106267029976e-05,
+      "loss": 1.9171,
+      "step": 2907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.508106267029973e-05,
+      "loss": 1.9361,
+      "step": 2908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5079019073569484e-05,
+      "loss": 1.8556,
+      "step": 2909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5076975476839236e-05,
+      "loss": 1.9141,
+      "step": 2910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5074931880108992e-05,
+      "loss": 1.7849,
+      "step": 2911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5072888283378747e-05,
+      "loss": 1.8268,
+      "step": 2912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.50708446866485e-05,
+      "loss": 1.7329,
+      "step": 2913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5068801089918255e-05,
+      "loss": 1.889,
+      "step": 2914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5066757493188014e-05,
+      "loss": 2.0065,
+      "step": 2915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5064713896457767e-05,
+      "loss": 1.9588,
+      "step": 2916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5062670299727522e-05,
+      "loss": 1.9255,
+      "step": 2917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5060626702997274e-05,
+      "loss": 1.9315,
+      "step": 2918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.505858310626703e-05,
+      "loss": 1.8886,
+      "step": 2919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5056539509536786e-05,
+      "loss": 1.8549,
+      "step": 2920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5054495912806538e-05,
+      "loss": 1.9659,
+      "step": 2921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5052452316076297e-05,
+      "loss": 1.9458,
+      "step": 2922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.505040871934605e-05,
+      "loss": 1.7982,
+      "step": 2923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5048365122615805e-05,
+      "loss": 1.8138,
+      "step": 2924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.504632152588556e-05,
+      "loss": 1.802,
+      "step": 2925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5044277929155313e-05,
+      "loss": 1.802,
+      "step": 2926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.504223433242507e-05,
+      "loss": 1.909,
+      "step": 2927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.504019073569482e-05,
+      "loss": 1.8483,
+      "step": 2928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5038147138964576e-05,
+      "loss": 1.9662,
+      "step": 2929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5036103542234335e-05,
+      "loss": 1.8589,
+      "step": 2930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5034059945504088e-05,
+      "loss": 1.861,
+      "step": 2931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5032016348773843e-05,
+      "loss": 1.9404,
+      "step": 2932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.50299727520436e-05,
+      "loss": 2.0483,
+      "step": 2933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.502792915531335e-05,
+      "loss": 1.725,
+      "step": 2934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5025885558583107e-05,
+      "loss": 1.855,
+      "step": 2935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.502384196185286e-05,
+      "loss": 1.9607,
+      "step": 2936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5021798365122618e-05,
+      "loss": 1.7916,
+      "step": 2937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5019754768392374e-05,
+      "loss": 1.7753,
+      "step": 2938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5017711171662126e-05,
+      "loss": 1.8606,
+      "step": 2939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.501566757493188e-05,
+      "loss": 1.8838,
+      "step": 2940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5013623978201634e-05,
+      "loss": 1.8004,
+      "step": 2941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.501158038147139e-05,
+      "loss": 1.9075,
+      "step": 2942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5009536784741145e-05,
+      "loss": 1.9042,
+      "step": 2943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.50074931880109e-05,
+      "loss": 1.9274,
+      "step": 2944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5005449591280656e-05,
+      "loss": 1.8007,
+      "step": 2945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.500340599455041e-05,
+      "loss": 1.7755,
+      "step": 2946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5001362397820164e-05,
+      "loss": 1.8383,
+      "step": 2947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.499931880108992e-05,
+      "loss": 1.9235,
+      "step": 2948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4997275204359672e-05,
+      "loss": 1.9372,
+      "step": 2949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4995231607629428e-05,
+      "loss": 1.9353,
+      "step": 2950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4993188010899183e-05,
+      "loss": 1.8784,
+      "step": 2951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.499114441416894e-05,
+      "loss": 1.928,
+      "step": 2952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4989100817438695e-05,
+      "loss": 1.7299,
+      "step": 2953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4987057220708447e-05,
+      "loss": 1.848,
+      "step": 2954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4985013623978202e-05,
+      "loss": 1.8364,
+      "step": 2955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4982970027247958e-05,
+      "loss": 1.8999,
+      "step": 2956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.498092643051771e-05,
+      "loss": 1.8346,
+      "step": 2957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4978882833787466e-05,
+      "loss": 1.8668,
+      "step": 2958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.497683923705722e-05,
+      "loss": 1.6696,
+      "step": 2959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4974795640326977e-05,
+      "loss": 1.8661,
+      "step": 2960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4972752043596733e-05,
+      "loss": 1.847,
+      "step": 2961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4970708446866485e-05,
+      "loss": 1.8901,
+      "step": 2962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.496866485013624e-05,
+      "loss": 1.77,
+      "step": 2963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4966621253405993e-05,
+      "loss": 1.8128,
+      "step": 2964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.496457765667575e-05,
+      "loss": 1.8574,
+      "step": 2965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4962534059945504e-05,
+      "loss": 1.929,
+      "step": 2966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.496049046321526e-05,
+      "loss": 1.8235,
+      "step": 2967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4958446866485015e-05,
+      "loss": 1.8644,
+      "step": 2968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.495640326975477e-05,
+      "loss": 1.8085,
+      "step": 2969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4954359673024523e-05,
+      "loss": 1.8415,
+      "step": 2970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.495231607629428e-05,
+      "loss": 1.8371,
+      "step": 2971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.495027247956403e-05,
+      "loss": 1.826,
+      "step": 2972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4948228882833787e-05,
+      "loss": 1.7719,
+      "step": 2973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4946185286103546e-05,
+      "loss": 1.9844,
+      "step": 2974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4944141689373298e-05,
+      "loss": 1.7186,
+      "step": 2975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4942098092643054e-05,
+      "loss": 1.833,
+      "step": 2976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4940054495912806e-05,
+      "loss": 1.9223,
+      "step": 2977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.493801089918256e-05,
+      "loss": 1.8527,
+      "step": 2978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4935967302452317e-05,
+      "loss": 1.8824,
+      "step": 2979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.493392370572207e-05,
+      "loss": 1.7963,
+      "step": 2980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4931880108991825e-05,
+      "loss": 1.9538,
+      "step": 2981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.492983651226158e-05,
+      "loss": 1.9435,
+      "step": 2982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4927792915531336e-05,
+      "loss": 1.9861,
+      "step": 2983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4925749318801092e-05,
+      "loss": 1.9479,
+      "step": 2984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4923705722070844e-05,
+      "loss": 1.8601,
+      "step": 2985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.49216621253406e-05,
+      "loss": 1.8966,
+      "step": 2986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4919618528610356e-05,
+      "loss": 1.8025,
+      "step": 2987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4917574931880108e-05,
+      "loss": 1.7515,
+      "step": 2988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4915531335149867e-05,
+      "loss": 1.7881,
+      "step": 2989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.491348773841962e-05,
+      "loss": 1.7903,
+      "step": 2990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4911444141689375e-05,
+      "loss": 1.8418,
+      "step": 2991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.490940054495913e-05,
+      "loss": 1.8539,
+      "step": 2992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4907356948228883e-05,
+      "loss": 1.8715,
+      "step": 2993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4905313351498638e-05,
+      "loss": 1.8859,
+      "step": 2994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.490326975476839e-05,
+      "loss": 1.9216,
+      "step": 2995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4901226158038146e-05,
+      "loss": 1.871,
+      "step": 2996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4899182561307905e-05,
+      "loss": 1.776,
+      "step": 2997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4897138964577657e-05,
+      "loss": 1.8618,
+      "step": 2998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4895095367847413e-05,
+      "loss": 1.85,
+      "step": 2999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4893051771117165e-05,
+      "loss": 1.7587,
+      "step": 3000
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 1.6506836414337158,
+      "eval_runtime": 721.2074,
+      "eval_samples_per_second": 5.754,
+      "eval_steps_per_second": 1.439,
+      "step": 3000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.489100817438692e-05,
+      "loss": 1.7116,
+      "step": 3001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4888964577656676e-05,
+      "loss": 1.8181,
+      "step": 3002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.488692098092643e-05,
+      "loss": 1.9326,
+      "step": 3003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4884877384196188e-05,
+      "loss": 1.9362,
+      "step": 3004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4882833787465943e-05,
+      "loss": 1.7848,
+      "step": 3005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4880790190735696e-05,
+      "loss": 1.7963,
+      "step": 3006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.487874659400545e-05,
+      "loss": 1.6909,
+      "step": 3007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4876702997275203e-05,
+      "loss": 1.8044,
+      "step": 3008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.487465940054496e-05,
+      "loss": 1.9262,
+      "step": 3009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4872615803814715e-05,
+      "loss": 1.8623,
+      "step": 3010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4870572207084467e-05,
+      "loss": 1.8107,
+      "step": 3011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4868528610354226e-05,
+      "loss": 1.7362,
+      "step": 3012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4866485013623978e-05,
+      "loss": 1.8874,
+      "step": 3013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4864441416893734e-05,
+      "loss": 1.8098,
+      "step": 3014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.486239782016349e-05,
+      "loss": 1.819,
+      "step": 3015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4860354223433242e-05,
+      "loss": 1.7337,
+      "step": 3016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4858310626702997e-05,
+      "loss": 1.8319,
+      "step": 3017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.485626702997275e-05,
+      "loss": 1.9152,
+      "step": 3018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.485422343324251e-05,
+      "loss": 1.8731,
+      "step": 3019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4852179836512264e-05,
+      "loss": 1.9075,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4850136239782017e-05,
+      "loss": 1.8459,
+      "step": 3021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4848092643051772e-05,
+      "loss": 1.8213,
+      "step": 3022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4846049046321528e-05,
+      "loss": 1.8101,
+      "step": 3023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.484400544959128e-05,
+      "loss": 1.8905,
+      "step": 3024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4841961852861036e-05,
+      "loss": 1.687,
+      "step": 3025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.483991825613079e-05,
+      "loss": 2.0097,
+      "step": 3026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4837874659400547e-05,
+      "loss": 1.8661,
+      "step": 3027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4835831062670303e-05,
+      "loss": 1.9119,
+      "step": 3028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4833787465940055e-05,
+      "loss": 1.9174,
+      "step": 3029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.483174386920981e-05,
+      "loss": 1.8813,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4829700272479563e-05,
+      "loss": 1.9667,
+      "step": 3031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.482765667574932e-05,
+      "loss": 1.8762,
+      "step": 3032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4825613079019074e-05,
+      "loss": 1.7923,
+      "step": 3033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.482356948228883e-05,
+      "loss": 1.8047,
+      "step": 3034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4821525885558585e-05,
+      "loss": 1.8721,
+      "step": 3035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4819482288828337e-05,
+      "loss": 1.9865,
+      "step": 3036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4817438692098093e-05,
+      "loss": 1.9556,
+      "step": 3037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.481539509536785e-05,
+      "loss": 1.8259,
+      "step": 3038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.48133514986376e-05,
+      "loss": 1.7393,
+      "step": 3039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4811307901907357e-05,
+      "loss": 1.9519,
+      "step": 3040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4809264305177116e-05,
+      "loss": 1.7872,
+      "step": 3041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4807220708446868e-05,
+      "loss": 1.7262,
+      "step": 3042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4805177111716624e-05,
+      "loss": 1.8065,
+      "step": 3043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4803133514986376e-05,
+      "loss": 1.875,
+      "step": 3044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.480108991825613e-05,
+      "loss": 1.831,
+      "step": 3045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4799046321525887e-05,
+      "loss": 1.7864,
+      "step": 3046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.479700272479564e-05,
+      "loss": 1.8745,
+      "step": 3047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4794959128065395e-05,
+      "loss": 1.898,
+      "step": 3048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.479291553133515e-05,
+      "loss": 1.8469,
+      "step": 3049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4790871934604906e-05,
+      "loss": 1.8568,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4788828337874662e-05,
+      "loss": 1.8207,
+      "step": 3051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4786784741144414e-05,
+      "loss": 1.9303,
+      "step": 3052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.478474114441417e-05,
+      "loss": 1.9107,
+      "step": 3053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4782697547683922e-05,
+      "loss": 1.7399,
+      "step": 3054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4780653950953678e-05,
+      "loss": 1.862,
+      "step": 3055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4778610354223437e-05,
+      "loss": 1.8449,
+      "step": 3056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.477656675749319e-05,
+      "loss": 1.9275,
+      "step": 3057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4774523160762944e-05,
+      "loss": 1.8048,
+      "step": 3058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.47724795640327e-05,
+      "loss": 1.7356,
+      "step": 3059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4770435967302452e-05,
+      "loss": 1.8948,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4768392370572208e-05,
+      "loss": 1.8974,
+      "step": 3061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.476634877384196e-05,
+      "loss": 1.7642,
+      "step": 3062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4764305177111716e-05,
+      "loss": 2.0615,
+      "step": 3063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4762261580381475e-05,
+      "loss": 1.8812,
+      "step": 3064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4760217983651227e-05,
+      "loss": 1.8731,
+      "step": 3065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4758174386920983e-05,
+      "loss": 1.8441,
+      "step": 3066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4756130790190735e-05,
+      "loss": 1.8323,
+      "step": 3067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.475408719346049e-05,
+      "loss": 1.905,
+      "step": 3068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4752043596730246e-05,
+      "loss": 1.7349,
+      "step": 3069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.475e-05,
+      "loss": 1.6809,
+      "step": 3070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4747956403269758e-05,
+      "loss": 1.8629,
+      "step": 3071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.474591280653951e-05,
+      "loss": 1.793,
+      "step": 3072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4743869209809265e-05,
+      "loss": 1.8629,
+      "step": 3073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.474182561307902e-05,
+      "loss": 1.7492,
+      "step": 3074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4739782016348773e-05,
+      "loss": 1.8259,
+      "step": 3075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.473773841961853e-05,
+      "loss": 1.8835,
+      "step": 3076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4735694822888285e-05,
+      "loss": 1.868,
+      "step": 3077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4733651226158037e-05,
+      "loss": 1.9401,
+      "step": 3078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4731607629427796e-05,
+      "loss": 1.9391,
+      "step": 3079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4729564032697548e-05,
+      "loss": 1.9142,
+      "step": 3080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4727520435967304e-05,
+      "loss": 1.9089,
+      "step": 3081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.472547683923706e-05,
+      "loss": 1.7066,
+      "step": 3082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.472343324250681e-05,
+      "loss": 1.9849,
+      "step": 3083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4721389645776567e-05,
+      "loss": 1.9348,
+      "step": 3084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.471934604904632e-05,
+      "loss": 1.688,
+      "step": 3085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.471730245231608e-05,
+      "loss": 1.7234,
+      "step": 3086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4715258855585834e-05,
+      "loss": 1.9126,
+      "step": 3087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4713215258855586e-05,
+      "loss": 1.7722,
+      "step": 3088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4711171662125342e-05,
+      "loss": 2.0764,
+      "step": 3089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4709128065395094e-05,
+      "loss": 1.8411,
+      "step": 3090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.470708446866485e-05,
+      "loss": 1.7946,
+      "step": 3091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4705040871934605e-05,
+      "loss": 1.7328,
+      "step": 3092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4702997275204358e-05,
+      "loss": 1.6291,
+      "step": 3093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4700953678474117e-05,
+      "loss": 1.7385,
+      "step": 3094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.469891008174387e-05,
+      "loss": 1.7643,
+      "step": 3095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4696866485013625e-05,
+      "loss": 1.78,
+      "step": 3096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.469482288828338e-05,
+      "loss": 1.7974,
+      "step": 3097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4692779291553133e-05,
+      "loss": 1.8907,
+      "step": 3098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4690735694822888e-05,
+      "loss": 1.7517,
+      "step": 3099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4688692098092644e-05,
+      "loss": 1.7887,
+      "step": 3100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.46866485013624e-05,
+      "loss": 1.7329,
+      "step": 3101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4684604904632155e-05,
+      "loss": 1.8082,
+      "step": 3102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4682561307901907e-05,
+      "loss": 1.8099,
+      "step": 3103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4680517711171663e-05,
+      "loss": 1.8521,
+      "step": 3104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.467847411444142e-05,
+      "loss": 1.9688,
+      "step": 3105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.467643051771117e-05,
+      "loss": 2.0059,
+      "step": 3106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4674386920980926e-05,
+      "loss": 1.8899,
+      "step": 3107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.467234332425068e-05,
+      "loss": 1.9273,
+      "step": 3108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4670299727520438e-05,
+      "loss": 1.9255,
+      "step": 3109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4668256130790193e-05,
+      "loss": 1.7115,
+      "step": 3110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4666212534059946e-05,
+      "loss": 1.8807,
+      "step": 3111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.46641689373297e-05,
+      "loss": 1.7739,
+      "step": 3112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4662125340599453e-05,
+      "loss": 1.8979,
+      "step": 3113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.466008174386921e-05,
+      "loss": 1.7576,
+      "step": 3114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4658038147138965e-05,
+      "loss": 1.909,
+      "step": 3115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.465599455040872e-05,
+      "loss": 1.7934,
+      "step": 3116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4653950953678476e-05,
+      "loss": 1.8485,
+      "step": 3117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.465190735694823e-05,
+      "loss": 1.794,
+      "step": 3118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4649863760217984e-05,
+      "loss": 1.6911,
+      "step": 3119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.464782016348774e-05,
+      "loss": 1.8315,
+      "step": 3120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4645776566757492e-05,
+      "loss": 1.7525,
+      "step": 3121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4643732970027247e-05,
+      "loss": 1.8588,
+      "step": 3122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4641689373297006e-05,
+      "loss": 1.8455,
+      "step": 3123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.463964577656676e-05,
+      "loss": 1.8506,
+      "step": 3124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4637602179836514e-05,
+      "loss": 1.833,
+      "step": 3125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4635558583106267e-05,
+      "loss": 1.8965,
+      "step": 3126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4633514986376022e-05,
+      "loss": 1.8536,
+      "step": 3127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4631471389645778e-05,
+      "loss": 1.8726,
+      "step": 3128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.462942779291553e-05,
+      "loss": 1.8821,
+      "step": 3129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4627384196185286e-05,
+      "loss": 1.9232,
+      "step": 3130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.462534059945504e-05,
+      "loss": 1.8505,
+      "step": 3131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4623297002724797e-05,
+      "loss": 1.8721,
+      "step": 3132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4621253405994553e-05,
+      "loss": 1.9308,
+      "step": 3133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4619209809264305e-05,
+      "loss": 1.8899,
+      "step": 3134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.461716621253406e-05,
+      "loss": 1.749,
+      "step": 3135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4615122615803816e-05,
+      "loss": 1.8345,
+      "step": 3136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4613079019073568e-05,
+      "loss": 1.8778,
+      "step": 3137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4611035422343327e-05,
+      "loss": 1.7276,
+      "step": 3138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.460899182561308e-05,
+      "loss": 1.808,
+      "step": 3139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4606948228882835e-05,
+      "loss": 1.8045,
+      "step": 3140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.460490463215259e-05,
+      "loss": 1.7804,
+      "step": 3141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4602861035422343e-05,
+      "loss": 2.0784,
+      "step": 3142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.46008174386921e-05,
+      "loss": 1.8876,
+      "step": 3143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.459877384196185e-05,
+      "loss": 1.9376,
+      "step": 3144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4596730245231607e-05,
+      "loss": 1.8602,
+      "step": 3145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4594686648501366e-05,
+      "loss": 1.8014,
+      "step": 3146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4592643051771118e-05,
+      "loss": 2.0056,
+      "step": 3147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4590599455040873e-05,
+      "loss": 1.9064,
+      "step": 3148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4588555858310626e-05,
+      "loss": 1.9183,
+      "step": 3149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.458651226158038e-05,
+      "loss": 1.7833,
+      "step": 3150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4584468664850137e-05,
+      "loss": 1.858,
+      "step": 3151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.458242506811989e-05,
+      "loss": 1.6812,
+      "step": 3152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4580381471389648e-05,
+      "loss": 1.8755,
+      "step": 3153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4578337874659404e-05,
+      "loss": 1.6947,
+      "step": 3154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4576294277929156e-05,
+      "loss": 1.922,
+      "step": 3155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4574250681198912e-05,
+      "loss": 1.8063,
+      "step": 3156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4572207084468664e-05,
+      "loss": 1.7792,
+      "step": 3157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.457016348773842e-05,
+      "loss": 1.8836,
+      "step": 3158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4568119891008175e-05,
+      "loss": 1.8455,
+      "step": 3159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4566076294277928e-05,
+      "loss": 1.8081,
+      "step": 3160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4564032697547687e-05,
+      "loss": 1.7894,
+      "step": 3161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.456198910081744e-05,
+      "loss": 1.8917,
+      "step": 3162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4559945504087194e-05,
+      "loss": 1.7518,
+      "step": 3163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.455790190735695e-05,
+      "loss": 1.9648,
+      "step": 3164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4555858310626702e-05,
+      "loss": 1.763,
+      "step": 3165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4553814713896458e-05,
+      "loss": 1.7271,
+      "step": 3166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.455177111716621e-05,
+      "loss": 1.8136,
+      "step": 3167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.454972752043597e-05,
+      "loss": 1.8632,
+      "step": 3168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4547683923705725e-05,
+      "loss": 1.8356,
+      "step": 3169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4545640326975477e-05,
+      "loss": 1.8019,
+      "step": 3170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4543596730245233e-05,
+      "loss": 1.9934,
+      "step": 3171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.454155313351499e-05,
+      "loss": 1.7821,
+      "step": 3172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.453950953678474e-05,
+      "loss": 1.804,
+      "step": 3173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4537465940054496e-05,
+      "loss": 1.792,
+      "step": 3174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.453542234332425e-05,
+      "loss": 1.8321,
+      "step": 3175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4533378746594007e-05,
+      "loss": 1.7964,
+      "step": 3176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4531335149863763e-05,
+      "loss": 1.8054,
+      "step": 3177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4529291553133515e-05,
+      "loss": 1.9963,
+      "step": 3178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.452724795640327e-05,
+      "loss": 1.8831,
+      "step": 3179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4525204359673023e-05,
+      "loss": 1.774,
+      "step": 3180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.452316076294278e-05,
+      "loss": 1.7137,
+      "step": 3181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4521117166212535e-05,
+      "loss": 1.7315,
+      "step": 3182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.451907356948229e-05,
+      "loss": 1.9055,
+      "step": 3183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4517029972752046e-05,
+      "loss": 1.884,
+      "step": 3184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4514986376021798e-05,
+      "loss": 1.7496,
+      "step": 3185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4512942779291554e-05,
+      "loss": 1.8541,
+      "step": 3186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.451089918256131e-05,
+      "loss": 1.9012,
+      "step": 3187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.450885558583106e-05,
+      "loss": 1.7934,
+      "step": 3188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4506811989100817e-05,
+      "loss": 1.7928,
+      "step": 3189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4504768392370573e-05,
+      "loss": 1.849,
+      "step": 3190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.450272479564033e-05,
+      "loss": 1.8462,
+      "step": 3191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4500681198910084e-05,
+      "loss": 1.9703,
+      "step": 3192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4498637602179836e-05,
+      "loss": 1.7577,
+      "step": 3193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4496594005449592e-05,
+      "loss": 1.7642,
+      "step": 3194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4494550408719348e-05,
+      "loss": 1.7602,
+      "step": 3195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.44925068119891e-05,
+      "loss": 1.8179,
+      "step": 3196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4490463215258855e-05,
+      "loss": 1.7788,
+      "step": 3197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.448841961852861e-05,
+      "loss": 1.7827,
+      "step": 3198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4486376021798367e-05,
+      "loss": 1.7786,
+      "step": 3199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4484332425068122e-05,
+      "loss": 1.9406,
+      "step": 3200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4482288828337875e-05,
+      "loss": 1.9015,
+      "step": 3201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.448024523160763e-05,
+      "loss": 1.8185,
+      "step": 3202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4478201634877382e-05,
+      "loss": 1.7936,
+      "step": 3203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4476158038147138e-05,
+      "loss": 1.7529,
+      "step": 3204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4474114441416894e-05,
+      "loss": 1.905,
+      "step": 3205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.447207084468665e-05,
+      "loss": 1.7039,
+      "step": 3206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4470027247956405e-05,
+      "loss": 1.7963,
+      "step": 3207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.446798365122616e-05,
+      "loss": 1.6868,
+      "step": 3208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4465940054495913e-05,
+      "loss": 1.6965,
+      "step": 3209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.446389645776567e-05,
+      "loss": 1.8769,
+      "step": 3210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.446185286103542e-05,
+      "loss": 1.8177,
+      "step": 3211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4459809264305176e-05,
+      "loss": 1.8843,
+      "step": 3212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4457765667574935e-05,
+      "loss": 1.9269,
+      "step": 3213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4455722070844688e-05,
+      "loss": 1.7211,
+      "step": 3214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4453678474114443e-05,
+      "loss": 1.7141,
+      "step": 3215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4451634877384196e-05,
+      "loss": 1.6693,
+      "step": 3216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.444959128065395e-05,
+      "loss": 1.7011,
+      "step": 3217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4447547683923707e-05,
+      "loss": 1.7591,
+      "step": 3218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.444550408719346e-05,
+      "loss": 1.7871,
+      "step": 3219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4443460490463218e-05,
+      "loss": 1.9057,
+      "step": 3220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.444141689373297e-05,
+      "loss": 1.8484,
+      "step": 3221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4439373297002726e-05,
+      "loss": 1.7713,
+      "step": 3222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.443732970027248e-05,
+      "loss": 1.7094,
+      "step": 3223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4435286103542234e-05,
+      "loss": 1.8655,
+      "step": 3224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.443324250681199e-05,
+      "loss": 1.8904,
+      "step": 3225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4431198910081745e-05,
+      "loss": 1.8168,
+      "step": 3226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4429155313351497e-05,
+      "loss": 1.8438,
+      "step": 3227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4427111716621256e-05,
+      "loss": 1.8209,
+      "step": 3228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.442506811989101e-05,
+      "loss": 1.8279,
+      "step": 3229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4423024523160764e-05,
+      "loss": 1.8654,
+      "step": 3230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.442098092643052e-05,
+      "loss": 1.8601,
+      "step": 3231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4418937329700272e-05,
+      "loss": 1.7892,
+      "step": 3232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4416893732970028e-05,
+      "loss": 1.9361,
+      "step": 3233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.441485013623978e-05,
+      "loss": 1.842,
+      "step": 3234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.441280653950954e-05,
+      "loss": 1.8943,
+      "step": 3235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4410762942779295e-05,
+      "loss": 1.7506,
+      "step": 3236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4408719346049047e-05,
+      "loss": 1.7454,
+      "step": 3237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4406675749318802e-05,
+      "loss": 1.7536,
+      "step": 3238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4404632152588555e-05,
+      "loss": 1.78,
+      "step": 3239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.440258855585831e-05,
+      "loss": 1.7911,
+      "step": 3240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4400544959128066e-05,
+      "loss": 1.828,
+      "step": 3241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4398501362397818e-05,
+      "loss": 1.8704,
+      "step": 3242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4396457765667577e-05,
+      "loss": 1.8686,
+      "step": 3243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4394414168937333e-05,
+      "loss": 1.8799,
+      "step": 3244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4392370572207085e-05,
+      "loss": 1.8603,
+      "step": 3245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.439032697547684e-05,
+      "loss": 1.7505,
+      "step": 3246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4388283378746593e-05,
+      "loss": 1.8435,
+      "step": 3247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.438623978201635e-05,
+      "loss": 1.7902,
+      "step": 3248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4384196185286104e-05,
+      "loss": 1.8728,
+      "step": 3249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.438215258855586e-05,
+      "loss": 1.825,
+      "step": 3250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4380108991825616e-05,
+      "loss": 1.82,
+      "step": 3251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4378065395095368e-05,
+      "loss": 1.8337,
+      "step": 3252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4376021798365123e-05,
+      "loss": 1.8063,
+      "step": 3253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.437397820163488e-05,
+      "loss": 2.0284,
+      "step": 3254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.437193460490463e-05,
+      "loss": 1.868,
+      "step": 3255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4369891008174387e-05,
+      "loss": 1.7934,
+      "step": 3256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.436784741144414e-05,
+      "loss": 1.9401,
+      "step": 3257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4365803814713898e-05,
+      "loss": 1.801,
+      "step": 3258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4363760217983654e-05,
+      "loss": 1.8509,
+      "step": 3259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4361716621253406e-05,
+      "loss": 1.7842,
+      "step": 3260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4359673024523162e-05,
+      "loss": 1.9316,
+      "step": 3261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4357629427792917e-05,
+      "loss": 1.7737,
+      "step": 3262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.435558583106267e-05,
+      "loss": 1.8871,
+      "step": 3263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.4353542234332425e-05,
+      "loss": 1.8466,
+      "step": 3264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.435149863760218e-05,
+      "loss": 1.8316,
+      "step": 3265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4349455040871936e-05,
+      "loss": 1.8042,
+      "step": 3266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4347411444141692e-05,
+      "loss": 1.8299,
+      "step": 3267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4345367847411444e-05,
+      "loss": 1.7386,
+      "step": 3268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.43433242506812e-05,
+      "loss": 1.7914,
+      "step": 3269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4341280653950952e-05,
+      "loss": 1.7945,
+      "step": 3270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4339237057220708e-05,
+      "loss": 1.8659,
+      "step": 3271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4337193460490464e-05,
+      "loss": 1.7622,
+      "step": 3272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.433514986376022e-05,
+      "loss": 1.7534,
+      "step": 3273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4333106267029975e-05,
+      "loss": 1.7674,
+      "step": 3274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4331062670299727e-05,
+      "loss": 1.883,
+      "step": 3275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4329019073569483e-05,
+      "loss": 1.9574,
+      "step": 3276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4326975476839238e-05,
+      "loss": 1.9008,
+      "step": 3277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.432493188010899e-05,
+      "loss": 1.8334,
+      "step": 3278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4322888283378746e-05,
+      "loss": 1.8205,
+      "step": 3279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4320844686648502e-05,
+      "loss": 2.0137,
+      "step": 3280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4318801089918257e-05,
+      "loss": 1.793,
+      "step": 3281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4316757493188013e-05,
+      "loss": 1.8878,
+      "step": 3282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4314713896457765e-05,
+      "loss": 1.8532,
+      "step": 3283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.431267029972752e-05,
+      "loss": 1.7467,
+      "step": 3284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4310626702997277e-05,
+      "loss": 1.8673,
+      "step": 3285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.430858310626703e-05,
+      "loss": 1.897,
+      "step": 3286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4306539509536784e-05,
+      "loss": 1.6602,
+      "step": 3287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.430449591280654e-05,
+      "loss": 1.8811,
+      "step": 3288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4302452316076296e-05,
+      "loss": 1.7691,
+      "step": 3289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.430040871934605e-05,
+      "loss": 1.8068,
+      "step": 3290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4298365122615804e-05,
+      "loss": 1.7672,
+      "step": 3291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.429632152588556e-05,
+      "loss": 1.8467,
+      "step": 3292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.429427792915531e-05,
+      "loss": 1.665,
+      "step": 3293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4292234332425067e-05,
+      "loss": 1.8997,
+      "step": 3294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4290190735694826e-05,
+      "loss": 1.7869,
+      "step": 3295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.428814713896458e-05,
+      "loss": 1.7522,
+      "step": 3296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4286103542234334e-05,
+      "loss": 1.9344,
+      "step": 3297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4284059945504086e-05,
+      "loss": 2.0681,
+      "step": 3298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4282016348773842e-05,
+      "loss": 1.6938,
+      "step": 3299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4279972752043598e-05,
+      "loss": 1.8551,
+      "step": 3300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.427792915531335e-05,
+      "loss": 1.9155,
+      "step": 3301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.427588555858311e-05,
+      "loss": 1.8921,
+      "step": 3302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4273841961852864e-05,
+      "loss": 1.8516,
+      "step": 3303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4271798365122617e-05,
+      "loss": 2.0202,
+      "step": 3304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4269754768392372e-05,
+      "loss": 1.7705,
+      "step": 3305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4267711171662125e-05,
+      "loss": 1.8354,
+      "step": 3306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.426566757493188e-05,
+      "loss": 1.7607,
+      "step": 3307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4263623978201636e-05,
+      "loss": 1.9354,
+      "step": 3308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4261580381471388e-05,
+      "loss": 1.7251,
+      "step": 3309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4259536784741147e-05,
+      "loss": 1.7979,
+      "step": 3310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.42574931880109e-05,
+      "loss": 2.016,
+      "step": 3311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4255449591280655e-05,
+      "loss": 1.7996,
+      "step": 3312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.425340599455041e-05,
+      "loss": 1.7828,
+      "step": 3313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.4251362397820163e-05,
+      "loss": 1.873,
+      "step": 3314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.424931880108992e-05,
+      "loss": 1.8686,
+      "step": 3315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.424727520435967e-05,
+      "loss": 1.6985,
+      "step": 3316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.424523160762943e-05,
+      "loss": 1.7953,
+      "step": 3317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4243188010899185e-05,
+      "loss": 1.9216,
+      "step": 3318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4241144414168938e-05,
+      "loss": 1.9129,
+      "step": 3319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4239100817438693e-05,
+      "loss": 1.8651,
+      "step": 3320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.423705722070845e-05,
+      "loss": 1.9064,
+      "step": 3321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.42350136239782e-05,
+      "loss": 1.7938,
+      "step": 3322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4232970027247957e-05,
+      "loss": 1.8902,
+      "step": 3323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.423092643051771e-05,
+      "loss": 1.8279,
+      "step": 3324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4228882833787468e-05,
+      "loss": 1.8536,
+      "step": 3325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4226839237057224e-05,
+      "loss": 1.8948,
+      "step": 3326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4224795640326976e-05,
+      "loss": 1.7488,
+      "step": 3327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.422275204359673e-05,
+      "loss": 1.7407,
+      "step": 3328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4220708446866484e-05,
+      "loss": 1.7811,
+      "step": 3329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.421866485013624e-05,
+      "loss": 1.7956,
+      "step": 3330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4216621253405995e-05,
+      "loss": 1.9207,
+      "step": 3331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.421457765667575e-05,
+      "loss": 1.8741,
+      "step": 3332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4212534059945506e-05,
+      "loss": 1.8176,
+      "step": 3333
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.421049046321526e-05,
+      "loss": 1.7984,
+      "step": 3334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4208446866485014e-05,
+      "loss": 1.8038,
+      "step": 3335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.420640326975477e-05,
+      "loss": 1.792,
+      "step": 3336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4204359673024522e-05,
+      "loss": 1.822,
+      "step": 3337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4202316076294278e-05,
+      "loss": 1.8459,
+      "step": 3338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4200272479564033e-05,
+      "loss": 1.7854,
+      "step": 3339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.419822888283379e-05,
+      "loss": 1.8866,
+      "step": 3340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4196185286103545e-05,
+      "loss": 1.6968,
+      "step": 3341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4194141689373297e-05,
+      "loss": 1.7944,
+      "step": 3342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4192098092643052e-05,
+      "loss": 1.8934,
+      "step": 3343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4190054495912808e-05,
+      "loss": 1.9224,
+      "step": 3344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.418801089918256e-05,
+      "loss": 1.7928,
+      "step": 3345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4185967302452316e-05,
+      "loss": 1.9107,
+      "step": 3346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.418392370572207e-05,
+      "loss": 1.8163,
+      "step": 3347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4181880108991827e-05,
+      "loss": 1.7305,
+      "step": 3348
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4179836512261583e-05,
+      "loss": 1.8154,
+      "step": 3349
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4177792915531335e-05,
+      "loss": 1.721,
+      "step": 3350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.417574931880109e-05,
+      "loss": 1.8366,
+      "step": 3351
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4173705722070843e-05,
+      "loss": 1.8034,
+      "step": 3352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.41716621253406e-05,
+      "loss": 1.8576,
+      "step": 3353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4169618528610354e-05,
+      "loss": 1.7269,
+      "step": 3354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.416757493188011e-05,
+      "loss": 1.7726,
+      "step": 3355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4165531335149866e-05,
+      "loss": 1.8736,
+      "step": 3356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.416348773841962e-05,
+      "loss": 1.8711,
+      "step": 3357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4161444141689373e-05,
+      "loss": 1.7969,
+      "step": 3358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.415940054495913e-05,
+      "loss": 1.8495,
+      "step": 3359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.415735694822888e-05,
+      "loss": 1.8297,
+      "step": 3360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4155313351498637e-05,
+      "loss": 1.7055,
+      "step": 3361
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4153269754768396e-05,
+      "loss": 1.8355,
+      "step": 3362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4151226158038148e-05,
+      "loss": 1.8699,
+      "step": 3363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4149182561307904e-05,
+      "loss": 1.8321,
+      "step": 3364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.4147138964577656e-05,
+      "loss": 1.8082,
+      "step": 3365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.414509536784741e-05,
+      "loss": 1.6669,
+      "step": 3366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4143051771117167e-05,
+      "loss": 1.9317,
+      "step": 3367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.414100817438692e-05,
+      "loss": 1.6778,
+      "step": 3368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4138964577656675e-05,
+      "loss": 1.9033,
+      "step": 3369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.413692098092643e-05,
+      "loss": 1.7996,
+      "step": 3370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4134877384196186e-05,
+      "loss": 1.8924,
+      "step": 3371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4132833787465942e-05,
+      "loss": 1.809,
+      "step": 3372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4130790190735694e-05,
+      "loss": 1.8953,
+      "step": 3373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.412874659400545e-05,
+      "loss": 1.7659,
+      "step": 3374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4126702997275206e-05,
+      "loss": 1.8693,
+      "step": 3375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4124659400544958e-05,
+      "loss": 1.8386,
+      "step": 3376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4122615803814717e-05,
+      "loss": 1.8632,
+      "step": 3377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.412057220708447e-05,
+      "loss": 1.7108,
+      "step": 3378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4118528610354225e-05,
+      "loss": 1.7322,
+      "step": 3379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.411648501362398e-05,
+      "loss": 1.8308,
+      "step": 3380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4114441416893733e-05,
+      "loss": 1.8785,
+      "step": 3381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4112397820163488e-05,
+      "loss": 1.8485,
+      "step": 3382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.411035422343324e-05,
+      "loss": 1.7723,
+      "step": 3383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4108310626702996e-05,
+      "loss": 1.8148,
+      "step": 3384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4106267029972755e-05,
+      "loss": 1.7651,
+      "step": 3385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4104223433242507e-05,
+      "loss": 1.9042,
+      "step": 3386
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4102179836512263e-05,
+      "loss": 1.8511,
+      "step": 3387
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4100136239782015e-05,
+      "loss": 1.827,
+      "step": 3388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.409809264305177e-05,
+      "loss": 1.8104,
+      "step": 3389
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4096049046321527e-05,
+      "loss": 1.9253,
+      "step": 3390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.409400544959128e-05,
+      "loss": 1.7376,
+      "step": 3391
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4091961852861038e-05,
+      "loss": 1.729,
+      "step": 3392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4089918256130793e-05,
+      "loss": 1.8026,
+      "step": 3393
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4087874659400546e-05,
+      "loss": 1.7689,
+      "step": 3394
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.40858310626703e-05,
+      "loss": 1.9525,
+      "step": 3395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4083787465940054e-05,
+      "loss": 1.8689,
+      "step": 3396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.408174386920981e-05,
+      "loss": 1.9146,
+      "step": 3397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4079700272479565e-05,
+      "loss": 1.779,
+      "step": 3398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.407765667574932e-05,
+      "loss": 1.9209,
+      "step": 3399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4075613079019076e-05,
+      "loss": 1.7975,
+      "step": 3400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.407356948228883e-05,
+      "loss": 1.7273,
+      "step": 3401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4071525885558584e-05,
+      "loss": 1.9118,
+      "step": 3402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.406948228882834e-05,
+      "loss": 1.8081,
+      "step": 3403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4067438692098092e-05,
+      "loss": 2.0161,
+      "step": 3404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4065395095367847e-05,
+      "loss": 1.83,
+      "step": 3405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.40633514986376e-05,
+      "loss": 1.9081,
+      "step": 3406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.406130790190736e-05,
+      "loss": 1.7126,
+      "step": 3407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4059264305177114e-05,
+      "loss": 1.8767,
+      "step": 3408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4057220708446867e-05,
+      "loss": 1.752,
+      "step": 3409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4055177111716622e-05,
+      "loss": 1.7486,
+      "step": 3410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4053133514986378e-05,
+      "loss": 1.8332,
+      "step": 3411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.405108991825613e-05,
+      "loss": 1.7601,
+      "step": 3412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4049046321525886e-05,
+      "loss": 1.8823,
+      "step": 3413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.404700272479564e-05,
+      "loss": 1.8719,
+      "step": 3414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4044959128065397e-05,
+      "loss": 1.8364,
+      "step": 3415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4042915531335153e-05,
+      "loss": 1.7765,
+      "step": 3416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4040871934604905e-05,
+      "loss": 1.8041,
+      "step": 3417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.403882833787466e-05,
+      "loss": 1.8491,
+      "step": 3418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4036784741144413e-05,
+      "loss": 1.7961,
+      "step": 3419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.403474114441417e-05,
+      "loss": 1.9162,
+      "step": 3420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4032697547683924e-05,
+      "loss": 1.8096,
+      "step": 3421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.403065395095368e-05,
+      "loss": 1.7396,
+      "step": 3422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4028610354223435e-05,
+      "loss": 1.8218,
+      "step": 3423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4026566757493188e-05,
+      "loss": 1.895,
+      "step": 3424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4024523160762943e-05,
+      "loss": 1.787,
+      "step": 3425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.40224795640327e-05,
+      "loss": 1.8698,
+      "step": 3426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.402043596730245e-05,
+      "loss": 1.7689,
+      "step": 3427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4018392370572207e-05,
+      "loss": 1.7181,
+      "step": 3428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4016348773841966e-05,
+      "loss": 1.7523,
+      "step": 3429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4014305177111718e-05,
+      "loss": 1.8338,
+      "step": 3430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4012261580381474e-05,
+      "loss": 1.8186,
+      "step": 3431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4010217983651226e-05,
+      "loss": 1.8592,
+      "step": 3432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.400817438692098e-05,
+      "loss": 1.7188,
+      "step": 3433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4006130790190737e-05,
+      "loss": 1.8385,
+      "step": 3434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.400408719346049e-05,
+      "loss": 1.8107,
+      "step": 3435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4002043596730245e-05,
+      "loss": 1.6376,
+      "step": 3436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4e-05,
+      "loss": 1.8018,
+      "step": 3437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3997956403269756e-05,
+      "loss": 1.8174,
+      "step": 3438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3995912806539512e-05,
+      "loss": 1.8421,
+      "step": 3439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3993869209809264e-05,
+      "loss": 1.7766,
+      "step": 3440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.399182561307902e-05,
+      "loss": 1.8475,
+      "step": 3441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3989782016348772e-05,
+      "loss": 1.829,
+      "step": 3442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3987738419618528e-05,
+      "loss": 1.8872,
+      "step": 3443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3985694822888287e-05,
+      "loss": 1.7385,
+      "step": 3444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.398365122615804e-05,
+      "loss": 1.7572,
+      "step": 3445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3981607629427795e-05,
+      "loss": 1.9298,
+      "step": 3446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.397956403269755e-05,
+      "loss": 1.714,
+      "step": 3447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3977520435967302e-05,
+      "loss": 1.7917,
+      "step": 3448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3975476839237058e-05,
+      "loss": 1.8118,
+      "step": 3449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.397343324250681e-05,
+      "loss": 1.7579,
+      "step": 3450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3971389645776566e-05,
+      "loss": 1.746,
+      "step": 3451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3969346049046325e-05,
+      "loss": 1.7129,
+      "step": 3452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3967302452316077e-05,
+      "loss": 1.873,
+      "step": 3453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3965258855585833e-05,
+      "loss": 1.7921,
+      "step": 3454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3963215258855585e-05,
+      "loss": 1.7257,
+      "step": 3455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.396117166212534e-05,
+      "loss": 1.7458,
+      "step": 3456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3959128065395096e-05,
+      "loss": 1.8341,
+      "step": 3457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395708446866485e-05,
+      "loss": 1.8498,
+      "step": 3458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3955040871934608e-05,
+      "loss": 1.8589,
+      "step": 3459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395299727520436e-05,
+      "loss": 1.7102,
+      "step": 3460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3950953678474115e-05,
+      "loss": 1.9239,
+      "step": 3461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.394891008174387e-05,
+      "loss": 1.8304,
+      "step": 3462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3946866485013623e-05,
+      "loss": 1.9073,
+      "step": 3463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.394482288828338e-05,
+      "loss": 1.6819,
+      "step": 3464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3942779291553135e-05,
+      "loss": 1.6659,
+      "step": 3465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3940735694822887e-05,
+      "loss": 1.7515,
+      "step": 3466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3938692098092646e-05,
+      "loss": 1.7731,
+      "step": 3467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3936648501362398e-05,
+      "loss": 1.9116,
+      "step": 3468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3934604904632154e-05,
+      "loss": 1.9336,
+      "step": 3469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.393256130790191e-05,
+      "loss": 1.6984,
+      "step": 3470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.393051771117166e-05,
+      "loss": 1.8578,
+      "step": 3471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3928474114441417e-05,
+      "loss": 1.7929,
+      "step": 3472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.392643051771117e-05,
+      "loss": 1.8274,
+      "step": 3473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.392438692098093e-05,
+      "loss": 1.7705,
+      "step": 3474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3922343324250684e-05,
+      "loss": 1.7292,
+      "step": 3475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3920299727520436e-05,
+      "loss": 1.7896,
+      "step": 3476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3918256130790192e-05,
+      "loss": 1.912,
+      "step": 3477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3916212534059944e-05,
+      "loss": 1.9169,
+      "step": 3478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.39141689373297e-05,
+      "loss": 1.8561,
+      "step": 3479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3912125340599456e-05,
+      "loss": 1.7441,
+      "step": 3480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3910081743869208e-05,
+      "loss": 1.7035,
+      "step": 3481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3908038147138967e-05,
+      "loss": 1.6931,
+      "step": 3482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.390599455040872e-05,
+      "loss": 1.8632,
+      "step": 3483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3903950953678475e-05,
+      "loss": 1.8178,
+      "step": 3484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.390190735694823e-05,
+      "loss": 1.9963,
+      "step": 3485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3899863760217983e-05,
+      "loss": 1.8379,
+      "step": 3486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3897820163487738e-05,
+      "loss": 1.8252,
+      "step": 3487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3895776566757494e-05,
+      "loss": 1.7452,
+      "step": 3488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.389373297002725e-05,
+      "loss": 1.9544,
+      "step": 3489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3891689373297005e-05,
+      "loss": 1.9264,
+      "step": 3490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3889645776566757e-05,
+      "loss": 1.8933,
+      "step": 3491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3887602179836513e-05,
+      "loss": 1.8974,
+      "step": 3492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.388555858310627e-05,
+      "loss": 1.6836,
+      "step": 3493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.388351498637602e-05,
+      "loss": 1.7306,
+      "step": 3494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3881471389645776e-05,
+      "loss": 1.6415,
+      "step": 3495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3879427792915532e-05,
+      "loss": 1.894,
+      "step": 3496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3877384196185288e-05,
+      "loss": 1.9822,
+      "step": 3497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3875340599455043e-05,
+      "loss": 1.8216,
+      "step": 3498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3873297002724796e-05,
+      "loss": 1.8221,
+      "step": 3499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.387125340599455e-05,
+      "loss": 1.7239,
+      "step": 3500
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 1.631633996963501,
+      "eval_runtime": 720.9282,
+      "eval_samples_per_second": 5.756,
+      "eval_steps_per_second": 1.44,
+      "step": 3500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3869209809264304e-05,
+      "loss": 1.7102,
+      "step": 3501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.386716621253406e-05,
+      "loss": 1.7791,
+      "step": 3502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3865122615803815e-05,
+      "loss": 1.889,
+      "step": 3503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.386307901907357e-05,
+      "loss": 1.9139,
+      "step": 3504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3861035422343326e-05,
+      "loss": 1.8779,
+      "step": 3505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.385899182561308e-05,
+      "loss": 1.843,
+      "step": 3506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3856948228882834e-05,
+      "loss": 1.8194,
+      "step": 3507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.385490463215259e-05,
+      "loss": 1.8175,
+      "step": 3508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3852861035422342e-05,
+      "loss": 1.7877,
+      "step": 3509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3850817438692097e-05,
+      "loss": 1.7332,
+      "step": 3510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3848773841961856e-05,
+      "loss": 1.9967,
+      "step": 3511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.384673024523161e-05,
+      "loss": 1.6937,
+      "step": 3512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3844686648501364e-05,
+      "loss": 1.8102,
+      "step": 3513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3842643051771117e-05,
+      "loss": 1.8388,
+      "step": 3514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3840599455040872e-05,
+      "loss": 1.9073,
+      "step": 3515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3838555858310628e-05,
+      "loss": 1.7648,
+      "step": 3516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383651226158038e-05,
+      "loss": 1.8781,
+      "step": 3517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3834468664850136e-05,
+      "loss": 1.8568,
+      "step": 3518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.383242506811989e-05,
+      "loss": 1.9687,
+      "step": 3519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3830381471389647e-05,
+      "loss": 1.8294,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3828337874659403e-05,
+      "loss": 1.7116,
+      "step": 3521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3826294277929155e-05,
+      "loss": 1.8082,
+      "step": 3522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.382425068119891e-05,
+      "loss": 1.814,
+      "step": 3523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3822207084468666e-05,
+      "loss": 1.8252,
+      "step": 3524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.382016348773842e-05,
+      "loss": 1.8043,
+      "step": 3525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3818119891008177e-05,
+      "loss": 1.7907,
+      "step": 3526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.381607629427793e-05,
+      "loss": 1.7593,
+      "step": 3527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3814032697547685e-05,
+      "loss": 1.7394,
+      "step": 3528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.381198910081744e-05,
+      "loss": 1.7769,
+      "step": 3529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3809945504087193e-05,
+      "loss": 1.7121,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.380790190735695e-05,
+      "loss": 1.8653,
+      "step": 3531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.38058583106267e-05,
+      "loss": 1.8702,
+      "step": 3532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3803814713896457e-05,
+      "loss": 1.8488,
+      "step": 3533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3801771117166216e-05,
+      "loss": 1.7706,
+      "step": 3534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3799727520435968e-05,
+      "loss": 1.8146,
+      "step": 3535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3797683923705724e-05,
+      "loss": 1.845,
+      "step": 3536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3795640326975476e-05,
+      "loss": 1.7752,
+      "step": 3537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.379359673024523e-05,
+      "loss": 1.8366,
+      "step": 3538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3791553133514987e-05,
+      "loss": 1.9039,
+      "step": 3539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.378950953678474e-05,
+      "loss": 1.8102,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.37874659400545e-05,
+      "loss": 1.7961,
+      "step": 3541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3785422343324254e-05,
+      "loss": 1.7585,
+      "step": 3542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3783378746594006e-05,
+      "loss": 1.8404,
+      "step": 3543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3781335149863762e-05,
+      "loss": 1.8112,
+      "step": 3544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3779291553133514e-05,
+      "loss": 1.7312,
+      "step": 3545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.377724795640327e-05,
+      "loss": 1.7439,
+      "step": 3546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3775204359673025e-05,
+      "loss": 1.7628,
+      "step": 3547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3773160762942778e-05,
+      "loss": 1.7095,
+      "step": 3548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3771117166212537e-05,
+      "loss": 1.703,
+      "step": 3549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.376907356948229e-05,
+      "loss": 1.7464,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3767029972752044e-05,
+      "loss": 1.8508,
+      "step": 3551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.37649863760218e-05,
+      "loss": 1.7815,
+      "step": 3552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3762942779291552e-05,
+      "loss": 1.7111,
+      "step": 3553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3760899182561308e-05,
+      "loss": 1.8356,
+      "step": 3554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.375885558583106e-05,
+      "loss": 1.6771,
+      "step": 3555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.375681198910082e-05,
+      "loss": 1.9269,
+      "step": 3556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3754768392370575e-05,
+      "loss": 1.8251,
+      "step": 3557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3752724795640327e-05,
+      "loss": 1.71,
+      "step": 3558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3750681198910083e-05,
+      "loss": 1.8108,
+      "step": 3559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.374863760217984e-05,
+      "loss": 1.9022,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.374659400544959e-05,
+      "loss": 1.7118,
+      "step": 3561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3744550408719346e-05,
+      "loss": 1.8286,
+      "step": 3562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.37425068119891e-05,
+      "loss": 1.7647,
+      "step": 3563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3740463215258858e-05,
+      "loss": 1.859,
+      "step": 3564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3738419618528613e-05,
+      "loss": 1.917,
+      "step": 3565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3736376021798365e-05,
+      "loss": 1.741,
+      "step": 3566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.373433242506812e-05,
+      "loss": 1.8752,
+      "step": 3567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3732288828337873e-05,
+      "loss": 1.8013,
+      "step": 3568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.373024523160763e-05,
+      "loss": 1.7566,
+      "step": 3569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3728201634877385e-05,
+      "loss": 1.8024,
+      "step": 3570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.372615803814714e-05,
+      "loss": 1.847,
+      "step": 3571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3724114441416896e-05,
+      "loss": 1.8225,
+      "step": 3572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3722070844686648e-05,
+      "loss": 1.8438,
+      "step": 3573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3720027247956404e-05,
+      "loss": 1.7053,
+      "step": 3574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.371798365122616e-05,
+      "loss": 1.7883,
+      "step": 3575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.371594005449591e-05,
+      "loss": 1.853,
+      "step": 3576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3713896457765667e-05,
+      "loss": 1.9315,
+      "step": 3577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3711852861035426e-05,
+      "loss": 1.8156,
+      "step": 3578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.370980926430518e-05,
+      "loss": 1.8039,
+      "step": 3579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3707765667574934e-05,
+      "loss": 1.8555,
+      "step": 3580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3705722070844686e-05,
+      "loss": 1.6924,
+      "step": 3581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3703678474114442e-05,
+      "loss": 1.8951,
+      "step": 3582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3701634877384198e-05,
+      "loss": 1.8288,
+      "step": 3583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.369959128065395e-05,
+      "loss": 1.8287,
+      "step": 3584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3697547683923705e-05,
+      "loss": 1.9634,
+      "step": 3585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.369550408719346e-05,
+      "loss": 1.7396,
+      "step": 3586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3693460490463217e-05,
+      "loss": 1.7308,
+      "step": 3587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3691416893732972e-05,
+      "loss": 1.7519,
+      "step": 3588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3689373297002725e-05,
+      "loss": 1.7576,
+      "step": 3589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.368732970027248e-05,
+      "loss": 1.7943,
+      "step": 3590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3685286103542233e-05,
+      "loss": 1.7825,
+      "step": 3591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3683242506811988e-05,
+      "loss": 1.8276,
+      "step": 3592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3681198910081747e-05,
+      "loss": 1.8498,
+      "step": 3593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.36791553133515e-05,
+      "loss": 1.8554,
+      "step": 3594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3677111716621255e-05,
+      "loss": 1.9021,
+      "step": 3595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.367506811989101e-05,
+      "loss": 1.7931,
+      "step": 3596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3673024523160763e-05,
+      "loss": 1.7903,
+      "step": 3597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.367098092643052e-05,
+      "loss": 1.974,
+      "step": 3598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.366893732970027e-05,
+      "loss": 1.9823,
+      "step": 3599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3666893732970026e-05,
+      "loss": 1.842,
+      "step": 3600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3664850136239785e-05,
+      "loss": 1.7381,
+      "step": 3601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3662806539509538e-05,
+      "loss": 1.7036,
+      "step": 3602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3660762942779293e-05,
+      "loss": 1.7747,
+      "step": 3603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3658719346049046e-05,
+      "loss": 1.8194,
+      "step": 3604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.36566757493188e-05,
+      "loss": 1.7456,
+      "step": 3605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3654632152588557e-05,
+      "loss": 1.6777,
+      "step": 3606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.365258855585831e-05,
+      "loss": 1.7062,
+      "step": 3607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3650544959128068e-05,
+      "loss": 1.6811,
+      "step": 3608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.364850136239782e-05,
+      "loss": 1.8268,
+      "step": 3609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3646457765667576e-05,
+      "loss": 1.8111,
+      "step": 3610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.364441416893733e-05,
+      "loss": 1.8895,
+      "step": 3611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3642370572207084e-05,
+      "loss": 1.8043,
+      "step": 3612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.364032697547684e-05,
+      "loss": 1.8079,
+      "step": 3613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3638283378746595e-05,
+      "loss": 1.8079,
+      "step": 3614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3636239782016347e-05,
+      "loss": 1.8096,
+      "step": 3615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3634196185286106e-05,
+      "loss": 1.8449,
+      "step": 3616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.363215258855586e-05,
+      "loss": 1.8369,
+      "step": 3617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3630108991825614e-05,
+      "loss": 1.8113,
+      "step": 3618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.362806539509537e-05,
+      "loss": 1.79,
+      "step": 3619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3626021798365122e-05,
+      "loss": 1.7859,
+      "step": 3620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3623978201634878e-05,
+      "loss": 1.8529,
+      "step": 3621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.362193460490463e-05,
+      "loss": 1.8758,
+      "step": 3622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.361989100817439e-05,
+      "loss": 1.999,
+      "step": 3623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3617847411444145e-05,
+      "loss": 1.7737,
+      "step": 3624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3615803814713897e-05,
+      "loss": 1.7102,
+      "step": 3625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3613760217983653e-05,
+      "loss": 1.6565,
+      "step": 3626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3611716621253405e-05,
+      "loss": 1.8678,
+      "step": 3627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.360967302452316e-05,
+      "loss": 1.7557,
+      "step": 3628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3607629427792916e-05,
+      "loss": 1.7517,
+      "step": 3629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.360558583106267e-05,
+      "loss": 1.7632,
+      "step": 3630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3603542234332427e-05,
+      "loss": 1.9586,
+      "step": 3631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3601498637602183e-05,
+      "loss": 1.8127,
+      "step": 3632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3599455040871935e-05,
+      "loss": 1.7714,
+      "step": 3633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.359741144414169e-05,
+      "loss": 1.7496,
+      "step": 3634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3595367847411443e-05,
+      "loss": 1.8376,
+      "step": 3635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.35933242506812e-05,
+      "loss": 1.9824,
+      "step": 3636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3591280653950954e-05,
+      "loss": 1.7409,
+      "step": 3637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.358923705722071e-05,
+      "loss": 1.8136,
+      "step": 3638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3587193460490466e-05,
+      "loss": 1.7067,
+      "step": 3639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3585149863760218e-05,
+      "loss": 1.8098,
+      "step": 3640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3583106267029973e-05,
+      "loss": 1.8537,
+      "step": 3641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.358106267029973e-05,
+      "loss": 1.8406,
+      "step": 3642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.357901907356948e-05,
+      "loss": 1.7994,
+      "step": 3643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3576975476839237e-05,
+      "loss": 1.8725,
+      "step": 3644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.357493188010899e-05,
+      "loss": 1.8818,
+      "step": 3645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3572888283378748e-05,
+      "loss": 1.7631,
+      "step": 3646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3570844686648504e-05,
+      "loss": 1.7927,
+      "step": 3647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3568801089918256e-05,
+      "loss": 1.8923,
+      "step": 3648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3566757493188012e-05,
+      "loss": 1.779,
+      "step": 3649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3564713896457767e-05,
+      "loss": 1.9571,
+      "step": 3650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.356267029972752e-05,
+      "loss": 1.6929,
+      "step": 3651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3560626702997275e-05,
+      "loss": 1.7585,
+      "step": 3652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.355858310626703e-05,
+      "loss": 1.796,
+      "step": 3653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3556539509536787e-05,
+      "loss": 1.8593,
+      "step": 3654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3554495912806542e-05,
+      "loss": 1.8778,
+      "step": 3655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3552452316076294e-05,
+      "loss": 1.8918,
+      "step": 3656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.355040871934605e-05,
+      "loss": 1.828,
+      "step": 3657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3548365122615802e-05,
+      "loss": 1.889,
+      "step": 3658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3546321525885558e-05,
+      "loss": 1.6811,
+      "step": 3659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3544277929155314e-05,
+      "loss": 1.7304,
+      "step": 3660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.354223433242507e-05,
+      "loss": 1.8937,
+      "step": 3661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3540190735694825e-05,
+      "loss": 1.8905,
+      "step": 3662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3538147138964577e-05,
+      "loss": 1.8209,
+      "step": 3663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3536103542234333e-05,
+      "loss": 1.8575,
+      "step": 3664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.353405994550409e-05,
+      "loss": 1.7782,
+      "step": 3665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.353201634877384e-05,
+      "loss": 1.8034,
+      "step": 3666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3529972752043596e-05,
+      "loss": 1.8524,
+      "step": 3667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3527929155313352e-05,
+      "loss": 1.8234,
+      "step": 3668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.3525885558583107e-05,
+      "loss": 1.8151,
+      "step": 3669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3523841961852863e-05,
+      "loss": 1.7985,
+      "step": 3670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3521798365122615e-05,
+      "loss": 1.8143,
+      "step": 3671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.351975476839237e-05,
+      "loss": 1.8426,
+      "step": 3672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3517711171662127e-05,
+      "loss": 1.9436,
+      "step": 3673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.351566757493188e-05,
+      "loss": 1.7735,
+      "step": 3674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3513623978201638e-05,
+      "loss": 1.8158,
+      "step": 3675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.351158038147139e-05,
+      "loss": 1.7879,
+      "step": 3676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3509536784741146e-05,
+      "loss": 1.6754,
+      "step": 3677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.35074931880109e-05,
+      "loss": 1.7431,
+      "step": 3678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3505449591280654e-05,
+      "loss": 1.78,
+      "step": 3679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.350340599455041e-05,
+      "loss": 1.8702,
+      "step": 3680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.350136239782016e-05,
+      "loss": 1.8819,
+      "step": 3681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3499318801089917e-05,
+      "loss": 1.752,
+      "step": 3682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3497275204359676e-05,
+      "loss": 1.7702,
+      "step": 3683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.349523160762943e-05,
+      "loss": 1.7789,
+      "step": 3684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3493188010899184e-05,
+      "loss": 1.7473,
+      "step": 3685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3491144414168936e-05,
+      "loss": 1.6978,
+      "step": 3686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3489100817438692e-05,
+      "loss": 1.8374,
+      "step": 3687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3487057220708448e-05,
+      "loss": 1.9496,
+      "step": 3688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.34850136239782e-05,
+      "loss": 1.741,
+      "step": 3689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.348297002724796e-05,
+      "loss": 1.8187,
+      "step": 3690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3480926430517714e-05,
+      "loss": 1.8136,
+      "step": 3691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3478882833787467e-05,
+      "loss": 1.796,
+      "step": 3692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3476839237057222e-05,
+      "loss": 1.7805,
+      "step": 3693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3474795640326975e-05,
+      "loss": 1.8254,
+      "step": 3694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.347275204359673e-05,
+      "loss": 1.838,
+      "step": 3695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3470708446866486e-05,
+      "loss": 1.9198,
+      "step": 3696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3468664850136238e-05,
+      "loss": 1.7636,
+      "step": 3697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3466621253405997e-05,
+      "loss": 1.7958,
+      "step": 3698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.346457765667575e-05,
+      "loss": 1.8104,
+      "step": 3699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3462534059945505e-05,
+      "loss": 1.7842,
+      "step": 3700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.346049046321526e-05,
+      "loss": 1.6538,
+      "step": 3701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3458446866485013e-05,
+      "loss": 1.6851,
+      "step": 3702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.345640326975477e-05,
+      "loss": 1.7961,
+      "step": 3703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.345435967302452e-05,
+      "loss": 1.7847,
+      "step": 3704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.345231607629428e-05,
+      "loss": 1.8237,
+      "step": 3705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3450272479564035e-05,
+      "loss": 1.7531,
+      "step": 3706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3448228882833788e-05,
+      "loss": 1.9107,
+      "step": 3707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3446185286103543e-05,
+      "loss": 1.7883,
+      "step": 3708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.34441416893733e-05,
+      "loss": 1.9039,
+      "step": 3709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.344209809264305e-05,
+      "loss": 1.7557,
+      "step": 3710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3440054495912807e-05,
+      "loss": 1.782,
+      "step": 3711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.343801089918256e-05,
+      "loss": 1.9047,
+      "step": 3712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3435967302452318e-05,
+      "loss": 1.7592,
+      "step": 3713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3433923705722074e-05,
+      "loss": 1.7371,
+      "step": 3714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3431880108991826e-05,
+      "loss": 1.7764,
+      "step": 3715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.342983651226158e-05,
+      "loss": 1.8446,
+      "step": 3716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3427792915531334e-05,
+      "loss": 1.97,
+      "step": 3717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.342574931880109e-05,
+      "loss": 1.6736,
+      "step": 3718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.3423705722070845e-05,
+      "loss": 1.7666,
+      "step": 3719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.34216621253406e-05,
+      "loss": 1.9497,
+      "step": 3720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3419618528610356e-05,
+      "loss": 1.7774,
+      "step": 3721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.341757493188011e-05,
+      "loss": 1.902,
+      "step": 3722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3415531335149864e-05,
+      "loss": 1.9015,
+      "step": 3723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.341348773841962e-05,
+      "loss": 1.8074,
+      "step": 3724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3411444141689372e-05,
+      "loss": 1.7603,
+      "step": 3725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3409400544959128e-05,
+      "loss": 1.7427,
+      "step": 3726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3407356948228883e-05,
+      "loss": 1.7169,
+      "step": 3727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.340531335149864e-05,
+      "loss": 1.8041,
+      "step": 3728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3403269754768395e-05,
+      "loss": 1.7677,
+      "step": 3729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3401226158038147e-05,
+      "loss": 1.9268,
+      "step": 3730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3399182561307903e-05,
+      "loss": 1.7828,
+      "step": 3731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3397138964577658e-05,
+      "loss": 1.8066,
+      "step": 3732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.339509536784741e-05,
+      "loss": 1.8427,
+      "step": 3733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3393051771117166e-05,
+      "loss": 1.7174,
+      "step": 3734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.339100817438692e-05,
+      "loss": 1.8366,
+      "step": 3735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3388964577656677e-05,
+      "loss": 1.8801,
+      "step": 3736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3386920980926433e-05,
+      "loss": 1.8168,
+      "step": 3737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3384877384196185e-05,
+      "loss": 1.7791,
+      "step": 3738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.338283378746594e-05,
+      "loss": 1.8427,
+      "step": 3739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3380790190735693e-05,
+      "loss": 1.6446,
+      "step": 3740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.337874659400545e-05,
+      "loss": 1.7099,
+      "step": 3741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3376702997275204e-05,
+      "loss": 1.7561,
+      "step": 3742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.337465940054496e-05,
+      "loss": 1.7876,
+      "step": 3743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3372615803814716e-05,
+      "loss": 1.8049,
+      "step": 3744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.337057220708447e-05,
+      "loss": 1.6425,
+      "step": 3745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3368528610354223e-05,
+      "loss": 1.7666,
+      "step": 3746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.336648501362398e-05,
+      "loss": 1.8531,
+      "step": 3747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.336444141689373e-05,
+      "loss": 1.8149,
+      "step": 3748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3362397820163487e-05,
+      "loss": 1.9529,
+      "step": 3749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3360354223433246e-05,
+      "loss": 1.8187,
+      "step": 3750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3358310626702998e-05,
+      "loss": 1.8421,
+      "step": 3751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3356267029972754e-05,
+      "loss": 1.7634,
+      "step": 3752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3354223433242506e-05,
+      "loss": 1.8457,
+      "step": 3753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3352179836512262e-05,
+      "loss": 1.7864,
+      "step": 3754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3350136239782017e-05,
+      "loss": 1.8187,
+      "step": 3755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.334809264305177e-05,
+      "loss": 1.8321,
+      "step": 3756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3346049046321525e-05,
+      "loss": 1.8252,
+      "step": 3757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.334400544959128e-05,
+      "loss": 1.9303,
+      "step": 3758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3341961852861037e-05,
+      "loss": 1.932,
+      "step": 3759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3339918256130792e-05,
+      "loss": 1.6768,
+      "step": 3760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3337874659400544e-05,
+      "loss": 1.7851,
+      "step": 3761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.33358310626703e-05,
+      "loss": 1.7636,
+      "step": 3762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3333787465940056e-05,
+      "loss": 1.8553,
+      "step": 3763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3331743869209808e-05,
+      "loss": 1.7393,
+      "step": 3764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3329700272479567e-05,
+      "loss": 1.7869,
+      "step": 3765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.332765667574932e-05,
+      "loss": 1.9184,
+      "step": 3766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3325613079019075e-05,
+      "loss": 1.8132,
+      "step": 3767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.332356948228883e-05,
+      "loss": 1.8873,
+      "step": 3768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3321525885558583e-05,
+      "loss": 1.7589,
+      "step": 3769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3319482288828338e-05,
+      "loss": 1.8336,
+      "step": 3770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.331743869209809e-05,
+      "loss": 1.8702,
+      "step": 3771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.331539509536785e-05,
+      "loss": 1.8836,
+      "step": 3772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3313351498637605e-05,
+      "loss": 1.6069,
+      "step": 3773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3311307901907357e-05,
+      "loss": 1.6792,
+      "step": 3774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3309264305177113e-05,
+      "loss": 1.7515,
+      "step": 3775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3307220708446865e-05,
+      "loss": 1.8592,
+      "step": 3776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.330517711171662e-05,
+      "loss": 1.741,
+      "step": 3777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3303133514986377e-05,
+      "loss": 1.9923,
+      "step": 3778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.330108991825613e-05,
+      "loss": 1.9138,
+      "step": 3779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3299046321525888e-05,
+      "loss": 1.7984,
+      "step": 3780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3297002724795643e-05,
+      "loss": 1.8487,
+      "step": 3781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3294959128065396e-05,
+      "loss": 1.77,
+      "step": 3782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.329291553133515e-05,
+      "loss": 1.7983,
+      "step": 3783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3290871934604904e-05,
+      "loss": 1.8109,
+      "step": 3784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.328882833787466e-05,
+      "loss": 1.7518,
+      "step": 3785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3286784741144415e-05,
+      "loss": 1.856,
+      "step": 3786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.328474114441417e-05,
+      "loss": 1.8535,
+      "step": 3787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3282697547683926e-05,
+      "loss": 1.7878,
+      "step": 3788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.328065395095368e-05,
+      "loss": 1.8604,
+      "step": 3789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3278610354223434e-05,
+      "loss": 1.8047,
+      "step": 3790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.327656675749319e-05,
+      "loss": 1.7588,
+      "step": 3791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3274523160762942e-05,
+      "loss": 1.7518,
+      "step": 3792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3272479564032698e-05,
+      "loss": 1.8381,
+      "step": 3793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.327043596730245e-05,
+      "loss": 1.827,
+      "step": 3794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.326839237057221e-05,
+      "loss": 1.8296,
+      "step": 3795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3266348773841964e-05,
+      "loss": 1.7427,
+      "step": 3796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3264305177111717e-05,
+      "loss": 1.8068,
+      "step": 3797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3262261580381472e-05,
+      "loss": 1.8295,
+      "step": 3798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3260217983651228e-05,
+      "loss": 1.7689,
+      "step": 3799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.325817438692098e-05,
+      "loss": 1.5907,
+      "step": 3800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3256130790190736e-05,
+      "loss": 1.7616,
+      "step": 3801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.325408719346049e-05,
+      "loss": 1.8346,
+      "step": 3802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3252043596730247e-05,
+      "loss": 1.6732,
+      "step": 3803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 1.7975,
+      "step": 3804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3247956403269755e-05,
+      "loss": 1.8354,
+      "step": 3805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.324591280653951e-05,
+      "loss": 1.9268,
+      "step": 3806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3243869209809263e-05,
+      "loss": 1.8783,
+      "step": 3807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.324182561307902e-05,
+      "loss": 1.7895,
+      "step": 3808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3239782016348774e-05,
+      "loss": 1.8663,
+      "step": 3809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.323773841961853e-05,
+      "loss": 1.7281,
+      "step": 3810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3235694822888285e-05,
+      "loss": 1.7944,
+      "step": 3811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3233651226158038e-05,
+      "loss": 1.8367,
+      "step": 3812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3231607629427793e-05,
+      "loss": 1.8306,
+      "step": 3813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.322956403269755e-05,
+      "loss": 1.7323,
+      "step": 3814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.32275204359673e-05,
+      "loss": 1.7928,
+      "step": 3815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3225476839237057e-05,
+      "loss": 1.7504,
+      "step": 3816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3223433242506816e-05,
+      "loss": 1.709,
+      "step": 3817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3221389645776568e-05,
+      "loss": 1.8092,
+      "step": 3818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3219346049046324e-05,
+      "loss": 1.9093,
+      "step": 3819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3217302452316076e-05,
+      "loss": 1.7755,
+      "step": 3820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.321525885558583e-05,
+      "loss": 1.8604,
+      "step": 3821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3213215258855587e-05,
+      "loss": 1.7269,
+      "step": 3822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.321117166212534e-05,
+      "loss": 1.7594,
+      "step": 3823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3209128065395095e-05,
+      "loss": 1.7205,
+      "step": 3824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.320708446866485e-05,
+      "loss": 1.7867,
+      "step": 3825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3205040871934606e-05,
+      "loss": 1.8366,
+      "step": 3826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3202997275204362e-05,
+      "loss": 1.7975,
+      "step": 3827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3200953678474114e-05,
+      "loss": 1.7772,
+      "step": 3828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.319891008174387e-05,
+      "loss": 1.8226,
+      "step": 3829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3196866485013622e-05,
+      "loss": 1.7376,
+      "step": 3830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3194822888283378e-05,
+      "loss": 1.7576,
+      "step": 3831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3192779291553137e-05,
+      "loss": 1.7371,
+      "step": 3832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.319073569482289e-05,
+      "loss": 1.7365,
+      "step": 3833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3188692098092645e-05,
+      "loss": 1.9082,
+      "step": 3834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.31866485013624e-05,
+      "loss": 1.8258,
+      "step": 3835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3184604904632152e-05,
+      "loss": 1.8034,
+      "step": 3836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3182561307901908e-05,
+      "loss": 1.8024,
+      "step": 3837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.318051771117166e-05,
+      "loss": 1.7018,
+      "step": 3838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3178474114441416e-05,
+      "loss": 1.6182,
+      "step": 3839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3176430517711175e-05,
+      "loss": 1.8342,
+      "step": 3840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3174386920980927e-05,
+      "loss": 1.8406,
+      "step": 3841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3172343324250683e-05,
+      "loss": 1.8192,
+      "step": 3842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3170299727520435e-05,
+      "loss": 1.7442,
+      "step": 3843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.316825613079019e-05,
+      "loss": 1.8927,
+      "step": 3844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3166212534059946e-05,
+      "loss": 1.9245,
+      "step": 3845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.31641689373297e-05,
+      "loss": 1.8474,
+      "step": 3846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3162125340599458e-05,
+      "loss": 1.8928,
+      "step": 3847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.316008174386921e-05,
+      "loss": 1.7336,
+      "step": 3848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3158038147138966e-05,
+      "loss": 1.7885,
+      "step": 3849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.315599455040872e-05,
+      "loss": 1.8236,
+      "step": 3850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3153950953678473e-05,
+      "loss": 1.9195,
+      "step": 3851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.315190735694823e-05,
+      "loss": 1.8603,
+      "step": 3852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3149863760217985e-05,
+      "loss": 1.8231,
+      "step": 3853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.314782016348774e-05,
+      "loss": 1.7187,
+      "step": 3854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3145776566757496e-05,
+      "loss": 1.8616,
+      "step": 3855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3143732970027248e-05,
+      "loss": 1.9118,
+      "step": 3856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3141689373297004e-05,
+      "loss": 1.5976,
+      "step": 3857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.313964577656676e-05,
+      "loss": 1.7448,
+      "step": 3858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.313760217983651e-05,
+      "loss": 1.7167,
+      "step": 3859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3135558583106267e-05,
+      "loss": 1.7298,
+      "step": 3860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.313351498637602e-05,
+      "loss": 1.7179,
+      "step": 3861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.313147138964578e-05,
+      "loss": 1.7655,
+      "step": 3862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3129427792915534e-05,
+      "loss": 1.6836,
+      "step": 3863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3127384196185286e-05,
+      "loss": 1.7207,
+      "step": 3864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3125340599455042e-05,
+      "loss": 1.8134,
+      "step": 3865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3123297002724794e-05,
+      "loss": 1.6948,
+      "step": 3866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.312125340599455e-05,
+      "loss": 1.8624,
+      "step": 3867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3119209809264306e-05,
+      "loss": 1.7733,
+      "step": 3868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.311716621253406e-05,
+      "loss": 1.7623,
+      "step": 3869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3115122615803817e-05,
+      "loss": 1.8124,
+      "step": 3870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.311307901907357e-05,
+      "loss": 1.7831,
+      "step": 3871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3111035422343325e-05,
+      "loss": 1.8094,
+      "step": 3872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.310899182561308e-05,
+      "loss": 1.6836,
+      "step": 3873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3106948228882833e-05,
+      "loss": 1.8186,
+      "step": 3874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3104904632152588e-05,
+      "loss": 1.8119,
+      "step": 3875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3102861035422344e-05,
+      "loss": 1.9037,
+      "step": 3876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.31008174386921e-05,
+      "loss": 1.8663,
+      "step": 3877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3098773841961855e-05,
+      "loss": 1.7766,
+      "step": 3878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3096730245231607e-05,
+      "loss": 1.8986,
+      "step": 3879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3094686648501363e-05,
+      "loss": 1.8038,
+      "step": 3880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.309264305177112e-05,
+      "loss": 1.7858,
+      "step": 3881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.309059945504087e-05,
+      "loss": 1.8011,
+      "step": 3882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3088555858310627e-05,
+      "loss": 1.7186,
+      "step": 3883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3086512261580382e-05,
+      "loss": 1.8981,
+      "step": 3884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3084468664850138e-05,
+      "loss": 1.9932,
+      "step": 3885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3082425068119893e-05,
+      "loss": 1.8862,
+      "step": 3886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3080381471389646e-05,
+      "loss": 1.8966,
+      "step": 3887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.30783378746594e-05,
+      "loss": 1.8673,
+      "step": 3888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3076294277929154e-05,
+      "loss": 1.7393,
+      "step": 3889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.307425068119891e-05,
+      "loss": 1.794,
+      "step": 3890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3072207084468665e-05,
+      "loss": 1.9616,
+      "step": 3891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.307016348773842e-05,
+      "loss": 1.6439,
+      "step": 3892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3068119891008176e-05,
+      "loss": 1.9642,
+      "step": 3893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3066076294277932e-05,
+      "loss": 1.6833,
+      "step": 3894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3064032697547684e-05,
+      "loss": 1.9059,
+      "step": 3895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.306198910081744e-05,
+      "loss": 1.8133,
+      "step": 3896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3059945504087192e-05,
+      "loss": 1.835,
+      "step": 3897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3057901907356947e-05,
+      "loss": 1.7584,
+      "step": 3898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3055858310626706e-05,
+      "loss": 1.6064,
+      "step": 3899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.305381471389646e-05,
+      "loss": 1.7272,
+      "step": 3900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3051771117166214e-05,
+      "loss": 1.8274,
+      "step": 3901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3049727520435967e-05,
+      "loss": 1.6367,
+      "step": 3902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3047683923705722e-05,
+      "loss": 1.8469,
+      "step": 3903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3045640326975478e-05,
+      "loss": 1.8618,
+      "step": 3904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.304359673024523e-05,
+      "loss": 1.8493,
+      "step": 3905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3041553133514986e-05,
+      "loss": 1.6918,
+      "step": 3906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.303950953678474e-05,
+      "loss": 1.8119,
+      "step": 3907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3037465940054497e-05,
+      "loss": 1.8447,
+      "step": 3908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3035422343324253e-05,
+      "loss": 1.6946,
+      "step": 3909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3033378746594005e-05,
+      "loss": 1.8346,
+      "step": 3910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.303133514986376e-05,
+      "loss": 1.8689,
+      "step": 3911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3029291553133516e-05,
+      "loss": 1.871,
+      "step": 3912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.302724795640327e-05,
+      "loss": 1.8064,
+      "step": 3913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3025204359673027e-05,
+      "loss": 1.7467,
+      "step": 3914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.302316076294278e-05,
+      "loss": 1.8133,
+      "step": 3915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3021117166212535e-05,
+      "loss": 1.7484,
+      "step": 3916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.301907356948229e-05,
+      "loss": 1.7517,
+      "step": 3917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3017029972752043e-05,
+      "loss": 1.7659,
+      "step": 3918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.30149863760218e-05,
+      "loss": 1.8049,
+      "step": 3919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.301294277929155e-05,
+      "loss": 1.8504,
+      "step": 3920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3010899182561307e-05,
+      "loss": 1.919,
+      "step": 3921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3008855585831066e-05,
+      "loss": 1.8499,
+      "step": 3922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3006811989100818e-05,
+      "loss": 1.7495,
+      "step": 3923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3004768392370574e-05,
+      "loss": 1.6484,
+      "step": 3924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3002724795640326e-05,
+      "loss": 1.7856,
+      "step": 3925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.300068119891008e-05,
+      "loss": 1.6729,
+      "step": 3926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2998637602179837e-05,
+      "loss": 1.6643,
+      "step": 3927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.299659400544959e-05,
+      "loss": 1.8213,
+      "step": 3928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.299455040871935e-05,
+      "loss": 1.7997,
+      "step": 3929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2992506811989104e-05,
+      "loss": 1.8297,
+      "step": 3930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2990463215258856e-05,
+      "loss": 1.9182,
+      "step": 3931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2988419618528612e-05,
+      "loss": 1.9272,
+      "step": 3932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2986376021798364e-05,
+      "loss": 1.795,
+      "step": 3933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.298433242506812e-05,
+      "loss": 1.8185,
+      "step": 3934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2982288828337875e-05,
+      "loss": 1.9322,
+      "step": 3935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2980245231607628e-05,
+      "loss": 1.7487,
+      "step": 3936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2978201634877387e-05,
+      "loss": 1.9447,
+      "step": 3937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.297615803814714e-05,
+      "loss": 2.0505,
+      "step": 3938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2974114441416895e-05,
+      "loss": 1.6705,
+      "step": 3939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.297207084468665e-05,
+      "loss": 1.8283,
+      "step": 3940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2970027247956402e-05,
+      "loss": 1.8661,
+      "step": 3941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2967983651226158e-05,
+      "loss": 1.7888,
+      "step": 3942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.296594005449591e-05,
+      "loss": 1.7255,
+      "step": 3943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.296389645776567e-05,
+      "loss": 1.8405,
+      "step": 3944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2961852861035425e-05,
+      "loss": 1.8231,
+      "step": 3945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2959809264305177e-05,
+      "loss": 1.9657,
+      "step": 3946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2957765667574933e-05,
+      "loss": 1.6724,
+      "step": 3947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.295572207084469e-05,
+      "loss": 1.8876,
+      "step": 3948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.295367847411444e-05,
+      "loss": 1.8602,
+      "step": 3949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2951634877384196e-05,
+      "loss": 1.6392,
+      "step": 3950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2949591280653952e-05,
+      "loss": 1.9361,
+      "step": 3951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2947547683923708e-05,
+      "loss": 1.8134,
+      "step": 3952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2945504087193463e-05,
+      "loss": 1.7731,
+      "step": 3953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2943460490463215e-05,
+      "loss": 1.8222,
+      "step": 3954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.294141689373297e-05,
+      "loss": 1.8074,
+      "step": 3955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2939373297002723e-05,
+      "loss": 1.924,
+      "step": 3956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.293732970027248e-05,
+      "loss": 1.626,
+      "step": 3957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2935286103542235e-05,
+      "loss": 1.8,
+      "step": 3958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.293324250681199e-05,
+      "loss": 1.676,
+      "step": 3959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2931198910081746e-05,
+      "loss": 1.7603,
+      "step": 3960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2929155313351498e-05,
+      "loss": 1.768,
+      "step": 3961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2927111716621254e-05,
+      "loss": 1.7622,
+      "step": 3962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.292506811989101e-05,
+      "loss": 1.6909,
+      "step": 3963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.292302452316076e-05,
+      "loss": 1.8037,
+      "step": 3964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2920980926430517e-05,
+      "loss": 1.816,
+      "step": 3965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2918937329700276e-05,
+      "loss": 1.7884,
+      "step": 3966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.291689373297003e-05,
+      "loss": 1.7772,
+      "step": 3967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2914850136239784e-05,
+      "loss": 1.7729,
+      "step": 3968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2912806539509536e-05,
+      "loss": 1.8977,
+      "step": 3969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2910762942779292e-05,
+      "loss": 1.7506,
+      "step": 3970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2908719346049048e-05,
+      "loss": 1.9231,
+      "step": 3971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.29066757493188e-05,
+      "loss": 1.7949,
+      "step": 3972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2904632152588556e-05,
+      "loss": 1.8748,
+      "step": 3973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290258855585831e-05,
+      "loss": 1.7928,
+      "step": 3974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2900544959128067e-05,
+      "loss": 1.9489,
+      "step": 3975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2898501362397822e-05,
+      "loss": 1.7525,
+      "step": 3976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2896457765667575e-05,
+      "loss": 1.9084,
+      "step": 3977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.289441416893733e-05,
+      "loss": 1.7487,
+      "step": 3978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2892370572207083e-05,
+      "loss": 1.7397,
+      "step": 3979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2890326975476838e-05,
+      "loss": 1.8861,
+      "step": 3980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2888283378746597e-05,
+      "loss": 1.8158,
+      "step": 3981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.288623978201635e-05,
+      "loss": 1.7831,
+      "step": 3982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2884196185286105e-05,
+      "loss": 1.7298,
+      "step": 3983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.288215258855586e-05,
+      "loss": 1.8743,
+      "step": 3984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2880108991825613e-05,
+      "loss": 1.8418,
+      "step": 3985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287806539509537e-05,
+      "loss": 1.7786,
+      "step": 3986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287602179836512e-05,
+      "loss": 1.8834,
+      "step": 3987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2873978201634876e-05,
+      "loss": 1.8402,
+      "step": 3988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2871934604904636e-05,
+      "loss": 1.8449,
+      "step": 3989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2869891008174388e-05,
+      "loss": 1.8534,
+      "step": 3990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2867847411444143e-05,
+      "loss": 1.8155,
+      "step": 3991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2865803814713896e-05,
+      "loss": 1.913,
+      "step": 3992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.286376021798365e-05,
+      "loss": 1.8599,
+      "step": 3993
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2861716621253407e-05,
+      "loss": 1.8079,
+      "step": 3994
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285967302452316e-05,
+      "loss": 2.0035,
+      "step": 3995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2857629427792918e-05,
+      "loss": 1.788,
+      "step": 3996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285558583106267e-05,
+      "loss": 1.7302,
+      "step": 3997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2853542234332426e-05,
+      "loss": 1.7236,
+      "step": 3998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285149863760218e-05,
+      "loss": 1.7179,
+      "step": 3999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2849455040871934e-05,
+      "loss": 1.7452,
+      "step": 4000
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 1.6136624813079834,
+      "eval_runtime": 720.9436,
+      "eval_samples_per_second": 5.756,
+      "eval_steps_per_second": 1.44,
+      "step": 4000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284741144414169e-05,
+      "loss": 1.72,
+      "step": 4001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2845367847411445e-05,
+      "loss": 1.6346,
+      "step": 4002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2843324250681197e-05,
+      "loss": 1.8181,
+      "step": 4003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2841280653950956e-05,
+      "loss": 1.7456,
+      "step": 4004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283923705722071e-05,
+      "loss": 1.7696,
+      "step": 4005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2837193460490464e-05,
+      "loss": 1.8526,
+      "step": 4006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283514986376022e-05,
+      "loss": 1.7657,
+      "step": 4007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2833106267029972e-05,
+      "loss": 1.73,
+      "step": 4008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2831062670299728e-05,
+      "loss": 1.8534,
+      "step": 4009
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282901907356948e-05,
+      "loss": 1.7809,
+      "step": 4010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282697547683924e-05,
+      "loss": 1.6673,
+      "step": 4011
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2824931880108995e-05,
+      "loss": 1.9246,
+      "step": 4012
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2822888283378747e-05,
+      "loss": 1.8404,
+      "step": 4013
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2820844686648503e-05,
+      "loss": 1.8211,
+      "step": 4014
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2818801089918255e-05,
+      "loss": 1.8328,
+      "step": 4015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281675749318801e-05,
+      "loss": 1.7595,
+      "step": 4016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2814713896457766e-05,
+      "loss": 1.8173,
+      "step": 4017
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281267029972752e-05,
+      "loss": 1.7996,
+      "step": 4018
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2810626702997277e-05,
+      "loss": 1.837,
+      "step": 4019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2808583106267033e-05,
+      "loss": 1.7753,
+      "step": 4020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2806539509536785e-05,
+      "loss": 2.0403,
+      "step": 4021
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280449591280654e-05,
+      "loss": 1.8371,
+      "step": 4022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2802452316076293e-05,
+      "loss": 1.7654,
+      "step": 4023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.280040871934605e-05,
+      "loss": 1.7609,
+      "step": 4024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2798365122615804e-05,
+      "loss": 1.805,
+      "step": 4025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.279632152588556e-05,
+      "loss": 1.6917,
+      "step": 4026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2794277929155316e-05,
+      "loss": 1.8163,
+      "step": 4027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2792234332425068e-05,
+      "loss": 1.8063,
+      "step": 4028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2790190735694824e-05,
+      "loss": 1.7332,
+      "step": 4029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.278814713896458e-05,
+      "loss": 1.7674,
+      "step": 4030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.278610354223433e-05,
+      "loss": 1.9338,
+      "step": 4031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2784059945504087e-05,
+      "loss": 1.7354,
+      "step": 4032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.278201634877384e-05,
+      "loss": 1.6672,
+      "step": 4033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.27799727520436e-05,
+      "loss": 1.6896,
+      "step": 4034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2777929155313354e-05,
+      "loss": 1.8232,
+      "step": 4035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2775885558583106e-05,
+      "loss": 1.8174,
+      "step": 4036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2773841961852862e-05,
+      "loss": 1.7876,
+      "step": 4037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2771798365122617e-05,
+      "loss": 1.8262,
+      "step": 4038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.276975476839237e-05,
+      "loss": 1.8391,
+      "step": 4039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2767711171662125e-05,
+      "loss": 1.6564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.276566757493188e-05,
+      "loss": 1.7803,
+      "step": 4041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2763623978201637e-05,
+      "loss": 1.9284,
+      "step": 4042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2761580381471392e-05,
+      "loss": 1.841,
+      "step": 4043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2759536784741144e-05,
+      "loss": 1.8586,
+      "step": 4044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.27574931880109e-05,
+      "loss": 1.8038,
+      "step": 4045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2755449591280652e-05,
+      "loss": 1.8005,
+      "step": 4046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2753405994550408e-05,
+      "loss": 1.8511,
+      "step": 4047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2751362397820167e-05,
+      "loss": 1.8145,
+      "step": 4048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.274931880108992e-05,
+      "loss": 1.6677,
+      "step": 4049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2747275204359675e-05,
+      "loss": 1.8012,
+      "step": 4050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2745231607629427e-05,
+      "loss": 1.8207,
+      "step": 4051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2743188010899183e-05,
+      "loss": 1.8059,
+      "step": 4052
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.274114441416894e-05,
+      "loss": 1.734,
+      "step": 4053
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.273910081743869e-05,
+      "loss": 1.8385,
+      "step": 4054
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2737057220708446e-05,
+      "loss": 1.9666,
+      "step": 4055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2735013623978202e-05,
+      "loss": 1.8989,
+      "step": 4056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2732970027247958e-05,
+      "loss": 1.8047,
+      "step": 4057
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2730926430517713e-05,
+      "loss": 1.9745,
+      "step": 4058
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2728882833787465e-05,
+      "loss": 1.734,
+      "step": 4059
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.272683923705722e-05,
+      "loss": 1.7894,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2724795640326977e-05,
+      "loss": 1.8617,
+      "step": 4061
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.272275204359673e-05,
+      "loss": 1.8337,
+      "step": 4062
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2720708446866488e-05,
+      "loss": 1.6621,
+      "step": 4063
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.271866485013624e-05,
+      "loss": 1.8565,
+      "step": 4064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2716621253405996e-05,
+      "loss": 1.833,
+      "step": 4065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.271457765667575e-05,
+      "loss": 1.7478,
+      "step": 4066
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2712534059945504e-05,
+      "loss": 1.8352,
+      "step": 4067
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.271049046321526e-05,
+      "loss": 1.7998,
+      "step": 4068
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.270844686648501e-05,
+      "loss": 1.7768,
+      "step": 4069
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2706403269754767e-05,
+      "loss": 1.7105,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2704359673024526e-05,
+      "loss": 1.726,
+      "step": 4071
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.270231607629428e-05,
+      "loss": 1.7105,
+      "step": 4072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2700272479564034e-05,
+      "loss": 1.8063,
+      "step": 4073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2698228882833786e-05,
+      "loss": 1.9001,
+      "step": 4074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2696185286103542e-05,
+      "loss": 1.8195,
+      "step": 4075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2694141689373298e-05,
+      "loss": 1.8096,
+      "step": 4076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.269209809264305e-05,
+      "loss": 1.7955,
+      "step": 4077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.269005449591281e-05,
+      "loss": 1.7961,
+      "step": 4078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2688010899182565e-05,
+      "loss": 1.7799,
+      "step": 4079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2685967302452317e-05,
+      "loss": 1.9053,
+      "step": 4080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2683923705722072e-05,
+      "loss": 1.8174,
+      "step": 4081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2681880108991825e-05,
+      "loss": 1.7647,
+      "step": 4082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.267983651226158e-05,
+      "loss": 1.7541,
+      "step": 4083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2677792915531336e-05,
+      "loss": 1.7229,
+      "step": 4084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2675749318801088e-05,
+      "loss": 1.8654,
+      "step": 4085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2673705722070847e-05,
+      "loss": 1.8702,
+      "step": 4086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.26716621253406e-05,
+      "loss": 1.7114,
+      "step": 4087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2669618528610355e-05,
+      "loss": 1.8495,
+      "step": 4088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.266757493188011e-05,
+      "loss": 1.8569,
+      "step": 4089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2665531335149863e-05,
+      "loss": 1.6842,
+      "step": 4090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.266348773841962e-05,
+      "loss": 1.9145,
+      "step": 4091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2661444141689374e-05,
+      "loss": 1.8486,
+      "step": 4092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.265940054495913e-05,
+      "loss": 1.8027,
+      "step": 4093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2657356948228885e-05,
+      "loss": 1.9229,
+      "step": 4094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2655313351498638e-05,
+      "loss": 1.6169,
+      "step": 4095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2653269754768393e-05,
+      "loss": 1.6957,
+      "step": 4096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.265122615803815e-05,
+      "loss": 1.7747,
+      "step": 4097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.26491825613079e-05,
+      "loss": 1.8056,
+      "step": 4098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2647138964577657e-05,
+      "loss": 1.7452,
+      "step": 4099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.264509536784741e-05,
+      "loss": 1.8812,
+      "step": 4100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2643051771117168e-05,
+      "loss": 1.9334,
+      "step": 4101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2641008174386924e-05,
+      "loss": 1.8114,
+      "step": 4102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2638964577656676e-05,
+      "loss": 1.7978,
+      "step": 4103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.263692098092643e-05,
+      "loss": 1.7346,
+      "step": 4104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2634877384196184e-05,
+      "loss": 1.7848,
+      "step": 4105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.263283378746594e-05,
+      "loss": 1.7639,
+      "step": 4106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2630790190735695e-05,
+      "loss": 1.6869,
+      "step": 4107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.262874659400545e-05,
+      "loss": 1.7628,
+      "step": 4108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2626702997275206e-05,
+      "loss": 1.9102,
+      "step": 4109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.262465940054496e-05,
+      "loss": 1.8076,
+      "step": 4110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2622615803814714e-05,
+      "loss": 1.8562,
+      "step": 4111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.262057220708447e-05,
+      "loss": 1.8431,
+      "step": 4112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2618528610354222e-05,
+      "loss": 1.6984,
+      "step": 4113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2616485013623978e-05,
+      "loss": 1.8496,
+      "step": 4114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2614441416893733e-05,
+      "loss": 1.8594,
+      "step": 4115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.261239782016349e-05,
+      "loss": 1.8713,
+      "step": 4116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2610354223433245e-05,
+      "loss": 1.9347,
+      "step": 4117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2608310626702997e-05,
+      "loss": 1.8511,
+      "step": 4118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2606267029972753e-05,
+      "loss": 1.8466,
+      "step": 4119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2604223433242508e-05,
+      "loss": 1.7502,
+      "step": 4120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.260217983651226e-05,
+      "loss": 1.8239,
+      "step": 4121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2600136239782016e-05,
+      "loss": 1.7728,
+      "step": 4122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2598092643051772e-05,
+      "loss": 1.7581,
+      "step": 4123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2596049046321527e-05,
+      "loss": 1.8189,
+      "step": 4124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2594005449591283e-05,
+      "loss": 1.8421,
+      "step": 4125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2591961852861035e-05,
+      "loss": 1.8007,
+      "step": 4126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.258991825613079e-05,
+      "loss": 1.8774,
+      "step": 4127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2587874659400543e-05,
+      "loss": 1.7585,
+      "step": 4128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.25858310626703e-05,
+      "loss": 1.8036,
+      "step": 4129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2583787465940058e-05,
+      "loss": 1.7623,
+      "step": 4130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.258174386920981e-05,
+      "loss": 1.5914,
+      "step": 4131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2579700272479566e-05,
+      "loss": 1.7819,
+      "step": 4132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.257765667574932e-05,
+      "loss": 1.7628,
+      "step": 4133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2575613079019074e-05,
+      "loss": 1.8235,
+      "step": 4134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.257356948228883e-05,
+      "loss": 1.8683,
+      "step": 4135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.257152588555858e-05,
+      "loss": 1.7788,
+      "step": 4136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2569482288828337e-05,
+      "loss": 1.7368,
+      "step": 4137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2567438692098096e-05,
+      "loss": 1.9614,
+      "step": 4138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2565395095367848e-05,
+      "loss": 1.7886,
+      "step": 4139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2563351498637604e-05,
+      "loss": 1.7286,
+      "step": 4140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2561307901907356e-05,
+      "loss": 1.8064,
+      "step": 4141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2559264305177112e-05,
+      "loss": 1.7948,
+      "step": 4142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2557220708446867e-05,
+      "loss": 1.7526,
+      "step": 4143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.255517711171662e-05,
+      "loss": 1.7457,
+      "step": 4144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.255313351498638e-05,
+      "loss": 1.7814,
+      "step": 4145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.255108991825613e-05,
+      "loss": 1.7468,
+      "step": 4146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2549046321525887e-05,
+      "loss": 1.7897,
+      "step": 4147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2547002724795642e-05,
+      "loss": 1.8498,
+      "step": 4148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2544959128065394e-05,
+      "loss": 1.9088,
+      "step": 4149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.254291553133515e-05,
+      "loss": 1.8015,
+      "step": 4150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2540871934604906e-05,
+      "loss": 1.7854,
+      "step": 4151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2538828337874658e-05,
+      "loss": 1.7656,
+      "step": 4152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2536784741144417e-05,
+      "loss": 1.6798,
+      "step": 4153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.253474114441417e-05,
+      "loss": 1.8559,
+      "step": 4154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2532697547683925e-05,
+      "loss": 1.6847,
+      "step": 4155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.253065395095368e-05,
+      "loss": 1.7925,
+      "step": 4156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2528610354223433e-05,
+      "loss": 1.7456,
+      "step": 4157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.252656675749319e-05,
+      "loss": 1.8484,
+      "step": 4158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.252452316076294e-05,
+      "loss": 1.8149,
+      "step": 4159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.25224795640327e-05,
+      "loss": 1.8868,
+      "step": 4160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2520435967302455e-05,
+      "loss": 1.6994,
+      "step": 4161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2518392370572207e-05,
+      "loss": 1.8343,
+      "step": 4162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2516348773841963e-05,
+      "loss": 1.7076,
+      "step": 4163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2514305177111715e-05,
+      "loss": 1.8912,
+      "step": 4164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.251226158038147e-05,
+      "loss": 1.8197,
+      "step": 4165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2510217983651227e-05,
+      "loss": 1.815,
+      "step": 4166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.250817438692098e-05,
+      "loss": 1.759,
+      "step": 4167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2506130790190738e-05,
+      "loss": 1.8201,
+      "step": 4168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2504087193460494e-05,
+      "loss": 1.7094,
+      "step": 4169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2502043596730246e-05,
+      "loss": 1.7193,
+      "step": 4170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.25e-05,
+      "loss": 1.8134,
+      "step": 4171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2497956403269754e-05,
+      "loss": 1.7869,
+      "step": 4172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.249591280653951e-05,
+      "loss": 1.8123,
+      "step": 4173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.2493869209809265e-05,
+      "loss": 1.7231,
+      "step": 4174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.249182561307902e-05,
+      "loss": 1.7289,
+      "step": 4175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2489782016348776e-05,
+      "loss": 1.8227,
+      "step": 4176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.248773841961853e-05,
+      "loss": 1.7618,
+      "step": 4177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2485694822888284e-05,
+      "loss": 1.7905,
+      "step": 4178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.248365122615804e-05,
+      "loss": 1.7491,
+      "step": 4179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2481607629427792e-05,
+      "loss": 1.8155,
+      "step": 4180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2479564032697548e-05,
+      "loss": 1.8833,
+      "step": 4181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.24775204359673e-05,
+      "loss": 1.7213,
+      "step": 4182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.247547683923706e-05,
+      "loss": 1.7848,
+      "step": 4183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2473433242506814e-05,
+      "loss": 1.9146,
+      "step": 4184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2471389645776567e-05,
+      "loss": 1.9118,
+      "step": 4185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2469346049046322e-05,
+      "loss": 1.792,
+      "step": 4186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2467302452316078e-05,
+      "loss": 1.8353,
+      "step": 4187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.246525885558583e-05,
+      "loss": 1.7875,
+      "step": 4188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2463215258855586e-05,
+      "loss": 1.9141,
+      "step": 4189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.246117166212534e-05,
+      "loss": 1.7728,
+      "step": 4190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2459128065395097e-05,
+      "loss": 1.7886,
+      "step": 4191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2457084468664853e-05,
+      "loss": 1.6588,
+      "step": 4192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2455040871934605e-05,
+      "loss": 1.7884,
+      "step": 4193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.245299727520436e-05,
+      "loss": 1.8363,
+      "step": 4194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2450953678474113e-05,
+      "loss": 1.9453,
+      "step": 4195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.244891008174387e-05,
+      "loss": 1.7759,
+      "step": 4196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2446866485013624e-05,
+      "loss": 1.711,
+      "step": 4197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.244482288828338e-05,
+      "loss": 1.8318,
+      "step": 4198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2442779291553135e-05,
+      "loss": 1.9701,
+      "step": 4199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2440735694822888e-05,
+      "loss": 1.8024,
+      "step": 4200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2438692098092643e-05,
+      "loss": 1.6596,
+      "step": 4201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.24366485013624e-05,
+      "loss": 1.733,
+      "step": 4202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.243460490463215e-05,
+      "loss": 1.7325,
+      "step": 4203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2432561307901907e-05,
+      "loss": 1.6733,
+      "step": 4204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2430517711171666e-05,
+      "loss": 1.8389,
+      "step": 4205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2428474114441418e-05,
+      "loss": 1.6386,
+      "step": 4206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2426430517711174e-05,
+      "loss": 1.7113,
+      "step": 4207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2424386920980926e-05,
+      "loss": 1.818,
+      "step": 4208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.242234332425068e-05,
+      "loss": 1.689,
+      "step": 4209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2420299727520437e-05,
+      "loss": 1.8267,
+      "step": 4210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.241825613079019e-05,
+      "loss": 1.7479,
+      "step": 4211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2416212534059945e-05,
+      "loss": 1.8846,
+      "step": 4212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.24141689373297e-05,
+      "loss": 1.8591,
+      "step": 4213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2412125340599456e-05,
+      "loss": 1.7638,
+      "step": 4214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2410081743869212e-05,
+      "loss": 1.7396,
+      "step": 4215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2408038147138964e-05,
+      "loss": 1.7212,
+      "step": 4216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.240599455040872e-05,
+      "loss": 1.857,
+      "step": 4217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2403950953678472e-05,
+      "loss": 1.7368,
+      "step": 4218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2401907356948228e-05,
+      "loss": 1.7317,
+      "step": 4219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2399863760217987e-05,
+      "loss": 1.7823,
+      "step": 4220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.239782016348774e-05,
+      "loss": 1.7985,
+      "step": 4221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2395776566757495e-05,
+      "loss": 1.6766,
+      "step": 4222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.239373297002725e-05,
+      "loss": 1.8094,
+      "step": 4223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2391689373297003e-05,
+      "loss": 1.8779,
+      "step": 4224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2389645776566758e-05,
+      "loss": 1.8378,
+      "step": 4225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.238760217983651e-05,
+      "loss": 1.784,
+      "step": 4226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.238555858310627e-05,
+      "loss": 1.7576,
+      "step": 4227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2383514986376025e-05,
+      "loss": 1.634,
+      "step": 4228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2381471389645777e-05,
+      "loss": 1.7976,
+      "step": 4229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2379427792915533e-05,
+      "loss": 1.8114,
+      "step": 4230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2377384196185285e-05,
+      "loss": 1.7117,
+      "step": 4231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.237534059945504e-05,
+      "loss": 1.775,
+      "step": 4232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2373297002724796e-05,
+      "loss": 1.7307,
+      "step": 4233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.237125340599455e-05,
+      "loss": 1.7206,
+      "step": 4234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2369209809264308e-05,
+      "loss": 1.916,
+      "step": 4235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.236716621253406e-05,
+      "loss": 1.9118,
+      "step": 4236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2365122615803816e-05,
+      "loss": 1.7805,
+      "step": 4237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.236307901907357e-05,
+      "loss": 1.7564,
+      "step": 4238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2361035422343323e-05,
+      "loss": 1.7437,
+      "step": 4239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.235899182561308e-05,
+      "loss": 1.7566,
+      "step": 4240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2356948228882835e-05,
+      "loss": 1.7726,
+      "step": 4241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.235490463215259e-05,
+      "loss": 1.7568,
+      "step": 4242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2352861035422346e-05,
+      "loss": 1.7154,
+      "step": 4243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2350817438692098e-05,
+      "loss": 1.7896,
+      "step": 4244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2348773841961854e-05,
+      "loss": 1.757,
+      "step": 4245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.234673024523161e-05,
+      "loss": 1.8321,
+      "step": 4246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2344686648501362e-05,
+      "loss": 1.6836,
+      "step": 4247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2342643051771117e-05,
+      "loss": 1.7592,
+      "step": 4248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.234059945504087e-05,
+      "loss": 1.8748,
+      "step": 4249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.233855585831063e-05,
+      "loss": 1.8248,
+      "step": 4250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2336512261580384e-05,
+      "loss": 1.8335,
+      "step": 4251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2334468664850137e-05,
+      "loss": 1.7445,
+      "step": 4252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2332425068119892e-05,
+      "loss": 1.8346,
+      "step": 4253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2330381471389644e-05,
+      "loss": 1.7946,
+      "step": 4254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.23283378746594e-05,
+      "loss": 1.7469,
+      "step": 4255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2326294277929156e-05,
+      "loss": 1.7221,
+      "step": 4256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.232425068119891e-05,
+      "loss": 1.8271,
+      "step": 4257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2322207084468667e-05,
+      "loss": 1.7752,
+      "step": 4258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.232016348773842e-05,
+      "loss": 1.7959,
+      "step": 4259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2318119891008175e-05,
+      "loss": 1.727,
+      "step": 4260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.231607629427793e-05,
+      "loss": 1.6505,
+      "step": 4261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2314032697547683e-05,
+      "loss": 1.7264,
+      "step": 4262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.231198910081744e-05,
+      "loss": 1.8531,
+      "step": 4263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2309945504087194e-05,
+      "loss": 1.785,
+      "step": 4264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.230790190735695e-05,
+      "loss": 1.8302,
+      "step": 4265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2305858310626705e-05,
+      "loss": 1.7959,
+      "step": 4266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2303814713896457e-05,
+      "loss": 1.7642,
+      "step": 4267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2301771117166213e-05,
+      "loss": 1.7074,
+      "step": 4268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.229972752043597e-05,
+      "loss": 1.7537,
+      "step": 4269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.229768392370572e-05,
+      "loss": 1.9246,
+      "step": 4270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2295640326975477e-05,
+      "loss": 1.7805,
+      "step": 4271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2293596730245232e-05,
+      "loss": 1.7866,
+      "step": 4272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2291553133514988e-05,
+      "loss": 1.7096,
+      "step": 4273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2289509536784743e-05,
+      "loss": 1.8701,
+      "step": 4274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2287465940054496e-05,
+      "loss": 1.8233,
+      "step": 4275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.228542234332425e-05,
+      "loss": 1.7632,
+      "step": 4276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2283378746594007e-05,
+      "loss": 1.7143,
+      "step": 4277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.228133514986376e-05,
+      "loss": 1.8292,
+      "step": 4278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2279291553133515e-05,
+      "loss": 1.8143,
+      "step": 4279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.227724795640327e-05,
+      "loss": 1.9071,
+      "step": 4280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2275204359673026e-05,
+      "loss": 1.8006,
+      "step": 4281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2273160762942782e-05,
+      "loss": 1.8245,
+      "step": 4282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2271117166212534e-05,
+      "loss": 1.7989,
+      "step": 4283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.226907356948229e-05,
+      "loss": 1.7949,
+      "step": 4284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2267029972752042e-05,
+      "loss": 1.8318,
+      "step": 4285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2264986376021798e-05,
+      "loss": 1.6765,
+      "step": 4286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2262942779291557e-05,
+      "loss": 1.8378,
+      "step": 4287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.226089918256131e-05,
+      "loss": 1.7984,
+      "step": 4288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2258855585831064e-05,
+      "loss": 1.7387,
+      "step": 4289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2256811989100817e-05,
+      "loss": 1.7952,
+      "step": 4290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2254768392370572e-05,
+      "loss": 1.6931,
+      "step": 4291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2252724795640328e-05,
+      "loss": 1.8161,
+      "step": 4292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.225068119891008e-05,
+      "loss": 1.6436,
+      "step": 4293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2248637602179836e-05,
+      "loss": 1.6701,
+      "step": 4294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.224659400544959e-05,
+      "loss": 1.7978,
+      "step": 4295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2244550408719347e-05,
+      "loss": 1.6654,
+      "step": 4296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2242506811989103e-05,
+      "loss": 1.7243,
+      "step": 4297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2240463215258855e-05,
+      "loss": 1.7971,
+      "step": 4298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.223841961852861e-05,
+      "loss": 1.7764,
+      "step": 4299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2236376021798366e-05,
+      "loss": 1.6744,
+      "step": 4300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.223433242506812e-05,
+      "loss": 1.8017,
+      "step": 4301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2232288828337877e-05,
+      "loss": 1.7532,
+      "step": 4302
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.223024523160763e-05,
+      "loss": 1.6274,
+      "step": 4303
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2228201634877385e-05,
+      "loss": 1.7708,
+      "step": 4304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.222615803814714e-05,
+      "loss": 1.7448,
+      "step": 4305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2224114441416893e-05,
+      "loss": 1.7222,
+      "step": 4306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.222207084468665e-05,
+      "loss": 1.7997,
+      "step": 4307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.22200272479564e-05,
+      "loss": 1.713,
+      "step": 4308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.221798365122616e-05,
+      "loss": 1.8034,
+      "step": 4309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2215940054495916e-05,
+      "loss": 1.7661,
+      "step": 4310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2213896457765668e-05,
+      "loss": 1.7362,
+      "step": 4311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2211852861035424e-05,
+      "loss": 1.8003,
+      "step": 4312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2209809264305176e-05,
+      "loss": 1.8168,
+      "step": 4313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.220776566757493e-05,
+      "loss": 1.7289,
+      "step": 4314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2205722070844687e-05,
+      "loss": 1.7107,
+      "step": 4315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.220367847411444e-05,
+      "loss": 1.7419,
+      "step": 4316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.22016348773842e-05,
+      "loss": 1.8687,
+      "step": 4317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2199591280653954e-05,
+      "loss": 1.8286,
+      "step": 4318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2197547683923706e-05,
+      "loss": 1.7627,
+      "step": 4319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2195504087193462e-05,
+      "loss": 1.7862,
+      "step": 4320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2193460490463214e-05,
+      "loss": 1.779,
+      "step": 4321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.219141689373297e-05,
+      "loss": 1.9301,
+      "step": 4322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2189373297002725e-05,
+      "loss": 1.7316,
+      "step": 4323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.218732970027248e-05,
+      "loss": 1.7809,
+      "step": 4324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2185286103542237e-05,
+      "loss": 1.8222,
+      "step": 4325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.218324250681199e-05,
+      "loss": 1.7898,
+      "step": 4326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2181198910081745e-05,
+      "loss": 1.7605,
+      "step": 4327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.21791553133515e-05,
+      "loss": 1.8194,
+      "step": 4328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2177111716621252e-05,
+      "loss": 1.8523,
+      "step": 4329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2175068119891008e-05,
+      "loss": 1.8604,
+      "step": 4330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.217302452316076e-05,
+      "loss": 1.7744,
+      "step": 4331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.217098092643052e-05,
+      "loss": 1.7511,
+      "step": 4332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2168937329700275e-05,
+      "loss": 1.8313,
+      "step": 4333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2166893732970027e-05,
+      "loss": 1.8282,
+      "step": 4334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2164850136239783e-05,
+      "loss": 1.7914,
+      "step": 4335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.216280653950954e-05,
+      "loss": 1.8049,
+      "step": 4336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.216076294277929e-05,
+      "loss": 1.8995,
+      "step": 4337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2158719346049046e-05,
+      "loss": 1.6672,
+      "step": 4338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2156675749318802e-05,
+      "loss": 1.8342,
+      "step": 4339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2154632152588558e-05,
+      "loss": 1.6669,
+      "step": 4340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2152588555858313e-05,
+      "loss": 1.9629,
+      "step": 4341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2150544959128066e-05,
+      "loss": 1.9481,
+      "step": 4342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.214850136239782e-05,
+      "loss": 1.6697,
+      "step": 4343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2146457765667573e-05,
+      "loss": 1.8419,
+      "step": 4344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.214441416893733e-05,
+      "loss": 1.7259,
+      "step": 4345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2142370572207085e-05,
+      "loss": 1.6789,
+      "step": 4346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.214032697547684e-05,
+      "loss": 1.7675,
+      "step": 4347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2138283378746596e-05,
+      "loss": 1.7442,
+      "step": 4348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2136239782016348e-05,
+      "loss": 1.7826,
+      "step": 4349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2134196185286104e-05,
+      "loss": 1.8059,
+      "step": 4350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.213215258855586e-05,
+      "loss": 1.9936,
+      "step": 4351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.213010899182561e-05,
+      "loss": 1.9598,
+      "step": 4352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2128065395095367e-05,
+      "loss": 1.819,
+      "step": 4353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2126021798365126e-05,
+      "loss": 1.7365,
+      "step": 4354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.212397820163488e-05,
+      "loss": 1.6735,
+      "step": 4355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2121934604904634e-05,
+      "loss": 1.886,
+      "step": 4356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2119891008174386e-05,
+      "loss": 1.8084,
+      "step": 4357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2117847411444142e-05,
+      "loss": 1.8194,
+      "step": 4358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2115803814713898e-05,
+      "loss": 1.7711,
+      "step": 4359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.211376021798365e-05,
+      "loss": 1.8172,
+      "step": 4360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2111716621253406e-05,
+      "loss": 1.9227,
+      "step": 4361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.210967302452316e-05,
+      "loss": 1.7367,
+      "step": 4362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2107629427792917e-05,
+      "loss": 1.6801,
+      "step": 4363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2105585831062673e-05,
+      "loss": 1.7446,
+      "step": 4364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2103542234332425e-05,
+      "loss": 1.7715,
+      "step": 4365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.210149863760218e-05,
+      "loss": 1.79,
+      "step": 4366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2099455040871933e-05,
+      "loss": 1.7573,
+      "step": 4367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2097411444141688e-05,
+      "loss": 1.8067,
+      "step": 4368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2095367847411447e-05,
+      "loss": 1.7652,
+      "step": 4369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.20933242506812e-05,
+      "loss": 1.7883,
+      "step": 4370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2091280653950955e-05,
+      "loss": 1.8204,
+      "step": 4371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.208923705722071e-05,
+      "loss": 1.7938,
+      "step": 4372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2087193460490463e-05,
+      "loss": 1.8858,
+      "step": 4373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.208514986376022e-05,
+      "loss": 1.6886,
+      "step": 4374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.208310626702997e-05,
+      "loss": 1.7364,
+      "step": 4375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2081062670299727e-05,
+      "loss": 1.8178,
+      "step": 4376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2079019073569486e-05,
+      "loss": 1.7997,
+      "step": 4377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2076975476839238e-05,
+      "loss": 1.8414,
+      "step": 4378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2074931880108993e-05,
+      "loss": 1.7596,
+      "step": 4379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2072888283378746e-05,
+      "loss": 1.8378,
+      "step": 4380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.20708446866485e-05,
+      "loss": 1.6878,
+      "step": 4381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2068801089918257e-05,
+      "loss": 1.7367,
+      "step": 4382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.206675749318801e-05,
+      "loss": 1.8628,
+      "step": 4383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2064713896457768e-05,
+      "loss": 1.7804,
+      "step": 4384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.206267029972752e-05,
+      "loss": 1.6714,
+      "step": 4385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2060626702997276e-05,
+      "loss": 1.7145,
+      "step": 4386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2058583106267032e-05,
+      "loss": 1.7732,
+      "step": 4387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2056539509536784e-05,
+      "loss": 1.5915,
+      "step": 4388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.205449591280654e-05,
+      "loss": 1.7623,
+      "step": 4389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2052452316076295e-05,
+      "loss": 1.7846,
+      "step": 4390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2050408719346047e-05,
+      "loss": 1.7911,
+      "step": 4391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2048365122615807e-05,
+      "loss": 1.8467,
+      "step": 4392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.204632152588556e-05,
+      "loss": 1.8545,
+      "step": 4393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2044277929155314e-05,
+      "loss": 1.6857,
+      "step": 4394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.204223433242507e-05,
+      "loss": 1.7142,
+      "step": 4395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2040190735694822e-05,
+      "loss": 1.8474,
+      "step": 4396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2038147138964578e-05,
+      "loss": 1.8056,
+      "step": 4397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.203610354223433e-05,
+      "loss": 1.7134,
+      "step": 4398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.203405994550409e-05,
+      "loss": 1.9089,
+      "step": 4399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2032016348773845e-05,
+      "loss": 1.8147,
+      "step": 4400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2029972752043597e-05,
+      "loss": 1.8084,
+      "step": 4401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2027929155313353e-05,
+      "loss": 1.7269,
+      "step": 4402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2025885558583105e-05,
+      "loss": 1.7758,
+      "step": 4403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.202384196185286e-05,
+      "loss": 1.7031,
+      "step": 4404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2021798365122616e-05,
+      "loss": 1.714,
+      "step": 4405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2019754768392372e-05,
+      "loss": 1.6983,
+      "step": 4406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2017711171662127e-05,
+      "loss": 1.6477,
+      "step": 4407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2015667574931883e-05,
+      "loss": 1.9577,
+      "step": 4408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2013623978201635e-05,
+      "loss": 1.6998,
+      "step": 4409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.201158038147139e-05,
+      "loss": 1.7963,
+      "step": 4410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2009536784741143e-05,
+      "loss": 1.6964,
+      "step": 4411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.20074931880109e-05,
+      "loss": 1.8763,
+      "step": 4412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2005449591280654e-05,
+      "loss": 1.8874,
+      "step": 4413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.200340599455041e-05,
+      "loss": 1.8286,
+      "step": 4414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2001362397820166e-05,
+      "loss": 1.9262,
+      "step": 4415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1999318801089918e-05,
+      "loss": 1.8189,
+      "step": 4416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1997275204359674e-05,
+      "loss": 1.7777,
+      "step": 4417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.199523160762943e-05,
+      "loss": 1.7954,
+      "step": 4418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.199318801089918e-05,
+      "loss": 1.7722,
+      "step": 4419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1991144414168937e-05,
+      "loss": 1.6345,
+      "step": 4420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1989100817438693e-05,
+      "loss": 1.9314,
+      "step": 4421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.198705722070845e-05,
+      "loss": 1.8397,
+      "step": 4422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1985013623978204e-05,
+      "loss": 1.7971,
+      "step": 4423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1982970027247956e-05,
+      "loss": 1.6892,
+      "step": 4424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1980926430517712e-05,
+      "loss": 1.6855,
+      "step": 4425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1978882833787468e-05,
+      "loss": 1.6576,
+      "step": 4426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.197683923705722e-05,
+      "loss": 1.7421,
+      "step": 4427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1974795640326975e-05,
+      "loss": 1.7518,
+      "step": 4428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.197275204359673e-05,
+      "loss": 1.7962,
+      "step": 4429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1970708446866487e-05,
+      "loss": 1.8269,
+      "step": 4430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1968664850136242e-05,
+      "loss": 1.9649,
+      "step": 4431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1966621253405995e-05,
+      "loss": 1.6957,
+      "step": 4432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.196457765667575e-05,
+      "loss": 1.8497,
+      "step": 4433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1962534059945502e-05,
+      "loss": 1.6069,
+      "step": 4434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1960490463215258e-05,
+      "loss": 1.9028,
+      "step": 4435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1958446866485017e-05,
+      "loss": 1.6764,
+      "step": 4436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.195640326975477e-05,
+      "loss": 1.9751,
+      "step": 4437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1954359673024525e-05,
+      "loss": 1.7248,
+      "step": 4438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1952316076294277e-05,
+      "loss": 1.7841,
+      "step": 4439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1950272479564033e-05,
+      "loss": 1.7897,
+      "step": 4440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.194822888283379e-05,
+      "loss": 1.7934,
+      "step": 4441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.194618528610354e-05,
+      "loss": 1.7706,
+      "step": 4442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1944141689373296e-05,
+      "loss": 1.6985,
+      "step": 4443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1942098092643055e-05,
+      "loss": 1.8168,
+      "step": 4444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1940054495912808e-05,
+      "loss": 1.7806,
+      "step": 4445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1938010899182563e-05,
+      "loss": 1.8195,
+      "step": 4446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1935967302452315e-05,
+      "loss": 1.6977,
+      "step": 4447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.193392370572207e-05,
+      "loss": 1.8361,
+      "step": 4448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1931880108991827e-05,
+      "loss": 1.7673,
+      "step": 4449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.192983651226158e-05,
+      "loss": 1.9001,
+      "step": 4450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1927792915531338e-05,
+      "loss": 1.8,
+      "step": 4451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.192574931880109e-05,
+      "loss": 1.7471,
+      "step": 4452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1923705722070846e-05,
+      "loss": 1.8388,
+      "step": 4453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.19216621253406e-05,
+      "loss": 1.7188,
+      "step": 4454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1919618528610354e-05,
+      "loss": 1.7108,
+      "step": 4455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.191757493188011e-05,
+      "loss": 1.7192,
+      "step": 4456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.191553133514986e-05,
+      "loss": 1.8505,
+      "step": 4457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1913487738419617e-05,
+      "loss": 1.7065,
+      "step": 4458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1911444141689376e-05,
+      "loss": 1.7107,
+      "step": 4459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.190940054495913e-05,
+      "loss": 1.7432,
+      "step": 4460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1907356948228884e-05,
+      "loss": 1.8465,
+      "step": 4461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.190531335149864e-05,
+      "loss": 1.6877,
+      "step": 4462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1903269754768392e-05,
+      "loss": 1.815,
+      "step": 4463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1901226158038148e-05,
+      "loss": 1.85,
+      "step": 4464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.18991825613079e-05,
+      "loss": 1.7457,
+      "step": 4465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.189713896457766e-05,
+      "loss": 1.8097,
+      "step": 4466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1895095367847415e-05,
+      "loss": 1.7197,
+      "step": 4467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1893051771117167e-05,
+      "loss": 1.7683,
+      "step": 4468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1891008174386922e-05,
+      "loss": 1.8221,
+      "step": 4469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1888964577656675e-05,
+      "loss": 1.7764,
+      "step": 4470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.188692098092643e-05,
+      "loss": 1.7784,
+      "step": 4471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1884877384196186e-05,
+      "loss": 1.883,
+      "step": 4472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1882833787465938e-05,
+      "loss": 1.745,
+      "step": 4473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1880790190735697e-05,
+      "loss": 1.8097,
+      "step": 4474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.187874659400545e-05,
+      "loss": 1.7914,
+      "step": 4475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1876702997275205e-05,
+      "loss": 1.7204,
+      "step": 4476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.187465940054496e-05,
+      "loss": 1.7689,
+      "step": 4477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1872615803814713e-05,
+      "loss": 1.7562,
+      "step": 4478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.187057220708447e-05,
+      "loss": 1.7618,
+      "step": 4479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1868528610354224e-05,
+      "loss": 1.8536,
+      "step": 4480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.186648501362398e-05,
+      "loss": 1.8673,
+      "step": 4481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1864441416893736e-05,
+      "loss": 1.8456,
+      "step": 4482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1862397820163488e-05,
+      "loss": 1.8704,
+      "step": 4483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1860354223433243e-05,
+      "loss": 1.7209,
+      "step": 4484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1858310626703e-05,
+      "loss": 1.8013,
+      "step": 4485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.185626702997275e-05,
+      "loss": 1.8656,
+      "step": 4486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1854223433242507e-05,
+      "loss": 1.8119,
+      "step": 4487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.185217983651226e-05,
+      "loss": 1.7607,
+      "step": 4488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1850136239782018e-05,
+      "loss": 1.6605,
+      "step": 4489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1848092643051774e-05,
+      "loss": 1.8212,
+      "step": 4490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1846049046321526e-05,
+      "loss": 1.7251,
+      "step": 4491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.184400544959128e-05,
+      "loss": 1.8006,
+      "step": 4492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1841961852861034e-05,
+      "loss": 1.8354,
+      "step": 4493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.183991825613079e-05,
+      "loss": 1.8308,
+      "step": 4494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1837874659400545e-05,
+      "loss": 1.7931,
+      "step": 4495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.18358310626703e-05,
+      "loss": 1.8658,
+      "step": 4496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1833787465940056e-05,
+      "loss": 1.8005,
+      "step": 4497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.183174386920981e-05,
+      "loss": 1.6745,
+      "step": 4498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1829700272479564e-05,
+      "loss": 1.7713,
+      "step": 4499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.182765667574932e-05,
+      "loss": 1.7415,
+      "step": 4500
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 1.5983014106750488,
+      "eval_runtime": 721.088,
+      "eval_samples_per_second": 5.755,
+      "eval_steps_per_second": 1.439,
+      "step": 4500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1825613079019072e-05,
+      "loss": 1.7649,
+      "step": 4501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1823569482288828e-05,
+      "loss": 1.791,
+      "step": 4502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1821525885558587e-05,
+      "loss": 1.7792,
+      "step": 4503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.181948228882834e-05,
+      "loss": 1.7344,
+      "step": 4504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1817438692098095e-05,
+      "loss": 1.7102,
+      "step": 4505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1815395095367847e-05,
+      "loss": 1.774,
+      "step": 4506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1813351498637603e-05,
+      "loss": 1.8395,
+      "step": 4507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1811307901907358e-05,
+      "loss": 1.758,
+      "step": 4508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.180926430517711e-05,
+      "loss": 1.7949,
+      "step": 4509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1807220708446866e-05,
+      "loss": 1.9096,
+      "step": 4510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1805177111716622e-05,
+      "loss": 1.6941,
+      "step": 4511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1803133514986377e-05,
+      "loss": 1.7245,
+      "step": 4512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1801089918256133e-05,
+      "loss": 1.8743,
+      "step": 4513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1799046321525885e-05,
+      "loss": 1.7786,
+      "step": 4514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.179700272479564e-05,
+      "loss": 1.7614,
+      "step": 4515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1794959128065393e-05,
+      "loss": 1.7571,
+      "step": 4516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.179291553133515e-05,
+      "loss": 1.827,
+      "step": 4517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1790871934604908e-05,
+      "loss": 1.8722,
+      "step": 4518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.178882833787466e-05,
+      "loss": 1.8141,
+      "step": 4519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1786784741144416e-05,
+      "loss": 1.7056,
+      "step": 4520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.178474114441417e-05,
+      "loss": 1.7444,
+      "step": 4521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1782697547683924e-05,
+      "loss": 1.7021,
+      "step": 4522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.178065395095368e-05,
+      "loss": 1.7453,
+      "step": 4523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.177861035422343e-05,
+      "loss": 1.7355,
+      "step": 4524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1776566757493187e-05,
+      "loss": 1.8584,
+      "step": 4525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1774523160762946e-05,
+      "loss": 1.9187,
+      "step": 4526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.17724795640327e-05,
+      "loss": 1.8447,
+      "step": 4527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1770435967302454e-05,
+      "loss": 1.8336,
+      "step": 4528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1768392370572206e-05,
+      "loss": 1.7716,
+      "step": 4529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1766348773841962e-05,
+      "loss": 1.8119,
+      "step": 4530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1764305177111717e-05,
+      "loss": 1.8233,
+      "step": 4531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.176226158038147e-05,
+      "loss": 1.6611,
+      "step": 4532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.176021798365123e-05,
+      "loss": 1.8915,
+      "step": 4533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.175817438692098e-05,
+      "loss": 1.7378,
+      "step": 4534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1756130790190737e-05,
+      "loss": 1.7237,
+      "step": 4535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1754087193460492e-05,
+      "loss": 1.7919,
+      "step": 4536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1752043596730244e-05,
+      "loss": 1.7231,
+      "step": 4537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.175e-05,
+      "loss": 1.719,
+      "step": 4538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1747956403269756e-05,
+      "loss": 1.8413,
+      "step": 4539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1745912806539508e-05,
+      "loss": 1.8137,
+      "step": 4540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1743869209809267e-05,
+      "loss": 1.79,
+      "step": 4541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.174182561307902e-05,
+      "loss": 1.9342,
+      "step": 4542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1739782016348775e-05,
+      "loss": 1.7659,
+      "step": 4543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.173773841961853e-05,
+      "loss": 1.811,
+      "step": 4544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1735694822888283e-05,
+      "loss": 1.7916,
+      "step": 4545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.173365122615804e-05,
+      "loss": 1.6441,
+      "step": 4546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.173160762942779e-05,
+      "loss": 1.6991,
+      "step": 4547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.172956403269755e-05,
+      "loss": 1.7701,
+      "step": 4548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1727520435967305e-05,
+      "loss": 1.7008,
+      "step": 4549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1725476839237058e-05,
+      "loss": 1.6471,
+      "step": 4550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1723433242506813e-05,
+      "loss": 1.8566,
+      "step": 4551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1721389645776565e-05,
+      "loss": 1.757,
+      "step": 4552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.171934604904632e-05,
+      "loss": 1.8187,
+      "step": 4553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1717302452316077e-05,
+      "loss": 1.7102,
+      "step": 4554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.171525885558583e-05,
+      "loss": 1.8384,
+      "step": 4555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1713215258855588e-05,
+      "loss": 1.825,
+      "step": 4556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1711171662125344e-05,
+      "loss": 1.6864,
+      "step": 4557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1709128065395096e-05,
+      "loss": 1.8187,
+      "step": 4558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.170708446866485e-05,
+      "loss": 1.7797,
+      "step": 4559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1705040871934604e-05,
+      "loss": 1.8226,
+      "step": 4560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.170299727520436e-05,
+      "loss": 1.6807,
+      "step": 4561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1700953678474115e-05,
+      "loss": 1.7264,
+      "step": 4562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.169891008174387e-05,
+      "loss": 1.8077,
+      "step": 4563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1696866485013626e-05,
+      "loss": 1.7683,
+      "step": 4564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.169482288828338e-05,
+      "loss": 1.8614,
+      "step": 4565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1692779291553134e-05,
+      "loss": 1.7935,
+      "step": 4566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.169073569482289e-05,
+      "loss": 1.8058,
+      "step": 4567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1688692098092642e-05,
+      "loss": 1.6778,
+      "step": 4568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1686648501362398e-05,
+      "loss": 1.7109,
+      "step": 4569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.168460490463215e-05,
+      "loss": 1.8076,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.168256130790191e-05,
+      "loss": 1.9163,
+      "step": 4571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1680517711171665e-05,
+      "loss": 1.7388,
+      "step": 4572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1678474114441417e-05,
+      "loss": 1.7379,
+      "step": 4573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1676430517711172e-05,
+      "loss": 1.7214,
+      "step": 4574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1674386920980928e-05,
+      "loss": 1.8055,
+      "step": 4575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.167234332425068e-05,
+      "loss": 1.9384,
+      "step": 4576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1670299727520436e-05,
+      "loss": 1.7888,
+      "step": 4577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.166825613079019e-05,
+      "loss": 1.8324,
+      "step": 4578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1666212534059947e-05,
+      "loss": 1.7484,
+      "step": 4579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1664168937329703e-05,
+      "loss": 1.7653,
+      "step": 4580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1662125340599455e-05,
+      "loss": 1.8593,
+      "step": 4581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.166008174386921e-05,
+      "loss": 1.7855,
+      "step": 4582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1658038147138963e-05,
+      "loss": 1.9285,
+      "step": 4583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.165599455040872e-05,
+      "loss": 1.7036,
+      "step": 4584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1653950953678478e-05,
+      "loss": 1.7658,
+      "step": 4585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.165190735694823e-05,
+      "loss": 1.794,
+      "step": 4586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1649863760217985e-05,
+      "loss": 1.7487,
+      "step": 4587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1647820163487738e-05,
+      "loss": 1.7871,
+      "step": 4588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1645776566757493e-05,
+      "loss": 1.7322,
+      "step": 4589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.164373297002725e-05,
+      "loss": 1.7875,
+      "step": 4590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1641689373297e-05,
+      "loss": 1.6934,
+      "step": 4591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1639645776566757e-05,
+      "loss": 1.6761,
+      "step": 4592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1637602179836516e-05,
+      "loss": 1.7751,
+      "step": 4593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1635558583106268e-05,
+      "loss": 1.7985,
+      "step": 4594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1633514986376024e-05,
+      "loss": 1.7237,
+      "step": 4595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1631471389645776e-05,
+      "loss": 1.7607,
+      "step": 4596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.162942779291553e-05,
+      "loss": 1.8631,
+      "step": 4597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1627384196185287e-05,
+      "loss": 1.8486,
+      "step": 4598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.162534059945504e-05,
+      "loss": 2.1674,
+      "step": 4599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.16232970027248e-05,
+      "loss": 1.6641,
+      "step": 4600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.162125340599455e-05,
+      "loss": 1.816,
+      "step": 4601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1619209809264306e-05,
+      "loss": 1.7915,
+      "step": 4602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1617166212534062e-05,
+      "loss": 1.7638,
+      "step": 4603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1615122615803814e-05,
+      "loss": 1.9034,
+      "step": 4604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.161307901907357e-05,
+      "loss": 1.8063,
+      "step": 4605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1611035422343322e-05,
+      "loss": 1.812,
+      "step": 4606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1608991825613078e-05,
+      "loss": 1.7175,
+      "step": 4607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1606948228882837e-05,
+      "loss": 1.8585,
+      "step": 4608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.160490463215259e-05,
+      "loss": 1.7556,
+      "step": 4609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1602861035422345e-05,
+      "loss": 1.816,
+      "step": 4610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.16008174386921e-05,
+      "loss": 1.6606,
+      "step": 4611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1598773841961853e-05,
+      "loss": 1.9618,
+      "step": 4612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1596730245231608e-05,
+      "loss": 1.8042,
+      "step": 4613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.159468664850136e-05,
+      "loss": 1.743,
+      "step": 4614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.159264305177112e-05,
+      "loss": 1.8301,
+      "step": 4615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1590599455040875e-05,
+      "loss": 1.8249,
+      "step": 4616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1588555858310627e-05,
+      "loss": 1.8098,
+      "step": 4617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1586512261580383e-05,
+      "loss": 1.8297,
+      "step": 4618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1584468664850135e-05,
+      "loss": 1.792,
+      "step": 4619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.158242506811989e-05,
+      "loss": 1.885,
+      "step": 4620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1580381471389646e-05,
+      "loss": 1.9493,
+      "step": 4621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.15783378746594e-05,
+      "loss": 1.8132,
+      "step": 4622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1576294277929158e-05,
+      "loss": 1.8362,
+      "step": 4623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.157425068119891e-05,
+      "loss": 1.7672,
+      "step": 4624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1572207084468666e-05,
+      "loss": 2.0144,
+      "step": 4625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.157016348773842e-05,
+      "loss": 1.8354,
+      "step": 4626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1568119891008174e-05,
+      "loss": 1.8231,
+      "step": 4627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.156607629427793e-05,
+      "loss": 1.7807,
+      "step": 4628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1564032697547685e-05,
+      "loss": 1.8047,
+      "step": 4629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.156198910081744e-05,
+      "loss": 1.7153,
+      "step": 4630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1559945504087196e-05,
+      "loss": 1.7835,
+      "step": 4631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1557901907356948e-05,
+      "loss": 1.7741,
+      "step": 4632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1555858310626704e-05,
+      "loss": 1.8203,
+      "step": 4633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.155381471389646e-05,
+      "loss": 1.9434,
+      "step": 4634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1551771117166212e-05,
+      "loss": 1.692,
+      "step": 4635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1549727520435967e-05,
+      "loss": 1.9232,
+      "step": 4636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.154768392370572e-05,
+      "loss": 1.7722,
+      "step": 4637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.154564032697548e-05,
+      "loss": 1.8423,
+      "step": 4638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1543596730245234e-05,
+      "loss": 1.7606,
+      "step": 4639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1541553133514987e-05,
+      "loss": 1.7662,
+      "step": 4640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1539509536784742e-05,
+      "loss": 1.8733,
+      "step": 4641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1537465940054494e-05,
+      "loss": 1.9207,
+      "step": 4642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.153542234332425e-05,
+      "loss": 1.7115,
+      "step": 4643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1533378746594006e-05,
+      "loss": 1.947,
+      "step": 4644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.153133514986376e-05,
+      "loss": 1.7871,
+      "step": 4645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1529291553133517e-05,
+      "loss": 1.8098,
+      "step": 4646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1527247956403273e-05,
+      "loss": 1.78,
+      "step": 4647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1525204359673025e-05,
+      "loss": 1.8527,
+      "step": 4648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.152316076294278e-05,
+      "loss": 1.7994,
+      "step": 4649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1521117166212533e-05,
+      "loss": 1.8466,
+      "step": 4650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.151907356948229e-05,
+      "loss": 1.7626,
+      "step": 4651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1517029972752044e-05,
+      "loss": 1.6579,
+      "step": 4652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.15149863760218e-05,
+      "loss": 1.7961,
+      "step": 4653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1512942779291555e-05,
+      "loss": 1.8354,
+      "step": 4654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1510899182561308e-05,
+      "loss": 1.7712,
+      "step": 4655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1508855585831063e-05,
+      "loss": 1.8305,
+      "step": 4656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.150681198910082e-05,
+      "loss": 1.846,
+      "step": 4657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.150476839237057e-05,
+      "loss": 1.8291,
+      "step": 4658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1502724795640327e-05,
+      "loss": 1.7971,
+      "step": 4659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1500681198910082e-05,
+      "loss": 1.8726,
+      "step": 4660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1498637602179838e-05,
+      "loss": 1.8668,
+      "step": 4661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1496594005449594e-05,
+      "loss": 1.7808,
+      "step": 4662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1494550408719346e-05,
+      "loss": 1.8457,
+      "step": 4663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.14925068119891e-05,
+      "loss": 1.6232,
+      "step": 4664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1490463215258857e-05,
+      "loss": 1.7999,
+      "step": 4665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.148841961852861e-05,
+      "loss": 1.9543,
+      "step": 4666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1486376021798365e-05,
+      "loss": 1.8235,
+      "step": 4667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.148433242506812e-05,
+      "loss": 1.817,
+      "step": 4668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1482288828337876e-05,
+      "loss": 1.8869,
+      "step": 4669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1480245231607632e-05,
+      "loss": 1.8447,
+      "step": 4670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1478201634877384e-05,
+      "loss": 1.8846,
+      "step": 4671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.147615803814714e-05,
+      "loss": 1.8182,
+      "step": 4672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1474114441416892e-05,
+      "loss": 1.8381,
+      "step": 4673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1472070844686648e-05,
+      "loss": 1.819,
+      "step": 4674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1470027247956407e-05,
+      "loss": 1.6571,
+      "step": 4675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.146798365122616e-05,
+      "loss": 1.762,
+      "step": 4676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1465940054495914e-05,
+      "loss": 1.7194,
+      "step": 4677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1463896457765667e-05,
+      "loss": 1.6961,
+      "step": 4678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1461852861035422e-05,
+      "loss": 1.9175,
+      "step": 4679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1459809264305178e-05,
+      "loss": 1.8495,
+      "step": 4680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.145776566757493e-05,
+      "loss": 1.7796,
+      "step": 4681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.145572207084469e-05,
+      "loss": 1.8897,
+      "step": 4682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.145367847411444e-05,
+      "loss": 1.8628,
+      "step": 4683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1451634877384197e-05,
+      "loss": 1.7633,
+      "step": 4684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1449591280653953e-05,
+      "loss": 1.7731,
+      "step": 4685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1447547683923705e-05,
+      "loss": 1.8955,
+      "step": 4686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.144550408719346e-05,
+      "loss": 1.6452,
+      "step": 4687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1443460490463216e-05,
+      "loss": 1.7078,
+      "step": 4688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.144141689373297e-05,
+      "loss": 1.7974,
+      "step": 4689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1439373297002728e-05,
+      "loss": 1.8219,
+      "step": 4690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.143732970027248e-05,
+      "loss": 1.6642,
+      "step": 4691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1435286103542235e-05,
+      "loss": 1.9244,
+      "step": 4692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.143324250681199e-05,
+      "loss": 1.7328,
+      "step": 4693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1431198910081743e-05,
+      "loss": 1.736,
+      "step": 4694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.14291553133515e-05,
+      "loss": 1.7822,
+      "step": 4695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.142711171662125e-05,
+      "loss": 1.7647,
+      "step": 4696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.142506811989101e-05,
+      "loss": 1.8671,
+      "step": 4697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1423024523160766e-05,
+      "loss": 1.8166,
+      "step": 4698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1420980926430518e-05,
+      "loss": 1.8553,
+      "step": 4699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1418937329700274e-05,
+      "loss": 1.7738,
+      "step": 4700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1416893732970026e-05,
+      "loss": 1.9366,
+      "step": 4701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.141485013623978e-05,
+      "loss": 1.8949,
+      "step": 4702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1412806539509537e-05,
+      "loss": 1.7032,
+      "step": 4703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.141076294277929e-05,
+      "loss": 1.8248,
+      "step": 4704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.140871934604905e-05,
+      "loss": 1.9088,
+      "step": 4705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1406675749318804e-05,
+      "loss": 1.7305,
+      "step": 4706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1404632152588556e-05,
+      "loss": 1.7641,
+      "step": 4707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1402588555858312e-05,
+      "loss": 1.8788,
+      "step": 4708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1400544959128064e-05,
+      "loss": 1.6975,
+      "step": 4709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.139850136239782e-05,
+      "loss": 1.7465,
+      "step": 4710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1396457765667576e-05,
+      "loss": 1.7916,
+      "step": 4711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.139441416893733e-05,
+      "loss": 1.8082,
+      "step": 4712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1392370572207087e-05,
+      "loss": 1.7222,
+      "step": 4713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.139032697547684e-05,
+      "loss": 1.8268,
+      "step": 4714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1388283378746595e-05,
+      "loss": 1.847,
+      "step": 4715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.138623978201635e-05,
+      "loss": 1.8924,
+      "step": 4716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1384196185286103e-05,
+      "loss": 1.834,
+      "step": 4717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1382152588555858e-05,
+      "loss": 1.7598,
+      "step": 4718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.138010899182561e-05,
+      "loss": 1.9012,
+      "step": 4719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.137806539509537e-05,
+      "loss": 1.7184,
+      "step": 4720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1376021798365125e-05,
+      "loss": 1.8646,
+      "step": 4721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1373978201634877e-05,
+      "loss": 1.7761,
+      "step": 4722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1371934604904633e-05,
+      "loss": 1.8101,
+      "step": 4723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.136989100817439e-05,
+      "loss": 1.7221,
+      "step": 4724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.136784741144414e-05,
+      "loss": 1.7849,
+      "step": 4725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1365803814713896e-05,
+      "loss": 1.7633,
+      "step": 4726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1363760217983652e-05,
+      "loss": 1.7231,
+      "step": 4727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1361716621253408e-05,
+      "loss": 1.6015,
+      "step": 4728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1359673024523163e-05,
+      "loss": 1.7842,
+      "step": 4729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1357629427792916e-05,
+      "loss": 1.7889,
+      "step": 4730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.135558583106267e-05,
+      "loss": 1.8108,
+      "step": 4731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1353542234332423e-05,
+      "loss": 1.808,
+      "step": 4732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.135149863760218e-05,
+      "loss": 1.6725,
+      "step": 4733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1349455040871935e-05,
+      "loss": 1.8007,
+      "step": 4734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.134741144414169e-05,
+      "loss": 1.853,
+      "step": 4735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1345367847411446e-05,
+      "loss": 1.8451,
+      "step": 4736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1343324250681198e-05,
+      "loss": 1.8552,
+      "step": 4737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1341280653950954e-05,
+      "loss": 1.7714,
+      "step": 4738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.133923705722071e-05,
+      "loss": 1.8551,
+      "step": 4739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1337193460490462e-05,
+      "loss": 1.7914,
+      "step": 4740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1335149863760217e-05,
+      "loss": 1.6828,
+      "step": 4741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1333106267029976e-05,
+      "loss": 1.6459,
+      "step": 4742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.133106267029973e-05,
+      "loss": 1.7264,
+      "step": 4743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1329019073569484e-05,
+      "loss": 1.7255,
+      "step": 4744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1326975476839237e-05,
+      "loss": 1.7481,
+      "step": 4745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1324931880108992e-05,
+      "loss": 1.9551,
+      "step": 4746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1322888283378748e-05,
+      "loss": 1.7765,
+      "step": 4747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.13208446866485e-05,
+      "loss": 1.6984,
+      "step": 4748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1318801089918256e-05,
+      "loss": 1.6768,
+      "step": 4749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.131675749318801e-05,
+      "loss": 1.6881,
+      "step": 4750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1314713896457767e-05,
+      "loss": 1.5851,
+      "step": 4751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1312670299727523e-05,
+      "loss": 1.7821,
+      "step": 4752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1310626702997275e-05,
+      "loss": 1.7535,
+      "step": 4753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.130858310626703e-05,
+      "loss": 1.8393,
+      "step": 4754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1306539509536783e-05,
+      "loss": 1.7913,
+      "step": 4755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.130449591280654e-05,
+      "loss": 1.7818,
+      "step": 4756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1302452316076297e-05,
+      "loss": 1.6359,
+      "step": 4757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.130040871934605e-05,
+      "loss": 1.7638,
+      "step": 4758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1298365122615805e-05,
+      "loss": 1.8005,
+      "step": 4759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.129632152588556e-05,
+      "loss": 1.7221,
+      "step": 4760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1294277929155313e-05,
+      "loss": 1.6652,
+      "step": 4761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.129223433242507e-05,
+      "loss": 1.9088,
+      "step": 4762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.129019073569482e-05,
+      "loss": 1.7234,
+      "step": 4763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1288147138964577e-05,
+      "loss": 1.6573,
+      "step": 4764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1286103542234336e-05,
+      "loss": 1.7129,
+      "step": 4765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1284059945504088e-05,
+      "loss": 1.8589,
+      "step": 4766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1282016348773844e-05,
+      "loss": 1.7936,
+      "step": 4767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1279972752043596e-05,
+      "loss": 1.7112,
+      "step": 4768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.127792915531335e-05,
+      "loss": 1.8935,
+      "step": 4769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1275885558583107e-05,
+      "loss": 1.7168,
+      "step": 4770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.127384196185286e-05,
+      "loss": 1.8794,
+      "step": 4771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1271798365122618e-05,
+      "loss": 1.8146,
+      "step": 4772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.126975476839237e-05,
+      "loss": 1.7794,
+      "step": 4773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1267711171662126e-05,
+      "loss": 1.7195,
+      "step": 4774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1265667574931882e-05,
+      "loss": 1.8197,
+      "step": 4775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1263623978201634e-05,
+      "loss": 1.8618,
+      "step": 4776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.126158038147139e-05,
+      "loss": 1.9037,
+      "step": 4777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1259536784741145e-05,
+      "loss": 1.745,
+      "step": 4778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.12574931880109e-05,
+      "loss": 1.6934,
+      "step": 4779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1255449591280657e-05,
+      "loss": 1.7805,
+      "step": 4780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.125340599455041e-05,
+      "loss": 1.7925,
+      "step": 4781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1251362397820164e-05,
+      "loss": 1.7901,
+      "step": 4782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.124931880108992e-05,
+      "loss": 1.7785,
+      "step": 4783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1247275204359672e-05,
+      "loss": 1.7982,
+      "step": 4784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1245231607629428e-05,
+      "loss": 1.7694,
+      "step": 4785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.124318801089918e-05,
+      "loss": 1.7545,
+      "step": 4786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.124114441416894e-05,
+      "loss": 1.7865,
+      "step": 4787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1239100817438695e-05,
+      "loss": 1.7682,
+      "step": 4788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1237057220708447e-05,
+      "loss": 1.8471,
+      "step": 4789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1235013623978203e-05,
+      "loss": 1.7267,
+      "step": 4790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1232970027247955e-05,
+      "loss": 1.9036,
+      "step": 4791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.123092643051771e-05,
+      "loss": 1.8436,
+      "step": 4792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1228882833787466e-05,
+      "loss": 1.8671,
+      "step": 4793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1226839237057222e-05,
+      "loss": 1.7849,
+      "step": 4794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1224795640326977e-05,
+      "loss": 1.927,
+      "step": 4795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1222752043596733e-05,
+      "loss": 2.0102,
+      "step": 4796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1220708446866485e-05,
+      "loss": 1.6812,
+      "step": 4797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121866485013624e-05,
+      "loss": 1.6746,
+      "step": 4798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1216621253405993e-05,
+      "loss": 1.6549,
+      "step": 4799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121457765667575e-05,
+      "loss": 1.9118,
+      "step": 4800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1212534059945505e-05,
+      "loss": 1.797,
+      "step": 4801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121049046321526e-05,
+      "loss": 1.8652,
+      "step": 4802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1208446866485016e-05,
+      "loss": 1.5752,
+      "step": 4803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1206403269754768e-05,
+      "loss": 1.7951,
+      "step": 4804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1204359673024524e-05,
+      "loss": 1.7463,
+      "step": 4805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.120231607629428e-05,
+      "loss": 1.7035,
+      "step": 4806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.120027247956403e-05,
+      "loss": 1.7911,
+      "step": 4807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1198228882833787e-05,
+      "loss": 1.7878,
+      "step": 4808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1196185286103543e-05,
+      "loss": 1.8227,
+      "step": 4809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.11941416893733e-05,
+      "loss": 1.7614,
+      "step": 4810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1192098092643054e-05,
+      "loss": 1.7376,
+      "step": 4811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1190054495912806e-05,
+      "loss": 1.7453,
+      "step": 4812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1188010899182562e-05,
+      "loss": 1.7636,
+      "step": 4813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1185967302452318e-05,
+      "loss": 1.7723,
+      "step": 4814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.118392370572207e-05,
+      "loss": 1.6737,
+      "step": 4815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1181880108991825e-05,
+      "loss": 1.7527,
+      "step": 4816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117983651226158e-05,
+      "loss": 1.7413,
+      "step": 4817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1177792915531337e-05,
+      "loss": 1.8516,
+      "step": 4818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1175749318801092e-05,
+      "loss": 1.6711,
+      "step": 4819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1173705722070845e-05,
+      "loss": 1.8048,
+      "step": 4820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.11716621253406e-05,
+      "loss": 1.8076,
+      "step": 4821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1169618528610352e-05,
+      "loss": 1.7279,
+      "step": 4822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1167574931880108e-05,
+      "loss": 1.6581,
+      "step": 4823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1165531335149867e-05,
+      "loss": 1.7342,
+      "step": 4824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.116348773841962e-05,
+      "loss": 1.6237,
+      "step": 4825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1161444141689375e-05,
+      "loss": 1.7928,
+      "step": 4826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1159400544959127e-05,
+      "loss": 1.7976,
+      "step": 4827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1157356948228883e-05,
+      "loss": 1.8152,
+      "step": 4828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.115531335149864e-05,
+      "loss": 1.7194,
+      "step": 4829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.115326975476839e-05,
+      "loss": 1.8209,
+      "step": 4830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1151226158038146e-05,
+      "loss": 1.7368,
+      "step": 4831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1149182561307905e-05,
+      "loss": 1.7312,
+      "step": 4832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1147138964577658e-05,
+      "loss": 1.7808,
+      "step": 4833
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1145095367847413e-05,
+      "loss": 1.7539,
+      "step": 4834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1143051771117166e-05,
+      "loss": 1.8221,
+      "step": 4835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.114100817438692e-05,
+      "loss": 1.7945,
+      "step": 4836
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1138964577656677e-05,
+      "loss": 1.5977,
+      "step": 4837
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.113692098092643e-05,
+      "loss": 1.7381,
+      "step": 4838
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1134877384196188e-05,
+      "loss": 1.8531,
+      "step": 4839
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.113283378746594e-05,
+      "loss": 1.7922,
+      "step": 4840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1130790190735696e-05,
+      "loss": 1.8416,
+      "step": 4841
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.112874659400545e-05,
+      "loss": 1.7405,
+      "step": 4842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1126702997275204e-05,
+      "loss": 1.7338,
+      "step": 4843
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.112465940054496e-05,
+      "loss": 1.7568,
+      "step": 4844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1122615803814712e-05,
+      "loss": 1.8483,
+      "step": 4845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1120572207084467e-05,
+      "loss": 1.6002,
+      "step": 4846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1118528610354226e-05,
+      "loss": 1.742,
+      "step": 4847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.111648501362398e-05,
+      "loss": 1.7294,
+      "step": 4848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1114441416893734e-05,
+      "loss": 1.8298,
+      "step": 4849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.111239782016349e-05,
+      "loss": 1.6709,
+      "step": 4850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1110354223433242e-05,
+      "loss": 1.7556,
+      "step": 4851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1108310626702998e-05,
+      "loss": 1.7074,
+      "step": 4852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.110626702997275e-05,
+      "loss": 1.7893,
+      "step": 4853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.110422343324251e-05,
+      "loss": 1.8448,
+      "step": 4854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1102179836512265e-05,
+      "loss": 1.8233,
+      "step": 4855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1100136239782017e-05,
+      "loss": 1.6988,
+      "step": 4856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1098092643051773e-05,
+      "loss": 1.6819,
+      "step": 4857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1096049046321525e-05,
+      "loss": 1.8364,
+      "step": 4858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.109400544959128e-05,
+      "loss": 1.7651,
+      "step": 4859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1091961852861036e-05,
+      "loss": 1.6608,
+      "step": 4860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.108991825613079e-05,
+      "loss": 1.7494,
+      "step": 4861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1087874659400547e-05,
+      "loss": 1.696,
+      "step": 4862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.10858310626703e-05,
+      "loss": 1.7928,
+      "step": 4863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1083787465940055e-05,
+      "loss": 1.7441,
+      "step": 4864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.108174386920981e-05,
+      "loss": 1.8684,
+      "step": 4865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1079700272479563e-05,
+      "loss": 1.6195,
+      "step": 4866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.107765667574932e-05,
+      "loss": 1.8291,
+      "step": 4867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1075613079019074e-05,
+      "loss": 1.6612,
+      "step": 4868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.107356948228883e-05,
+      "loss": 1.7251,
+      "step": 4869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1071525885558586e-05,
+      "loss": 1.6516,
+      "step": 4870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1069482288828338e-05,
+      "loss": 1.7739,
+      "step": 4871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1067438692098093e-05,
+      "loss": 1.7676,
+      "step": 4872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.106539509536785e-05,
+      "loss": 1.6621,
+      "step": 4873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.10633514986376e-05,
+      "loss": 1.7601,
+      "step": 4874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1061307901907357e-05,
+      "loss": 1.8086,
+      "step": 4875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1059264305177113e-05,
+      "loss": 1.7519,
+      "step": 4876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1057220708446868e-05,
+      "loss": 1.8161,
+      "step": 4877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1055177111716624e-05,
+      "loss": 1.7723,
+      "step": 4878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1053133514986376e-05,
+      "loss": 1.7136,
+      "step": 4879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1051089918256132e-05,
+      "loss": 1.88,
+      "step": 4880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1049046321525884e-05,
+      "loss": 1.81,
+      "step": 4881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.104700272479564e-05,
+      "loss": 1.7059,
+      "step": 4882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1044959128065395e-05,
+      "loss": 1.843,
+      "step": 4883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.104291553133515e-05,
+      "loss": 1.7468,
+      "step": 4884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1040871934604907e-05,
+      "loss": 1.7117,
+      "step": 4885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.103882833787466e-05,
+      "loss": 1.7109,
+      "step": 4886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1036784741144414e-05,
+      "loss": 1.7321,
+      "step": 4887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.103474114441417e-05,
+      "loss": 1.754,
+      "step": 4888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1032697547683922e-05,
+      "loss": 1.7911,
+      "step": 4889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1030653950953678e-05,
+      "loss": 1.8203,
+      "step": 4890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1028610354223437e-05,
+      "loss": 1.8533,
+      "step": 4891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.102656675749319e-05,
+      "loss": 1.8214,
+      "step": 4892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1024523160762945e-05,
+      "loss": 1.6708,
+      "step": 4893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1022479564032697e-05,
+      "loss": 1.77,
+      "step": 4894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1020435967302453e-05,
+      "loss": 1.7382,
+      "step": 4895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.101839237057221e-05,
+      "loss": 1.6884,
+      "step": 4896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.101634877384196e-05,
+      "loss": 1.8607,
+      "step": 4897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1014305177111716e-05,
+      "loss": 1.876,
+      "step": 4898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1012261580381472e-05,
+      "loss": 1.7962,
+      "step": 4899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1010217983651227e-05,
+      "loss": 1.8352,
+      "step": 4900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1008174386920983e-05,
+      "loss": 1.784,
+      "step": 4901
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1006130790190735e-05,
+      "loss": 1.7768,
+      "step": 4902
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.100408719346049e-05,
+      "loss": 1.8135,
+      "step": 4903
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1002043596730243e-05,
+      "loss": 1.7107,
+      "step": 4904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1e-05,
+      "loss": 1.7747,
+      "step": 4905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0997956403269758e-05,
+      "loss": 1.7113,
+      "step": 4906
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.099591280653951e-05,
+      "loss": 1.7991,
+      "step": 4907
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0993869209809266e-05,
+      "loss": 1.7281,
+      "step": 4908
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.099182561307902e-05,
+      "loss": 1.9188,
+      "step": 4909
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0989782016348774e-05,
+      "loss": 1.7143,
+      "step": 4910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.098773841961853e-05,
+      "loss": 1.8043,
+      "step": 4911
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.098569482288828e-05,
+      "loss": 1.8291,
+      "step": 4912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0983651226158037e-05,
+      "loss": 1.6851,
+      "step": 4913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0981607629427796e-05,
+      "loss": 1.6939,
+      "step": 4914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.097956403269755e-05,
+      "loss": 1.97,
+      "step": 4915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0977520435967304e-05,
+      "loss": 1.7375,
+      "step": 4916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0975476839237056e-05,
+      "loss": 1.7365,
+      "step": 4917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0973433242506812e-05,
+      "loss": 1.9376,
+      "step": 4918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0971389645776568e-05,
+      "loss": 1.7509,
+      "step": 4919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.096934604904632e-05,
+      "loss": 1.7886,
+      "step": 4920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.096730245231608e-05,
+      "loss": 1.7977,
+      "step": 4921
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.096525885558583e-05,
+      "loss": 1.7649,
+      "step": 4922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0963215258855587e-05,
+      "loss": 1.8422,
+      "step": 4923
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0961171662125342e-05,
+      "loss": 1.7324,
+      "step": 4924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0959128065395095e-05,
+      "loss": 1.7895,
+      "step": 4925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.095708446866485e-05,
+      "loss": 1.7613,
+      "step": 4926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0955040871934606e-05,
+      "loss": 1.7803,
+      "step": 4927
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0952997275204358e-05,
+      "loss": 1.7628,
+      "step": 4928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0950953678474117e-05,
+      "loss": 1.803,
+      "step": 4929
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.094891008174387e-05,
+      "loss": 1.902,
+      "step": 4930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0946866485013625e-05,
+      "loss": 1.7439,
+      "step": 4931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.094482288828338e-05,
+      "loss": 1.6906,
+      "step": 4932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0942779291553133e-05,
+      "loss": 1.764,
+      "step": 4933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.094073569482289e-05,
+      "loss": 1.8052,
+      "step": 4934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.093869209809264e-05,
+      "loss": 1.8639,
+      "step": 4935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.09366485013624e-05,
+      "loss": 1.7317,
+      "step": 4936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0934604904632155e-05,
+      "loss": 1.6371,
+      "step": 4937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0932561307901908e-05,
+      "loss": 1.7342,
+      "step": 4938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0930517711171663e-05,
+      "loss": 1.9161,
+      "step": 4939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0928474114441415e-05,
+      "loss": 1.9318,
+      "step": 4940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.092643051771117e-05,
+      "loss": 1.7634,
+      "step": 4941
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0924386920980927e-05,
+      "loss": 2.0282,
+      "step": 4942
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.092234332425068e-05,
+      "loss": 1.719,
+      "step": 4943
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0920299727520438e-05,
+      "loss": 1.8024,
+      "step": 4944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0918256130790194e-05,
+      "loss": 1.8218,
+      "step": 4945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0916212534059946e-05,
+      "loss": 1.7507,
+      "step": 4946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.09141689373297e-05,
+      "loss": 1.9251,
+      "step": 4947
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0912125340599454e-05,
+      "loss": 1.806,
+      "step": 4948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.091008174386921e-05,
+      "loss": 1.6505,
+      "step": 4949
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0908038147138965e-05,
+      "loss": 1.8198,
+      "step": 4950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.090599455040872e-05,
+      "loss": 1.8138,
+      "step": 4951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0903950953678476e-05,
+      "loss": 1.7638,
+      "step": 4952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.090190735694823e-05,
+      "loss": 1.6558,
+      "step": 4953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0899863760217984e-05,
+      "loss": 1.8905,
+      "step": 4954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.089782016348774e-05,
+      "loss": 1.8493,
+      "step": 4955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0895776566757492e-05,
+      "loss": 1.747,
+      "step": 4956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0893732970027248e-05,
+      "loss": 1.7466,
+      "step": 4957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0891689373297003e-05,
+      "loss": 1.7602,
+      "step": 4958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.088964577656676e-05,
+      "loss": 1.7565,
+      "step": 4959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0887602179836515e-05,
+      "loss": 1.8125,
+      "step": 4960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0885558583106267e-05,
+      "loss": 1.6522,
+      "step": 4961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0883514986376022e-05,
+      "loss": 1.7643,
+      "step": 4962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0881471389645778e-05,
+      "loss": 1.7629,
+      "step": 4963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.087942779291553e-05,
+      "loss": 1.8444,
+      "step": 4964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0877384196185286e-05,
+      "loss": 1.7976,
+      "step": 4965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.087534059945504e-05,
+      "loss": 1.9175,
+      "step": 4966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0873297002724797e-05,
+      "loss": 1.7953,
+      "step": 4967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0871253405994553e-05,
+      "loss": 1.7735,
+      "step": 4968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0869209809264305e-05,
+      "loss": 1.8218,
+      "step": 4969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.086716621253406e-05,
+      "loss": 1.7627,
+      "step": 4970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0865122615803813e-05,
+      "loss": 1.7681,
+      "step": 4971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.086307901907357e-05,
+      "loss": 1.7517,
+      "step": 4972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0861035422343328e-05,
+      "loss": 1.8265,
+      "step": 4973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.085899182561308e-05,
+      "loss": 1.7802,
+      "step": 4974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0856948228882836e-05,
+      "loss": 1.6184,
+      "step": 4975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0854904632152588e-05,
+      "loss": 1.7599,
+      "step": 4976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0852861035422343e-05,
+      "loss": 2.0221,
+      "step": 4977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.08508174386921e-05,
+      "loss": 1.7909,
+      "step": 4978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.084877384196185e-05,
+      "loss": 1.7771,
+      "step": 4979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0846730245231607e-05,
+      "loss": 1.7623,
+      "step": 4980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0844686648501366e-05,
+      "loss": 1.7965,
+      "step": 4981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0842643051771118e-05,
+      "loss": 1.6058,
+      "step": 4982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0840599455040874e-05,
+      "loss": 1.811,
+      "step": 4983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0838555858310626e-05,
+      "loss": 1.7619,
+      "step": 4984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.083651226158038e-05,
+      "loss": 1.7893,
+      "step": 4985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0834468664850137e-05,
+      "loss": 1.8131,
+      "step": 4986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.083242506811989e-05,
+      "loss": 1.5929,
+      "step": 4987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.083038147138965e-05,
+      "loss": 1.8205,
+      "step": 4988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.08283378746594e-05,
+      "loss": 1.7674,
+      "step": 4989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0826294277929156e-05,
+      "loss": 1.805,
+      "step": 4990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0824250681198912e-05,
+      "loss": 1.8025,
+      "step": 4991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0822207084468664e-05,
+      "loss": 1.7821,
+      "step": 4992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.082016348773842e-05,
+      "loss": 1.7666,
+      "step": 4993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0818119891008172e-05,
+      "loss": 1.6771,
+      "step": 4994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0816076294277928e-05,
+      "loss": 1.8462,
+      "step": 4995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0814032697547687e-05,
+      "loss": 1.7491,
+      "step": 4996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.081198910081744e-05,
+      "loss": 1.7327,
+      "step": 4997
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0809945504087195e-05,
+      "loss": 1.8127,
+      "step": 4998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.080790190735695e-05,
+      "loss": 1.818,
+      "step": 4999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0805858310626703e-05,
+      "loss": 1.7733,
+      "step": 5000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 1.5829943418502808,
+      "eval_runtime": 721.3542,
+      "eval_samples_per_second": 5.753,
+      "eval_steps_per_second": 1.439,
+      "step": 5000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0803814713896458e-05,
+      "loss": 1.6934,
+      "step": 5001
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.080177111716621e-05,
+      "loss": 1.7066,
+      "step": 5002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.079972752043597e-05,
+      "loss": 1.7929,
+      "step": 5003
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0797683923705725e-05,
+      "loss": 1.8004,
+      "step": 5004
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0795640326975477e-05,
+      "loss": 1.7157,
+      "step": 5005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0793596730245233e-05,
+      "loss": 1.6556,
+      "step": 5006
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0791553133514985e-05,
+      "loss": 1.8544,
+      "step": 5007
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.078950953678474e-05,
+      "loss": 1.6502,
+      "step": 5008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0787465940054497e-05,
+      "loss": 1.6816,
+      "step": 5009
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.078542234332425e-05,
+      "loss": 1.7774,
+      "step": 5010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0783378746594008e-05,
+      "loss": 1.7399,
+      "step": 5011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.078133514986376e-05,
+      "loss": 1.8242,
+      "step": 5012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0779291553133516e-05,
+      "loss": 1.7684,
+      "step": 5013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.077724795640327e-05,
+      "loss": 1.7869,
+      "step": 5014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0775204359673024e-05,
+      "loss": 1.7436,
+      "step": 5015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.077316076294278e-05,
+      "loss": 1.7483,
+      "step": 5016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0771117166212535e-05,
+      "loss": 1.6484,
+      "step": 5017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.076907356948229e-05,
+      "loss": 1.7599,
+      "step": 5018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0767029972752046e-05,
+      "loss": 1.7713,
+      "step": 5019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.07649863760218e-05,
+      "loss": 1.7822,
+      "step": 5020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0762942779291554e-05,
+      "loss": 1.6991,
+      "step": 5021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.076089918256131e-05,
+      "loss": 1.8135,
+      "step": 5022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0758855585831062e-05,
+      "loss": 1.7987,
+      "step": 5023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0756811989100817e-05,
+      "loss": 1.7187,
+      "step": 5024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.075476839237057e-05,
+      "loss": 1.9121,
+      "step": 5025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.075272479564033e-05,
+      "loss": 1.7642,
+      "step": 5026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0750681198910084e-05,
+      "loss": 1.6999,
+      "step": 5027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0748637602179837e-05,
+      "loss": 1.7438,
+      "step": 5028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0746594005449592e-05,
+      "loss": 1.6895,
+      "step": 5029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0744550408719345e-05,
+      "loss": 1.7855,
+      "step": 5030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.07425068119891e-05,
+      "loss": 1.7711,
+      "step": 5031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0740463215258856e-05,
+      "loss": 1.8222,
+      "step": 5032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.073841961852861e-05,
+      "loss": 1.787,
+      "step": 5033
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0736376021798367e-05,
+      "loss": 1.9114,
+      "step": 5034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0734332425068123e-05,
+      "loss": 1.8537,
+      "step": 5035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0732288828337875e-05,
+      "loss": 1.8327,
+      "step": 5036
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.073024523160763e-05,
+      "loss": 1.8178,
+      "step": 5037
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0728201634877383e-05,
+      "loss": 1.7908,
+      "step": 5038
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.072615803814714e-05,
+      "loss": 1.7654,
+      "step": 5039
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0724114441416894e-05,
+      "loss": 1.8752,
+      "step": 5040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.072207084468665e-05,
+      "loss": 1.7936,
+      "step": 5041
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0720027247956405e-05,
+      "loss": 1.7921,
+      "step": 5042
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0717983651226158e-05,
+      "loss": 1.7371,
+      "step": 5043
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0715940054495913e-05,
+      "loss": 1.7574,
+      "step": 5044
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.071389645776567e-05,
+      "loss": 1.6754,
+      "step": 5045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.071185286103542e-05,
+      "loss": 1.8809,
+      "step": 5046
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0709809264305177e-05,
+      "loss": 1.6993,
+      "step": 5047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0707765667574932e-05,
+      "loss": 1.7574,
+      "step": 5048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0705722070844688e-05,
+      "loss": 1.7478,
+      "step": 5049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0703678474114444e-05,
+      "loss": 1.7272,
+      "step": 5050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0701634877384196e-05,
+      "loss": 1.6683,
+      "step": 5051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.069959128065395e-05,
+      "loss": 1.7295,
+      "step": 5052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0697547683923707e-05,
+      "loss": 1.7993,
+      "step": 5053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.069550408719346e-05,
+      "loss": 1.655,
+      "step": 5054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.069346049046322e-05,
+      "loss": 1.8849,
+      "step": 5055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.069141689373297e-05,
+      "loss": 1.7962,
+      "step": 5056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0689373297002726e-05,
+      "loss": 1.717,
+      "step": 5057
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0687329700272482e-05,
+      "loss": 1.7764,
+      "step": 5058
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0685286103542234e-05,
+      "loss": 1.9798,
+      "step": 5059
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.068324250681199e-05,
+      "loss": 1.7553,
+      "step": 5060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0681198910081742e-05,
+      "loss": 3.0186,
+      "step": 5061
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0679155313351498e-05,
+      "loss": 1.7608,
+      "step": 5062
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0677111716621257e-05,
+      "loss": 1.7174,
+      "step": 5063
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.067506811989101e-05,
+      "loss": 1.7072,
+      "step": 5064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0673024523160765e-05,
+      "loss": 1.7446,
+      "step": 5065
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0670980926430517e-05,
+      "loss": 1.6865,
+      "step": 5066
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0668937329700272e-05,
+      "loss": 1.7789,
+      "step": 5067
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0666893732970028e-05,
+      "loss": 1.8378,
+      "step": 5068
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.066485013623978e-05,
+      "loss": 1.8311,
+      "step": 5069
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.066280653950954e-05,
+      "loss": 1.7102,
+      "step": 5070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.066076294277929e-05,
+      "loss": 1.6766,
+      "step": 5071
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0658719346049047e-05,
+      "loss": 1.7541,
+      "step": 5072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0656675749318803e-05,
+      "loss": 1.9377,
+      "step": 5073
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0654632152588555e-05,
+      "loss": 1.757,
+      "step": 5074
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.065258855585831e-05,
+      "loss": 1.7206,
+      "step": 5075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0650544959128066e-05,
+      "loss": 1.7647,
+      "step": 5076
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.064850136239782e-05,
+      "loss": 1.5984,
+      "step": 5077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0646457765667578e-05,
+      "loss": 1.7273,
+      "step": 5078
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.064441416893733e-05,
+      "loss": 1.7183,
+      "step": 5079
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0642370572207085e-05,
+      "loss": 1.8371,
+      "step": 5080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.064032697547684e-05,
+      "loss": 1.9178,
+      "step": 5081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0638283378746593e-05,
+      "loss": 1.9338,
+      "step": 5082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.063623978201635e-05,
+      "loss": 1.6727,
+      "step": 5083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.06341961852861e-05,
+      "loss": 1.7955,
+      "step": 5084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.063215258855586e-05,
+      "loss": 1.6847,
+      "step": 5085
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0630108991825616e-05,
+      "loss": 1.6635,
+      "step": 5086
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0628065395095368e-05,
+      "loss": 1.758,
+      "step": 5087
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0626021798365124e-05,
+      "loss": 1.7339,
+      "step": 5088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0623978201634876e-05,
+      "loss": 1.8763,
+      "step": 5089
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.062193460490463e-05,
+      "loss": 1.8862,
+      "step": 5090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0619891008174387e-05,
+      "loss": 1.8743,
+      "step": 5091
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.061784741144414e-05,
+      "loss": 1.6117,
+      "step": 5092
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.06158038147139e-05,
+      "loss": 1.7353,
+      "step": 5093
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0613760217983654e-05,
+      "loss": 1.7725,
+      "step": 5094
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0611716621253406e-05,
+      "loss": 1.7896,
+      "step": 5095
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0609673024523162e-05,
+      "loss": 1.8852,
+      "step": 5096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0607629427792914e-05,
+      "loss": 1.7667,
+      "step": 5097
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.060558583106267e-05,
+      "loss": 1.8223,
+      "step": 5098
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0603542234332426e-05,
+      "loss": 1.666,
+      "step": 5099
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.060149863760218e-05,
+      "loss": 1.6964,
+      "step": 5100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0599455040871937e-05,
+      "loss": 1.732,
+      "step": 5101
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.059741144414169e-05,
+      "loss": 1.9249,
+      "step": 5102
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0595367847411445e-05,
+      "loss": 1.6874,
+      "step": 5103
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.05933242506812e-05,
+      "loss": 1.7427,
+      "step": 5104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0591280653950953e-05,
+      "loss": 1.8906,
+      "step": 5105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0589237057220708e-05,
+      "loss": 1.7699,
+      "step": 5106
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.058719346049046e-05,
+      "loss": 1.823,
+      "step": 5107
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.058514986376022e-05,
+      "loss": 1.7673,
+      "step": 5108
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0583106267029975e-05,
+      "loss": 1.865,
+      "step": 5109
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0581062670299727e-05,
+      "loss": 1.7713,
+      "step": 5110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0579019073569483e-05,
+      "loss": 1.7543,
+      "step": 5111
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.057697547683924e-05,
+      "loss": 1.8518,
+      "step": 5112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.057493188010899e-05,
+      "loss": 1.7642,
+      "step": 5113
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0572888283378746e-05,
+      "loss": 1.7644,
+      "step": 5114
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0570844686648502e-05,
+      "loss": 1.7774,
+      "step": 5115
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0568801089918258e-05,
+      "loss": 1.7332,
+      "step": 5116
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0566757493188013e-05,
+      "loss": 1.7367,
+      "step": 5117
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0564713896457766e-05,
+      "loss": 1.6256,
+      "step": 5118
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.056267029972752e-05,
+      "loss": 1.6768,
+      "step": 5119
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0560626702997274e-05,
+      "loss": 1.7163,
+      "step": 5120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.055858310626703e-05,
+      "loss": 1.828,
+      "step": 5121
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0556539509536785e-05,
+      "loss": 1.7535,
+      "step": 5122
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.055449591280654e-05,
+      "loss": 1.7387,
+      "step": 5123
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0552452316076296e-05,
+      "loss": 1.7513,
+      "step": 5124
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0550408719346048e-05,
+      "loss": 1.7569,
+      "step": 5125
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0548365122615804e-05,
+      "loss": 1.7023,
+      "step": 5126
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.054632152588556e-05,
+      "loss": 1.7523,
+      "step": 5127
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0544277929155312e-05,
+      "loss": 1.7409,
+      "step": 5128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0542234332425067e-05,
+      "loss": 1.7251,
+      "step": 5129
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0540190735694826e-05,
+      "loss": 1.7407,
+      "step": 5130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.053814713896458e-05,
+      "loss": 1.717,
+      "step": 5131
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0536103542234334e-05,
+      "loss": 1.6723,
+      "step": 5132
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0534059945504087e-05,
+      "loss": 1.8064,
+      "step": 5133
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0532016348773842e-05,
+      "loss": 1.76,
+      "step": 5134
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0529972752043598e-05,
+      "loss": 1.786,
+      "step": 5135
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.052792915531335e-05,
+      "loss": 1.995,
+      "step": 5136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.052588555858311e-05,
+      "loss": 1.7012,
+      "step": 5137
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.052384196185286e-05,
+      "loss": 1.6698,
+      "step": 5138
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0521798365122617e-05,
+      "loss": 1.785,
+      "step": 5139
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0519754768392373e-05,
+      "loss": 1.8138,
+      "step": 5140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0517711171662125e-05,
+      "loss": 1.7698,
+      "step": 5141
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.051566757493188e-05,
+      "loss": 1.6437,
+      "step": 5142
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0513623978201633e-05,
+      "loss": 1.7571,
+      "step": 5143
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.051158038147139e-05,
+      "loss": 1.7949,
+      "step": 5144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0509536784741147e-05,
+      "loss": 1.7272,
+      "step": 5145
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.05074931880109e-05,
+      "loss": 1.827,
+      "step": 5146
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0505449591280655e-05,
+      "loss": 1.7855,
+      "step": 5147
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.050340599455041e-05,
+      "loss": 1.7193,
+      "step": 5148
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0501362397820163e-05,
+      "loss": 1.7106,
+      "step": 5149
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.049931880108992e-05,
+      "loss": 1.6877,
+      "step": 5150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.049727520435967e-05,
+      "loss": 1.6763,
+      "step": 5151
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.049523160762943e-05,
+      "loss": 1.8472,
+      "step": 5152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0493188010899186e-05,
+      "loss": 1.8274,
+      "step": 5153
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0491144414168938e-05,
+      "loss": 1.8882,
+      "step": 5154
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0489100817438694e-05,
+      "loss": 1.7322,
+      "step": 5155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0487057220708446e-05,
+      "loss": 1.9212,
+      "step": 5156
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.04850136239782e-05,
+      "loss": 1.7511,
+      "step": 5157
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0482970027247957e-05,
+      "loss": 1.9282,
+      "step": 5158
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.048092643051771e-05,
+      "loss": 1.6323,
+      "step": 5159
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.047888283378747e-05,
+      "loss": 1.7363,
+      "step": 5160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.047683923705722e-05,
+      "loss": 1.6226,
+      "step": 5161
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0474795640326976e-05,
+      "loss": 1.725,
+      "step": 5162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0472752043596732e-05,
+      "loss": 1.6508,
+      "step": 5163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0470708446866484e-05,
+      "loss": 1.7823,
+      "step": 5164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.046866485013624e-05,
+      "loss": 1.7105,
+      "step": 5165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0466621253405995e-05,
+      "loss": 1.7461,
+      "step": 5166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.046457765667575e-05,
+      "loss": 1.829,
+      "step": 5167
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0462534059945507e-05,
+      "loss": 1.7065,
+      "step": 5168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.046049046321526e-05,
+      "loss": 1.7901,
+      "step": 5169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0458446866485014e-05,
+      "loss": 1.7522,
+      "step": 5170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.045640326975477e-05,
+      "loss": 1.6607,
+      "step": 5171
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0454359673024522e-05,
+      "loss": 1.7273,
+      "step": 5172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0452316076294278e-05,
+      "loss": 1.8241,
+      "step": 5173
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.045027247956403e-05,
+      "loss": 1.9106,
+      "step": 5174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.044822888283379e-05,
+      "loss": 1.6543,
+      "step": 5175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0446185286103545e-05,
+      "loss": 1.7924,
+      "step": 5176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0444141689373297e-05,
+      "loss": 1.7964,
+      "step": 5177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0442098092643053e-05,
+      "loss": 1.731,
+      "step": 5178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0440054495912805e-05,
+      "loss": 1.7075,
+      "step": 5179
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.043801089918256e-05,
+      "loss": 1.7231,
+      "step": 5180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0435967302452316e-05,
+      "loss": 1.7464,
+      "step": 5181
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0433923705722072e-05,
+      "loss": 1.8147,
+      "step": 5182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0431880108991828e-05,
+      "loss": 1.732,
+      "step": 5183
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0429836512261583e-05,
+      "loss": 1.7438,
+      "step": 5184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0427792915531335e-05,
+      "loss": 1.7466,
+      "step": 5185
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.042574931880109e-05,
+      "loss": 1.8831,
+      "step": 5186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0423705722070843e-05,
+      "loss": 1.7337,
+      "step": 5187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.04216621253406e-05,
+      "loss": 1.7817,
+      "step": 5188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0419618528610355e-05,
+      "loss": 1.8278,
+      "step": 5189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.041757493188011e-05,
+      "loss": 1.6941,
+      "step": 5190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0415531335149866e-05,
+      "loss": 1.876,
+      "step": 5191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0413487738419618e-05,
+      "loss": 1.6731,
+      "step": 5192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0411444141689374e-05,
+      "loss": 1.7084,
+      "step": 5193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.040940054495913e-05,
+      "loss": 1.9017,
+      "step": 5194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.040735694822888e-05,
+      "loss": 1.6012,
+      "step": 5195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0405313351498637e-05,
+      "loss": 1.778,
+      "step": 5196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0403269754768393e-05,
+      "loss": 1.8557,
+      "step": 5197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.040122615803815e-05,
+      "loss": 1.7627,
+      "step": 5198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0399182561307904e-05,
+      "loss": 1.8455,
+      "step": 5199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0397138964577656e-05,
+      "loss": 1.8274,
+      "step": 5200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0395095367847412e-05,
+      "loss": 1.7324,
+      "step": 5201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0393051771117168e-05,
+      "loss": 1.7801,
+      "step": 5202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.039100817438692e-05,
+      "loss": 1.7621,
+      "step": 5203
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0388964577656676e-05,
+      "loss": 1.6632,
+      "step": 5204
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.038692098092643e-05,
+      "loss": 1.8617,
+      "step": 5205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0384877384196187e-05,
+      "loss": 1.7934,
+      "step": 5206
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0382833787465942e-05,
+      "loss": 1.6487,
+      "step": 5207
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0380790190735695e-05,
+      "loss": 1.7132,
+      "step": 5208
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.037874659400545e-05,
+      "loss": 1.722,
+      "step": 5209
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0376702997275203e-05,
+      "loss": 1.5637,
+      "step": 5210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0374659400544958e-05,
+      "loss": 1.7574,
+      "step": 5211
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0372615803814717e-05,
+      "loss": 1.7527,
+      "step": 5212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.037057220708447e-05,
+      "loss": 1.8089,
+      "step": 5213
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0368528610354225e-05,
+      "loss": 1.677,
+      "step": 5214
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0366485013623977e-05,
+      "loss": 1.7439,
+      "step": 5215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0364441416893733e-05,
+      "loss": 1.7159,
+      "step": 5216
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.036239782016349e-05,
+      "loss": 1.7783,
+      "step": 5217
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.036035422343324e-05,
+      "loss": 1.7508,
+      "step": 5218
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0358310626702996e-05,
+      "loss": 1.7417,
+      "step": 5219
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0356267029972755e-05,
+      "loss": 1.6437,
+      "step": 5220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0354223433242508e-05,
+      "loss": 1.7633,
+      "step": 5221
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0352179836512263e-05,
+      "loss": 1.795,
+      "step": 5222
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0350136239782016e-05,
+      "loss": 1.7899,
+      "step": 5223
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.034809264305177e-05,
+      "loss": 1.9291,
+      "step": 5224
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0346049046321527e-05,
+      "loss": 1.8404,
+      "step": 5225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.034400544959128e-05,
+      "loss": 1.6699,
+      "step": 5226
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0341961852861038e-05,
+      "loss": 1.6931,
+      "step": 5227
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.033991825613079e-05,
+      "loss": 1.7409,
+      "step": 5228
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0337874659400546e-05,
+      "loss": 1.7407,
+      "step": 5229
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.03358310626703e-05,
+      "loss": 1.9276,
+      "step": 5230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0333787465940054e-05,
+      "loss": 1.6902,
+      "step": 5231
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.033174386920981e-05,
+      "loss": 1.6634,
+      "step": 5232
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0329700272479562e-05,
+      "loss": 1.7036,
+      "step": 5233
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.032765667574932e-05,
+      "loss": 1.6779,
+      "step": 5234
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0325613079019076e-05,
+      "loss": 1.5677,
+      "step": 5235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.032356948228883e-05,
+      "loss": 1.7858,
+      "step": 5236
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0321525885558584e-05,
+      "loss": 1.7543,
+      "step": 5237
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.031948228882834e-05,
+      "loss": 1.7533,
+      "step": 5238
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0317438692098092e-05,
+      "loss": 1.7555,
+      "step": 5239
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0315395095367848e-05,
+      "loss": 1.6835,
+      "step": 5240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.03133514986376e-05,
+      "loss": 1.8116,
+      "step": 5241
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.031130790190736e-05,
+      "loss": 1.8167,
+      "step": 5242
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0309264305177115e-05,
+      "loss": 1.7493,
+      "step": 5243
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0307220708446867e-05,
+      "loss": 1.7529,
+      "step": 5244
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0305177111716623e-05,
+      "loss": 1.6833,
+      "step": 5245
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0303133514986375e-05,
+      "loss": 1.8771,
+      "step": 5246
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.030108991825613e-05,
+      "loss": 1.7211,
+      "step": 5247
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0299046321525886e-05,
+      "loss": 1.8063,
+      "step": 5248
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0297002724795642e-05,
+      "loss": 1.8549,
+      "step": 5249
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0294959128065397e-05,
+      "loss": 1.7037,
+      "step": 5250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.029291553133515e-05,
+      "loss": 1.7552,
+      "step": 5251
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0290871934604905e-05,
+      "loss": 1.7186,
+      "step": 5252
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.028882833787466e-05,
+      "loss": 1.812,
+      "step": 5253
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0286784741144413e-05,
+      "loss": 1.7481,
+      "step": 5254
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.028474114441417e-05,
+      "loss": 1.7475,
+      "step": 5255
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0282697547683924e-05,
+      "loss": 1.6941,
+      "step": 5256
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.028065395095368e-05,
+      "loss": 1.6855,
+      "step": 5257
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0278610354223436e-05,
+      "loss": 1.9214,
+      "step": 5258
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0276566757493188e-05,
+      "loss": 1.954,
+      "step": 5259
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0274523160762944e-05,
+      "loss": 1.6739,
+      "step": 5260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.02724795640327e-05,
+      "loss": 1.6643,
+      "step": 5261
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.027043596730245e-05,
+      "loss": 1.7577,
+      "step": 5262
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0268392370572207e-05,
+      "loss": 1.8144,
+      "step": 5263
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0266348773841963e-05,
+      "loss": 1.6949,
+      "step": 5264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0264305177111718e-05,
+      "loss": 1.8112,
+      "step": 5265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0262261580381474e-05,
+      "loss": 1.7121,
+      "step": 5266
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0260217983651226e-05,
+      "loss": 1.8592,
+      "step": 5267
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0258174386920982e-05,
+      "loss": 1.817,
+      "step": 5268
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0256130790190734e-05,
+      "loss": 1.6469,
+      "step": 5269
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.025408719346049e-05,
+      "loss": 1.7795,
+      "step": 5270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0252043596730245e-05,
+      "loss": 1.8244,
+      "step": 5271
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.025e-05,
+      "loss": 1.6816,
+      "step": 5272
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0247956403269757e-05,
+      "loss": 1.7714,
+      "step": 5273
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.024591280653951e-05,
+      "loss": 1.8546,
+      "step": 5274
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0243869209809264e-05,
+      "loss": 1.7742,
+      "step": 5275
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.024182561307902e-05,
+      "loss": 1.7035,
+      "step": 5276
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0239782016348772e-05,
+      "loss": 1.8312,
+      "step": 5277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0237738419618528e-05,
+      "loss": 1.7781,
+      "step": 5278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0235694822888287e-05,
+      "loss": 1.7808,
+      "step": 5279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.023365122615804e-05,
+      "loss": 1.7265,
+      "step": 5280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0231607629427795e-05,
+      "loss": 1.7164,
+      "step": 5281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0229564032697547e-05,
+      "loss": 1.7466,
+      "step": 5282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0227520435967303e-05,
+      "loss": 1.7222,
+      "step": 5283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.022547683923706e-05,
+      "loss": 1.739,
+      "step": 5284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.022343324250681e-05,
+      "loss": 1.8395,
+      "step": 5285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0221389645776566e-05,
+      "loss": 1.7421,
+      "step": 5286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.0219346049046322e-05,
+      "loss": 1.8455,
+      "step": 5287
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0217302452316078e-05,
+      "loss": 1.6861,
+      "step": 5288
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0215258855585833e-05,
+      "loss": 1.846,
+      "step": 5289
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0213215258855585e-05,
+      "loss": 1.7463,
+      "step": 5290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.021117166212534e-05,
+      "loss": 1.7482,
+      "step": 5291
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0209128065395093e-05,
+      "loss": 1.8091,
+      "step": 5292
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.020708446866485e-05,
+      "loss": 1.6084,
+      "step": 5293
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0205040871934608e-05,
+      "loss": 1.7389,
+      "step": 5294
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.020299727520436e-05,
+      "loss": 1.7664,
+      "step": 5295
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0200953678474116e-05,
+      "loss": 1.9614,
+      "step": 5296
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.019891008174387e-05,
+      "loss": 1.6682,
+      "step": 5297
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0196866485013624e-05,
+      "loss": 1.7193,
+      "step": 5298
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.019482288828338e-05,
+      "loss": 1.8294,
+      "step": 5299
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.019277929155313e-05,
+      "loss": 1.7007,
+      "step": 5300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0190735694822887e-05,
+      "loss": 1.7417,
+      "step": 5301
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0188692098092646e-05,
+      "loss": 1.9064,
+      "step": 5302
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.01866485013624e-05,
+      "loss": 1.7009,
+      "step": 5303
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0184604904632154e-05,
+      "loss": 1.7442,
+      "step": 5304
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0182561307901906e-05,
+      "loss": 1.7409,
+      "step": 5305
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0180517711171662e-05,
+      "loss": 1.7233,
+      "step": 5306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0178474114441418e-05,
+      "loss": 1.6831,
+      "step": 5307
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.017643051771117e-05,
+      "loss": 1.5807,
+      "step": 5308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.017438692098093e-05,
+      "loss": 1.6981,
+      "step": 5309
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.017234332425068e-05,
+      "loss": 1.7914,
+      "step": 5310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0170299727520437e-05,
+      "loss": 1.7134,
+      "step": 5311
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0168256130790192e-05,
+      "loss": 1.8287,
+      "step": 5312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0166212534059945e-05,
+      "loss": 1.8534,
+      "step": 5313
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.01641689373297e-05,
+      "loss": 1.8048,
+      "step": 5314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0162125340599456e-05,
+      "loss": 1.7565,
+      "step": 5315
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0160081743869208e-05,
+      "loss": 1.7041,
+      "step": 5316
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0158038147138967e-05,
+      "loss": 1.6877,
+      "step": 5317
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.015599455040872e-05,
+      "loss": 1.7576,
+      "step": 5318
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0153950953678475e-05,
+      "loss": 1.7039,
+      "step": 5319
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.015190735694823e-05,
+      "loss": 1.659,
+      "step": 5320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0149863760217983e-05,
+      "loss": 1.7198,
+      "step": 5321
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.014782016348774e-05,
+      "loss": 1.6815,
+      "step": 5322
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.014577656675749e-05,
+      "loss": 1.7401,
+      "step": 5323
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.014373297002725e-05,
+      "loss": 1.7349,
+      "step": 5324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0141689373297005e-05,
+      "loss": 1.6118,
+      "step": 5325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0139645776566758e-05,
+      "loss": 1.7791,
+      "step": 5326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0137602179836513e-05,
+      "loss": 1.7918,
+      "step": 5327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0135558583106266e-05,
+      "loss": 1.528,
+      "step": 5328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.013351498637602e-05,
+      "loss": 1.7433,
+      "step": 5329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0131471389645777e-05,
+      "loss": 1.6957,
+      "step": 5330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0129427792915532e-05,
+      "loss": 1.926,
+      "step": 5331
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0127384196185288e-05,
+      "loss": 1.8048,
+      "step": 5332
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0125340599455044e-05,
+      "loss": 1.6959,
+      "step": 5333
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0123297002724796e-05,
+      "loss": 1.8237,
+      "step": 5334
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.012125340599455e-05,
+      "loss": 1.7491,
+      "step": 5335
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0119209809264304e-05,
+      "loss": 1.6965,
+      "step": 5336
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.011716621253406e-05,
+      "loss": 1.769,
+      "step": 5337
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.0115122615803815e-05,
+      "loss": 1.7257,
+      "step": 5338
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.011307901907357e-05,
+      "loss": 1.6312,
+      "step": 5339
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0111035422343326e-05,
+      "loss": 1.7933,
+      "step": 5340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.010899182561308e-05,
+      "loss": 1.6746,
+      "step": 5341
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0106948228882834e-05,
+      "loss": 1.7522,
+      "step": 5342
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.010490463215259e-05,
+      "loss": 1.7268,
+      "step": 5343
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0102861035422342e-05,
+      "loss": 1.8788,
+      "step": 5344
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0100817438692098e-05,
+      "loss": 1.6914,
+      "step": 5345
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0098773841961853e-05,
+      "loss": 1.8354,
+      "step": 5346
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.009673024523161e-05,
+      "loss": 1.7761,
+      "step": 5347
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0094686648501365e-05,
+      "loss": 1.7961,
+      "step": 5348
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0092643051771117e-05,
+      "loss": 1.7902,
+      "step": 5349
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0090599455040873e-05,
+      "loss": 1.7025,
+      "step": 5350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0088555858310628e-05,
+      "loss": 1.874,
+      "step": 5351
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.008651226158038e-05,
+      "loss": 1.7083,
+      "step": 5352
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0084468664850136e-05,
+      "loss": 1.872,
+      "step": 5353
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.008242506811989e-05,
+      "loss": 1.8019,
+      "step": 5354
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0080381471389647e-05,
+      "loss": 1.8498,
+      "step": 5355
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0078337874659403e-05,
+      "loss": 1.7442,
+      "step": 5356
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0076294277929155e-05,
+      "loss": 1.8023,
+      "step": 5357
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.007425068119891e-05,
+      "loss": 1.6745,
+      "step": 5358
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0072207084468663e-05,
+      "loss": 1.6471,
+      "step": 5359
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.007016348773842e-05,
+      "loss": 1.7557,
+      "step": 5360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0068119891008178e-05,
+      "loss": 1.7451,
+      "step": 5361
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.006607629427793e-05,
+      "loss": 1.6986,
+      "step": 5362
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0064032697547686e-05,
+      "loss": 1.8283,
+      "step": 5363
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0061989100817438e-05,
+      "loss": 1.6916,
+      "step": 5364
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0059945504087193e-05,
+      "loss": 1.8394,
+      "step": 5365
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.005790190735695e-05,
+      "loss": 1.7534,
+      "step": 5366
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.00558583106267e-05,
+      "loss": 1.7439,
+      "step": 5367
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0053814713896457e-05,
+      "loss": 1.7704,
+      "step": 5368
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0051771117166216e-05,
+      "loss": 1.7807,
+      "step": 5369
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0049727520435968e-05,
+      "loss": 1.8712,
+      "step": 5370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0047683923705724e-05,
+      "loss": 1.6013,
+      "step": 5371
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0045640326975476e-05,
+      "loss": 1.7822,
+      "step": 5372
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0043596730245232e-05,
+      "loss": 1.6878,
+      "step": 5373
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0041553133514987e-05,
+      "loss": 1.5529,
+      "step": 5374
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.003950953678474e-05,
+      "loss": 1.7033,
+      "step": 5375
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.00374659400545e-05,
+      "loss": 1.8639,
+      "step": 5376
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.003542234332425e-05,
+      "loss": 1.6038,
+      "step": 5377
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0033378746594007e-05,
+      "loss": 1.7789,
+      "step": 5378
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0031335149863762e-05,
+      "loss": 1.8104,
+      "step": 5379
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0029291553133514e-05,
+      "loss": 1.7821,
+      "step": 5380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.002724795640327e-05,
+      "loss": 1.7637,
+      "step": 5381
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0025204359673022e-05,
+      "loss": 1.8952,
+      "step": 5382
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0023160762942778e-05,
+      "loss": 1.588,
+      "step": 5383
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0021117166212537e-05,
+      "loss": 1.7075,
+      "step": 5384
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.001907356948229e-05,
+      "loss": 1.7272,
+      "step": 5385
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0017029972752045e-05,
+      "loss": 1.5939,
+      "step": 5386
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.00149863760218e-05,
+      "loss": 1.7027,
+      "step": 5387
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.0012942779291553e-05,
+      "loss": 1.7843,
+      "step": 5388
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.001089918256131e-05,
+      "loss": 1.7319,
+      "step": 5389
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.000885558583106e-05,
+      "loss": 1.7612,
+      "step": 5390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.000681198910082e-05,
+      "loss": 1.6279,
+      "step": 5391
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.0004768392370575e-05,
+      "loss": 1.7382,
+      "step": 5392
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.0002724795640327e-05,
+      "loss": 1.7527,
+      "step": 5393
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.0000681198910083e-05,
+      "loss": 1.6665,
+      "step": 5394
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9998637602179835e-05,
+      "loss": 1.7708,
+      "step": 5395
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999659400544959e-05,
+      "loss": 1.7473,
+      "step": 5396
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9994550408719347e-05,
+      "loss": 1.8894,
+      "step": 5397
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.99925068119891e-05,
+      "loss": 1.8053,
+      "step": 5398
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9990463215258858e-05,
+      "loss": 1.6426,
+      "step": 5399
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.998841961852861e-05,
+      "loss": 1.6563,
+      "step": 5400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9986376021798366e-05,
+      "loss": 1.6808,
+      "step": 5401
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.998433242506812e-05,
+      "loss": 1.7392,
+      "step": 5402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9982288828337874e-05,
+      "loss": 1.6519,
+      "step": 5403
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.998024523160763e-05,
+      "loss": 1.7402,
+      "step": 5404
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9978201634877385e-05,
+      "loss": 1.7977,
+      "step": 5405
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997615803814714e-05,
+      "loss": 1.8106,
+      "step": 5406
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9974114441416896e-05,
+      "loss": 1.6379,
+      "step": 5407
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.997207084468665e-05,
+      "loss": 1.7935,
+      "step": 5408
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9970027247956404e-05,
+      "loss": 1.708,
+      "step": 5409
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.996798365122616e-05,
+      "loss": 1.7204,
+      "step": 5410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9965940054495912e-05,
+      "loss": 1.7184,
+      "step": 5411
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9963896457765668e-05,
+      "loss": 1.8162,
+      "step": 5412
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9961852861035423e-05,
+      "loss": 1.7098,
+      "step": 5413
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.995980926430518e-05,
+      "loss": 1.7583,
+      "step": 5414
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9957765667574934e-05,
+      "loss": 1.6808,
+      "step": 5415
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9955722070844687e-05,
+      "loss": 1.8589,
+      "step": 5416
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9953678474114442e-05,
+      "loss": 1.7547,
+      "step": 5417
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9951634877384195e-05,
+      "loss": 1.684,
+      "step": 5418
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.994959128065395e-05,
+      "loss": 1.5661,
+      "step": 5419
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9947547683923706e-05,
+      "loss": 1.7132,
+      "step": 5420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.994550408719346e-05,
+      "loss": 1.8077,
+      "step": 5421
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9943460490463217e-05,
+      "loss": 1.8977,
+      "step": 5422
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9941416893732973e-05,
+      "loss": 1.8949,
+      "step": 5423
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9939373297002725e-05,
+      "loss": 1.7542,
+      "step": 5424
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.993732970027248e-05,
+      "loss": 1.7237,
+      "step": 5425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9935286103542233e-05,
+      "loss": 1.9057,
+      "step": 5426
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.993324250681199e-05,
+      "loss": 1.8655,
+      "step": 5427
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9931198910081747e-05,
+      "loss": 1.7347,
+      "step": 5428
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.99291553133515e-05,
+      "loss": 1.9744,
+      "step": 5429
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9927111716621255e-05,
+      "loss": 1.8106,
+      "step": 5430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9925068119891008e-05,
+      "loss": 1.703,
+      "step": 5431
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9923024523160763e-05,
+      "loss": 1.783,
+      "step": 5432
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.992098092643052e-05,
+      "loss": 1.8159,
+      "step": 5433
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.991893732970027e-05,
+      "loss": 1.805,
+      "step": 5434
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9916893732970027e-05,
+      "loss": 1.752,
+      "step": 5435
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9914850136239782e-05,
+      "loss": 1.6628,
+      "step": 5436
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9912806539509538e-05,
+      "loss": 1.7488,
+      "step": 5437
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9910762942779294e-05,
+      "loss": 1.7395,
+      "step": 5438
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9908719346049046e-05,
+      "loss": 1.7334,
+      "step": 5439
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.99066757493188e-05,
+      "loss": 1.819,
+      "step": 5440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9904632152588557e-05,
+      "loss": 1.7215,
+      "step": 5441
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.990258855585831e-05,
+      "loss": 1.7333,
+      "step": 5442
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.990054495912807e-05,
+      "loss": 1.6721,
+      "step": 5443
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.989850136239782e-05,
+      "loss": 1.7241,
+      "step": 5444
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9896457765667576e-05,
+      "loss": 1.6735,
+      "step": 5445
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9894414168937332e-05,
+      "loss": 1.7875,
+      "step": 5446
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9892370572207084e-05,
+      "loss": 1.8164,
+      "step": 5447
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.989032697547684e-05,
+      "loss": 1.7424,
+      "step": 5448
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9888283378746592e-05,
+      "loss": 1.6237,
+      "step": 5449
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9886239782016348e-05,
+      "loss": 1.7287,
+      "step": 5450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9884196185286107e-05,
+      "loss": 1.8415,
+      "step": 5451
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.988215258855586e-05,
+      "loss": 1.751,
+      "step": 5452
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9880108991825615e-05,
+      "loss": 1.7017,
+      "step": 5453
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9878065395095367e-05,
+      "loss": 1.7214,
+      "step": 5454
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9876021798365122e-05,
+      "loss": 1.8776,
+      "step": 5455
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9873978201634878e-05,
+      "loss": 1.8831,
+      "step": 5456
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.987193460490463e-05,
+      "loss": 1.7037,
+      "step": 5457
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.986989100817439e-05,
+      "loss": 1.5765,
+      "step": 5458
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.986784741144414e-05,
+      "loss": 1.7436,
+      "step": 5459
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9865803814713897e-05,
+      "loss": 1.6769,
+      "step": 5460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9863760217983653e-05,
+      "loss": 1.6733,
+      "step": 5461
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9861716621253405e-05,
+      "loss": 1.8879,
+      "step": 5462
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.985967302452316e-05,
+      "loss": 1.767,
+      "step": 5463
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9857629427792916e-05,
+      "loss": 1.7168,
+      "step": 5464
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.985558583106267e-05,
+      "loss": 1.8238,
+      "step": 5465
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9853542234332428e-05,
+      "loss": 1.7933,
+      "step": 5466
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.985149863760218e-05,
+      "loss": 1.8017,
+      "step": 5467
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9849455040871936e-05,
+      "loss": 1.6758,
+      "step": 5468
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.984741144414169e-05,
+      "loss": 1.6471,
+      "step": 5469
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9845367847411443e-05,
+      "loss": 1.8127,
+      "step": 5470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.98433242506812e-05,
+      "loss": 1.8162,
+      "step": 5471
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.984128065395095e-05,
+      "loss": 1.6462,
+      "step": 5472
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.983923705722071e-05,
+      "loss": 1.7622,
+      "step": 5473
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9837193460490466e-05,
+      "loss": 1.8078,
+      "step": 5474
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9835149863760218e-05,
+      "loss": 1.7877,
+      "step": 5475
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9833106267029974e-05,
+      "loss": 1.9052,
+      "step": 5476
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9831062670299726e-05,
+      "loss": 1.7371,
+      "step": 5477
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9829019073569482e-05,
+      "loss": 1.8287,
+      "step": 5478
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9826975476839237e-05,
+      "loss": 1.7679,
+      "step": 5479
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.982493188010899e-05,
+      "loss": 1.8042,
+      "step": 5480
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.982288828337875e-05,
+      "loss": 1.8849,
+      "step": 5481
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9820844686648504e-05,
+      "loss": 1.7266,
+      "step": 5482
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9818801089918256e-05,
+      "loss": 1.7679,
+      "step": 5483
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9816757493188012e-05,
+      "loss": 1.6701,
+      "step": 5484
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9814713896457764e-05,
+      "loss": 1.8178,
+      "step": 5485
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.981267029972752e-05,
+      "loss": 1.8561,
+      "step": 5486
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9810626702997276e-05,
+      "loss": 1.744,
+      "step": 5487
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.980858310626703e-05,
+      "loss": 1.7644,
+      "step": 5488
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9806539509536787e-05,
+      "loss": 1.8709,
+      "step": 5489
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.980449591280654e-05,
+      "loss": 1.8952,
+      "step": 5490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9802452316076295e-05,
+      "loss": 1.803,
+      "step": 5491
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.980040871934605e-05,
+      "loss": 1.5999,
+      "step": 5492
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9798365122615803e-05,
+      "loss": 1.7674,
+      "step": 5493
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9796321525885558e-05,
+      "loss": 1.6961,
+      "step": 5494
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.979427792915531e-05,
+      "loss": 1.6443,
+      "step": 5495
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.979223433242507e-05,
+      "loss": 1.8052,
+      "step": 5496
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9790190735694825e-05,
+      "loss": 1.7402,
+      "step": 5497
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9788147138964577e-05,
+      "loss": 1.6495,
+      "step": 5498
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9786103542234333e-05,
+      "loss": 1.7847,
+      "step": 5499
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.978405994550409e-05,
+      "loss": 1.7656,
+      "step": 5500
+    },
+    {
+      "epoch": 1.09,
+      "eval_loss": 1.5735337734222412,
+      "eval_runtime": 721.0228,
+      "eval_samples_per_second": 5.756,
+      "eval_steps_per_second": 1.44,
+      "step": 5500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.978201634877384e-05,
+      "loss": 1.7295,
+      "step": 5501
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9779972752043597e-05,
+      "loss": 1.7258,
+      "step": 5502
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9777929155313352e-05,
+      "loss": 1.7551,
+      "step": 5503
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9775885558583108e-05,
+      "loss": 1.7875,
+      "step": 5504
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9773841961852863e-05,
+      "loss": 1.7591,
+      "step": 5505
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9771798365122616e-05,
+      "loss": 1.7434,
+      "step": 5506
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.976975476839237e-05,
+      "loss": 1.6422,
+      "step": 5507
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9767711171662124e-05,
+      "loss": 1.7267,
+      "step": 5508
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.976566757493188e-05,
+      "loss": 1.7833,
+      "step": 5509
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9763623978201638e-05,
+      "loss": 1.7297,
+      "step": 5510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.976158038147139e-05,
+      "loss": 1.6948,
+      "step": 5511
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9759536784741146e-05,
+      "loss": 1.7888,
+      "step": 5512
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.97574931880109e-05,
+      "loss": 1.7488,
+      "step": 5513
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9755449591280654e-05,
+      "loss": 1.7839,
+      "step": 5514
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.975340599455041e-05,
+      "loss": 1.8495,
+      "step": 5515
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9751362397820162e-05,
+      "loss": 1.7079,
+      "step": 5516
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9749318801089917e-05,
+      "loss": 1.8308,
+      "step": 5517
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9747275204359677e-05,
+      "loss": 1.7949,
+      "step": 5518
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.974523160762943e-05,
+      "loss": 1.6652,
+      "step": 5519
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9743188010899184e-05,
+      "loss": 1.6323,
+      "step": 5520
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9741144414168937e-05,
+      "loss": 1.7532,
+      "step": 5521
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9739100817438692e-05,
+      "loss": 1.833,
+      "step": 5522
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9737057220708448e-05,
+      "loss": 1.5963,
+      "step": 5523
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.97350136239782e-05,
+      "loss": 1.7422,
+      "step": 5524
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.973297002724796e-05,
+      "loss": 1.7543,
+      "step": 5525
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.973092643051771e-05,
+      "loss": 1.7073,
+      "step": 5526
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9728882833787467e-05,
+      "loss": 1.729,
+      "step": 5527
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9726839237057223e-05,
+      "loss": 1.8303,
+      "step": 5528
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9724795640326975e-05,
+      "loss": 1.7538,
+      "step": 5529
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.972275204359673e-05,
+      "loss": 1.6719,
+      "step": 5530
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9720708446866483e-05,
+      "loss": 1.7694,
+      "step": 5531
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.971866485013624e-05,
+      "loss": 1.9156,
+      "step": 5532
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9716621253405997e-05,
+      "loss": 1.6698,
+      "step": 5533
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.971457765667575e-05,
+      "loss": 1.7343,
+      "step": 5534
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9712534059945505e-05,
+      "loss": 1.8066,
+      "step": 5535
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.971049046321526e-05,
+      "loss": 1.7511,
+      "step": 5536
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9708446866485013e-05,
+      "loss": 1.808,
+      "step": 5537
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.970640326975477e-05,
+      "loss": 1.6359,
+      "step": 5538
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.970435967302452e-05,
+      "loss": 1.7083,
+      "step": 5539
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.970231607629428e-05,
+      "loss": 1.6943,
+      "step": 5540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9700272479564036e-05,
+      "loss": 1.7156,
+      "step": 5541
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9698228882833788e-05,
+      "loss": 1.6587,
+      "step": 5542
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9696185286103544e-05,
+      "loss": 1.6926,
+      "step": 5543
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9694141689373296e-05,
+      "loss": 1.7862,
+      "step": 5544
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.969209809264305e-05,
+      "loss": 1.7335,
+      "step": 5545
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9690054495912807e-05,
+      "loss": 1.6403,
+      "step": 5546
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.968801089918256e-05,
+      "loss": 1.8063,
+      "step": 5547
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.968596730245232e-05,
+      "loss": 1.7884,
+      "step": 5548
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.968392370572207e-05,
+      "loss": 1.7692,
+      "step": 5549
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9681880108991826e-05,
+      "loss": 1.8154,
+      "step": 5550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9679836512261582e-05,
+      "loss": 1.7252,
+      "step": 5551
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9677792915531334e-05,
+      "loss": 1.7033,
+      "step": 5552
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.967574931880109e-05,
+      "loss": 1.7744,
+      "step": 5553
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9673705722070845e-05,
+      "loss": 1.7996,
+      "step": 5554
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.96716621253406e-05,
+      "loss": 1.6774,
+      "step": 5555
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9669618528610357e-05,
+      "loss": 1.6816,
+      "step": 5556
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.966757493188011e-05,
+      "loss": 1.775,
+      "step": 5557
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9665531335149865e-05,
+      "loss": 1.8023,
+      "step": 5558
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.966348773841962e-05,
+      "loss": 1.7322,
+      "step": 5559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9661444141689372e-05,
+      "loss": 1.7835,
+      "step": 5560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9659400544959128e-05,
+      "loss": 1.6549,
+      "step": 5561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.965735694822888e-05,
+      "loss": 1.7898,
+      "step": 5562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.965531335149864e-05,
+      "loss": 1.7394,
+      "step": 5563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9653269754768395e-05,
+      "loss": 1.754,
+      "step": 5564
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9651226158038147e-05,
+      "loss": 1.6607,
+      "step": 5565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9649182561307903e-05,
+      "loss": 1.6822,
+      "step": 5566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9647138964577655e-05,
+      "loss": 1.6466,
+      "step": 5567
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.964509536784741e-05,
+      "loss": 1.7501,
+      "step": 5568
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9643051771117166e-05,
+      "loss": 1.7151,
+      "step": 5569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9641008174386922e-05,
+      "loss": 1.6767,
+      "step": 5570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9638964577656678e-05,
+      "loss": 1.7619,
+      "step": 5571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9636920980926433e-05,
+      "loss": 1.9286,
+      "step": 5572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634877384196185e-05,
+      "loss": 1.6859,
+      "step": 5573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963283378746594e-05,
+      "loss": 1.6961,
+      "step": 5574
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9630790190735693e-05,
+      "loss": 1.8823,
+      "step": 5575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.962874659400545e-05,
+      "loss": 1.7004,
+      "step": 5576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9626702997275205e-05,
+      "loss": 1.7744,
+      "step": 5577
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.962465940054496e-05,
+      "loss": 1.6469,
+      "step": 5578
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9622615803814716e-05,
+      "loss": 1.7082,
+      "step": 5579
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9620572207084468e-05,
+      "loss": 1.7225,
+      "step": 5580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9618528610354224e-05,
+      "loss": 1.7739,
+      "step": 5581
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.961648501362398e-05,
+      "loss": 1.9291,
+      "step": 5582
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.961444141689373e-05,
+      "loss": 1.7247,
+      "step": 5583
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9612397820163487e-05,
+      "loss": 1.7885,
+      "step": 5584
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9610354223433243e-05,
+      "loss": 1.7681,
+      "step": 5585
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9608310626703e-05,
+      "loss": 1.7769,
+      "step": 5586
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9606267029972754e-05,
+      "loss": 1.7753,
+      "step": 5587
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9604223433242506e-05,
+      "loss": 1.745,
+      "step": 5588
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9602179836512262e-05,
+      "loss": 1.7479,
+      "step": 5589
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9600136239782018e-05,
+      "loss": 1.6853,
+      "step": 5590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.959809264305177e-05,
+      "loss": 1.7281,
+      "step": 5591
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9596049046321526e-05,
+      "loss": 1.7925,
+      "step": 5592
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.959400544959128e-05,
+      "loss": 1.9002,
+      "step": 5593
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9591961852861037e-05,
+      "loss": 1.6541,
+      "step": 5594
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9589918256130792e-05,
+      "loss": 1.7782,
+      "step": 5595
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9587874659400545e-05,
+      "loss": 1.8292,
+      "step": 5596
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.95858310626703e-05,
+      "loss": 1.7437,
+      "step": 5597
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9583787465940053e-05,
+      "loss": 1.7765,
+      "step": 5598
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9581743869209808e-05,
+      "loss": 1.8338,
+      "step": 5599
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9579700272479567e-05,
+      "loss": 1.6503,
+      "step": 5600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.957765667574932e-05,
+      "loss": 1.7282,
+      "step": 5601
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9575613079019075e-05,
+      "loss": 1.8248,
+      "step": 5602
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9573569482288827e-05,
+      "loss": 1.6879,
+      "step": 5603
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9571525885558583e-05,
+      "loss": 1.6617,
+      "step": 5604
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.956948228882834e-05,
+      "loss": 1.7944,
+      "step": 5605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.956743869209809e-05,
+      "loss": 1.713,
+      "step": 5606
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.956539509536785e-05,
+      "loss": 1.666,
+      "step": 5607
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9563351498637606e-05,
+      "loss": 1.8447,
+      "step": 5608
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9561307901907358e-05,
+      "loss": 1.7596,
+      "step": 5609
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9559264305177113e-05,
+      "loss": 1.704,
+      "step": 5610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9557220708446866e-05,
+      "loss": 1.9063,
+      "step": 5611
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.955517711171662e-05,
+      "loss": 1.7124,
+      "step": 5612
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9553133514986377e-05,
+      "loss": 1.6929,
+      "step": 5613
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.955108991825613e-05,
+      "loss": 1.7434,
+      "step": 5614
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9549046321525888e-05,
+      "loss": 1.7003,
+      "step": 5615
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.954700272479564e-05,
+      "loss": 1.8291,
+      "step": 5616
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9544959128065396e-05,
+      "loss": 1.6917,
+      "step": 5617
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.954291553133515e-05,
+      "loss": 1.7018,
+      "step": 5618
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9540871934604904e-05,
+      "loss": 1.704,
+      "step": 5619
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.953882833787466e-05,
+      "loss": 1.6629,
+      "step": 5620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9536784741144412e-05,
+      "loss": 1.623,
+      "step": 5621
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.953474114441417e-05,
+      "loss": 1.7707,
+      "step": 5622
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9532697547683926e-05,
+      "loss": 1.73,
+      "step": 5623
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.953065395095368e-05,
+      "loss": 1.507,
+      "step": 5624
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9528610354223434e-05,
+      "loss": 1.9029,
+      "step": 5625
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.952656675749319e-05,
+      "loss": 1.7053,
+      "step": 5626
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9524523160762942e-05,
+      "loss": 1.9016,
+      "step": 5627
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9522479564032698e-05,
+      "loss": 1.7,
+      "step": 5628
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.952043596730245e-05,
+      "loss": 1.7578,
+      "step": 5629
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.951839237057221e-05,
+      "loss": 1.7612,
+      "step": 5630
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9516348773841965e-05,
+      "loss": 1.6631,
+      "step": 5631
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9514305177111717e-05,
+      "loss": 1.7095,
+      "step": 5632
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9512261580381473e-05,
+      "loss": 1.7786,
+      "step": 5633
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9510217983651225e-05,
+      "loss": 1.7695,
+      "step": 5634
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.950817438692098e-05,
+      "loss": 1.8025,
+      "step": 5635
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9506130790190736e-05,
+      "loss": 1.6937,
+      "step": 5636
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9504087193460492e-05,
+      "loss": 1.7642,
+      "step": 5637
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9502043596730247e-05,
+      "loss": 1.7384,
+      "step": 5638
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.95e-05,
+      "loss": 1.7844,
+      "step": 5639
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9497956403269755e-05,
+      "loss": 1.7784,
+      "step": 5640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.949591280653951e-05,
+      "loss": 1.7045,
+      "step": 5641
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9493869209809263e-05,
+      "loss": 1.8176,
+      "step": 5642
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.949182561307902e-05,
+      "loss": 1.6782,
+      "step": 5643
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9489782016348774e-05,
+      "loss": 1.8476,
+      "step": 5644
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.948773841961853e-05,
+      "loss": 2.0027,
+      "step": 5645
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9485694822888286e-05,
+      "loss": 1.7143,
+      "step": 5646
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9483651226158038e-05,
+      "loss": 1.844,
+      "step": 5647
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9481607629427794e-05,
+      "loss": 1.726,
+      "step": 5648
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.947956403269755e-05,
+      "loss": 1.6978,
+      "step": 5649
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.94775204359673e-05,
+      "loss": 1.8286,
+      "step": 5650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9475476839237057e-05,
+      "loss": 1.7472,
+      "step": 5651
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9473433242506813e-05,
+      "loss": 1.7911,
+      "step": 5652
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.947138964577657e-05,
+      "loss": 1.756,
+      "step": 5653
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9469346049046324e-05,
+      "loss": 1.7457,
+      "step": 5654
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9467302452316076e-05,
+      "loss": 1.8437,
+      "step": 5655
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9465258855585832e-05,
+      "loss": 1.6276,
+      "step": 5656
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9463215258855584e-05,
+      "loss": 1.6739,
+      "step": 5657
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.946117166212534e-05,
+      "loss": 1.7769,
+      "step": 5658
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9459128065395095e-05,
+      "loss": 1.6679,
+      "step": 5659
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.945708446866485e-05,
+      "loss": 1.724,
+      "step": 5660
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9455040871934607e-05,
+      "loss": 1.7477,
+      "step": 5661
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.945299727520436e-05,
+      "loss": 1.8945,
+      "step": 5662
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9450953678474115e-05,
+      "loss": 1.7379,
+      "step": 5663
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.944891008174387e-05,
+      "loss": 1.8069,
+      "step": 5664
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9446866485013622e-05,
+      "loss": 1.5513,
+      "step": 5665
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9444822888283378e-05,
+      "loss": 1.6945,
+      "step": 5666
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9442779291553137e-05,
+      "loss": 1.7786,
+      "step": 5667
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.944073569482289e-05,
+      "loss": 1.7171,
+      "step": 5668
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9438692098092645e-05,
+      "loss": 1.6529,
+      "step": 5669
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9436648501362397e-05,
+      "loss": 1.7398,
+      "step": 5670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9434604904632153e-05,
+      "loss": 1.8099,
+      "step": 5671
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.943256130790191e-05,
+      "loss": 1.6863,
+      "step": 5672
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.943051771117166e-05,
+      "loss": 1.872,
+      "step": 5673
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9428474114441416e-05,
+      "loss": 1.7027,
+      "step": 5674
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9426430517711172e-05,
+      "loss": 1.8409,
+      "step": 5675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9424386920980928e-05,
+      "loss": 1.6276,
+      "step": 5676
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9422343324250683e-05,
+      "loss": 1.7675,
+      "step": 5677
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9420299727520435e-05,
+      "loss": 1.7378,
+      "step": 5678
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.941825613079019e-05,
+      "loss": 1.7762,
+      "step": 5679
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9416212534059943e-05,
+      "loss": 1.812,
+      "step": 5680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.94141689373297e-05,
+      "loss": 1.7975,
+      "step": 5681
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9412125340599458e-05,
+      "loss": 1.7889,
+      "step": 5682
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.941008174386921e-05,
+      "loss": 1.6943,
+      "step": 5683
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9408038147138966e-05,
+      "loss": 1.7122,
+      "step": 5684
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.940599455040872e-05,
+      "loss": 1.694,
+      "step": 5685
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9403950953678474e-05,
+      "loss": 1.7328,
+      "step": 5686
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.940190735694823e-05,
+      "loss": 1.8095,
+      "step": 5687
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.939986376021798e-05,
+      "loss": 1.728,
+      "step": 5688
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.939782016348774e-05,
+      "loss": 1.696,
+      "step": 5689
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9395776566757496e-05,
+      "loss": 1.8411,
+      "step": 5690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.939373297002725e-05,
+      "loss": 1.7574,
+      "step": 5691
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9391689373297004e-05,
+      "loss": 1.7196,
+      "step": 5692
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9389645776566756e-05,
+      "loss": 1.7618,
+      "step": 5693
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9387602179836512e-05,
+      "loss": 1.8591,
+      "step": 5694
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9385558583106268e-05,
+      "loss": 1.7552,
+      "step": 5695
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.938351498637602e-05,
+      "loss": 1.5885,
+      "step": 5696
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.938147138964578e-05,
+      "loss": 1.8892,
+      "step": 5697
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.937942779291553e-05,
+      "loss": 1.7282,
+      "step": 5698
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9377384196185287e-05,
+      "loss": 1.7073,
+      "step": 5699
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9375340599455042e-05,
+      "loss": 1.7823,
+      "step": 5700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9373297002724795e-05,
+      "loss": 1.7993,
+      "step": 5701
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.937125340599455e-05,
+      "loss": 1.4869,
+      "step": 5702
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9369209809264306e-05,
+      "loss": 1.7595,
+      "step": 5703
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.936716621253406e-05,
+      "loss": 1.7049,
+      "step": 5704
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9365122615803817e-05,
+      "loss": 1.7268,
+      "step": 5705
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.936307901907357e-05,
+      "loss": 1.7754,
+      "step": 5706
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9361035422343325e-05,
+      "loss": 1.7898,
+      "step": 5707
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.935899182561308e-05,
+      "loss": 1.7634,
+      "step": 5708
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9356948228882833e-05,
+      "loss": 1.7806,
+      "step": 5709
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.935490463215259e-05,
+      "loss": 1.8135,
+      "step": 5710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.935286103542234e-05,
+      "loss": 1.7624,
+      "step": 5711
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.93508174386921e-05,
+      "loss": 1.7408,
+      "step": 5712
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9348773841961855e-05,
+      "loss": 1.8674,
+      "step": 5713
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9346730245231608e-05,
+      "loss": 1.7143,
+      "step": 5714
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9344686648501363e-05,
+      "loss": 1.7532,
+      "step": 5715
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9342643051771116e-05,
+      "loss": 1.8226,
+      "step": 5716
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.934059945504087e-05,
+      "loss": 1.5961,
+      "step": 5717
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9338555858310627e-05,
+      "loss": 1.8232,
+      "step": 5718
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9336512261580382e-05,
+      "loss": 1.7374,
+      "step": 5719
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9334468664850138e-05,
+      "loss": 1.8658,
+      "step": 5720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9332425068119894e-05,
+      "loss": 1.7691,
+      "step": 5721
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9330381471389646e-05,
+      "loss": 1.8799,
+      "step": 5722
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.93283378746594e-05,
+      "loss": 1.7386,
+      "step": 5723
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9326294277929154e-05,
+      "loss": 1.7546,
+      "step": 5724
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.932425068119891e-05,
+      "loss": 1.918,
+      "step": 5725
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9322207084468665e-05,
+      "loss": 1.6939,
+      "step": 5726
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.932016348773842e-05,
+      "loss": 1.8635,
+      "step": 5727
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9318119891008176e-05,
+      "loss": 1.7672,
+      "step": 5728
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.931607629427793e-05,
+      "loss": 1.7256,
+      "step": 5729
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9314032697547684e-05,
+      "loss": 1.7964,
+      "step": 5730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.931198910081744e-05,
+      "loss": 1.671,
+      "step": 5731
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9309945504087192e-05,
+      "loss": 1.7682,
+      "step": 5732
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9307901907356948e-05,
+      "loss": 1.793,
+      "step": 5733
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9305858310626703e-05,
+      "loss": 1.7689,
+      "step": 5734
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.930381471389646e-05,
+      "loss": 1.7315,
+      "step": 5735
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9301771117166215e-05,
+      "loss": 1.7766,
+      "step": 5736
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9299727520435967e-05,
+      "loss": 1.7394,
+      "step": 5737
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9297683923705723e-05,
+      "loss": 1.7674,
+      "step": 5738
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9295640326975478e-05,
+      "loss": 1.7887,
+      "step": 5739
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.929359673024523e-05,
+      "loss": 1.7646,
+      "step": 5740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9291553133514986e-05,
+      "loss": 1.7449,
+      "step": 5741
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9289509536784742e-05,
+      "loss": 1.859,
+      "step": 5742
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9287465940054497e-05,
+      "loss": 1.7163,
+      "step": 5743
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9285422343324253e-05,
+      "loss": 1.8495,
+      "step": 5744
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9283378746594005e-05,
+      "loss": 1.7082,
+      "step": 5745
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.928133514986376e-05,
+      "loss": 1.7887,
+      "step": 5746
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9279291553133513e-05,
+      "loss": 1.7171,
+      "step": 5747
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.927724795640327e-05,
+      "loss": 1.7716,
+      "step": 5748
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9275204359673028e-05,
+      "loss": 1.7387,
+      "step": 5749
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.927316076294278e-05,
+      "loss": 1.7737,
+      "step": 5750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9271117166212536e-05,
+      "loss": 1.7219,
+      "step": 5751
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9269073569482288e-05,
+      "loss": 1.6352,
+      "step": 5752
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9267029972752044e-05,
+      "loss": 1.6968,
+      "step": 5753
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.92649863760218e-05,
+      "loss": 1.7232,
+      "step": 5754
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.926294277929155e-05,
+      "loss": 1.5852,
+      "step": 5755
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9260899182561307e-05,
+      "loss": 1.8488,
+      "step": 5756
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9258855585831066e-05,
+      "loss": 1.7042,
+      "step": 5757
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9256811989100818e-05,
+      "loss": 1.6506,
+      "step": 5758
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9254768392370574e-05,
+      "loss": 1.7942,
+      "step": 5759
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9252724795640326e-05,
+      "loss": 1.766,
+      "step": 5760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9250681198910082e-05,
+      "loss": 1.8509,
+      "step": 5761
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9248637602179837e-05,
+      "loss": 1.7359,
+      "step": 5762
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.924659400544959e-05,
+      "loss": 1.7715,
+      "step": 5763
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.924455040871935e-05,
+      "loss": 1.7786,
+      "step": 5764
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.92425068119891e-05,
+      "loss": 1.7194,
+      "step": 5765
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9240463215258857e-05,
+      "loss": 1.7206,
+      "step": 5766
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9238419618528612e-05,
+      "loss": 1.7331,
+      "step": 5767
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9236376021798364e-05,
+      "loss": 1.8887,
+      "step": 5768
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.923433242506812e-05,
+      "loss": 1.6345,
+      "step": 5769
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9232288828337872e-05,
+      "loss": 1.7361,
+      "step": 5770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9230245231607628e-05,
+      "loss": 1.8397,
+      "step": 5771
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9228201634877387e-05,
+      "loss": 1.8812,
+      "step": 5772
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.922615803814714e-05,
+      "loss": 1.7057,
+      "step": 5773
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9224114441416895e-05,
+      "loss": 1.9329,
+      "step": 5774
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.922207084468665e-05,
+      "loss": 1.6719,
+      "step": 5775
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9220027247956403e-05,
+      "loss": 1.8002,
+      "step": 5776
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.921798365122616e-05,
+      "loss": 1.7135,
+      "step": 5777
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.921594005449591e-05,
+      "loss": 1.6316,
+      "step": 5778
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.921389645776567e-05,
+      "loss": 1.7577,
+      "step": 5779
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9211852861035425e-05,
+      "loss": 1.7691,
+      "step": 5780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9209809264305178e-05,
+      "loss": 1.7958,
+      "step": 5781
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9207765667574933e-05,
+      "loss": 1.8307,
+      "step": 5782
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9205722070844685e-05,
+      "loss": 1.8598,
+      "step": 5783
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.920367847411444e-05,
+      "loss": 1.7126,
+      "step": 5784
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9201634877384197e-05,
+      "loss": 1.7706,
+      "step": 5785
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9199591280653952e-05,
+      "loss": 1.7631,
+      "step": 5786
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9197547683923708e-05,
+      "loss": 1.7341,
+      "step": 5787
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.919550408719346e-05,
+      "loss": 1.7684,
+      "step": 5788
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9193460490463216e-05,
+      "loss": 1.8163,
+      "step": 5789
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.919141689373297e-05,
+      "loss": 1.8025,
+      "step": 5790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9189373297002724e-05,
+      "loss": 1.6764,
+      "step": 5791
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.918732970027248e-05,
+      "loss": 1.8028,
+      "step": 5792
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9185286103542235e-05,
+      "loss": 1.7197,
+      "step": 5793
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.918324250681199e-05,
+      "loss": 1.7383,
+      "step": 5794
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9181198910081746e-05,
+      "loss": 1.6769,
+      "step": 5795
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.91791553133515e-05,
+      "loss": 1.6674,
+      "step": 5796
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9177111716621254e-05,
+      "loss": 1.6855,
+      "step": 5797
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.917506811989101e-05,
+      "loss": 1.7858,
+      "step": 5798
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9173024523160762e-05,
+      "loss": 1.9076,
+      "step": 5799
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9170980926430518e-05,
+      "loss": 1.6282,
+      "step": 5800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9168937329700273e-05,
+      "loss": 1.9052,
+      "step": 5801
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.916689373297003e-05,
+      "loss": 1.7423,
+      "step": 5802
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9164850136239784e-05,
+      "loss": 1.6888,
+      "step": 5803
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9162806539509537e-05,
+      "loss": 1.9193,
+      "step": 5804
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9160762942779292e-05,
+      "loss": 1.8627,
+      "step": 5805
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9158719346049045e-05,
+      "loss": 1.7369,
+      "step": 5806
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.91566757493188e-05,
+      "loss": 1.8253,
+      "step": 5807
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9154632152588556e-05,
+      "loss": 1.7679,
+      "step": 5808
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.915258855585831e-05,
+      "loss": 1.6958,
+      "step": 5809
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9150544959128067e-05,
+      "loss": 1.9097,
+      "step": 5810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9148501362397823e-05,
+      "loss": 1.7816,
+      "step": 5811
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9146457765667575e-05,
+      "loss": 1.7227,
+      "step": 5812
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.914441416893733e-05,
+      "loss": 1.6351,
+      "step": 5813
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9142370572207083e-05,
+      "loss": 1.6886,
+      "step": 5814
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.914032697547684e-05,
+      "loss": 1.6724,
+      "step": 5815
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9138283378746598e-05,
+      "loss": 1.7306,
+      "step": 5816
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.913623978201635e-05,
+      "loss": 1.8417,
+      "step": 5817
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9134196185286105e-05,
+      "loss": 1.6955,
+      "step": 5818
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9132152588555858e-05,
+      "loss": 1.7036,
+      "step": 5819
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9130108991825613e-05,
+      "loss": 1.7387,
+      "step": 5820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.912806539509537e-05,
+      "loss": 1.7719,
+      "step": 5821
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.912602179836512e-05,
+      "loss": 1.4384,
+      "step": 5822
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9123978201634877e-05,
+      "loss": 1.8912,
+      "step": 5823
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9121934604904632e-05,
+      "loss": 1.9256,
+      "step": 5824
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9119891008174388e-05,
+      "loss": 1.6412,
+      "step": 5825
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9117847411444144e-05,
+      "loss": 1.7634,
+      "step": 5826
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9115803814713896e-05,
+      "loss": 1.669,
+      "step": 5827
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.911376021798365e-05,
+      "loss": 1.8747,
+      "step": 5828
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9111716621253407e-05,
+      "loss": 1.7906,
+      "step": 5829
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.910967302452316e-05,
+      "loss": 1.6722,
+      "step": 5830
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.910762942779292e-05,
+      "loss": 1.812,
+      "step": 5831
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.910558583106267e-05,
+      "loss": 1.9018,
+      "step": 5832
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9103542234332426e-05,
+      "loss": 1.7443,
+      "step": 5833
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9101498637602182e-05,
+      "loss": 1.8903,
+      "step": 5834
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9099455040871934e-05,
+      "loss": 1.8218,
+      "step": 5835
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.909741144414169e-05,
+      "loss": 1.6743,
+      "step": 5836
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9095367847411442e-05,
+      "loss": 1.7512,
+      "step": 5837
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9093324250681198e-05,
+      "loss": 1.7779,
+      "step": 5838
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9091280653950957e-05,
+      "loss": 1.7236,
+      "step": 5839
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.908923705722071e-05,
+      "loss": 1.6681,
+      "step": 5840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9087193460490465e-05,
+      "loss": 1.7082,
+      "step": 5841
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9085149863760217e-05,
+      "loss": 1.7434,
+      "step": 5842
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9083106267029973e-05,
+      "loss": 1.7164,
+      "step": 5843
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9081062670299728e-05,
+      "loss": 1.9211,
+      "step": 5844
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.907901907356948e-05,
+      "loss": 1.6544,
+      "step": 5845
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.907697547683924e-05,
+      "loss": 1.65,
+      "step": 5846
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.907493188010899e-05,
+      "loss": 1.8821,
+      "step": 5847
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9072888283378747e-05,
+      "loss": 1.7671,
+      "step": 5848
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9070844686648503e-05,
+      "loss": 1.8101,
+      "step": 5849
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9068801089918255e-05,
+      "loss": 1.6178,
+      "step": 5850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.906675749318801e-05,
+      "loss": 1.6957,
+      "step": 5851
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9064713896457766e-05,
+      "loss": 1.6792,
+      "step": 5852
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.906267029972752e-05,
+      "loss": 1.7902,
+      "step": 5853
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9060626702997278e-05,
+      "loss": 1.7292,
+      "step": 5854
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.905858310626703e-05,
+      "loss": 1.743,
+      "step": 5855
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9056539509536786e-05,
+      "loss": 1.7033,
+      "step": 5856
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.905449591280654e-05,
+      "loss": 1.9491,
+      "step": 5857
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9052452316076293e-05,
+      "loss": 1.73,
+      "step": 5858
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.905040871934605e-05,
+      "loss": 1.5607,
+      "step": 5859
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.90483651226158e-05,
+      "loss": 1.7936,
+      "step": 5860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.904632152588556e-05,
+      "loss": 1.7644,
+      "step": 5861
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9044277929155316e-05,
+      "loss": 1.7647,
+      "step": 5862
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9042234332425068e-05,
+      "loss": 1.8274,
+      "step": 5863
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9040190735694824e-05,
+      "loss": 1.8118,
+      "step": 5864
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9038147138964576e-05,
+      "loss": 1.7046,
+      "step": 5865
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9036103542234332e-05,
+      "loss": 1.7502,
+      "step": 5866
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9034059945504087e-05,
+      "loss": 1.7834,
+      "step": 5867
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.903201634877384e-05,
+      "loss": 1.72,
+      "step": 5868
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.90299727520436e-05,
+      "loss": 1.6349,
+      "step": 5869
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9027929155313354e-05,
+      "loss": 1.8704,
+      "step": 5870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9025885558583107e-05,
+      "loss": 1.7405,
+      "step": 5871
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9023841961852862e-05,
+      "loss": 1.7374,
+      "step": 5872
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9021798365122614e-05,
+      "loss": 1.6965,
+      "step": 5873
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.901975476839237e-05,
+      "loss": 1.728,
+      "step": 5874
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9017711171662126e-05,
+      "loss": 1.7161,
+      "step": 5875
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.901566757493188e-05,
+      "loss": 1.6233,
+      "step": 5876
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9013623978201637e-05,
+      "loss": 1.6943,
+      "step": 5877
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.901158038147139e-05,
+      "loss": 1.7295,
+      "step": 5878
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9009536784741145e-05,
+      "loss": 1.8258,
+      "step": 5879
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.90074931880109e-05,
+      "loss": 1.682,
+      "step": 5880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9005449591280653e-05,
+      "loss": 1.6688,
+      "step": 5881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.900340599455041e-05,
+      "loss": 1.7382,
+      "step": 5882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9001362397820164e-05,
+      "loss": 1.8062,
+      "step": 5883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.899931880108992e-05,
+      "loss": 1.8732,
+      "step": 5884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8997275204359675e-05,
+      "loss": 1.8064,
+      "step": 5885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8995231607629427e-05,
+      "loss": 1.7041,
+      "step": 5886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8993188010899183e-05,
+      "loss": 1.8014,
+      "step": 5887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.899114441416894e-05,
+      "loss": 1.637,
+      "step": 5888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.898910081743869e-05,
+      "loss": 1.6612,
+      "step": 5889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8987057220708447e-05,
+      "loss": 1.8008,
+      "step": 5890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8985013623978202e-05,
+      "loss": 1.6594,
+      "step": 5891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8982970027247958e-05,
+      "loss": 1.7807,
+      "step": 5892
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8980926430517714e-05,
+      "loss": 1.825,
+      "step": 5893
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8978882833787466e-05,
+      "loss": 1.8497,
+      "step": 5894
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.897683923705722e-05,
+      "loss": 1.7107,
+      "step": 5895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8974795640326974e-05,
+      "loss": 1.8468,
+      "step": 5896
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.897275204359673e-05,
+      "loss": 1.8296,
+      "step": 5897
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8970708446866488e-05,
+      "loss": 1.6745,
+      "step": 5898
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.896866485013624e-05,
+      "loss": 1.8297,
+      "step": 5899
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8966621253405996e-05,
+      "loss": 1.7765,
+      "step": 5900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.896457765667575e-05,
+      "loss": 1.7413,
+      "step": 5901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8962534059945504e-05,
+      "loss": 1.804,
+      "step": 5902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.896049046321526e-05,
+      "loss": 1.6147,
+      "step": 5903
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8958446866485012e-05,
+      "loss": 1.8773,
+      "step": 5904
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8956403269754768e-05,
+      "loss": 1.6127,
+      "step": 5905
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8954359673024527e-05,
+      "loss": 1.7787,
+      "step": 5906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.895231607629428e-05,
+      "loss": 1.7823,
+      "step": 5907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8950272479564034e-05,
+      "loss": 1.7312,
+      "step": 5908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8948228882833787e-05,
+      "loss": 1.6428,
+      "step": 5909
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8946185286103542e-05,
+      "loss": 1.6969,
+      "step": 5910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8944141689373298e-05,
+      "loss": 1.7353,
+      "step": 5911
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.894209809264305e-05,
+      "loss": 1.7932,
+      "step": 5912
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.894005449591281e-05,
+      "loss": 1.6739,
+      "step": 5913
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893801089918256e-05,
+      "loss": 1.7579,
+      "step": 5914
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8935967302452317e-05,
+      "loss": 1.7115,
+      "step": 5915
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8933923705722073e-05,
+      "loss": 1.7064,
+      "step": 5916
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8931880108991825e-05,
+      "loss": 1.8762,
+      "step": 5917
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.892983651226158e-05,
+      "loss": 1.6344,
+      "step": 5918
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8927792915531333e-05,
+      "loss": 1.7299,
+      "step": 5919
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.892574931880109e-05,
+      "loss": 1.7499,
+      "step": 5920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8923705722070848e-05,
+      "loss": 1.8589,
+      "step": 5921
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.89216621253406e-05,
+      "loss": 1.7795,
+      "step": 5922
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8919618528610355e-05,
+      "loss": 1.6418,
+      "step": 5923
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891757493188011e-05,
+      "loss": 1.7102,
+      "step": 5924
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8915531335149863e-05,
+      "loss": 1.6447,
+      "step": 5925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891348773841962e-05,
+      "loss": 1.8922,
+      "step": 5926
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891144414168937e-05,
+      "loss": 1.7558,
+      "step": 5927
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.890940054495913e-05,
+      "loss": 1.8304,
+      "step": 5928
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8907356948228886e-05,
+      "loss": 1.6753,
+      "step": 5929
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8905313351498638e-05,
+      "loss": 1.7799,
+      "step": 5930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8903269754768394e-05,
+      "loss": 1.7224,
+      "step": 5931
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8901226158038146e-05,
+      "loss": 1.7417,
+      "step": 5932
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.88991825613079e-05,
+      "loss": 1.6005,
+      "step": 5933
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8897138964577657e-05,
+      "loss": 1.678,
+      "step": 5934
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.889509536784741e-05,
+      "loss": 1.6335,
+      "step": 5935
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.889305177111717e-05,
+      "loss": 1.9157,
+      "step": 5936
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.889100817438692e-05,
+      "loss": 1.6928,
+      "step": 5937
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8888964577656676e-05,
+      "loss": 1.7264,
+      "step": 5938
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8886920980926432e-05,
+      "loss": 1.7143,
+      "step": 5939
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8884877384196184e-05,
+      "loss": 1.7209,
+      "step": 5940
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.888283378746594e-05,
+      "loss": 1.7739,
+      "step": 5941
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8880790190735695e-05,
+      "loss": 1.6338,
+      "step": 5942
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.887874659400545e-05,
+      "loss": 1.7775,
+      "step": 5943
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8876702997275207e-05,
+      "loss": 1.862,
+      "step": 5944
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.887465940054496e-05,
+      "loss": 1.7976,
+      "step": 5945
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8872615803814715e-05,
+      "loss": 1.5793,
+      "step": 5946
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.887057220708447e-05,
+      "loss": 1.7625,
+      "step": 5947
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8868528610354222e-05,
+      "loss": 1.8014,
+      "step": 5948
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8866485013623978e-05,
+      "loss": 1.7885,
+      "step": 5949
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.886444141689373e-05,
+      "loss": 1.8608,
+      "step": 5950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.886239782016349e-05,
+      "loss": 1.7929,
+      "step": 5951
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8860354223433245e-05,
+      "loss": 1.6559,
+      "step": 5952
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8858310626702997e-05,
+      "loss": 1.684,
+      "step": 5953
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8856267029972753e-05,
+      "loss": 1.8224,
+      "step": 5954
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8854223433242505e-05,
+      "loss": 1.713,
+      "step": 5955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.885217983651226e-05,
+      "loss": 1.828,
+      "step": 5956
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8850136239782016e-05,
+      "loss": 1.6434,
+      "step": 5957
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8848092643051772e-05,
+      "loss": 1.6499,
+      "step": 5958
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8846049046321528e-05,
+      "loss": 1.7555,
+      "step": 5959
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8844005449591283e-05,
+      "loss": 1.7656,
+      "step": 5960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8841961852861036e-05,
+      "loss": 1.8417,
+      "step": 5961
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.883991825613079e-05,
+      "loss": 1.8357,
+      "step": 5962
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8837874659400543e-05,
+      "loss": 1.7481,
+      "step": 5963
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.88358310626703e-05,
+      "loss": 1.6723,
+      "step": 5964
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8833787465940058e-05,
+      "loss": 1.7418,
+      "step": 5965
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.883174386920981e-05,
+      "loss": 1.7378,
+      "step": 5966
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8829700272479566e-05,
+      "loss": 1.8034,
+      "step": 5967
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8827656675749318e-05,
+      "loss": 1.6596,
+      "step": 5968
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8825613079019074e-05,
+      "loss": 1.7236,
+      "step": 5969
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.882356948228883e-05,
+      "loss": 1.8008,
+      "step": 5970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8821525885558582e-05,
+      "loss": 1.7585,
+      "step": 5971
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8819482288828337e-05,
+      "loss": 1.6291,
+      "step": 5972
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8817438692098093e-05,
+      "loss": 1.7421,
+      "step": 5973
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.881539509536785e-05,
+      "loss": 1.7716,
+      "step": 5974
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8813351498637604e-05,
+      "loss": 1.6425,
+      "step": 5975
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8811307901907356e-05,
+      "loss": 1.7075,
+      "step": 5976
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8809264305177112e-05,
+      "loss": 1.8987,
+      "step": 5977
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8807220708446868e-05,
+      "loss": 1.6668,
+      "step": 5978
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.880517711171662e-05,
+      "loss": 1.6392,
+      "step": 5979
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.880313351498638e-05,
+      "loss": 1.8285,
+      "step": 5980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.880108991825613e-05,
+      "loss": 1.6689,
+      "step": 5981
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8799046321525887e-05,
+      "loss": 1.7332,
+      "step": 5982
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8797002724795643e-05,
+      "loss": 1.6978,
+      "step": 5983
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8794959128065395e-05,
+      "loss": 1.8237,
+      "step": 5984
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.879291553133515e-05,
+      "loss": 1.7735,
+      "step": 5985
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8790871934604903e-05,
+      "loss": 1.6817,
+      "step": 5986
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8788828337874658e-05,
+      "loss": 1.7218,
+      "step": 5987
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8786784741144417e-05,
+      "loss": 1.712,
+      "step": 5988
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.878474114441417e-05,
+      "loss": 1.6227,
+      "step": 5989
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8782697547683925e-05,
+      "loss": 1.7178,
+      "step": 5990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8780653950953677e-05,
+      "loss": 1.7286,
+      "step": 5991
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8778610354223433e-05,
+      "loss": 1.6931,
+      "step": 5992
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.877656675749319e-05,
+      "loss": 1.8002,
+      "step": 5993
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.877452316076294e-05,
+      "loss": 1.7913,
+      "step": 5994
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.87724795640327e-05,
+      "loss": 1.7388,
+      "step": 5995
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8770435967302456e-05,
+      "loss": 1.7703,
+      "step": 5996
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8768392370572208e-05,
+      "loss": 1.684,
+      "step": 5997
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8766348773841963e-05,
+      "loss": 1.735,
+      "step": 5998
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8764305177111716e-05,
+      "loss": 1.678,
+      "step": 5999
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.876226158038147e-05,
+      "loss": 1.6543,
+      "step": 6000
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 1.564336895942688,
+      "eval_runtime": 720.7697,
+      "eval_samples_per_second": 5.758,
+      "eval_steps_per_second": 1.44,
+      "step": 6000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8760217983651227e-05,
+      "loss": 1.7636,
+      "step": 6001
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.875817438692098e-05,
+      "loss": 1.6118,
+      "step": 6002
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8756130790190738e-05,
+      "loss": 1.6114,
+      "step": 6003
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.875408719346049e-05,
+      "loss": 1.6088,
+      "step": 6004
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8752043596730246e-05,
+      "loss": 1.6847,
+      "step": 6005
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.9878,
+      "step": 6006
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8747956403269754e-05,
+      "loss": 1.75,
+      "step": 6007
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.874591280653951e-05,
+      "loss": 1.8074,
+      "step": 6008
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8743869209809262e-05,
+      "loss": 1.7129,
+      "step": 6009
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.874182561307902e-05,
+      "loss": 1.8002,
+      "step": 6010
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8739782016348777e-05,
+      "loss": 1.594,
+      "step": 6011
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.873773841961853e-05,
+      "loss": 1.7349,
+      "step": 6012
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8735694822888284e-05,
+      "loss": 1.7141,
+      "step": 6013
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.873365122615804e-05,
+      "loss": 1.8154,
+      "step": 6014
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8731607629427792e-05,
+      "loss": 1.6939,
+      "step": 6015
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8729564032697548e-05,
+      "loss": 1.7392,
+      "step": 6016
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.87275204359673e-05,
+      "loss": 1.811,
+      "step": 6017
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.872547683923706e-05,
+      "loss": 1.7685,
+      "step": 6018
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8723433242506815e-05,
+      "loss": 1.6806,
+      "step": 6019
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8721389645776567e-05,
+      "loss": 1.7987,
+      "step": 6020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8719346049046323e-05,
+      "loss": 1.855,
+      "step": 6021
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8717302452316075e-05,
+      "loss": 1.7673,
+      "step": 6022
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.871525885558583e-05,
+      "loss": 1.8324,
+      "step": 6023
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8713215258855586e-05,
+      "loss": 1.6868,
+      "step": 6024
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8711171662125342e-05,
+      "loss": 1.7205,
+      "step": 6025
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8709128065395097e-05,
+      "loss": 1.9823,
+      "step": 6026
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.870708446866485e-05,
+      "loss": 1.7875,
+      "step": 6027
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8705040871934605e-05,
+      "loss": 1.7034,
+      "step": 6028
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.870299727520436e-05,
+      "loss": 1.6421,
+      "step": 6029
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8700953678474113e-05,
+      "loss": 1.6982,
+      "step": 6030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.869891008174387e-05,
+      "loss": 1.7344,
+      "step": 6031
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8696866485013624e-05,
+      "loss": 1.6096,
+      "step": 6032
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.869482288828338e-05,
+      "loss": 1.8827,
+      "step": 6033
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8692779291553136e-05,
+      "loss": 1.6708,
+      "step": 6034
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8690735694822888e-05,
+      "loss": 1.6792,
+      "step": 6035
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8688692098092644e-05,
+      "loss": 1.7443,
+      "step": 6036
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.86866485013624e-05,
+      "loss": 1.6231,
+      "step": 6037
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.868460490463215e-05,
+      "loss": 1.638,
+      "step": 6038
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8682561307901907e-05,
+      "loss": 1.7417,
+      "step": 6039
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8680517711171663e-05,
+      "loss": 1.7634,
+      "step": 6040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.867847411444142e-05,
+      "loss": 1.688,
+      "step": 6041
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8676430517711174e-05,
+      "loss": 1.771,
+      "step": 6042
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8674386920980926e-05,
+      "loss": 1.7541,
+      "step": 6043
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8672343324250682e-05,
+      "loss": 1.7517,
+      "step": 6044
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8670299727520434e-05,
+      "loss": 1.6987,
+      "step": 6045
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.866825613079019e-05,
+      "loss": 1.713,
+      "step": 6046
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8666212534059945e-05,
+      "loss": 1.6778,
+      "step": 6047
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.86641689373297e-05,
+      "loss": 1.6808,
+      "step": 6048
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8662125340599457e-05,
+      "loss": 1.7429,
+      "step": 6049
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.866008174386921e-05,
+      "loss": 1.894,
+      "step": 6050
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8658038147138965e-05,
+      "loss": 1.8189,
+      "step": 6051
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.865599455040872e-05,
+      "loss": 1.83,
+      "step": 6052
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8653950953678472e-05,
+      "loss": 1.8768,
+      "step": 6053
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8651907356948228e-05,
+      "loss": 1.7083,
+      "step": 6054
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8649863760217987e-05,
+      "loss": 1.767,
+      "step": 6055
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.864782016348774e-05,
+      "loss": 1.668,
+      "step": 6056
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8645776566757495e-05,
+      "loss": 1.6331,
+      "step": 6057
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8643732970027247e-05,
+      "loss": 1.7805,
+      "step": 6058
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8641689373297003e-05,
+      "loss": 1.9128,
+      "step": 6059
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.863964577656676e-05,
+      "loss": 1.7937,
+      "step": 6060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.863760217983651e-05,
+      "loss": 1.9049,
+      "step": 6061
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.863555858310627e-05,
+      "loss": 1.809,
+      "step": 6062
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8633514986376022e-05,
+      "loss": 1.816,
+      "step": 6063
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8631471389645778e-05,
+      "loss": 1.7779,
+      "step": 6064
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8629427792915533e-05,
+      "loss": 1.6508,
+      "step": 6065
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8627384196185285e-05,
+      "loss": 1.841,
+      "step": 6066
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.862534059945504e-05,
+      "loss": 1.9575,
+      "step": 6067
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8623297002724797e-05,
+      "loss": 1.7701,
+      "step": 6068
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.862125340599455e-05,
+      "loss": 1.7647,
+      "step": 6069
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8619209809264308e-05,
+      "loss": 1.7764,
+      "step": 6070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.861716621253406e-05,
+      "loss": 1.7747,
+      "step": 6071
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8615122615803816e-05,
+      "loss": 1.7342,
+      "step": 6072
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.861307901907357e-05,
+      "loss": 1.806,
+      "step": 6073
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8611035422343324e-05,
+      "loss": 1.6463,
+      "step": 6074
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.860899182561308e-05,
+      "loss": 1.7084,
+      "step": 6075
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.860694822888283e-05,
+      "loss": 1.7822,
+      "step": 6076
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.860490463215259e-05,
+      "loss": 1.663,
+      "step": 6077
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8602861035422346e-05,
+      "loss": 1.808,
+      "step": 6078
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.86008174386921e-05,
+      "loss": 1.6666,
+      "step": 6079
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8598773841961854e-05,
+      "loss": 1.8792,
+      "step": 6080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8596730245231606e-05,
+      "loss": 1.7144,
+      "step": 6081
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8594686648501362e-05,
+      "loss": 1.5938,
+      "step": 6082
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8592643051771118e-05,
+      "loss": 1.6946,
+      "step": 6083
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.859059945504087e-05,
+      "loss": 1.6865,
+      "step": 6084
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.858855585831063e-05,
+      "loss": 1.7811,
+      "step": 6085
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.858651226158038e-05,
+      "loss": 1.7893,
+      "step": 6086
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8584468664850137e-05,
+      "loss": 1.7023,
+      "step": 6087
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8582425068119892e-05,
+      "loss": 1.7101,
+      "step": 6088
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8580381471389645e-05,
+      "loss": 1.7834,
+      "step": 6089
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.85783378746594e-05,
+      "loss": 1.6887,
+      "step": 6090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8576294277929156e-05,
+      "loss": 1.684,
+      "step": 6091
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.857425068119891e-05,
+      "loss": 1.7316,
+      "step": 6092
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8572207084468667e-05,
+      "loss": 1.7669,
+      "step": 6093
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.857016348773842e-05,
+      "loss": 1.6515,
+      "step": 6094
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8568119891008175e-05,
+      "loss": 1.8395,
+      "step": 6095
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.856607629427793e-05,
+      "loss": 1.8004,
+      "step": 6096
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8564032697547683e-05,
+      "loss": 1.8839,
+      "step": 6097
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.856198910081744e-05,
+      "loss": 1.7218,
+      "step": 6098
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.855994550408719e-05,
+      "loss": 1.6978,
+      "step": 6099
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.855790190735695e-05,
+      "loss": 1.8657,
+      "step": 6100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8555858310626706e-05,
+      "loss": 1.7247,
+      "step": 6101
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8553814713896458e-05,
+      "loss": 1.7328,
+      "step": 6102
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8551771117166213e-05,
+      "loss": 1.6319,
+      "step": 6103
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8549727520435966e-05,
+      "loss": 1.6429,
+      "step": 6104
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.854768392370572e-05,
+      "loss": 1.7041,
+      "step": 6105
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8545640326975477e-05,
+      "loss": 1.7632,
+      "step": 6106
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8543596730245233e-05,
+      "loss": 1.7765,
+      "step": 6107
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8541553133514988e-05,
+      "loss": 1.8382,
+      "step": 6108
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8539509536784744e-05,
+      "loss": 1.7046,
+      "step": 6109
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8537465940054496e-05,
+      "loss": 1.7402,
+      "step": 6110
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8535422343324252e-05,
+      "loss": 1.68,
+      "step": 6111
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8533378746594004e-05,
+      "loss": 1.833,
+      "step": 6112
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.853133514986376e-05,
+      "loss": 1.7467,
+      "step": 6113
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8529291553133515e-05,
+      "loss": 1.5509,
+      "step": 6114
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.852724795640327e-05,
+      "loss": 1.743,
+      "step": 6115
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8525204359673026e-05,
+      "loss": 1.7381,
+      "step": 6116
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.852316076294278e-05,
+      "loss": 1.8154,
+      "step": 6117
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8521117166212534e-05,
+      "loss": 1.7088,
+      "step": 6118
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.851907356948229e-05,
+      "loss": 1.7768,
+      "step": 6119
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8517029972752042e-05,
+      "loss": 1.7726,
+      "step": 6120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8514986376021798e-05,
+      "loss": 1.7352,
+      "step": 6121
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8512942779291553e-05,
+      "loss": 1.6888,
+      "step": 6122
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.851089918256131e-05,
+      "loss": 1.7639,
+      "step": 6123
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8508855585831065e-05,
+      "loss": 1.7178,
+      "step": 6124
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8506811989100817e-05,
+      "loss": 1.7397,
+      "step": 6125
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8504768392370573e-05,
+      "loss": 1.6199,
+      "step": 6126
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8502724795640328e-05,
+      "loss": 1.678,
+      "step": 6127
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.850068119891008e-05,
+      "loss": 1.7002,
+      "step": 6128
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8498637602179836e-05,
+      "loss": 1.732,
+      "step": 6129
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8496594005449592e-05,
+      "loss": 1.766,
+      "step": 6130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8494550408719347e-05,
+      "loss": 1.8264,
+      "step": 6131
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8492506811989103e-05,
+      "loss": 1.6192,
+      "step": 6132
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8490463215258855e-05,
+      "loss": 1.8284,
+      "step": 6133
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.848841961852861e-05,
+      "loss": 1.723,
+      "step": 6134
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8486376021798363e-05,
+      "loss": 1.7457,
+      "step": 6135
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.848433242506812e-05,
+      "loss": 1.7383,
+      "step": 6136
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8482288828337878e-05,
+      "loss": 1.7691,
+      "step": 6137
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.848024523160763e-05,
+      "loss": 1.5841,
+      "step": 6138
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8478201634877386e-05,
+      "loss": 1.702,
+      "step": 6139
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8476158038147138e-05,
+      "loss": 1.8002,
+      "step": 6140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8474114441416894e-05,
+      "loss": 1.742,
+      "step": 6141
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.847207084468665e-05,
+      "loss": 1.649,
+      "step": 6142
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.84700272479564e-05,
+      "loss": 1.7832,
+      "step": 6143
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.846798365122616e-05,
+      "loss": 1.7444,
+      "step": 6144
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8465940054495916e-05,
+      "loss": 1.7656,
+      "step": 6145
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.846389645776567e-05,
+      "loss": 1.7917,
+      "step": 6146
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8461852861035424e-05,
+      "loss": 1.8163,
+      "step": 6147
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8459809264305176e-05,
+      "loss": 1.9405,
+      "step": 6148
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8457765667574932e-05,
+      "loss": 1.7273,
+      "step": 6149
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8455722070844687e-05,
+      "loss": 1.6783,
+      "step": 6150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.845367847411444e-05,
+      "loss": 1.6434,
+      "step": 6151
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.84516348773842e-05,
+      "loss": 1.7252,
+      "step": 6152
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.844959128065395e-05,
+      "loss": 1.6322,
+      "step": 6153
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8447547683923707e-05,
+      "loss": 1.7711,
+      "step": 6154
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8445504087193462e-05,
+      "loss": 1.8197,
+      "step": 6155
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8443460490463215e-05,
+      "loss": 1.6899,
+      "step": 6156
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.844141689373297e-05,
+      "loss": 1.7345,
+      "step": 6157
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8439373297002722e-05,
+      "loss": 1.7981,
+      "step": 6158
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.843732970027248e-05,
+      "loss": 1.8357,
+      "step": 6159
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8435286103542237e-05,
+      "loss": 1.7297,
+      "step": 6160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.843324250681199e-05,
+      "loss": 1.7039,
+      "step": 6161
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8431198910081745e-05,
+      "loss": 1.7517,
+      "step": 6162
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.84291553133515e-05,
+      "loss": 1.7553,
+      "step": 6163
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8427111716621253e-05,
+      "loss": 1.7131,
+      "step": 6164
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.842506811989101e-05,
+      "loss": 1.6899,
+      "step": 6165
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.842302452316076e-05,
+      "loss": 1.6162,
+      "step": 6166
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.842098092643052e-05,
+      "loss": 1.8017,
+      "step": 6167
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8418937329700275e-05,
+      "loss": 1.6529,
+      "step": 6168
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8416893732970028e-05,
+      "loss": 1.8321,
+      "step": 6169
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8414850136239783e-05,
+      "loss": 1.7053,
+      "step": 6170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8412806539509535e-05,
+      "loss": 1.6661,
+      "step": 6171
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.841076294277929e-05,
+      "loss": 1.597,
+      "step": 6172
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8408719346049047e-05,
+      "loss": 1.7454,
+      "step": 6173
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8406675749318802e-05,
+      "loss": 1.7063,
+      "step": 6174
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8404632152588558e-05,
+      "loss": 1.7556,
+      "step": 6175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.840258855585831e-05,
+      "loss": 1.6562,
+      "step": 6176
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8400544959128066e-05,
+      "loss": 1.7075,
+      "step": 6177
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.839850136239782e-05,
+      "loss": 1.803,
+      "step": 6178
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8396457765667574e-05,
+      "loss": 1.7771,
+      "step": 6179
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.839441416893733e-05,
+      "loss": 1.8347,
+      "step": 6180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8392370572207085e-05,
+      "loss": 1.755,
+      "step": 6181
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.839032697547684e-05,
+      "loss": 1.5681,
+      "step": 6182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8388283378746596e-05,
+      "loss": 1.7486,
+      "step": 6183
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.838623978201635e-05,
+      "loss": 1.633,
+      "step": 6184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8384196185286104e-05,
+      "loss": 1.4687,
+      "step": 6185
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.838215258855586e-05,
+      "loss": 1.6116,
+      "step": 6186
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8380108991825612e-05,
+      "loss": 1.7941,
+      "step": 6187
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8378065395095368e-05,
+      "loss": 1.658,
+      "step": 6188
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8376021798365123e-05,
+      "loss": 1.8193,
+      "step": 6189
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.837397820163488e-05,
+      "loss": 1.7781,
+      "step": 6190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8371934604904635e-05,
+      "loss": 1.9378,
+      "step": 6191
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8369891008174387e-05,
+      "loss": 1.8053,
+      "step": 6192
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8367847411444142e-05,
+      "loss": 1.7525,
+      "step": 6193
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8365803814713895e-05,
+      "loss": 1.8851,
+      "step": 6194
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.836376021798365e-05,
+      "loss": 1.7384,
+      "step": 6195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8361716621253406e-05,
+      "loss": 1.7722,
+      "step": 6196
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.835967302452316e-05,
+      "loss": 1.7301,
+      "step": 6197
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8357629427792917e-05,
+      "loss": 1.7109,
+      "step": 6198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8355585831062673e-05,
+      "loss": 1.6947,
+      "step": 6199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8353542234332425e-05,
+      "loss": 1.801,
+      "step": 6200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.835149863760218e-05,
+      "loss": 1.9595,
+      "step": 6201
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8349455040871933e-05,
+      "loss": 1.7555,
+      "step": 6202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.834741144414169e-05,
+      "loss": 1.7835,
+      "step": 6203
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8345367847411448e-05,
+      "loss": 1.698,
+      "step": 6204
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.83433242506812e-05,
+      "loss": 1.8285,
+      "step": 6205
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8341280653950955e-05,
+      "loss": 1.6449,
+      "step": 6206
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8339237057220708e-05,
+      "loss": 1.7191,
+      "step": 6207
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8337193460490463e-05,
+      "loss": 1.8034,
+      "step": 6208
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.833514986376022e-05,
+      "loss": 1.8842,
+      "step": 6209
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.833310626702997e-05,
+      "loss": 1.7286,
+      "step": 6210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8331062670299727e-05,
+      "loss": 1.6223,
+      "step": 6211
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8329019073569483e-05,
+      "loss": 1.8677,
+      "step": 6212
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8326975476839238e-05,
+      "loss": 1.7768,
+      "step": 6213
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8324931880108994e-05,
+      "loss": 1.8845,
+      "step": 6214
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8322888283378746e-05,
+      "loss": 1.8052,
+      "step": 6215
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.83208446866485e-05,
+      "loss": 1.913,
+      "step": 6216
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8318801089918257e-05,
+      "loss": 1.7908,
+      "step": 6217
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.831675749318801e-05,
+      "loss": 1.8318,
+      "step": 6218
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.831471389645777e-05,
+      "loss": 1.7759,
+      "step": 6219
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.831267029972752e-05,
+      "loss": 1.6714,
+      "step": 6220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8310626702997276e-05,
+      "loss": 1.7272,
+      "step": 6221
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8308583106267032e-05,
+      "loss": 1.6859,
+      "step": 6222
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8306539509536784e-05,
+      "loss": 1.6782,
+      "step": 6223
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.830449591280654e-05,
+      "loss": 1.7955,
+      "step": 6224
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8302452316076292e-05,
+      "loss": 1.4965,
+      "step": 6225
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8300408719346048e-05,
+      "loss": 1.6735,
+      "step": 6226
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8298365122615807e-05,
+      "loss": 1.8487,
+      "step": 6227
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.829632152588556e-05,
+      "loss": 1.7332,
+      "step": 6228
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8294277929155315e-05,
+      "loss": 1.7733,
+      "step": 6229
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8292234332425067e-05,
+      "loss": 1.6657,
+      "step": 6230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8290190735694823e-05,
+      "loss": 1.7989,
+      "step": 6231
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8288147138964578e-05,
+      "loss": 1.6911,
+      "step": 6232
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.828610354223433e-05,
+      "loss": 1.6182,
+      "step": 6233
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.828405994550409e-05,
+      "loss": 1.8464,
+      "step": 6234
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8282016348773845e-05,
+      "loss": 1.8682,
+      "step": 6235
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8279972752043597e-05,
+      "loss": 1.7099,
+      "step": 6236
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8277929155313353e-05,
+      "loss": 1.689,
+      "step": 6237
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8275885558583105e-05,
+      "loss": 1.7775,
+      "step": 6238
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.827384196185286e-05,
+      "loss": 1.7,
+      "step": 6239
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8271798365122617e-05,
+      "loss": 1.7229,
+      "step": 6240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8269754768392372e-05,
+      "loss": 1.645,
+      "step": 6241
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8267711171662128e-05,
+      "loss": 1.7014,
+      "step": 6242
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.826566757493188e-05,
+      "loss": 1.703,
+      "step": 6243
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8263623978201636e-05,
+      "loss": 1.7347,
+      "step": 6244
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.826158038147139e-05,
+      "loss": 1.5431,
+      "step": 6245
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8259536784741144e-05,
+      "loss": 1.7107,
+      "step": 6246
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.82574931880109e-05,
+      "loss": 1.6807,
+      "step": 6247
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.825544959128065e-05,
+      "loss": 1.6776,
+      "step": 6248
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.825340599455041e-05,
+      "loss": 1.7046,
+      "step": 6249
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8251362397820166e-05,
+      "loss": 1.6635,
+      "step": 6250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8249318801089918e-05,
+      "loss": 1.8428,
+      "step": 6251
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8247275204359674e-05,
+      "loss": 1.7343,
+      "step": 6252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.824523160762943e-05,
+      "loss": 1.7825,
+      "step": 6253
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8243188010899182e-05,
+      "loss": 1.6538,
+      "step": 6254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8241144414168937e-05,
+      "loss": 1.7167,
+      "step": 6255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8239100817438693e-05,
+      "loss": 1.688,
+      "step": 6256
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.823705722070845e-05,
+      "loss": 1.7859,
+      "step": 6257
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8235013623978204e-05,
+      "loss": 1.6481,
+      "step": 6258
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8232970027247957e-05,
+      "loss": 1.6304,
+      "step": 6259
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8230926430517712e-05,
+      "loss": 1.645,
+      "step": 6260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8228882833787464e-05,
+      "loss": 1.6226,
+      "step": 6261
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.822683923705722e-05,
+      "loss": 1.7402,
+      "step": 6262
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8224795640326976e-05,
+      "loss": 1.6938,
+      "step": 6263
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.822275204359673e-05,
+      "loss": 1.8034,
+      "step": 6264
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8220708446866487e-05,
+      "loss": 1.8628,
+      "step": 6265
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.821866485013624e-05,
+      "loss": 1.7881,
+      "step": 6266
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8216621253405995e-05,
+      "loss": 1.6674,
+      "step": 6267
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.821457765667575e-05,
+      "loss": 1.7912,
+      "step": 6268
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8212534059945503e-05,
+      "loss": 1.6818,
+      "step": 6269
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.821049046321526e-05,
+      "loss": 1.8707,
+      "step": 6270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8208446866485014e-05,
+      "loss": 1.7367,
+      "step": 6271
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.820640326975477e-05,
+      "loss": 1.6745,
+      "step": 6272
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8204359673024525e-05,
+      "loss": 1.8466,
+      "step": 6273
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8202316076294278e-05,
+      "loss": 1.6852,
+      "step": 6274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8200272479564033e-05,
+      "loss": 1.7254,
+      "step": 6275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.819822888283379e-05,
+      "loss": 1.655,
+      "step": 6276
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.819618528610354e-05,
+      "loss": 1.7593,
+      "step": 6277
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8194141689373297e-05,
+      "loss": 1.7723,
+      "step": 6278
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8192098092643052e-05,
+      "loss": 1.642,
+      "step": 6279
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8190054495912808e-05,
+      "loss": 1.8114,
+      "step": 6280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8188010899182564e-05,
+      "loss": 1.5558,
+      "step": 6281
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8185967302452316e-05,
+      "loss": 1.6763,
+      "step": 6282
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.818392370572207e-05,
+      "loss": 1.7859,
+      "step": 6283
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8181880108991824e-05,
+      "loss": 1.7637,
+      "step": 6284
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.817983651226158e-05,
+      "loss": 1.8969,
+      "step": 6285
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.817779291553134e-05,
+      "loss": 1.6366,
+      "step": 6286
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.817574931880109e-05,
+      "loss": 1.8023,
+      "step": 6287
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8173705722070846e-05,
+      "loss": 1.7246,
+      "step": 6288
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.81716621253406e-05,
+      "loss": 1.6933,
+      "step": 6289
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8169618528610354e-05,
+      "loss": 1.7085,
+      "step": 6290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.816757493188011e-05,
+      "loss": 1.6677,
+      "step": 6291
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8165531335149862e-05,
+      "loss": 1.7294,
+      "step": 6292
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8163487738419618e-05,
+      "loss": 1.8446,
+      "step": 6293
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8161444141689377e-05,
+      "loss": 1.7468,
+      "step": 6294
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.815940054495913e-05,
+      "loss": 1.7416,
+      "step": 6295
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8157356948228885e-05,
+      "loss": 1.7414,
+      "step": 6296
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8155313351498637e-05,
+      "loss": 1.7698,
+      "step": 6297
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8153269754768392e-05,
+      "loss": 1.7651,
+      "step": 6298
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8151226158038148e-05,
+      "loss": 1.8389,
+      "step": 6299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.81491825613079e-05,
+      "loss": 1.739,
+      "step": 6300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.814713896457766e-05,
+      "loss": 1.6763,
+      "step": 6301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.814509536784741e-05,
+      "loss": 1.7352,
+      "step": 6302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8143051771117167e-05,
+      "loss": 1.6028,
+      "step": 6303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8141008174386923e-05,
+      "loss": 1.7059,
+      "step": 6304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8138964577656675e-05,
+      "loss": 1.7668,
+      "step": 6305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.813692098092643e-05,
+      "loss": 1.7461,
+      "step": 6306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8134877384196183e-05,
+      "loss": 1.7169,
+      "step": 6307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.813283378746594e-05,
+      "loss": 1.7689,
+      "step": 6308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8130790190735698e-05,
+      "loss": 1.7403,
+      "step": 6309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.812874659400545e-05,
+      "loss": 1.6989,
+      "step": 6310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8126702997275205e-05,
+      "loss": 1.6481,
+      "step": 6311
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.812465940054496e-05,
+      "loss": 1.6788,
+      "step": 6312
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8122615803814713e-05,
+      "loss": 1.6499,
+      "step": 6313
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.812057220708447e-05,
+      "loss": 1.8638,
+      "step": 6314
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.811852861035422e-05,
+      "loss": 1.736,
+      "step": 6315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.811648501362398e-05,
+      "loss": 1.732,
+      "step": 6316
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8114441416893736e-05,
+      "loss": 1.7304,
+      "step": 6317
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8112397820163488e-05,
+      "loss": 1.8393,
+      "step": 6318
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8110354223433244e-05,
+      "loss": 1.5845,
+      "step": 6319
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8108310626702996e-05,
+      "loss": 1.5748,
+      "step": 6320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.810626702997275e-05,
+      "loss": 1.6253,
+      "step": 6321
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8104223433242507e-05,
+      "loss": 1.6671,
+      "step": 6322
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.810217983651226e-05,
+      "loss": 1.8483,
+      "step": 6323
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.810013623978202e-05,
+      "loss": 1.7654,
+      "step": 6324
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.809809264305177e-05,
+      "loss": 1.8002,
+      "step": 6325
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8096049046321526e-05,
+      "loss": 1.6547,
+      "step": 6326
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8094005449591282e-05,
+      "loss": 1.7022,
+      "step": 6327
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8091961852861034e-05,
+      "loss": 1.848,
+      "step": 6328
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.808991825613079e-05,
+      "loss": 1.6729,
+      "step": 6329
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8087874659400546e-05,
+      "loss": 1.618,
+      "step": 6330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.80858310626703e-05,
+      "loss": 1.8173,
+      "step": 6331
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8083787465940057e-05,
+      "loss": 1.7472,
+      "step": 6332
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.808174386920981e-05,
+      "loss": 1.6776,
+      "step": 6333
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8079700272479565e-05,
+      "loss": 1.6401,
+      "step": 6334
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.807765667574932e-05,
+      "loss": 1.681,
+      "step": 6335
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8075613079019073e-05,
+      "loss": 1.677,
+      "step": 6336
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8073569482288828e-05,
+      "loss": 1.6284,
+      "step": 6337
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8071525885558584e-05,
+      "loss": 1.6984,
+      "step": 6338
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.806948228882834e-05,
+      "loss": 1.6393,
+      "step": 6339
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8067438692098095e-05,
+      "loss": 1.7079,
+      "step": 6340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8065395095367847e-05,
+      "loss": 1.7867,
+      "step": 6341
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8063351498637603e-05,
+      "loss": 1.7655,
+      "step": 6342
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8061307901907355e-05,
+      "loss": 1.6547,
+      "step": 6343
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.805926430517711e-05,
+      "loss": 1.6947,
+      "step": 6344
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8057220708446866e-05,
+      "loss": 1.76,
+      "step": 6345
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8055177111716622e-05,
+      "loss": 1.7749,
+      "step": 6346
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8053133514986378e-05,
+      "loss": 1.7097,
+      "step": 6347
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8051089918256133e-05,
+      "loss": 1.768,
+      "step": 6348
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8049046321525886e-05,
+      "loss": 1.7324,
+      "step": 6349
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.804700272479564e-05,
+      "loss": 1.7309,
+      "step": 6350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8044959128065393e-05,
+      "loss": 1.728,
+      "step": 6351
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.804291553133515e-05,
+      "loss": 1.8009,
+      "step": 6352
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8040871934604908e-05,
+      "loss": 1.7841,
+      "step": 6353
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.803882833787466e-05,
+      "loss": 1.8169,
+      "step": 6354
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8036784741144416e-05,
+      "loss": 1.6785,
+      "step": 6355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8034741144414168e-05,
+      "loss": 1.616,
+      "step": 6356
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8032697547683924e-05,
+      "loss": 1.7151,
+      "step": 6357
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.803065395095368e-05,
+      "loss": 1.6377,
+      "step": 6358
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8028610354223432e-05,
+      "loss": 1.7089,
+      "step": 6359
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8026566757493187e-05,
+      "loss": 1.7268,
+      "step": 6360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8024523160762943e-05,
+      "loss": 1.6822,
+      "step": 6361
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.80224795640327e-05,
+      "loss": 1.7568,
+      "step": 6362
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8020435967302454e-05,
+      "loss": 1.6141,
+      "step": 6363
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8018392370572207e-05,
+      "loss": 1.718,
+      "step": 6364
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8016348773841962e-05,
+      "loss": 1.784,
+      "step": 6365
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8014305177111718e-05,
+      "loss": 1.7333,
+      "step": 6366
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.801226158038147e-05,
+      "loss": 1.7533,
+      "step": 6367
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.801021798365123e-05,
+      "loss": 1.6581,
+      "step": 6368
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.800817438692098e-05,
+      "loss": 1.7957,
+      "step": 6369
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8006130790190737e-05,
+      "loss": 1.6999,
+      "step": 6370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8004087193460493e-05,
+      "loss": 1.7561,
+      "step": 6371
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8002043596730245e-05,
+      "loss": 1.8079,
+      "step": 6372
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8e-05,
+      "loss": 1.8147,
+      "step": 6373
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7997956403269753e-05,
+      "loss": 1.6232,
+      "step": 6374
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.799591280653951e-05,
+      "loss": 1.6574,
+      "step": 6375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7993869209809267e-05,
+      "loss": 1.6391,
+      "step": 6376
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.799182561307902e-05,
+      "loss": 1.6306,
+      "step": 6377
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7989782016348775e-05,
+      "loss": 1.7379,
+      "step": 6378
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7987738419618527e-05,
+      "loss": 1.7575,
+      "step": 6379
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7985694822888283e-05,
+      "loss": 1.7759,
+      "step": 6380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.798365122615804e-05,
+      "loss": 1.6943,
+      "step": 6381
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.798160762942779e-05,
+      "loss": 1.7015,
+      "step": 6382
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.797956403269755e-05,
+      "loss": 1.5897,
+      "step": 6383
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7977520435967306e-05,
+      "loss": 1.7256,
+      "step": 6384
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7975476839237058e-05,
+      "loss": 1.6659,
+      "step": 6385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7973433242506814e-05,
+      "loss": 1.5777,
+      "step": 6386
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7971389645776566e-05,
+      "loss": 1.826,
+      "step": 6387
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.796934604904632e-05,
+      "loss": 1.6435,
+      "step": 6388
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7967302452316077e-05,
+      "loss": 1.6928,
+      "step": 6389
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.796525885558583e-05,
+      "loss": 1.6793,
+      "step": 6390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7963215258855588e-05,
+      "loss": 1.6555,
+      "step": 6391
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.796117166212534e-05,
+      "loss": 1.7292,
+      "step": 6392
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7959128065395096e-05,
+      "loss": 1.7497,
+      "step": 6393
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7957084468664852e-05,
+      "loss": 1.6411,
+      "step": 6394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7955040871934604e-05,
+      "loss": 1.6751,
+      "step": 6395
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.795299727520436e-05,
+      "loss": 1.7235,
+      "step": 6396
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7950953678474112e-05,
+      "loss": 1.692,
+      "step": 6397
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.794891008174387e-05,
+      "loss": 1.826,
+      "step": 6398
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7946866485013627e-05,
+      "loss": 1.5085,
+      "step": 6399
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.794482288828338e-05,
+      "loss": 1.7826,
+      "step": 6400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7942779291553134e-05,
+      "loss": 1.7297,
+      "step": 6401
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.794073569482289e-05,
+      "loss": 1.7939,
+      "step": 6402
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7938692098092642e-05,
+      "loss": 1.6679,
+      "step": 6403
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7936648501362398e-05,
+      "loss": 1.829,
+      "step": 6404
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.793460490463215e-05,
+      "loss": 1.8375,
+      "step": 6405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.793256130790191e-05,
+      "loss": 1.773,
+      "step": 6406
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7930517711171665e-05,
+      "loss": 1.719,
+      "step": 6407
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7928474114441417e-05,
+      "loss": 1.7239,
+      "step": 6408
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7926430517711173e-05,
+      "loss": 1.8913,
+      "step": 6409
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7924386920980925e-05,
+      "loss": 1.6612,
+      "step": 6410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.792234332425068e-05,
+      "loss": 1.7559,
+      "step": 6411
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7920299727520436e-05,
+      "loss": 1.6583,
+      "step": 6412
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7918256130790192e-05,
+      "loss": 1.7012,
+      "step": 6413
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7916212534059948e-05,
+      "loss": 1.7344,
+      "step": 6414
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.79141689373297e-05,
+      "loss": 1.7095,
+      "step": 6415
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7912125340599455e-05,
+      "loss": 1.733,
+      "step": 6416
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.791008174386921e-05,
+      "loss": 1.7558,
+      "step": 6417
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7908038147138963e-05,
+      "loss": 1.6971,
+      "step": 6418
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.790599455040872e-05,
+      "loss": 1.6762,
+      "step": 6419
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7903950953678478e-05,
+      "loss": 1.8042,
+      "step": 6420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.790190735694823e-05,
+      "loss": 1.6872,
+      "step": 6421
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7899863760217986e-05,
+      "loss": 1.8157,
+      "step": 6422
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7897820163487738e-05,
+      "loss": 1.7925,
+      "step": 6423
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7895776566757494e-05,
+      "loss": 1.7319,
+      "step": 6424
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.789373297002725e-05,
+      "loss": 1.7067,
+      "step": 6425
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7891689373297e-05,
+      "loss": 1.5981,
+      "step": 6426
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7889645776566757e-05,
+      "loss": 1.7341,
+      "step": 6427
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7887602179836513e-05,
+      "loss": 1.7436,
+      "step": 6428
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.788555858310627e-05,
+      "loss": 1.6766,
+      "step": 6429
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7883514986376024e-05,
+      "loss": 1.7699,
+      "step": 6430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7881471389645776e-05,
+      "loss": 1.8423,
+      "step": 6431
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7879427792915532e-05,
+      "loss": 1.6352,
+      "step": 6432
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7877384196185284e-05,
+      "loss": 1.7688,
+      "step": 6433
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.787534059945504e-05,
+      "loss": 1.8409,
+      "step": 6434
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.78732970027248e-05,
+      "loss": 1.7078,
+      "step": 6435
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.787125340599455e-05,
+      "loss": 1.5993,
+      "step": 6436
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7869209809264307e-05,
+      "loss": 1.7898,
+      "step": 6437
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7867166212534062e-05,
+      "loss": 1.7281,
+      "step": 6438
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7865122615803815e-05,
+      "loss": 1.8005,
+      "step": 6439
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.786307901907357e-05,
+      "loss": 1.6573,
+      "step": 6440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7861035422343322e-05,
+      "loss": 1.7088,
+      "step": 6441
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7858991825613078e-05,
+      "loss": 1.8033,
+      "step": 6442
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7856948228882837e-05,
+      "loss": 1.67,
+      "step": 6443
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.785490463215259e-05,
+      "loss": 1.7401,
+      "step": 6444
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7852861035422345e-05,
+      "loss": 1.7597,
+      "step": 6445
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7850817438692097e-05,
+      "loss": 1.7757,
+      "step": 6446
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7848773841961853e-05,
+      "loss": 1.8107,
+      "step": 6447
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.784673024523161e-05,
+      "loss": 1.6612,
+      "step": 6448
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.784468664850136e-05,
+      "loss": 1.6624,
+      "step": 6449
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.784264305177112e-05,
+      "loss": 1.6788,
+      "step": 6450
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7840599455040872e-05,
+      "loss": 1.6088,
+      "step": 6451
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7838555858310628e-05,
+      "loss": 1.7857,
+      "step": 6452
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7836512261580383e-05,
+      "loss": 1.7482,
+      "step": 6453
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7834468664850136e-05,
+      "loss": 1.8494,
+      "step": 6454
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.783242506811989e-05,
+      "loss": 1.8242,
+      "step": 6455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7830381471389647e-05,
+      "loss": 1.7433,
+      "step": 6456
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.78283378746594e-05,
+      "loss": 1.7325,
+      "step": 6457
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7826294277929158e-05,
+      "loss": 1.7765,
+      "step": 6458
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.782425068119891e-05,
+      "loss": 1.8324,
+      "step": 6459
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7822207084468666e-05,
+      "loss": 1.7371,
+      "step": 6460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.782016348773842e-05,
+      "loss": 1.725,
+      "step": 6461
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7818119891008174e-05,
+      "loss": 1.6378,
+      "step": 6462
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.781607629427793e-05,
+      "loss": 1.8219,
+      "step": 6463
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7814032697547682e-05,
+      "loss": 1.582,
+      "step": 6464
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.781198910081744e-05,
+      "loss": 1.678,
+      "step": 6465
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7809945504087196e-05,
+      "loss": 1.7213,
+      "step": 6466
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.780790190735695e-05,
+      "loss": 1.8102,
+      "step": 6467
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7805858310626704e-05,
+      "loss": 1.7577,
+      "step": 6468
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7803814713896456e-05,
+      "loss": 1.8005,
+      "step": 6469
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7801771117166212e-05,
+      "loss": 1.6994,
+      "step": 6470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7799727520435968e-05,
+      "loss": 1.8058,
+      "step": 6471
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.779768392370572e-05,
+      "loss": 1.7344,
+      "step": 6472
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.779564032697548e-05,
+      "loss": 1.7384,
+      "step": 6473
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.779359673024523e-05,
+      "loss": 1.7955,
+      "step": 6474
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7791553133514987e-05,
+      "loss": 1.6906,
+      "step": 6475
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7789509536784743e-05,
+      "loss": 1.7072,
+      "step": 6476
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7787465940054495e-05,
+      "loss": 1.7141,
+      "step": 6477
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.778542234332425e-05,
+      "loss": 1.6223,
+      "step": 6478
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7783378746594006e-05,
+      "loss": 1.7012,
+      "step": 6479
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.778133514986376e-05,
+      "loss": 1.7587,
+      "step": 6480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7779291553133517e-05,
+      "loss": 1.6172,
+      "step": 6481
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.777724795640327e-05,
+      "loss": 1.5908,
+      "step": 6482
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7775204359673025e-05,
+      "loss": 1.7461,
+      "step": 6483
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.777316076294278e-05,
+      "loss": 1.7122,
+      "step": 6484
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7771117166212533e-05,
+      "loss": 1.5648,
+      "step": 6485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.776907356948229e-05,
+      "loss": 1.8513,
+      "step": 6486
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.776702997275204e-05,
+      "loss": 1.5807,
+      "step": 6487
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.77649863760218e-05,
+      "loss": 1.6211,
+      "step": 6488
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7762942779291556e-05,
+      "loss": 1.576,
+      "step": 6489
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7760899182561308e-05,
+      "loss": 1.7028,
+      "step": 6490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7758855585831063e-05,
+      "loss": 1.7382,
+      "step": 6491
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7756811989100816e-05,
+      "loss": 1.6665,
+      "step": 6492
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.775476839237057e-05,
+      "loss": 1.8117,
+      "step": 6493
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7752724795640327e-05,
+      "loss": 1.787,
+      "step": 6494
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7750681198910083e-05,
+      "loss": 1.6729,
+      "step": 6495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7748637602179838e-05,
+      "loss": 1.6901,
+      "step": 6496
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7746594005449594e-05,
+      "loss": 1.6929,
+      "step": 6497
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7744550408719346e-05,
+      "loss": 1.8976,
+      "step": 6498
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7742506811989102e-05,
+      "loss": 1.6946,
+      "step": 6499
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7740463215258854e-05,
+      "loss": 1.7131,
+      "step": 6500
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 1.554611086845398,
+      "eval_runtime": 721.4054,
+      "eval_samples_per_second": 5.753,
+      "eval_steps_per_second": 1.439,
+      "step": 6500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.773841961852861e-05,
+      "loss": 1.7255,
+      "step": 6501
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7736376021798365e-05,
+      "loss": 1.7214,
+      "step": 6502
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.773433242506812e-05,
+      "loss": 1.826,
+      "step": 6503
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7732288828337877e-05,
+      "loss": 1.8563,
+      "step": 6504
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.773024523160763e-05,
+      "loss": 1.648,
+      "step": 6505
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7728201634877384e-05,
+      "loss": 1.8185,
+      "step": 6506
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.772615803814714e-05,
+      "loss": 1.6976,
+      "step": 6507
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7724114441416892e-05,
+      "loss": 1.84,
+      "step": 6508
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7722070844686648e-05,
+      "loss": 1.5517,
+      "step": 6509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7720027247956404e-05,
+      "loss": 1.7315,
+      "step": 6510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.771798365122616e-05,
+      "loss": 1.7981,
+      "step": 6511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7715940054495915e-05,
+      "loss": 1.7803,
+      "step": 6512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7713896457765667e-05,
+      "loss": 1.6931,
+      "step": 6513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7711852861035423e-05,
+      "loss": 1.7636,
+      "step": 6514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.770980926430518e-05,
+      "loss": 1.7115,
+      "step": 6515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.770776566757493e-05,
+      "loss": 1.8885,
+      "step": 6516
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.770572207084469e-05,
+      "loss": 1.6706,
+      "step": 6517
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7703678474114442e-05,
+      "loss": 1.7863,
+      "step": 6518
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7701634877384197e-05,
+      "loss": 1.6698,
+      "step": 6519
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7699591280653953e-05,
+      "loss": 1.6418,
+      "step": 6520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7697547683923705e-05,
+      "loss": 1.7201,
+      "step": 6521
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.769550408719346e-05,
+      "loss": 1.7166,
+      "step": 6522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7693460490463213e-05,
+      "loss": 1.8992,
+      "step": 6523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.769141689373297e-05,
+      "loss": 1.7526,
+      "step": 6524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7689373297002728e-05,
+      "loss": 1.6803,
+      "step": 6525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.768732970027248e-05,
+      "loss": 1.801,
+      "step": 6526
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7685286103542236e-05,
+      "loss": 1.7455,
+      "step": 6527
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7683242506811988e-05,
+      "loss": 1.727,
+      "step": 6528
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7681198910081744e-05,
+      "loss": 1.7115,
+      "step": 6529
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.76791553133515e-05,
+      "loss": 1.781,
+      "step": 6530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.767711171662125e-05,
+      "loss": 1.7144,
+      "step": 6531
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.767506811989101e-05,
+      "loss": 1.8038,
+      "step": 6532
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7673024523160766e-05,
+      "loss": 1.7639,
+      "step": 6533
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.767098092643052e-05,
+      "loss": 1.7332,
+      "step": 6534
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7668937329700274e-05,
+      "loss": 1.6604,
+      "step": 6535
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7666893732970026e-05,
+      "loss": 1.6753,
+      "step": 6536
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7664850136239782e-05,
+      "loss": 1.6441,
+      "step": 6537
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7662806539509538e-05,
+      "loss": 1.9233,
+      "step": 6538
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.766076294277929e-05,
+      "loss": 1.666,
+      "step": 6539
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.765871934604905e-05,
+      "loss": 1.7226,
+      "step": 6540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.76566757493188e-05,
+      "loss": 1.7615,
+      "step": 6541
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7654632152588557e-05,
+      "loss": 1.761,
+      "step": 6542
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7652588555858312e-05,
+      "loss": 1.8743,
+      "step": 6543
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7650544959128065e-05,
+      "loss": 1.731,
+      "step": 6544
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.764850136239782e-05,
+      "loss": 1.7508,
+      "step": 6545
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7646457765667572e-05,
+      "loss": 1.765,
+      "step": 6546
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.764441416893733e-05,
+      "loss": 1.6742,
+      "step": 6547
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7642370572207087e-05,
+      "loss": 1.782,
+      "step": 6548
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.764032697547684e-05,
+      "loss": 1.7667,
+      "step": 6549
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7638283378746595e-05,
+      "loss": 1.6383,
+      "step": 6550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.763623978201635e-05,
+      "loss": 1.7443,
+      "step": 6551
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7634196185286103e-05,
+      "loss": 1.6756,
+      "step": 6552
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.763215258855586e-05,
+      "loss": 1.7154,
+      "step": 6553
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.763010899182561e-05,
+      "loss": 1.5923,
+      "step": 6554
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.762806539509537e-05,
+      "loss": 1.7815,
+      "step": 6555
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7626021798365125e-05,
+      "loss": 1.6214,
+      "step": 6556
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7623978201634878e-05,
+      "loss": 1.8363,
+      "step": 6557
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7621934604904633e-05,
+      "loss": 1.8771,
+      "step": 6558
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7619891008174386e-05,
+      "loss": 1.6137,
+      "step": 6559
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.761784741144414e-05,
+      "loss": 1.8318,
+      "step": 6560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7615803814713897e-05,
+      "loss": 1.6923,
+      "step": 6561
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7613760217983652e-05,
+      "loss": 1.6881,
+      "step": 6562
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7611716621253408e-05,
+      "loss": 1.7979,
+      "step": 6563
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.760967302452316e-05,
+      "loss": 1.8157,
+      "step": 6564
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7607629427792916e-05,
+      "loss": 1.818,
+      "step": 6565
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.760558583106267e-05,
+      "loss": 1.658,
+      "step": 6566
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7603542234332424e-05,
+      "loss": 1.6858,
+      "step": 6567
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.760149863760218e-05,
+      "loss": 1.6966,
+      "step": 6568
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7599455040871935e-05,
+      "loss": 1.7611,
+      "step": 6569
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.759741144414169e-05,
+      "loss": 1.6854,
+      "step": 6570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7595367847411446e-05,
+      "loss": 1.6581,
+      "step": 6571
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.75933242506812e-05,
+      "loss": 1.7294,
+      "step": 6572
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7591280653950954e-05,
+      "loss": 1.7615,
+      "step": 6573
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.758923705722071e-05,
+      "loss": 1.7448,
+      "step": 6574
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7587193460490462e-05,
+      "loss": 1.6705,
+      "step": 6575
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7585149863760218e-05,
+      "loss": 1.8125,
+      "step": 6576
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7583106267029973e-05,
+      "loss": 1.6878,
+      "step": 6577
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.758106267029973e-05,
+      "loss": 1.6673,
+      "step": 6578
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7579019073569485e-05,
+      "loss": 1.7779,
+      "step": 6579
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7576975476839237e-05,
+      "loss": 1.8427,
+      "step": 6580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7574931880108992e-05,
+      "loss": 1.8189,
+      "step": 6581
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7572888283378745e-05,
+      "loss": 1.7019,
+      "step": 6582
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.75708446866485e-05,
+      "loss": 1.7096,
+      "step": 6583
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7568801089918256e-05,
+      "loss": 1.6637,
+      "step": 6584
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.756675749318801e-05,
+      "loss": 1.6236,
+      "step": 6585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7564713896457767e-05,
+      "loss": 1.7401,
+      "step": 6586
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7562670299727523e-05,
+      "loss": 1.7681,
+      "step": 6587
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7560626702997275e-05,
+      "loss": 1.6341,
+      "step": 6588
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.755858310626703e-05,
+      "loss": 1.7517,
+      "step": 6589
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7556539509536783e-05,
+      "loss": 1.7796,
+      "step": 6590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.755449591280654e-05,
+      "loss": 1.7921,
+      "step": 6591
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7552452316076298e-05,
+      "loss": 1.7142,
+      "step": 6592
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.755040871934605e-05,
+      "loss": 1.5622,
+      "step": 6593
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7548365122615806e-05,
+      "loss": 1.7306,
+      "step": 6594
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7546321525885558e-05,
+      "loss": 1.6297,
+      "step": 6595
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7544277929155313e-05,
+      "loss": 1.7465,
+      "step": 6596
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.754223433242507e-05,
+      "loss": 1.7221,
+      "step": 6597
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.754019073569482e-05,
+      "loss": 1.6878,
+      "step": 6598
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7538147138964577e-05,
+      "loss": 1.784,
+      "step": 6599
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7536103542234333e-05,
+      "loss": 1.6396,
+      "step": 6600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7534059945504088e-05,
+      "loss": 1.5566,
+      "step": 6601
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7532016348773844e-05,
+      "loss": 1.8324,
+      "step": 6602
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7529972752043596e-05,
+      "loss": 1.6054,
+      "step": 6603
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7527929155313352e-05,
+      "loss": 1.6449,
+      "step": 6604
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7525885558583107e-05,
+      "loss": 1.605,
+      "step": 6605
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.752384196185286e-05,
+      "loss": 1.6901,
+      "step": 6606
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.752179836512262e-05,
+      "loss": 1.7478,
+      "step": 6607
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.751975476839237e-05,
+      "loss": 1.81,
+      "step": 6608
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7517711171662126e-05,
+      "loss": 1.6728,
+      "step": 6609
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7515667574931882e-05,
+      "loss": 1.6386,
+      "step": 6610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7513623978201634e-05,
+      "loss": 1.7286,
+      "step": 6611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.751158038147139e-05,
+      "loss": 1.729,
+      "step": 6612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7509536784741142e-05,
+      "loss": 1.7566,
+      "step": 6613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.75074931880109e-05,
+      "loss": 1.7172,
+      "step": 6614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7505449591280657e-05,
+      "loss": 1.7201,
+      "step": 6615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.750340599455041e-05,
+      "loss": 1.896,
+      "step": 6616
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7501362397820165e-05,
+      "loss": 1.7454,
+      "step": 6617
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7499318801089917e-05,
+      "loss": 1.6334,
+      "step": 6618
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7497275204359673e-05,
+      "loss": 1.6101,
+      "step": 6619
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7495231607629428e-05,
+      "loss": 1.7327,
+      "step": 6620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.749318801089918e-05,
+      "loss": 1.7004,
+      "step": 6621
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.749114441416894e-05,
+      "loss": 1.6879,
+      "step": 6622
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7489100817438695e-05,
+      "loss": 1.7722,
+      "step": 6623
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7487057220708447e-05,
+      "loss": 1.7065,
+      "step": 6624
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7485013623978203e-05,
+      "loss": 1.7363,
+      "step": 6625
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7482970027247955e-05,
+      "loss": 1.8288,
+      "step": 6626
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.748092643051771e-05,
+      "loss": 1.7646,
+      "step": 6627
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7478882833787467e-05,
+      "loss": 1.6874,
+      "step": 6628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7476839237057222e-05,
+      "loss": 1.715,
+      "step": 6629
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7474795640326978e-05,
+      "loss": 1.5977,
+      "step": 6630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.747275204359673e-05,
+      "loss": 1.7592,
+      "step": 6631
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7470708446866486e-05,
+      "loss": 1.6315,
+      "step": 6632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746866485013624e-05,
+      "loss": 1.6625,
+      "step": 6633
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7466621253405994e-05,
+      "loss": 1.7334,
+      "step": 6634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746457765667575e-05,
+      "loss": 1.7041,
+      "step": 6635
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.74625340599455e-05,
+      "loss": 1.6582,
+      "step": 6636
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746049046321526e-05,
+      "loss": 1.7475,
+      "step": 6637
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7458446866485016e-05,
+      "loss": 1.9962,
+      "step": 6638
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745640326975477e-05,
+      "loss": 1.8498,
+      "step": 6639
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7454359673024524e-05,
+      "loss": 1.6908,
+      "step": 6640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745231607629428e-05,
+      "loss": 1.6754,
+      "step": 6641
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7450272479564032e-05,
+      "loss": 1.6615,
+      "step": 6642
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7448228882833787e-05,
+      "loss": 1.7295,
+      "step": 6643
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7446185286103543e-05,
+      "loss": 1.556,
+      "step": 6644
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.74441416893733e-05,
+      "loss": 1.5959,
+      "step": 6645
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7442098092643054e-05,
+      "loss": 1.7916,
+      "step": 6646
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7440054495912807e-05,
+      "loss": 1.6665,
+      "step": 6647
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7438010899182562e-05,
+      "loss": 1.7305,
+      "step": 6648
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7435967302452315e-05,
+      "loss": 1.7548,
+      "step": 6649
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.743392370572207e-05,
+      "loss": 1.8484,
+      "step": 6650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7431880108991826e-05,
+      "loss": 1.7534,
+      "step": 6651
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.742983651226158e-05,
+      "loss": 1.6924,
+      "step": 6652
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7427792915531337e-05,
+      "loss": 1.7423,
+      "step": 6653
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.742574931880109e-05,
+      "loss": 1.5835,
+      "step": 6654
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7423705722070845e-05,
+      "loss": 1.7556,
+      "step": 6655
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.74216621253406e-05,
+      "loss": 1.7401,
+      "step": 6656
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7419618528610353e-05,
+      "loss": 1.7771,
+      "step": 6657
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.741757493188011e-05,
+      "loss": 1.7802,
+      "step": 6658
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7415531335149864e-05,
+      "loss": 1.7492,
+      "step": 6659
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.741348773841962e-05,
+      "loss": 1.8209,
+      "step": 6660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7411444141689375e-05,
+      "loss": 1.5903,
+      "step": 6661
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7409400544959128e-05,
+      "loss": 1.732,
+      "step": 6662
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7407356948228883e-05,
+      "loss": 1.6632,
+      "step": 6663
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.740531335149864e-05,
+      "loss": 1.6963,
+      "step": 6664
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.740326975476839e-05,
+      "loss": 1.8492,
+      "step": 6665
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7401226158038147e-05,
+      "loss": 1.8053,
+      "step": 6666
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7399182561307902e-05,
+      "loss": 1.6922,
+      "step": 6667
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7397138964577658e-05,
+      "loss": 1.7625,
+      "step": 6668
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7395095367847414e-05,
+      "loss": 1.6634,
+      "step": 6669
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7393051771117166e-05,
+      "loss": 1.7101,
+      "step": 6670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.739100817438692e-05,
+      "loss": 1.6344,
+      "step": 6671
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7388964577656674e-05,
+      "loss": 1.774,
+      "step": 6672
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.738692098092643e-05,
+      "loss": 1.8099,
+      "step": 6673
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.738487738419619e-05,
+      "loss": 1.6152,
+      "step": 6674
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.738283378746594e-05,
+      "loss": 1.7358,
+      "step": 6675
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7380790190735696e-05,
+      "loss": 1.6394,
+      "step": 6676
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.737874659400545e-05,
+      "loss": 1.7372,
+      "step": 6677
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7376702997275204e-05,
+      "loss": 1.8229,
+      "step": 6678
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.737465940054496e-05,
+      "loss": 1.6156,
+      "step": 6679
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7372615803814712e-05,
+      "loss": 1.795,
+      "step": 6680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7370572207084468e-05,
+      "loss": 1.7798,
+      "step": 6681
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7368528610354227e-05,
+      "loss": 1.6815,
+      "step": 6682
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.736648501362398e-05,
+      "loss": 1.6864,
+      "step": 6683
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7364441416893735e-05,
+      "loss": 1.8685,
+      "step": 6684
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7362397820163487e-05,
+      "loss": 1.6772,
+      "step": 6685
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7360354223433242e-05,
+      "loss": 1.6289,
+      "step": 6686
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7360354223433242e-05,
+      "loss": 1.6616,
+      "step": 6687
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7358310626702998e-05,
+      "loss": 1.7521,
+      "step": 6688
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735626702997275e-05,
+      "loss": 1.7546,
+      "step": 6689
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735422343324251e-05,
+      "loss": 1.8035,
+      "step": 6690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735217983651226e-05,
+      "loss": 1.8322,
+      "step": 6691
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7350136239782017e-05,
+      "loss": 1.7207,
+      "step": 6692
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7348092643051773e-05,
+      "loss": 1.6817,
+      "step": 6693
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7346049046321525e-05,
+      "loss": 1.7937,
+      "step": 6694
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.734400544959128e-05,
+      "loss": 1.7424,
+      "step": 6695
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7341961852861033e-05,
+      "loss": 1.6679,
+      "step": 6696
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7339918256130792e-05,
+      "loss": 1.6522,
+      "step": 6697
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7337874659400548e-05,
+      "loss": 1.6573,
+      "step": 6698
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.73358310626703e-05,
+      "loss": 1.7896,
+      "step": 6699
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7333787465940055e-05,
+      "loss": 1.7721,
+      "step": 6700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.733174386920981e-05,
+      "loss": 1.6755,
+      "step": 6701
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7329700272479563e-05,
+      "loss": 1.7466,
+      "step": 6702
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.732765667574932e-05,
+      "loss": 1.6388,
+      "step": 6703
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.732561307901907e-05,
+      "loss": 1.7712,
+      "step": 6704
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.732356948228883e-05,
+      "loss": 1.7315,
+      "step": 6705
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7321525885558586e-05,
+      "loss": 1.7038,
+      "step": 6706
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7319482288828338e-05,
+      "loss": 1.7107,
+      "step": 6707
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7317438692098094e-05,
+      "loss": 1.7755,
+      "step": 6708
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7315395095367846e-05,
+      "loss": 1.7273,
+      "step": 6709
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.73133514986376e-05,
+      "loss": 1.6818,
+      "step": 6710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7311307901907357e-05,
+      "loss": 1.6938,
+      "step": 6711
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7309264305177113e-05,
+      "loss": 1.7099,
+      "step": 6712
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.730722070844687e-05,
+      "loss": 1.7255,
+      "step": 6713
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.730517711171662e-05,
+      "loss": 1.5981,
+      "step": 6714
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7303133514986376e-05,
+      "loss": 1.8033,
+      "step": 6715
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7301089918256132e-05,
+      "loss": 1.8265,
+      "step": 6716
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7299046321525884e-05,
+      "loss": 1.6989,
+      "step": 6717
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.729700272479564e-05,
+      "loss": 1.8111,
+      "step": 6718
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7294959128065396e-05,
+      "loss": 1.6916,
+      "step": 6719
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.729291553133515e-05,
+      "loss": 1.7975,
+      "step": 6720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7290871934604907e-05,
+      "loss": 1.789,
+      "step": 6721
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.728882833787466e-05,
+      "loss": 1.8584,
+      "step": 6722
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7286784741144415e-05,
+      "loss": 1.6509,
+      "step": 6723
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.728474114441417e-05,
+      "loss": 1.7034,
+      "step": 6724
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7282697547683923e-05,
+      "loss": 1.7967,
+      "step": 6725
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7280653950953678e-05,
+      "loss": 1.9633,
+      "step": 6726
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7278610354223434e-05,
+      "loss": 1.8316,
+      "step": 6727
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.727656675749319e-05,
+      "loss": 1.635,
+      "step": 6728
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7274523160762945e-05,
+      "loss": 1.7049,
+      "step": 6729
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7272479564032697e-05,
+      "loss": 1.7297,
+      "step": 6730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7270435967302453e-05,
+      "loss": 1.7831,
+      "step": 6731
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7268392370572205e-05,
+      "loss": 1.7708,
+      "step": 6732
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.726634877384196e-05,
+      "loss": 1.738,
+      "step": 6733
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7264305177111717e-05,
+      "loss": 1.7566,
+      "step": 6734
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7262261580381472e-05,
+      "loss": 1.6868,
+      "step": 6735
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7260217983651228e-05,
+      "loss": 1.7133,
+      "step": 6736
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7258174386920983e-05,
+      "loss": 1.8037,
+      "step": 6737
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7256130790190736e-05,
+      "loss": 1.6708,
+      "step": 6738
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.725408719346049e-05,
+      "loss": 1.7078,
+      "step": 6739
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7252043596730244e-05,
+      "loss": 1.8005,
+      "step": 6740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.725e-05,
+      "loss": 1.701,
+      "step": 6741
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7247956403269758e-05,
+      "loss": 1.732,
+      "step": 6742
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.724591280653951e-05,
+      "loss": 1.8847,
+      "step": 6743
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7243869209809266e-05,
+      "loss": 1.6839,
+      "step": 6744
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.724182561307902e-05,
+      "loss": 1.6674,
+      "step": 6745
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7239782016348774e-05,
+      "loss": 1.7768,
+      "step": 6746
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723773841961853e-05,
+      "loss": 1.6729,
+      "step": 6747
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7235694822888282e-05,
+      "loss": 1.7545,
+      "step": 6748
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7233651226158037e-05,
+      "loss": 1.5889,
+      "step": 6749
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7231607629427793e-05,
+      "loss": 1.7358,
+      "step": 6750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722956403269755e-05,
+      "loss": 1.7197,
+      "step": 6751
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7227520435967304e-05,
+      "loss": 1.555,
+      "step": 6752
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7225476839237057e-05,
+      "loss": 1.6876,
+      "step": 6753
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7223433242506812e-05,
+      "loss": 1.7128,
+      "step": 6754
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7221389645776568e-05,
+      "loss": 1.7666,
+      "step": 6755
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.721934604904632e-05,
+      "loss": 1.6696,
+      "step": 6756
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.721730245231608e-05,
+      "loss": 1.6917,
+      "step": 6757
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.721525885558583e-05,
+      "loss": 1.8634,
+      "step": 6758
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7213215258855587e-05,
+      "loss": 1.6939,
+      "step": 6759
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7211171662125343e-05,
+      "loss": 1.7407,
+      "step": 6760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7209128065395095e-05,
+      "loss": 1.5728,
+      "step": 6761
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.720708446866485e-05,
+      "loss": 1.7055,
+      "step": 6762
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7205040871934603e-05,
+      "loss": 1.627,
+      "step": 6763
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.720299727520436e-05,
+      "loss": 1.7877,
+      "step": 6764
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7200953678474117e-05,
+      "loss": 1.697,
+      "step": 6765
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719891008174387e-05,
+      "loss": 1.6759,
+      "step": 6766
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7196866485013625e-05,
+      "loss": 1.797,
+      "step": 6767
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194822888283378e-05,
+      "loss": 1.7424,
+      "step": 6768
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7192779291553133e-05,
+      "loss": 1.8318,
+      "step": 6769
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719073569482289e-05,
+      "loss": 1.7944,
+      "step": 6770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718869209809264e-05,
+      "loss": 1.7102,
+      "step": 6771
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71866485013624e-05,
+      "loss": 1.779,
+      "step": 6772
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7184604904632156e-05,
+      "loss": 1.7828,
+      "step": 6773
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7182561307901908e-05,
+      "loss": 1.6365,
+      "step": 6774
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180517711171664e-05,
+      "loss": 1.6373,
+      "step": 6775
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7178474114441416e-05,
+      "loss": 1.8144,
+      "step": 6776
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717643051771117e-05,
+      "loss": 1.6947,
+      "step": 6777
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7174386920980927e-05,
+      "loss": 1.7128,
+      "step": 6778
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717234332425068e-05,
+      "loss": 1.8011,
+      "step": 6779
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717029972752044e-05,
+      "loss": 1.6719,
+      "step": 6780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716825613079019e-05,
+      "loss": 1.7945,
+      "step": 6781
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7166212534059946e-05,
+      "loss": 1.7909,
+      "step": 6782
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7164168937329702e-05,
+      "loss": 1.8658,
+      "step": 6783
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7162125340599454e-05,
+      "loss": 1.6139,
+      "step": 6784
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716008174386921e-05,
+      "loss": 1.9506,
+      "step": 6785
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7158038147138962e-05,
+      "loss": 1.6154,
+      "step": 6786
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715599455040872e-05,
+      "loss": 1.832,
+      "step": 6787
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7153950953678477e-05,
+      "loss": 1.7546,
+      "step": 6788
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715190735694823e-05,
+      "loss": 1.7352,
+      "step": 6789
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7149863760217985e-05,
+      "loss": 1.7654,
+      "step": 6790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.714782016348774e-05,
+      "loss": 1.8319,
+      "step": 6791
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7145776566757492e-05,
+      "loss": 1.5938,
+      "step": 6792
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7143732970027248e-05,
+      "loss": 1.7447,
+      "step": 6793
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7141689373297004e-05,
+      "loss": 1.699,
+      "step": 6794
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.713964577656676e-05,
+      "loss": 1.651,
+      "step": 6795
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7137602179836515e-05,
+      "loss": 1.7201,
+      "step": 6796
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7135558583106267e-05,
+      "loss": 1.8098,
+      "step": 6797
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7133514986376023e-05,
+      "loss": 1.7637,
+      "step": 6798
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7131471389645775e-05,
+      "loss": 1.8766,
+      "step": 6799
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.712942779291553e-05,
+      "loss": 1.7682,
+      "step": 6800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7127384196185286e-05,
+      "loss": 1.6096,
+      "step": 6801
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7125340599455042e-05,
+      "loss": 1.7337,
+      "step": 6802
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7123297002724798e-05,
+      "loss": 1.7458,
+      "step": 6803
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.712125340599455e-05,
+      "loss": 1.6873,
+      "step": 6804
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7119209809264305e-05,
+      "loss": 1.7282,
+      "step": 6805
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711716621253406e-05,
+      "loss": 1.7997,
+      "step": 6806
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7115122615803813e-05,
+      "loss": 1.6557,
+      "step": 6807
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711307901907357e-05,
+      "loss": 1.6378,
+      "step": 6808
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7111035422343328e-05,
+      "loss": 1.6393,
+      "step": 6809
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710899182561308e-05,
+      "loss": 1.6773,
+      "step": 6810
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7106948228882836e-05,
+      "loss": 1.7556,
+      "step": 6811
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7104904632152588e-05,
+      "loss": 1.8393,
+      "step": 6812
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7102861035422344e-05,
+      "loss": 1.7718,
+      "step": 6813
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.71008174386921e-05,
+      "loss": 1.8612,
+      "step": 6814
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.709877384196185e-05,
+      "loss": 1.6744,
+      "step": 6815
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7096730245231607e-05,
+      "loss": 1.6423,
+      "step": 6816
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7094686648501363e-05,
+      "loss": 1.7621,
+      "step": 6817
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.709264305177112e-05,
+      "loss": 1.6418,
+      "step": 6818
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7090599455040874e-05,
+      "loss": 1.7125,
+      "step": 6819
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7088555858310626e-05,
+      "loss": 1.6931,
+      "step": 6820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7086512261580382e-05,
+      "loss": 1.8996,
+      "step": 6821
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7084468664850134e-05,
+      "loss": 1.722,
+      "step": 6822
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.708242506811989e-05,
+      "loss": 1.7533,
+      "step": 6823
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.708038147138965e-05,
+      "loss": 1.7316,
+      "step": 6824
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.70783378746594e-05,
+      "loss": 1.6318,
+      "step": 6825
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7076294277929157e-05,
+      "loss": 1.7433,
+      "step": 6826
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7074250681198912e-05,
+      "loss": 1.5921,
+      "step": 6827
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7072207084468665e-05,
+      "loss": 1.791,
+      "step": 6828
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.707016348773842e-05,
+      "loss": 1.6536,
+      "step": 6829
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7068119891008173e-05,
+      "loss": 1.6461,
+      "step": 6830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7066076294277928e-05,
+      "loss": 1.6433,
+      "step": 6831
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7064032697547687e-05,
+      "loss": 1.5607,
+      "step": 6832
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.706198910081744e-05,
+      "loss": 1.5855,
+      "step": 6833
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7059945504087195e-05,
+      "loss": 1.7307,
+      "step": 6834
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7057901907356947e-05,
+      "loss": 1.7206,
+      "step": 6835
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7055858310626703e-05,
+      "loss": 1.7232,
+      "step": 6836
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.705381471389646e-05,
+      "loss": 1.7368,
+      "step": 6837
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.705177111716621e-05,
+      "loss": 1.7406,
+      "step": 6838
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.704972752043597e-05,
+      "loss": 1.6245,
+      "step": 6839
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7047683923705722e-05,
+      "loss": 1.6965,
+      "step": 6840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7045640326975478e-05,
+      "loss": 1.6004,
+      "step": 6841
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7043596730245233e-05,
+      "loss": 1.8684,
+      "step": 6842
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7041553133514986e-05,
+      "loss": 1.7477,
+      "step": 6843
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.703950953678474e-05,
+      "loss": 1.7998,
+      "step": 6844
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7037465940054497e-05,
+      "loss": 1.6796,
+      "step": 6845
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.703542234332425e-05,
+      "loss": 1.7595,
+      "step": 6846
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7033378746594008e-05,
+      "loss": 1.7881,
+      "step": 6847
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.703133514986376e-05,
+      "loss": 1.7719,
+      "step": 6848
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7029291553133516e-05,
+      "loss": 1.6344,
+      "step": 6849
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.702724795640327e-05,
+      "loss": 1.652,
+      "step": 6850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7025204359673024e-05,
+      "loss": 1.7454,
+      "step": 6851
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.702316076294278e-05,
+      "loss": 1.5984,
+      "step": 6852
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7021117166212532e-05,
+      "loss": 1.6618,
+      "step": 6853
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.701907356948229e-05,
+      "loss": 1.7569,
+      "step": 6854
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7017029972752046e-05,
+      "loss": 1.7415,
+      "step": 6855
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.70149863760218e-05,
+      "loss": 1.6868,
+      "step": 6856
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7012942779291554e-05,
+      "loss": 1.7292,
+      "step": 6857
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7010899182561307e-05,
+      "loss": 1.6914,
+      "step": 6858
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7008855585831062e-05,
+      "loss": 1.7373,
+      "step": 6859
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7006811989100818e-05,
+      "loss": 1.6448,
+      "step": 6860
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.700476839237057e-05,
+      "loss": 1.7095,
+      "step": 6861
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.700272479564033e-05,
+      "loss": 1.7837,
+      "step": 6862
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.700068119891008e-05,
+      "loss": 1.7166,
+      "step": 6863
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6998637602179837e-05,
+      "loss": 1.6836,
+      "step": 6864
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6996594005449593e-05,
+      "loss": 1.7919,
+      "step": 6865
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6994550408719345e-05,
+      "loss": 1.5759,
+      "step": 6866
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.69925068119891e-05,
+      "loss": 1.7418,
+      "step": 6867
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6990463215258856e-05,
+      "loss": 1.635,
+      "step": 6868
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6988419618528612e-05,
+      "loss": 1.6614,
+      "step": 6869
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6986376021798367e-05,
+      "loss": 1.5939,
+      "step": 6870
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.698433242506812e-05,
+      "loss": 1.6558,
+      "step": 6871
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6982288828337875e-05,
+      "loss": 1.7605,
+      "step": 6872
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.698024523160763e-05,
+      "loss": 1.648,
+      "step": 6873
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6978201634877383e-05,
+      "loss": 1.899,
+      "step": 6874
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.697615803814714e-05,
+      "loss": 1.6355,
+      "step": 6875
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.697411444141689e-05,
+      "loss": 1.7234,
+      "step": 6876
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.697207084468665e-05,
+      "loss": 1.8588,
+      "step": 6877
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6970027247956406e-05,
+      "loss": 1.6673,
+      "step": 6878
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6967983651226158e-05,
+      "loss": 1.6967,
+      "step": 6879
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6965940054495914e-05,
+      "loss": 1.6276,
+      "step": 6880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6963896457765666e-05,
+      "loss": 1.5744,
+      "step": 6881
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.696185286103542e-05,
+      "loss": 1.6855,
+      "step": 6882
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6959809264305177e-05,
+      "loss": 1.7553,
+      "step": 6883
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6957765667574933e-05,
+      "loss": 1.674,
+      "step": 6884
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6955722070844688e-05,
+      "loss": 1.4975,
+      "step": 6885
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6953678474114444e-05,
+      "loss": 1.7529,
+      "step": 6886
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6951634877384196e-05,
+      "loss": 1.8415,
+      "step": 6887
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6949591280653952e-05,
+      "loss": 1.6564,
+      "step": 6888
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6947547683923704e-05,
+      "loss": 1.5975,
+      "step": 6889
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.694550408719346e-05,
+      "loss": 1.6642,
+      "step": 6890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.694346049046322e-05,
+      "loss": 1.677,
+      "step": 6891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.694141689373297e-05,
+      "loss": 1.6751,
+      "step": 6892
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6939373297002727e-05,
+      "loss": 1.6827,
+      "step": 6893
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.693732970027248e-05,
+      "loss": 1.8162,
+      "step": 6894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6935286103542234e-05,
+      "loss": 1.9078,
+      "step": 6895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.693324250681199e-05,
+      "loss": 1.6645,
+      "step": 6896
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6931198910081742e-05,
+      "loss": 1.847,
+      "step": 6897
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6929155313351498e-05,
+      "loss": 1.6774,
+      "step": 6898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6927111716621254e-05,
+      "loss": 1.7386,
+      "step": 6899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.692506811989101e-05,
+      "loss": 1.6885,
+      "step": 6900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6923024523160765e-05,
+      "loss": 1.5986,
+      "step": 6901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6920980926430517e-05,
+      "loss": 1.7342,
+      "step": 6902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6918937329700273e-05,
+      "loss": 1.8108,
+      "step": 6903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.691689373297003e-05,
+      "loss": 1.7783,
+      "step": 6904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.691485013623978e-05,
+      "loss": 1.705,
+      "step": 6905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.691280653950954e-05,
+      "loss": 1.7526,
+      "step": 6906
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6910762942779292e-05,
+      "loss": 1.8059,
+      "step": 6907
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6908719346049048e-05,
+      "loss": 1.5664,
+      "step": 6908
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6906675749318803e-05,
+      "loss": 1.5979,
+      "step": 6909
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6904632152588555e-05,
+      "loss": 1.619,
+      "step": 6910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.690258855585831e-05,
+      "loss": 1.7204,
+      "step": 6911
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6900544959128063e-05,
+      "loss": 1.6956,
+      "step": 6912
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.689850136239782e-05,
+      "loss": 1.879,
+      "step": 6913
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6896457765667578e-05,
+      "loss": 1.7718,
+      "step": 6914
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.689441416893733e-05,
+      "loss": 1.7153,
+      "step": 6915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6892370572207086e-05,
+      "loss": 1.6932,
+      "step": 6916
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6890326975476838e-05,
+      "loss": 1.7332,
+      "step": 6917
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6888283378746594e-05,
+      "loss": 1.7393,
+      "step": 6918
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.688623978201635e-05,
+      "loss": 1.8977,
+      "step": 6919
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.68841961852861e-05,
+      "loss": 1.7457,
+      "step": 6920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.688215258855586e-05,
+      "loss": 1.656,
+      "step": 6921
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6880108991825616e-05,
+      "loss": 1.736,
+      "step": 6922
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.687806539509537e-05,
+      "loss": 1.7679,
+      "step": 6923
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6876021798365124e-05,
+      "loss": 1.7297,
+      "step": 6924
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6873978201634876e-05,
+      "loss": 1.6969,
+      "step": 6925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6871934604904632e-05,
+      "loss": 1.804,
+      "step": 6926
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6869891008174388e-05,
+      "loss": 1.6849,
+      "step": 6927
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.686784741144414e-05,
+      "loss": 1.7153,
+      "step": 6928
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.68658038147139e-05,
+      "loss": 1.8518,
+      "step": 6929
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.686376021798365e-05,
+      "loss": 1.8129,
+      "step": 6930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6861716621253407e-05,
+      "loss": 1.6968,
+      "step": 6931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6859673024523162e-05,
+      "loss": 1.7157,
+      "step": 6932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6857629427792915e-05,
+      "loss": 1.7113,
+      "step": 6933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.685558583106267e-05,
+      "loss": 1.7981,
+      "step": 6934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6853542234332423e-05,
+      "loss": 1.7534,
+      "step": 6935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.685149863760218e-05,
+      "loss": 1.7535,
+      "step": 6936
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6849455040871937e-05,
+      "loss": 1.8136,
+      "step": 6937
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.684741144414169e-05,
+      "loss": 1.6523,
+      "step": 6938
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6845367847411445e-05,
+      "loss": 1.9757,
+      "step": 6939
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.68433242506812e-05,
+      "loss": 1.7738,
+      "step": 6940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6841280653950953e-05,
+      "loss": 1.7904,
+      "step": 6941
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.683923705722071e-05,
+      "loss": 1.7409,
+      "step": 6942
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.683719346049046e-05,
+      "loss": 1.7617,
+      "step": 6943
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.683514986376022e-05,
+      "loss": 1.7255,
+      "step": 6944
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6833106267029975e-05,
+      "loss": 1.6655,
+      "step": 6945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6831062670299728e-05,
+      "loss": 1.7373,
+      "step": 6946
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6829019073569483e-05,
+      "loss": 1.6633,
+      "step": 6947
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6826975476839236e-05,
+      "loss": 1.5901,
+      "step": 6948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.682493188010899e-05,
+      "loss": 1.7238,
+      "step": 6949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6822888283378747e-05,
+      "loss": 1.7789,
+      "step": 6950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6820844686648502e-05,
+      "loss": 1.7978,
+      "step": 6951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6818801089918258e-05,
+      "loss": 1.6273,
+      "step": 6952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.681675749318801e-05,
+      "loss": 1.6231,
+      "step": 6953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6814713896457766e-05,
+      "loss": 1.7723,
+      "step": 6954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.681267029972752e-05,
+      "loss": 1.67,
+      "step": 6955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6810626702997274e-05,
+      "loss": 1.8838,
+      "step": 6956
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.680858310626703e-05,
+      "loss": 1.6914,
+      "step": 6957
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6806539509536785e-05,
+      "loss": 1.8238,
+      "step": 6958
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.680449591280654e-05,
+      "loss": 1.7185,
+      "step": 6959
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6802452316076296e-05,
+      "loss": 1.6146,
+      "step": 6960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.680040871934605e-05,
+      "loss": 1.7053,
+      "step": 6961
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6798365122615804e-05,
+      "loss": 1.8042,
+      "step": 6962
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.679632152588556e-05,
+      "loss": 1.6715,
+      "step": 6963
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6794277929155312e-05,
+      "loss": 1.7317,
+      "step": 6964
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6792234332425068e-05,
+      "loss": 1.7087,
+      "step": 6965
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6790190735694823e-05,
+      "loss": 1.7882,
+      "step": 6966
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.678814713896458e-05,
+      "loss": 1.6467,
+      "step": 6967
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6786103542234335e-05,
+      "loss": 1.786,
+      "step": 6968
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6784059945504087e-05,
+      "loss": 1.677,
+      "step": 6969
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6782016348773843e-05,
+      "loss": 1.7125,
+      "step": 6970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6779972752043595e-05,
+      "loss": 1.7537,
+      "step": 6971
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.677792915531335e-05,
+      "loss": 1.6339,
+      "step": 6972
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.677588555858311e-05,
+      "loss": 1.7707,
+      "step": 6973
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.677384196185286e-05,
+      "loss": 1.6648,
+      "step": 6974
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6771798365122617e-05,
+      "loss": 1.6277,
+      "step": 6975
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6769754768392373e-05,
+      "loss": 1.7545,
+      "step": 6976
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6767711171662125e-05,
+      "loss": 1.7853,
+      "step": 6977
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.676566757493188e-05,
+      "loss": 1.6463,
+      "step": 6978
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6763623978201633e-05,
+      "loss": 1.7206,
+      "step": 6979
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.676158038147139e-05,
+      "loss": 1.6296,
+      "step": 6980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6759536784741148e-05,
+      "loss": 1.7563,
+      "step": 6981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.67574931880109e-05,
+      "loss": 1.7195,
+      "step": 6982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6755449591280656e-05,
+      "loss": 1.6723,
+      "step": 6983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6753405994550408e-05,
+      "loss": 1.6253,
+      "step": 6984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6751362397820163e-05,
+      "loss": 1.7662,
+      "step": 6985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.674931880108992e-05,
+      "loss": 1.6529,
+      "step": 6986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.674727520435967e-05,
+      "loss": 1.8238,
+      "step": 6987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.674523160762943e-05,
+      "loss": 1.832,
+      "step": 6988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6743188010899183e-05,
+      "loss": 1.6965,
+      "step": 6989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6741144414168938e-05,
+      "loss": 1.6447,
+      "step": 6990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6739100817438694e-05,
+      "loss": 1.7616,
+      "step": 6991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6737057220708446e-05,
+      "loss": 1.6968,
+      "step": 6992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6735013623978202e-05,
+      "loss": 1.7398,
+      "step": 6993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6732970027247957e-05,
+      "loss": 1.7526,
+      "step": 6994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.673092643051771e-05,
+      "loss": 1.6558,
+      "step": 6995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.672888283378747e-05,
+      "loss": 1.6948,
+      "step": 6996
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.672683923705722e-05,
+      "loss": 1.7585,
+      "step": 6997
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6724795640326977e-05,
+      "loss": 1.7579,
+      "step": 6998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6722752043596732e-05,
+      "loss": 1.6738,
+      "step": 6999
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6720708446866484e-05,
+      "loss": 1.6456,
+      "step": 7000
+    },
+    {
+      "epoch": 1.38,
+      "eval_loss": 1.5502526760101318,
+      "eval_runtime": 721.5901,
+      "eval_samples_per_second": 5.751,
+      "eval_steps_per_second": 1.438,
+      "step": 7000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.671866485013624e-05,
+      "loss": 1.5936,
+      "step": 7001
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6716621253405992e-05,
+      "loss": 1.6406,
+      "step": 7002
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.671457765667575e-05,
+      "loss": 1.825,
+      "step": 7003
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6712534059945507e-05,
+      "loss": 1.6502,
+      "step": 7004
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.671049046321526e-05,
+      "loss": 1.8908,
+      "step": 7005
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6708446866485015e-05,
+      "loss": 1.6052,
+      "step": 7006
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6706403269754767e-05,
+      "loss": 1.8205,
+      "step": 7007
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6704359673024523e-05,
+      "loss": 1.7334,
+      "step": 7008
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.670231607629428e-05,
+      "loss": 1.5751,
+      "step": 7009
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.670027247956403e-05,
+      "loss": 1.7235,
+      "step": 7010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.669822888283379e-05,
+      "loss": 1.5999,
+      "step": 7011
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6696185286103545e-05,
+      "loss": 1.8465,
+      "step": 7012
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6694141689373297e-05,
+      "loss": 1.6764,
+      "step": 7013
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6692098092643053e-05,
+      "loss": 1.6197,
+      "step": 7014
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6690054495912805e-05,
+      "loss": 1.8343,
+      "step": 7015
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.668801089918256e-05,
+      "loss": 1.7384,
+      "step": 7016
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6685967302452317e-05,
+      "loss": 1.6533,
+      "step": 7017
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6683923705722072e-05,
+      "loss": 1.7445,
+      "step": 7018
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6681880108991828e-05,
+      "loss": 1.7848,
+      "step": 7019
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.667983651226158e-05,
+      "loss": 1.8169,
+      "step": 7020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6677792915531336e-05,
+      "loss": 1.721,
+      "step": 7021
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.667574931880109e-05,
+      "loss": 1.8014,
+      "step": 7022
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6673705722070844e-05,
+      "loss": 1.798,
+      "step": 7023
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.66716621253406e-05,
+      "loss": 1.6716,
+      "step": 7024
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.666961852861035e-05,
+      "loss": 1.6803,
+      "step": 7025
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.666757493188011e-05,
+      "loss": 1.6946,
+      "step": 7026
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6665531335149866e-05,
+      "loss": 1.7341,
+      "step": 7027
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.666348773841962e-05,
+      "loss": 1.7275,
+      "step": 7028
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6661444141689374e-05,
+      "loss": 1.7361,
+      "step": 7029
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.665940054495913e-05,
+      "loss": 1.595,
+      "step": 7030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6657356948228882e-05,
+      "loss": 1.7377,
+      "step": 7031
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6655313351498638e-05,
+      "loss": 1.6985,
+      "step": 7032
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6653269754768393e-05,
+      "loss": 1.7934,
+      "step": 7033
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.665122615803815e-05,
+      "loss": 1.6473,
+      "step": 7034
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6649182561307904e-05,
+      "loss": 1.7296,
+      "step": 7035
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6647138964577657e-05,
+      "loss": 1.7099,
+      "step": 7036
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6645095367847412e-05,
+      "loss": 1.706,
+      "step": 7037
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6643051771117165e-05,
+      "loss": 1.6238,
+      "step": 7038
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.664100817438692e-05,
+      "loss": 1.6604,
+      "step": 7039
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6638964577656676e-05,
+      "loss": 1.7275,
+      "step": 7040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.663692098092643e-05,
+      "loss": 1.673,
+      "step": 7041
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6634877384196187e-05,
+      "loss": 1.7057,
+      "step": 7042
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.663283378746594e-05,
+      "loss": 1.6452,
+      "step": 7043
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6630790190735695e-05,
+      "loss": 1.6602,
+      "step": 7044
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.662874659400545e-05,
+      "loss": 1.6415,
+      "step": 7045
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6626702997275203e-05,
+      "loss": 1.651,
+      "step": 7046
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.662465940054496e-05,
+      "loss": 1.6475,
+      "step": 7047
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6622615803814714e-05,
+      "loss": 1.8498,
+      "step": 7048
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.662057220708447e-05,
+      "loss": 1.7815,
+      "step": 7049
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6618528610354225e-05,
+      "loss": 1.7115,
+      "step": 7050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6616485013623978e-05,
+      "loss": 1.7196,
+      "step": 7051
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6614441416893733e-05,
+      "loss": 1.7401,
+      "step": 7052
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.661239782016349e-05,
+      "loss": 1.7119,
+      "step": 7053
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.661035422343324e-05,
+      "loss": 1.5881,
+      "step": 7054
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6608310626702997e-05,
+      "loss": 1.7616,
+      "step": 7055
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6606267029972752e-05,
+      "loss": 1.6069,
+      "step": 7056
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6604223433242508e-05,
+      "loss": 1.6989,
+      "step": 7057
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6602179836512264e-05,
+      "loss": 1.7323,
+      "step": 7058
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6600136239782016e-05,
+      "loss": 1.843,
+      "step": 7059
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.659809264305177e-05,
+      "loss": 1.728,
+      "step": 7060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6596049046321524e-05,
+      "loss": 1.8286,
+      "step": 7061
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.659400544959128e-05,
+      "loss": 1.6789,
+      "step": 7062
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.659196185286104e-05,
+      "loss": 1.762,
+      "step": 7063
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.658991825613079e-05,
+      "loss": 1.7047,
+      "step": 7064
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6587874659400546e-05,
+      "loss": 1.6662,
+      "step": 7065
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.65858310626703e-05,
+      "loss": 1.7346,
+      "step": 7066
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6583787465940054e-05,
+      "loss": 1.7723,
+      "step": 7067
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.658174386920981e-05,
+      "loss": 1.9675,
+      "step": 7068
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6579700272479562e-05,
+      "loss": 1.7579,
+      "step": 7069
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.657765667574932e-05,
+      "loss": 1.6586,
+      "step": 7070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6575613079019077e-05,
+      "loss": 1.7803,
+      "step": 7071
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.657356948228883e-05,
+      "loss": 1.6523,
+      "step": 7072
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6571525885558585e-05,
+      "loss": 1.7314,
+      "step": 7073
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6569482288828337e-05,
+      "loss": 1.5049,
+      "step": 7074
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6567438692098092e-05,
+      "loss": 1.6741,
+      "step": 7075
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6565395095367848e-05,
+      "loss": 1.8241,
+      "step": 7076
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.65633514986376e-05,
+      "loss": 1.8556,
+      "step": 7077
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.656130790190736e-05,
+      "loss": 1.8583,
+      "step": 7078
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.655926430517711e-05,
+      "loss": 1.6949,
+      "step": 7079
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6557220708446867e-05,
+      "loss": 1.6812,
+      "step": 7080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6555177111716623e-05,
+      "loss": 1.5575,
+      "step": 7081
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6553133514986375e-05,
+      "loss": 1.8301,
+      "step": 7082
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.655108991825613e-05,
+      "loss": 1.7519,
+      "step": 7083
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6549046321525883e-05,
+      "loss": 1.6462,
+      "step": 7084
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6547002724795642e-05,
+      "loss": 1.6513,
+      "step": 7085
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6544959128065398e-05,
+      "loss": 1.7287,
+      "step": 7086
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.654291553133515e-05,
+      "loss": 1.7074,
+      "step": 7087
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6540871934604906e-05,
+      "loss": 1.6688,
+      "step": 7088
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.653882833787466e-05,
+      "loss": 1.6972,
+      "step": 7089
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6536784741144413e-05,
+      "loss": 1.7459,
+      "step": 7090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.653474114441417e-05,
+      "loss": 1.5934,
+      "step": 7091
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.653269754768392e-05,
+      "loss": 1.6406,
+      "step": 7092
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.653065395095368e-05,
+      "loss": 1.8154,
+      "step": 7093
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6528610354223436e-05,
+      "loss": 1.6644,
+      "step": 7094
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6526566757493188e-05,
+      "loss": 1.5478,
+      "step": 7095
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6524523160762944e-05,
+      "loss": 1.7447,
+      "step": 7096
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6522479564032696e-05,
+      "loss": 1.6582,
+      "step": 7097
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6520435967302452e-05,
+      "loss": 1.7066,
+      "step": 7098
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6518392370572207e-05,
+      "loss": 1.6606,
+      "step": 7099
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6516348773841963e-05,
+      "loss": 1.6995,
+      "step": 7100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.651430517711172e-05,
+      "loss": 1.6537,
+      "step": 7101
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.651226158038147e-05,
+      "loss": 1.6703,
+      "step": 7102
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6510217983651226e-05,
+      "loss": 1.8124,
+      "step": 7103
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6508174386920982e-05,
+      "loss": 1.6675,
+      "step": 7104
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6506130790190734e-05,
+      "loss": 1.8326,
+      "step": 7105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.650408719346049e-05,
+      "loss": 1.7142,
+      "step": 7106
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6502043596730246e-05,
+      "loss": 1.7296,
+      "step": 7107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.65e-05,
+      "loss": 1.6948,
+      "step": 7108
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6497956403269757e-05,
+      "loss": 1.6993,
+      "step": 7109
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.649591280653951e-05,
+      "loss": 1.5845,
+      "step": 7110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6493869209809265e-05,
+      "loss": 1.7535,
+      "step": 7111
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.649182561307902e-05,
+      "loss": 1.786,
+      "step": 7112
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6489782016348773e-05,
+      "loss": 1.8466,
+      "step": 7113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6487738419618528e-05,
+      "loss": 1.7797,
+      "step": 7114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6485694822888284e-05,
+      "loss": 1.7824,
+      "step": 7115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.648365122615804e-05,
+      "loss": 1.6929,
+      "step": 7116
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6481607629427795e-05,
+      "loss": 1.7198,
+      "step": 7117
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6479564032697547e-05,
+      "loss": 1.7822,
+      "step": 7118
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6477520435967303e-05,
+      "loss": 1.7481,
+      "step": 7119
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6475476839237055e-05,
+      "loss": 1.7082,
+      "step": 7120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.647343324250681e-05,
+      "loss": 1.7651,
+      "step": 7121
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6471389645776567e-05,
+      "loss": 1.7834,
+      "step": 7122
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6469346049046322e-05,
+      "loss": 1.7168,
+      "step": 7123
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6467302452316078e-05,
+      "loss": 1.6917,
+      "step": 7124
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6465258855585833e-05,
+      "loss": 1.7037,
+      "step": 7125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6463215258855586e-05,
+      "loss": 1.7199,
+      "step": 7126
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.646117166212534e-05,
+      "loss": 1.7738,
+      "step": 7127
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6459128065395094e-05,
+      "loss": 1.7119,
+      "step": 7128
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.645708446866485e-05,
+      "loss": 1.651,
+      "step": 7129
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6455040871934608e-05,
+      "loss": 1.5618,
+      "step": 7130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.645299727520436e-05,
+      "loss": 1.7944,
+      "step": 7131
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6450953678474116e-05,
+      "loss": 1.7797,
+      "step": 7132
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.644891008174387e-05,
+      "loss": 1.6868,
+      "step": 7133
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6446866485013624e-05,
+      "loss": 1.7334,
+      "step": 7134
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.644482288828338e-05,
+      "loss": 1.708,
+      "step": 7135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6442779291553132e-05,
+      "loss": 1.6486,
+      "step": 7136
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6440735694822888e-05,
+      "loss": 1.6857,
+      "step": 7137
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6438692098092643e-05,
+      "loss": 1.6683,
+      "step": 7138
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.64366485013624e-05,
+      "loss": 1.7174,
+      "step": 7139
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6434604904632154e-05,
+      "loss": 1.6281,
+      "step": 7140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6432561307901907e-05,
+      "loss": 1.8625,
+      "step": 7141
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6430517711171662e-05,
+      "loss": 1.7614,
+      "step": 7142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6428474114441418e-05,
+      "loss": 1.6477,
+      "step": 7143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.642643051771117e-05,
+      "loss": 1.6785,
+      "step": 7144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.642438692098093e-05,
+      "loss": 1.7241,
+      "step": 7145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.642234332425068e-05,
+      "loss": 1.7844,
+      "step": 7146
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6420299727520437e-05,
+      "loss": 1.7129,
+      "step": 7147
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6418256130790193e-05,
+      "loss": 1.67,
+      "step": 7148
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6416212534059945e-05,
+      "loss": 1.696,
+      "step": 7149
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.64141689373297e-05,
+      "loss": 1.7266,
+      "step": 7150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6412125340599453e-05,
+      "loss": 1.8173,
+      "step": 7151
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.641008174386921e-05,
+      "loss": 1.7292,
+      "step": 7152
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6408038147138967e-05,
+      "loss": 1.6641,
+      "step": 7153
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.640599455040872e-05,
+      "loss": 1.7771,
+      "step": 7154
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6403950953678475e-05,
+      "loss": 1.7217,
+      "step": 7155
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6401907356948228e-05,
+      "loss": 1.7119,
+      "step": 7156
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6399863760217983e-05,
+      "loss": 1.8508,
+      "step": 7157
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.639782016348774e-05,
+      "loss": 1.7209,
+      "step": 7158
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.639577656675749e-05,
+      "loss": 1.8619,
+      "step": 7159
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.639373297002725e-05,
+      "loss": 1.6812,
+      "step": 7160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6391689373297006e-05,
+      "loss": 1.7462,
+      "step": 7161
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6389645776566758e-05,
+      "loss": 1.682,
+      "step": 7162
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6387602179836514e-05,
+      "loss": 1.657,
+      "step": 7163
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6385558583106266e-05,
+      "loss": 1.6783,
+      "step": 7164
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.638351498637602e-05,
+      "loss": 1.6113,
+      "step": 7165
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6381471389645777e-05,
+      "loss": 1.7085,
+      "step": 7166
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6379427792915533e-05,
+      "loss": 1.6658,
+      "step": 7167
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.637738419618529e-05,
+      "loss": 1.7182,
+      "step": 7168
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.637534059945504e-05,
+      "loss": 1.7324,
+      "step": 7169
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6373297002724796e-05,
+      "loss": 1.7078,
+      "step": 7170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6371253405994552e-05,
+      "loss": 1.8928,
+      "step": 7171
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6369209809264304e-05,
+      "loss": 1.7751,
+      "step": 7172
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.636716621253406e-05,
+      "loss": 1.6373,
+      "step": 7173
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6365122615803812e-05,
+      "loss": 1.7541,
+      "step": 7174
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.636307901907357e-05,
+      "loss": 1.7374,
+      "step": 7175
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6361035422343327e-05,
+      "loss": 1.7824,
+      "step": 7176
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.635899182561308e-05,
+      "loss": 1.6911,
+      "step": 7177
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6356948228882835e-05,
+      "loss": 1.6574,
+      "step": 7178
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.635490463215259e-05,
+      "loss": 1.7787,
+      "step": 7179
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6352861035422342e-05,
+      "loss": 1.6838,
+      "step": 7180
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6350817438692098e-05,
+      "loss": 1.658,
+      "step": 7181
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6348773841961854e-05,
+      "loss": 1.6931,
+      "step": 7182
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.634673024523161e-05,
+      "loss": 1.6809,
+      "step": 7183
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6344686648501365e-05,
+      "loss": 1.6661,
+      "step": 7184
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6342643051771117e-05,
+      "loss": 1.8061,
+      "step": 7185
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6340599455040873e-05,
+      "loss": 1.6798,
+      "step": 7186
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6338555858310625e-05,
+      "loss": 1.7764,
+      "step": 7187
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.633651226158038e-05,
+      "loss": 1.7568,
+      "step": 7188
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6334468664850136e-05,
+      "loss": 1.6706,
+      "step": 7189
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6332425068119892e-05,
+      "loss": 1.7037,
+      "step": 7190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6330381471389648e-05,
+      "loss": 1.6791,
+      "step": 7191
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.63283378746594e-05,
+      "loss": 1.6936,
+      "step": 7192
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6326294277929156e-05,
+      "loss": 1.756,
+      "step": 7193
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.632425068119891e-05,
+      "loss": 1.7312,
+      "step": 7194
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6322207084468663e-05,
+      "loss": 1.5874,
+      "step": 7195
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.632016348773842e-05,
+      "loss": 1.7413,
+      "step": 7196
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6318119891008178e-05,
+      "loss": 1.7783,
+      "step": 7197
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.631607629427793e-05,
+      "loss": 1.7726,
+      "step": 7198
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6314032697547686e-05,
+      "loss": 1.6097,
+      "step": 7199
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6311989100817438e-05,
+      "loss": 1.6048,
+      "step": 7200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6309945504087194e-05,
+      "loss": 1.7215,
+      "step": 7201
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.630790190735695e-05,
+      "loss": 1.6931,
+      "step": 7202
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.63058583106267e-05,
+      "loss": 1.8957,
+      "step": 7203
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6303814713896457e-05,
+      "loss": 1.7615,
+      "step": 7204
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6301771117166213e-05,
+      "loss": 1.6559,
+      "step": 7205
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.629972752043597e-05,
+      "loss": 1.878,
+      "step": 7206
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6297683923705724e-05,
+      "loss": 1.8177,
+      "step": 7207
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6295640326975476e-05,
+      "loss": 1.7498,
+      "step": 7208
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6293596730245232e-05,
+      "loss": 1.6849,
+      "step": 7209
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6291553133514984e-05,
+      "loss": 1.7933,
+      "step": 7210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.628950953678474e-05,
+      "loss": 1.8206,
+      "step": 7211
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.62874659400545e-05,
+      "loss": 1.7536,
+      "step": 7212
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.628542234332425e-05,
+      "loss": 1.7607,
+      "step": 7213
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6283378746594007e-05,
+      "loss": 1.6619,
+      "step": 7214
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6281335149863762e-05,
+      "loss": 1.7653,
+      "step": 7215
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6279291553133515e-05,
+      "loss": 1.6956,
+      "step": 7216
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.627724795640327e-05,
+      "loss": 1.6858,
+      "step": 7217
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6275204359673023e-05,
+      "loss": 1.6476,
+      "step": 7218
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6273160762942778e-05,
+      "loss": 1.7409,
+      "step": 7219
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6271117166212537e-05,
+      "loss": 1.6354,
+      "step": 7220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.626907356948229e-05,
+      "loss": 1.5934,
+      "step": 7221
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6267029972752045e-05,
+      "loss": 1.7523,
+      "step": 7222
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6264986376021797e-05,
+      "loss": 1.73,
+      "step": 7223
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6262942779291553e-05,
+      "loss": 1.6939,
+      "step": 7224
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.626089918256131e-05,
+      "loss": 1.6566,
+      "step": 7225
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.625885558583106e-05,
+      "loss": 1.7034,
+      "step": 7226
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.625681198910082e-05,
+      "loss": 1.6174,
+      "step": 7227
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6254768392370572e-05,
+      "loss": 1.8289,
+      "step": 7228
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6252724795640328e-05,
+      "loss": 1.7773,
+      "step": 7229
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6250681198910083e-05,
+      "loss": 1.6456,
+      "step": 7230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6248637602179836e-05,
+      "loss": 1.7416,
+      "step": 7231
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.624659400544959e-05,
+      "loss": 1.654,
+      "step": 7232
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6244550408719347e-05,
+      "loss": 1.6992,
+      "step": 7233
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.62425068119891e-05,
+      "loss": 1.6723,
+      "step": 7234
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6240463215258858e-05,
+      "loss": 1.7081,
+      "step": 7235
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.623841961852861e-05,
+      "loss": 1.7798,
+      "step": 7236
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6236376021798366e-05,
+      "loss": 1.7887,
+      "step": 7237
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6234332425068122e-05,
+      "loss": 1.7403,
+      "step": 7238
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6232288828337874e-05,
+      "loss": 1.6891,
+      "step": 7239
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.623024523160763e-05,
+      "loss": 1.6773,
+      "step": 7240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6228201634877382e-05,
+      "loss": 1.7052,
+      "step": 7241
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.622615803814714e-05,
+      "loss": 1.5943,
+      "step": 7242
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6224114441416896e-05,
+      "loss": 1.6537,
+      "step": 7243
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.622207084468665e-05,
+      "loss": 1.8665,
+      "step": 7244
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6220027247956404e-05,
+      "loss": 1.8681,
+      "step": 7245
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6217983651226157e-05,
+      "loss": 1.6835,
+      "step": 7246
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6215940054495912e-05,
+      "loss": 1.535,
+      "step": 7247
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6213896457765668e-05,
+      "loss": 1.6815,
+      "step": 7248
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6211852861035424e-05,
+      "loss": 1.7862,
+      "step": 7249
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.620980926430518e-05,
+      "loss": 1.7839,
+      "step": 7250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.620776566757493e-05,
+      "loss": 1.7525,
+      "step": 7251
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6205722070844687e-05,
+      "loss": 1.7462,
+      "step": 7252
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6203678474114443e-05,
+      "loss": 1.635,
+      "step": 7253
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6201634877384195e-05,
+      "loss": 1.6675,
+      "step": 7254
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.619959128065395e-05,
+      "loss": 1.8064,
+      "step": 7255
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6197547683923706e-05,
+      "loss": 1.82,
+      "step": 7256
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6195504087193462e-05,
+      "loss": 1.5453,
+      "step": 7257
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6193460490463217e-05,
+      "loss": 1.751,
+      "step": 7258
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.619141689373297e-05,
+      "loss": 1.6548,
+      "step": 7259
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6189373297002725e-05,
+      "loss": 1.7142,
+      "step": 7260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.618732970027248e-05,
+      "loss": 1.6937,
+      "step": 7261
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6185286103542233e-05,
+      "loss": 1.8377,
+      "step": 7262
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.618324250681199e-05,
+      "loss": 1.7348,
+      "step": 7263
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6181198910081744e-05,
+      "loss": 1.7446,
+      "step": 7264
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.61791553133515e-05,
+      "loss": 1.8385,
+      "step": 7265
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6177111716621256e-05,
+      "loss": 1.6745,
+      "step": 7266
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6175068119891008e-05,
+      "loss": 1.7855,
+      "step": 7267
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6173024523160764e-05,
+      "loss": 1.7263,
+      "step": 7268
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6170980926430516e-05,
+      "loss": 1.8135,
+      "step": 7269
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.616893732970027e-05,
+      "loss": 1.6631,
+      "step": 7270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6166893732970027e-05,
+      "loss": 1.8021,
+      "step": 7271
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6164850136239783e-05,
+      "loss": 1.8156,
+      "step": 7272
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.616280653950954e-05,
+      "loss": 1.6156,
+      "step": 7273
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6160762942779294e-05,
+      "loss": 1.7737,
+      "step": 7274
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6158719346049046e-05,
+      "loss": 1.6632,
+      "step": 7275
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6156675749318802e-05,
+      "loss": 1.7545,
+      "step": 7276
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6154632152588554e-05,
+      "loss": 1.7135,
+      "step": 7277
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.615258855585831e-05,
+      "loss": 1.7091,
+      "step": 7278
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.615054495912807e-05,
+      "loss": 1.8293,
+      "step": 7279
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.614850136239782e-05,
+      "loss": 1.6594,
+      "step": 7280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6146457765667577e-05,
+      "loss": 1.8348,
+      "step": 7281
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.614441416893733e-05,
+      "loss": 1.7908,
+      "step": 7282
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6142370572207085e-05,
+      "loss": 1.6828,
+      "step": 7283
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.614032697547684e-05,
+      "loss": 1.9885,
+      "step": 7284
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6138283378746592e-05,
+      "loss": 1.7789,
+      "step": 7285
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6136239782016348e-05,
+      "loss": 1.5693,
+      "step": 7286
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6134196185286104e-05,
+      "loss": 1.788,
+      "step": 7287
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.613215258855586e-05,
+      "loss": 1.705,
+      "step": 7288
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6130108991825615e-05,
+      "loss": 1.6539,
+      "step": 7289
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6128065395095367e-05,
+      "loss": 1.7659,
+      "step": 7290
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6126021798365123e-05,
+      "loss": 1.8065,
+      "step": 7291
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.612397820163488e-05,
+      "loss": 1.8541,
+      "step": 7292
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.612193460490463e-05,
+      "loss": 1.7183,
+      "step": 7293
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.611989100817439e-05,
+      "loss": 1.761,
+      "step": 7294
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6117847411444142e-05,
+      "loss": 1.7222,
+      "step": 7295
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6115803814713898e-05,
+      "loss": 1.6643,
+      "step": 7296
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6113760217983653e-05,
+      "loss": 1.5612,
+      "step": 7297
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6111716621253405e-05,
+      "loss": 1.7875,
+      "step": 7298
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.610967302452316e-05,
+      "loss": 1.7317,
+      "step": 7299
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6107629427792913e-05,
+      "loss": 1.5049,
+      "step": 7300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.610558583106267e-05,
+      "loss": 1.6889,
+      "step": 7301
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6103542234332428e-05,
+      "loss": 1.7645,
+      "step": 7302
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.610149863760218e-05,
+      "loss": 1.7246,
+      "step": 7303
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6099455040871936e-05,
+      "loss": 1.6869,
+      "step": 7304
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6097411444141688e-05,
+      "loss": 1.7234,
+      "step": 7305
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6095367847411444e-05,
+      "loss": 1.8384,
+      "step": 7306
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.60933242506812e-05,
+      "loss": 1.6849,
+      "step": 7307
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.609128065395095e-05,
+      "loss": 1.7029,
+      "step": 7308
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.608923705722071e-05,
+      "loss": 1.7039,
+      "step": 7309
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6087193460490466e-05,
+      "loss": 1.6951,
+      "step": 7310
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.608514986376022e-05,
+      "loss": 1.7107,
+      "step": 7311
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6083106267029974e-05,
+      "loss": 1.6325,
+      "step": 7312
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6081062670299726e-05,
+      "loss": 1.6041,
+      "step": 7313
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6079019073569482e-05,
+      "loss": 1.689,
+      "step": 7314
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6076975476839238e-05,
+      "loss": 1.8559,
+      "step": 7315
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.607493188010899e-05,
+      "loss": 1.6568,
+      "step": 7316
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.607288828337875e-05,
+      "loss": 1.733,
+      "step": 7317
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.60708446866485e-05,
+      "loss": 1.6635,
+      "step": 7318
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6068801089918257e-05,
+      "loss": 1.6374,
+      "step": 7319
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6066757493188012e-05,
+      "loss": 1.6957,
+      "step": 7320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6064713896457765e-05,
+      "loss": 1.6565,
+      "step": 7321
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.606267029972752e-05,
+      "loss": 1.7272,
+      "step": 7322
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6060626702997273e-05,
+      "loss": 1.7973,
+      "step": 7323
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.605858310626703e-05,
+      "loss": 1.7288,
+      "step": 7324
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6056539509536787e-05,
+      "loss": 1.6946,
+      "step": 7325
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.605449591280654e-05,
+      "loss": 1.7962,
+      "step": 7326
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6052452316076295e-05,
+      "loss": 1.7397,
+      "step": 7327
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.605040871934605e-05,
+      "loss": 1.6084,
+      "step": 7328
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6048365122615803e-05,
+      "loss": 1.8558,
+      "step": 7329
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.604632152588556e-05,
+      "loss": 1.8523,
+      "step": 7330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.604427792915531e-05,
+      "loss": 1.6758,
+      "step": 7331
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.604223433242507e-05,
+      "loss": 1.7559,
+      "step": 7332
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6040190735694825e-05,
+      "loss": 1.6443,
+      "step": 7333
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6038147138964578e-05,
+      "loss": 1.8406,
+      "step": 7334
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6036103542234333e-05,
+      "loss": 1.7346,
+      "step": 7335
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6034059945504086e-05,
+      "loss": 1.6879,
+      "step": 7336
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.603201634877384e-05,
+      "loss": 1.8111,
+      "step": 7337
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6029972752043597e-05,
+      "loss": 1.6203,
+      "step": 7338
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6027929155313353e-05,
+      "loss": 1.7318,
+      "step": 7339
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6025885558583108e-05,
+      "loss": 1.776,
+      "step": 7340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.602384196185286e-05,
+      "loss": 1.6865,
+      "step": 7341
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6021798365122616e-05,
+      "loss": 1.737,
+      "step": 7342
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.601975476839237e-05,
+      "loss": 1.6823,
+      "step": 7343
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6017711171662124e-05,
+      "loss": 1.6869,
+      "step": 7344
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.601566757493188e-05,
+      "loss": 1.7173,
+      "step": 7345
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.601362397820164e-05,
+      "loss": 1.8251,
+      "step": 7346
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.601158038147139e-05,
+      "loss": 1.6313,
+      "step": 7347
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6009536784741146e-05,
+      "loss": 1.6875,
+      "step": 7348
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.60074931880109e-05,
+      "loss": 1.8825,
+      "step": 7349
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6005449591280654e-05,
+      "loss": 1.6738,
+      "step": 7350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.600340599455041e-05,
+      "loss": 1.6971,
+      "step": 7351
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6001362397820162e-05,
+      "loss": 1.5347,
+      "step": 7352
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5999318801089918e-05,
+      "loss": 1.8723,
+      "step": 7353
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5997275204359673e-05,
+      "loss": 1.7623,
+      "step": 7354
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.599523160762943e-05,
+      "loss": 1.8051,
+      "step": 7355
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5993188010899185e-05,
+      "loss": 1.7163,
+      "step": 7356
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5991144414168937e-05,
+      "loss": 1.7242,
+      "step": 7357
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5989100817438693e-05,
+      "loss": 1.7414,
+      "step": 7358
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5987057220708445e-05,
+      "loss": 1.6451,
+      "step": 7359
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.59850136239782e-05,
+      "loss": 1.6675,
+      "step": 7360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.598297002724796e-05,
+      "loss": 1.6662,
+      "step": 7361
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5980926430517712e-05,
+      "loss": 1.6257,
+      "step": 7362
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5978882833787467e-05,
+      "loss": 1.7551,
+      "step": 7363
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5976839237057223e-05,
+      "loss": 1.7076,
+      "step": 7364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5974795640326975e-05,
+      "loss": 1.797,
+      "step": 7365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.597275204359673e-05,
+      "loss": 1.6245,
+      "step": 7366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5970708446866483e-05,
+      "loss": 1.69,
+      "step": 7367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.596866485013624e-05,
+      "loss": 1.682,
+      "step": 7368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5966621253405998e-05,
+      "loss": 1.7648,
+      "step": 7369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.596457765667575e-05,
+      "loss": 1.6585,
+      "step": 7370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5962534059945506e-05,
+      "loss": 1.8493,
+      "step": 7371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5960490463215258e-05,
+      "loss": 1.8223,
+      "step": 7372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5958446866485014e-05,
+      "loss": 1.6951,
+      "step": 7373
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.595640326975477e-05,
+      "loss": 1.6574,
+      "step": 7374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.595435967302452e-05,
+      "loss": 1.8525,
+      "step": 7375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.595231607629428e-05,
+      "loss": 1.5748,
+      "step": 7376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5950272479564033e-05,
+      "loss": 1.7134,
+      "step": 7377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.594822888283379e-05,
+      "loss": 1.6502,
+      "step": 7378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5946185286103544e-05,
+      "loss": 1.7258,
+      "step": 7379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5944141689373296e-05,
+      "loss": 1.7873,
+      "step": 7380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5942098092643052e-05,
+      "loss": 1.5357,
+      "step": 7381
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5940054495912807e-05,
+      "loss": 1.7319,
+      "step": 7382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.593801089918256e-05,
+      "loss": 1.5956,
+      "step": 7383
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.593596730245232e-05,
+      "loss": 1.7185,
+      "step": 7384
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.593392370572207e-05,
+      "loss": 1.6869,
+      "step": 7385
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5931880108991827e-05,
+      "loss": 1.7313,
+      "step": 7386
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5929836512261582e-05,
+      "loss": 1.7529,
+      "step": 7387
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5927792915531334e-05,
+      "loss": 1.7146,
+      "step": 7388
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.592574931880109e-05,
+      "loss": 1.6333,
+      "step": 7389
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5923705722070842e-05,
+      "loss": 1.7326,
+      "step": 7390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.59216621253406e-05,
+      "loss": 1.7646,
+      "step": 7391
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5919618528610357e-05,
+      "loss": 1.6797,
+      "step": 7392
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.591757493188011e-05,
+      "loss": 1.7507,
+      "step": 7393
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5915531335149865e-05,
+      "loss": 1.654,
+      "step": 7394
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5913487738419617e-05,
+      "loss": 1.7057,
+      "step": 7395
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5911444141689373e-05,
+      "loss": 1.7707,
+      "step": 7396
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.590940054495913e-05,
+      "loss": 1.7167,
+      "step": 7397
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.590735694822888e-05,
+      "loss": 1.7564,
+      "step": 7398
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.590531335149864e-05,
+      "loss": 1.6464,
+      "step": 7399
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5903269754768395e-05,
+      "loss": 1.6995,
+      "step": 7400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5901226158038148e-05,
+      "loss": 1.5465,
+      "step": 7401
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5899182561307903e-05,
+      "loss": 1.7196,
+      "step": 7402
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5897138964577655e-05,
+      "loss": 1.8538,
+      "step": 7403
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.589509536784741e-05,
+      "loss": 1.6725,
+      "step": 7404
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5893051771117167e-05,
+      "loss": 1.6561,
+      "step": 7405
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5891008174386922e-05,
+      "loss": 1.7903,
+      "step": 7406
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5888964577656678e-05,
+      "loss": 1.7275,
+      "step": 7407
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.588692098092643e-05,
+      "loss": 1.6261,
+      "step": 7408
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5884877384196186e-05,
+      "loss": 1.6959,
+      "step": 7409
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.588283378746594e-05,
+      "loss": 1.768,
+      "step": 7410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5880790190735694e-05,
+      "loss": 1.8084,
+      "step": 7411
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.587874659400545e-05,
+      "loss": 1.8343,
+      "step": 7412
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.58767029972752e-05,
+      "loss": 1.7357,
+      "step": 7413
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.587465940054496e-05,
+      "loss": 1.635,
+      "step": 7414
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5872615803814716e-05,
+      "loss": 1.7805,
+      "step": 7415
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.587057220708447e-05,
+      "loss": 1.7636,
+      "step": 7416
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5868528610354224e-05,
+      "loss": 1.7096,
+      "step": 7417
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.586648501362398e-05,
+      "loss": 1.7719,
+      "step": 7418
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5864441416893732e-05,
+      "loss": 1.6265,
+      "step": 7419
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5862397820163488e-05,
+      "loss": 1.6198,
+      "step": 7420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5860354223433243e-05,
+      "loss": 1.6168,
+      "step": 7421
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5858310626703e-05,
+      "loss": 1.7872,
+      "step": 7422
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5856267029972755e-05,
+      "loss": 1.6208,
+      "step": 7423
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5854223433242507e-05,
+      "loss": 1.5961,
+      "step": 7424
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5852179836512262e-05,
+      "loss": 1.5464,
+      "step": 7425
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5850136239782015e-05,
+      "loss": 1.7451,
+      "step": 7426
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.584809264305177e-05,
+      "loss": 1.7767,
+      "step": 7427
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5846049046321526e-05,
+      "loss": 1.7327,
+      "step": 7428
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.584400544959128e-05,
+      "loss": 1.5809,
+      "step": 7429
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5841961852861037e-05,
+      "loss": 1.9036,
+      "step": 7430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.583991825613079e-05,
+      "loss": 1.7578,
+      "step": 7431
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5837874659400545e-05,
+      "loss": 1.7644,
+      "step": 7432
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.58358310626703e-05,
+      "loss": 1.7051,
+      "step": 7433
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5833787465940053e-05,
+      "loss": 1.6538,
+      "step": 7434
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.583174386920981e-05,
+      "loss": 1.6525,
+      "step": 7435
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5829700272479564e-05,
+      "loss": 1.7356,
+      "step": 7436
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.582765667574932e-05,
+      "loss": 1.6198,
+      "step": 7437
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5825613079019075e-05,
+      "loss": 1.6831,
+      "step": 7438
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5823569482288828e-05,
+      "loss": 1.7098,
+      "step": 7439
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5821525885558583e-05,
+      "loss": 1.7794,
+      "step": 7440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.581948228882834e-05,
+      "loss": 1.6317,
+      "step": 7441
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.581743869209809e-05,
+      "loss": 1.6995,
+      "step": 7442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.581539509536785e-05,
+      "loss": 1.7316,
+      "step": 7443
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5813351498637602e-05,
+      "loss": 1.7388,
+      "step": 7444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5811307901907358e-05,
+      "loss": 1.7094,
+      "step": 7445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5809264305177114e-05,
+      "loss": 1.8013,
+      "step": 7446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5807220708446866e-05,
+      "loss": 1.727,
+      "step": 7447
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.580517711171662e-05,
+      "loss": 1.7932,
+      "step": 7448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5803133514986374e-05,
+      "loss": 1.713,
+      "step": 7449
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.580108991825613e-05,
+      "loss": 1.7332,
+      "step": 7450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.579904632152589e-05,
+      "loss": 1.7981,
+      "step": 7451
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.579700272479564e-05,
+      "loss": 1.6536,
+      "step": 7452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5794959128065396e-05,
+      "loss": 1.6844,
+      "step": 7453
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.579291553133515e-05,
+      "loss": 1.809,
+      "step": 7454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5790871934604904e-05,
+      "loss": 1.7618,
+      "step": 7455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.578882833787466e-05,
+      "loss": 1.7089,
+      "step": 7456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5786784741144412e-05,
+      "loss": 1.6484,
+      "step": 7457
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.578474114441417e-05,
+      "loss": 1.8403,
+      "step": 7458
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5782697547683927e-05,
+      "loss": 1.7272,
+      "step": 7459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.578065395095368e-05,
+      "loss": 1.7372,
+      "step": 7460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5778610354223435e-05,
+      "loss": 1.6877,
+      "step": 7461
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5776566757493187e-05,
+      "loss": 1.7199,
+      "step": 7462
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5774523160762943e-05,
+      "loss": 1.7152,
+      "step": 7463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5772479564032698e-05,
+      "loss": 1.5978,
+      "step": 7464
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.577043596730245e-05,
+      "loss": 1.6002,
+      "step": 7465
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.576839237057221e-05,
+      "loss": 1.6719,
+      "step": 7466
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.576634877384196e-05,
+      "loss": 1.8314,
+      "step": 7467
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5764305177111717e-05,
+      "loss": 1.6653,
+      "step": 7468
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5762261580381473e-05,
+      "loss": 1.7818,
+      "step": 7469
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5760217983651225e-05,
+      "loss": 1.6419,
+      "step": 7470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.575817438692098e-05,
+      "loss": 1.7997,
+      "step": 7471
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5756130790190733e-05,
+      "loss": 1.5801,
+      "step": 7472
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5754087193460492e-05,
+      "loss": 1.7784,
+      "step": 7473
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5752043596730248e-05,
+      "loss": 1.732,
+      "step": 7474
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.575e-05,
+      "loss": 1.7067,
+      "step": 7475
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5747956403269756e-05,
+      "loss": 1.7446,
+      "step": 7476
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.574591280653951e-05,
+      "loss": 1.8021,
+      "step": 7477
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5743869209809263e-05,
+      "loss": 1.5931,
+      "step": 7478
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.574182561307902e-05,
+      "loss": 1.7054,
+      "step": 7479
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.573978201634877e-05,
+      "loss": 1.6499,
+      "step": 7480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.573773841961853e-05,
+      "loss": 1.7118,
+      "step": 7481
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5735694822888286e-05,
+      "loss": 1.6765,
+      "step": 7482
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5733651226158038e-05,
+      "loss": 1.6629,
+      "step": 7483
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5731607629427794e-05,
+      "loss": 1.7011,
+      "step": 7484
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5729564032697546e-05,
+      "loss": 1.6295,
+      "step": 7485
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5727520435967302e-05,
+      "loss": 1.7041,
+      "step": 7486
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5725476839237057e-05,
+      "loss": 1.7647,
+      "step": 7487
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5723433242506813e-05,
+      "loss": 1.6872,
+      "step": 7488
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.572138964577657e-05,
+      "loss": 1.7616,
+      "step": 7489
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.571934604904632e-05,
+      "loss": 1.7672,
+      "step": 7490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5717302452316077e-05,
+      "loss": 1.7868,
+      "step": 7491
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5715258855585832e-05,
+      "loss": 1.9092,
+      "step": 7492
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5713215258855584e-05,
+      "loss": 1.68,
+      "step": 7493
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.571117166212534e-05,
+      "loss": 1.62,
+      "step": 7494
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5709128065395096e-05,
+      "loss": 1.6153,
+      "step": 7495
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.570708446866485e-05,
+      "loss": 1.7844,
+      "step": 7496
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5705040871934607e-05,
+      "loss": 1.8102,
+      "step": 7497
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.570299727520436e-05,
+      "loss": 1.7841,
+      "step": 7498
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5700953678474115e-05,
+      "loss": 1.7789,
+      "step": 7499
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.569891008174387e-05,
+      "loss": 1.716,
+      "step": 7500
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 1.5421665906906128,
+      "eval_runtime": 721.3558,
+      "eval_samples_per_second": 5.753,
+      "eval_steps_per_second": 1.439,
+      "step": 7500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5696866485013623e-05,
+      "loss": 1.6304,
+      "step": 7501
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.569482288828338e-05,
+      "loss": 1.8076,
+      "step": 7502
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5692779291553134e-05,
+      "loss": 1.5892,
+      "step": 7503
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.569073569482289e-05,
+      "loss": 1.7142,
+      "step": 7504
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5688692098092645e-05,
+      "loss": 1.6732,
+      "step": 7505
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5686648501362397e-05,
+      "loss": 1.7037,
+      "step": 7506
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5684604904632153e-05,
+      "loss": 1.6586,
+      "step": 7507
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5682561307901905e-05,
+      "loss": 1.6958,
+      "step": 7508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.568051771117166e-05,
+      "loss": 1.9406,
+      "step": 7509
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5678474114441417e-05,
+      "loss": 1.8595,
+      "step": 7510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5676430517711172e-05,
+      "loss": 1.6358,
+      "step": 7511
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5674386920980928e-05,
+      "loss": 1.695,
+      "step": 7512
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5672343324250684e-05,
+      "loss": 1.7617,
+      "step": 7513
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5670299727520436e-05,
+      "loss": 1.8139,
+      "step": 7514
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.566825613079019e-05,
+      "loss": 1.5454,
+      "step": 7515
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5666212534059944e-05,
+      "loss": 1.7533,
+      "step": 7516
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.56641689373297e-05,
+      "loss": 1.76,
+      "step": 7517
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5662125340599458e-05,
+      "loss": 1.7282,
+      "step": 7518
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.566008174386921e-05,
+      "loss": 1.7502,
+      "step": 7519
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5658038147138966e-05,
+      "loss": 1.7793,
+      "step": 7520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.565599455040872e-05,
+      "loss": 1.7085,
+      "step": 7521
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5653950953678474e-05,
+      "loss": 1.801,
+      "step": 7522
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.565190735694823e-05,
+      "loss": 1.5915,
+      "step": 7523
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5649863760217982e-05,
+      "loss": 1.7045,
+      "step": 7524
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.564782016348774e-05,
+      "loss": 1.7795,
+      "step": 7525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5645776566757493e-05,
+      "loss": 1.7218,
+      "step": 7526
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.564373297002725e-05,
+      "loss": 1.7042,
+      "step": 7527
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5641689373297004e-05,
+      "loss": 1.6044,
+      "step": 7528
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5639645776566757e-05,
+      "loss": 1.6473,
+      "step": 7529
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5637602179836512e-05,
+      "loss": 1.6716,
+      "step": 7530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5635558583106268e-05,
+      "loss": 1.7293,
+      "step": 7531
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.563351498637602e-05,
+      "loss": 1.7124,
+      "step": 7532
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.563147138964578e-05,
+      "loss": 1.6383,
+      "step": 7533
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.562942779291553e-05,
+      "loss": 1.7257,
+      "step": 7534
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5627384196185287e-05,
+      "loss": 1.6734,
+      "step": 7535
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5625340599455043e-05,
+      "loss": 1.8369,
+      "step": 7536
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5623297002724795e-05,
+      "loss": 1.676,
+      "step": 7537
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.562125340599455e-05,
+      "loss": 1.6806,
+      "step": 7538
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5619209809264303e-05,
+      "loss": 1.7514,
+      "step": 7539
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5617166212534062e-05,
+      "loss": 1.6972,
+      "step": 7540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5615122615803818e-05,
+      "loss": 1.7018,
+      "step": 7541
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.561307901907357e-05,
+      "loss": 1.774,
+      "step": 7542
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5611035422343325e-05,
+      "loss": 1.6598,
+      "step": 7543
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5608991825613078e-05,
+      "loss": 1.6626,
+      "step": 7544
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5606948228882833e-05,
+      "loss": 1.7508,
+      "step": 7545
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.560490463215259e-05,
+      "loss": 1.8687,
+      "step": 7546
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.560286103542234e-05,
+      "loss": 1.6757,
+      "step": 7547
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.56008174386921e-05,
+      "loss": 1.812,
+      "step": 7548
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5598773841961856e-05,
+      "loss": 1.7223,
+      "step": 7549
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5596730245231608e-05,
+      "loss": 1.6951,
+      "step": 7550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5594686648501364e-05,
+      "loss": 1.7232,
+      "step": 7551
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5592643051771116e-05,
+      "loss": 1.7712,
+      "step": 7552
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.559059945504087e-05,
+      "loss": 1.6965,
+      "step": 7553
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5588555858310627e-05,
+      "loss": 1.7042,
+      "step": 7554
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5586512261580383e-05,
+      "loss": 1.7677,
+      "step": 7555
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.558446866485014e-05,
+      "loss": 1.7489,
+      "step": 7556
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.558242506811989e-05,
+      "loss": 1.8797,
+      "step": 7557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5580381471389646e-05,
+      "loss": 1.793,
+      "step": 7558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5578337874659402e-05,
+      "loss": 1.7341,
+      "step": 7559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5576294277929154e-05,
+      "loss": 1.7843,
+      "step": 7560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.557425068119891e-05,
+      "loss": 1.7466,
+      "step": 7561
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5572207084468662e-05,
+      "loss": 1.5977,
+      "step": 7562
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.557016348773842e-05,
+      "loss": 1.7036,
+      "step": 7563
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5568119891008177e-05,
+      "loss": 1.6529,
+      "step": 7564
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.556607629427793e-05,
+      "loss": 1.8123,
+      "step": 7565
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5564032697547685e-05,
+      "loss": 1.7701,
+      "step": 7566
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.556198910081744e-05,
+      "loss": 1.7043,
+      "step": 7567
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5559945504087192e-05,
+      "loss": 1.6501,
+      "step": 7568
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5557901907356948e-05,
+      "loss": 1.6874,
+      "step": 7569
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5555858310626704e-05,
+      "loss": 1.5837,
+      "step": 7570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.555381471389646e-05,
+      "loss": 1.6515,
+      "step": 7571
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5551771117166215e-05,
+      "loss": 1.7056,
+      "step": 7572
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5549727520435967e-05,
+      "loss": 1.6585,
+      "step": 7573
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5547683923705723e-05,
+      "loss": 1.7548,
+      "step": 7574
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5545640326975475e-05,
+      "loss": 1.7163,
+      "step": 7575
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.554359673024523e-05,
+      "loss": 1.6854,
+      "step": 7576
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5541553133514986e-05,
+      "loss": 1.7822,
+      "step": 7577
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5539509536784742e-05,
+      "loss": 1.6806,
+      "step": 7578
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5537465940054498e-05,
+      "loss": 1.8178,
+      "step": 7579
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.553542234332425e-05,
+      "loss": 1.6628,
+      "step": 7580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5533378746594006e-05,
+      "loss": 1.6742,
+      "step": 7581
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.553133514986376e-05,
+      "loss": 1.7322,
+      "step": 7582
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5529291553133513e-05,
+      "loss": 1.6863,
+      "step": 7583
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.552724795640327e-05,
+      "loss": 1.6559,
+      "step": 7584
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5525204359673028e-05,
+      "loss": 1.7317,
+      "step": 7585
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.552316076294278e-05,
+      "loss": 1.8747,
+      "step": 7586
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5521117166212536e-05,
+      "loss": 1.8813,
+      "step": 7587
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5519073569482288e-05,
+      "loss": 1.7259,
+      "step": 7588
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5517029972752044e-05,
+      "loss": 1.7802,
+      "step": 7589
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.55149863760218e-05,
+      "loss": 1.7686,
+      "step": 7590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5512942779291552e-05,
+      "loss": 1.7378,
+      "step": 7591
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5510899182561307e-05,
+      "loss": 1.6419,
+      "step": 7592
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5508855585831063e-05,
+      "loss": 1.8009,
+      "step": 7593
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.550681198910082e-05,
+      "loss": 1.7116,
+      "step": 7594
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5504768392370574e-05,
+      "loss": 1.7278,
+      "step": 7595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5502724795640326e-05,
+      "loss": 1.6903,
+      "step": 7596
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5500681198910082e-05,
+      "loss": 1.7283,
+      "step": 7597
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5498637602179834e-05,
+      "loss": 1.7775,
+      "step": 7598
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.549659400544959e-05,
+      "loss": 1.8403,
+      "step": 7599
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.549455040871935e-05,
+      "loss": 1.6289,
+      "step": 7600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.54925068119891e-05,
+      "loss": 1.6729,
+      "step": 7601
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5490463215258857e-05,
+      "loss": 1.6738,
+      "step": 7602
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5488419618528613e-05,
+      "loss": 1.6733,
+      "step": 7603
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5486376021798365e-05,
+      "loss": 1.684,
+      "step": 7604
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.548433242506812e-05,
+      "loss": 1.6926,
+      "step": 7605
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5482288828337873e-05,
+      "loss": 1.7239,
+      "step": 7606
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5480245231607628e-05,
+      "loss": 1.7895,
+      "step": 7607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5478201634877387e-05,
+      "loss": 1.7377,
+      "step": 7608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.547615803814714e-05,
+      "loss": 1.7888,
+      "step": 7609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5474114441416895e-05,
+      "loss": 1.7346,
+      "step": 7610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5472070844686647e-05,
+      "loss": 1.8066,
+      "step": 7611
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5470027247956403e-05,
+      "loss": 1.7855,
+      "step": 7612
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.546798365122616e-05,
+      "loss": 1.6954,
+      "step": 7613
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.546594005449591e-05,
+      "loss": 1.7966,
+      "step": 7614
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.546389645776567e-05,
+      "loss": 1.6991,
+      "step": 7615
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5461852861035422e-05,
+      "loss": 1.8061,
+      "step": 7616
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5459809264305178e-05,
+      "loss": 1.7897,
+      "step": 7617
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5457765667574933e-05,
+      "loss": 1.6314,
+      "step": 7618
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5455722070844686e-05,
+      "loss": 1.6807,
+      "step": 7619
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.545367847411444e-05,
+      "loss": 1.6925,
+      "step": 7620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5451634877384197e-05,
+      "loss": 1.812,
+      "step": 7621
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5449591280653953e-05,
+      "loss": 1.7135,
+      "step": 7622
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5447547683923708e-05,
+      "loss": 1.6252,
+      "step": 7623
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.544550408719346e-05,
+      "loss": 1.705,
+      "step": 7624
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5443460490463216e-05,
+      "loss": 1.6322,
+      "step": 7625
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5441416893732972e-05,
+      "loss": 1.6729,
+      "step": 7626
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5439373297002724e-05,
+      "loss": 1.7636,
+      "step": 7627
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.543732970027248e-05,
+      "loss": 1.7445,
+      "step": 7628
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5435286103542232e-05,
+      "loss": 1.7886,
+      "step": 7629
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.543324250681199e-05,
+      "loss": 1.809,
+      "step": 7630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5431198910081747e-05,
+      "loss": 1.8752,
+      "step": 7631
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.54291553133515e-05,
+      "loss": 1.6256,
+      "step": 7632
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5427111716621254e-05,
+      "loss": 1.7222,
+      "step": 7633
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5425068119891007e-05,
+      "loss": 1.7371,
+      "step": 7634
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5423024523160762e-05,
+      "loss": 1.7515,
+      "step": 7635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5420980926430518e-05,
+      "loss": 1.7338,
+      "step": 7636
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5418937329700274e-05,
+      "loss": 1.681,
+      "step": 7637
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.541689373297003e-05,
+      "loss": 1.8195,
+      "step": 7638
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.541485013623978e-05,
+      "loss": 1.8099,
+      "step": 7639
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5412806539509537e-05,
+      "loss": 1.7277,
+      "step": 7640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5410762942779293e-05,
+      "loss": 1.7272,
+      "step": 7641
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5408719346049045e-05,
+      "loss": 1.7626,
+      "step": 7642
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.54066757493188e-05,
+      "loss": 1.7321,
+      "step": 7643
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5404632152588556e-05,
+      "loss": 1.7094,
+      "step": 7644
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5402588555858312e-05,
+      "loss": 1.7266,
+      "step": 7645
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5400544959128067e-05,
+      "loss": 1.6095,
+      "step": 7646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.539850136239782e-05,
+      "loss": 1.7561,
+      "step": 7647
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5396457765667575e-05,
+      "loss": 1.7518,
+      "step": 7648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.539441416893733e-05,
+      "loss": 1.9166,
+      "step": 7649
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5392370572207083e-05,
+      "loss": 1.8085,
+      "step": 7650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.539032697547684e-05,
+      "loss": 1.767,
+      "step": 7651
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5388283378746594e-05,
+      "loss": 1.5904,
+      "step": 7652
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.538623978201635e-05,
+      "loss": 1.7572,
+      "step": 7653
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5384196185286106e-05,
+      "loss": 1.6257,
+      "step": 7654
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5382152588555858e-05,
+      "loss": 1.6595,
+      "step": 7655
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5380108991825614e-05,
+      "loss": 1.6111,
+      "step": 7656
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5378065395095366e-05,
+      "loss": 1.8371,
+      "step": 7657
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.537602179836512e-05,
+      "loss": 1.6465,
+      "step": 7658
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5373978201634877e-05,
+      "loss": 1.6275,
+      "step": 7659
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5371934604904633e-05,
+      "loss": 1.7652,
+      "step": 7660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.536989100817439e-05,
+      "loss": 1.719,
+      "step": 7661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5367847411444144e-05,
+      "loss": 1.8437,
+      "step": 7662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5365803814713896e-05,
+      "loss": 1.7994,
+      "step": 7663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5363760217983652e-05,
+      "loss": 1.7627,
+      "step": 7664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5361716621253404e-05,
+      "loss": 1.8499,
+      "step": 7665
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.535967302452316e-05,
+      "loss": 1.7391,
+      "step": 7666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.535762942779292e-05,
+      "loss": 1.6449,
+      "step": 7667
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.535558583106267e-05,
+      "loss": 1.6963,
+      "step": 7668
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5353542234332427e-05,
+      "loss": 1.7081,
+      "step": 7669
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.535149863760218e-05,
+      "loss": 1.7349,
+      "step": 7670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5349455040871935e-05,
+      "loss": 1.8384,
+      "step": 7671
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.534741144414169e-05,
+      "loss": 1.652,
+      "step": 7672
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5345367847411442e-05,
+      "loss": 1.7782,
+      "step": 7673
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5343324250681198e-05,
+      "loss": 1.7202,
+      "step": 7674
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5341280653950954e-05,
+      "loss": 1.6657,
+      "step": 7675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.533923705722071e-05,
+      "loss": 1.7587,
+      "step": 7676
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5337193460490465e-05,
+      "loss": 1.6391,
+      "step": 7677
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5335149863760217e-05,
+      "loss": 1.7172,
+      "step": 7678
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5333106267029973e-05,
+      "loss": 1.7398,
+      "step": 7679
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.533106267029973e-05,
+      "loss": 1.7597,
+      "step": 7680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.532901907356948e-05,
+      "loss": 1.6542,
+      "step": 7681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.532697547683924e-05,
+      "loss": 1.8789,
+      "step": 7682
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5324931880108992e-05,
+      "loss": 1.7306,
+      "step": 7683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5322888283378748e-05,
+      "loss": 1.8579,
+      "step": 7684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5320844686648503e-05,
+      "loss": 1.866,
+      "step": 7685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5318801089918256e-05,
+      "loss": 1.642,
+      "step": 7686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.531675749318801e-05,
+      "loss": 1.6933,
+      "step": 7687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5314713896457763e-05,
+      "loss": 1.6236,
+      "step": 7688
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.531267029972752e-05,
+      "loss": 1.7931,
+      "step": 7689
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5310626702997278e-05,
+      "loss": 1.6413,
+      "step": 7690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.530858310626703e-05,
+      "loss": 1.6817,
+      "step": 7691
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5306539509536786e-05,
+      "loss": 1.8274,
+      "step": 7692
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5304495912806538e-05,
+      "loss": 1.6661,
+      "step": 7693
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5302452316076294e-05,
+      "loss": 1.7147,
+      "step": 7694
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.530040871934605e-05,
+      "loss": 1.72,
+      "step": 7695
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.52983651226158e-05,
+      "loss": 1.6742,
+      "step": 7696
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.529632152588556e-05,
+      "loss": 1.6607,
+      "step": 7697
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5294277929155316e-05,
+      "loss": 1.8669,
+      "step": 7698
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.529223433242507e-05,
+      "loss": 1.6408,
+      "step": 7699
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5290190735694824e-05,
+      "loss": 1.7282,
+      "step": 7700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5288147138964576e-05,
+      "loss": 1.7542,
+      "step": 7701
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5286103542234332e-05,
+      "loss": 1.7643,
+      "step": 7702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5284059945504088e-05,
+      "loss": 1.7125,
+      "step": 7703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.528201634877384e-05,
+      "loss": 1.6007,
+      "step": 7704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.52799727520436e-05,
+      "loss": 1.7961,
+      "step": 7705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.527792915531335e-05,
+      "loss": 1.7132,
+      "step": 7706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5275885558583107e-05,
+      "loss": 1.6758,
+      "step": 7707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5273841961852862e-05,
+      "loss": 1.6706,
+      "step": 7708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5271798365122615e-05,
+      "loss": 1.6807,
+      "step": 7709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.526975476839237e-05,
+      "loss": 1.7208,
+      "step": 7710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5267711171662123e-05,
+      "loss": 1.7331,
+      "step": 7711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.526566757493188e-05,
+      "loss": 1.7082,
+      "step": 7712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5263623978201637e-05,
+      "loss": 1.6757,
+      "step": 7713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.526158038147139e-05,
+      "loss": 1.6268,
+      "step": 7714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5259536784741145e-05,
+      "loss": 1.632,
+      "step": 7715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5257493188010899e-05,
+      "loss": 1.6701,
+      "step": 7716
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5255449591280653e-05,
+      "loss": 1.5413,
+      "step": 7717
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5253405994550409e-05,
+      "loss": 1.6538,
+      "step": 7718
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5251362397820166e-05,
+      "loss": 1.7889,
+      "step": 7719
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.524931880108992e-05,
+      "loss": 1.7245,
+      "step": 7720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5247275204359674e-05,
+      "loss": 1.6437,
+      "step": 7721
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5245231607629428e-05,
+      "loss": 1.7899,
+      "step": 7722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5243188010899183e-05,
+      "loss": 1.6321,
+      "step": 7723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5241144414168937e-05,
+      "loss": 1.7026,
+      "step": 7724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5239100817438691e-05,
+      "loss": 1.826,
+      "step": 7725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5237057220708445e-05,
+      "loss": 1.7007,
+      "step": 7726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5235013623978203e-05,
+      "loss": 1.8285,
+      "step": 7727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5232970027247958e-05,
+      "loss": 1.7713,
+      "step": 7728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5230926430517712e-05,
+      "loss": 1.6121,
+      "step": 7729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5228882833787466e-05,
+      "loss": 1.6137,
+      "step": 7730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.522683923705722e-05,
+      "loss": 1.6879,
+      "step": 7731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5224795640326976e-05,
+      "loss": 1.5393,
+      "step": 7732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.522275204359673e-05,
+      "loss": 1.7099,
+      "step": 7733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5220708446866487e-05,
+      "loss": 1.7077,
+      "step": 7734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5218664850136241e-05,
+      "loss": 1.7975,
+      "step": 7735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5216621253405995e-05,
+      "loss": 1.74,
+      "step": 7736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.521457765667575e-05,
+      "loss": 1.6468,
+      "step": 7737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5212534059945504e-05,
+      "loss": 1.6813,
+      "step": 7738
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5210490463215258e-05,
+      "loss": 1.8255,
+      "step": 7739
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5208446866485012e-05,
+      "loss": 1.7551,
+      "step": 7740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5206403269754768e-05,
+      "loss": 1.6708,
+      "step": 7741
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5204359673024525e-05,
+      "loss": 1.6786,
+      "step": 7742
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5202316076294279e-05,
+      "loss": 1.7388,
+      "step": 7743
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5200272479564033e-05,
+      "loss": 1.7177,
+      "step": 7744
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5198228882833787e-05,
+      "loss": 1.6739,
+      "step": 7745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5196185286103543e-05,
+      "loss": 1.6344,
+      "step": 7746
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5194141689373297e-05,
+      "loss": 1.7625,
+      "step": 7747
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.519209809264305e-05,
+      "loss": 1.8164,
+      "step": 7748
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5190054495912808e-05,
+      "loss": 1.7531,
+      "step": 7749
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5188010899182563e-05,
+      "loss": 1.8186,
+      "step": 7750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5188010899182563e-05,
+      "loss": 1.79,
+      "step": 7751
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5185967302452317e-05,
+      "loss": 1.7018,
+      "step": 7752
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5183923705722071e-05,
+      "loss": 1.6673,
+      "step": 7753
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5181880108991825e-05,
+      "loss": 1.636,
+      "step": 7754
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5179836512261581e-05,
+      "loss": 1.6612,
+      "step": 7755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5177792915531335e-05,
+      "loss": 1.6993,
+      "step": 7756
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5175749318801089e-05,
+      "loss": 1.7724,
+      "step": 7757
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5173705722070846e-05,
+      "loss": 1.6587,
+      "step": 7758
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.51716621253406e-05,
+      "loss": 1.8544,
+      "step": 7759
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5169618528610356e-05,
+      "loss": 1.6524,
+      "step": 7760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.516757493188011e-05,
+      "loss": 1.8841,
+      "step": 7761
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5165531335149864e-05,
+      "loss": 1.735,
+      "step": 7762
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5163487738419618e-05,
+      "loss": 1.8879,
+      "step": 7763
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5161444141689373e-05,
+      "loss": 1.8132,
+      "step": 7764
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.515940054495913e-05,
+      "loss": 1.9678,
+      "step": 7765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5157356948228884e-05,
+      "loss": 1.6635,
+      "step": 7766
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5155313351498638e-05,
+      "loss": 1.7605,
+      "step": 7767
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5153269754768392e-05,
+      "loss": 1.6335,
+      "step": 7768
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5151226158038148e-05,
+      "loss": 1.6579,
+      "step": 7769
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5149182561307902e-05,
+      "loss": 1.7212,
+      "step": 7770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5147138964577656e-05,
+      "loss": 1.635,
+      "step": 7771
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.514509536784741e-05,
+      "loss": 1.7575,
+      "step": 7772
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5143051771117167e-05,
+      "loss": 1.674,
+      "step": 7773
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5141008174386923e-05,
+      "loss": 1.7304,
+      "step": 7774
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5138964577656677e-05,
+      "loss": 1.8215,
+      "step": 7775
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.513692098092643e-05,
+      "loss": 1.6226,
+      "step": 7776
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5134877384196185e-05,
+      "loss": 1.7761,
+      "step": 7777
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.513283378746594e-05,
+      "loss": 1.6143,
+      "step": 7778
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5130790190735694e-05,
+      "loss": 1.7725,
+      "step": 7779
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5128746594005451e-05,
+      "loss": 1.727,
+      "step": 7780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5126702997275205e-05,
+      "loss": 1.7526,
+      "step": 7781
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.512465940054496e-05,
+      "loss": 1.7337,
+      "step": 7782
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5122615803814715e-05,
+      "loss": 1.7643,
+      "step": 7783
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5120572207084469e-05,
+      "loss": 1.6098,
+      "step": 7784
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5118528610354223e-05,
+      "loss": 1.5846,
+      "step": 7785
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5116485013623977e-05,
+      "loss": 1.7112,
+      "step": 7786
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5114441416893732e-05,
+      "loss": 1.7214,
+      "step": 7787
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.511239782016349e-05,
+      "loss": 1.6345,
+      "step": 7788
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5110354223433244e-05,
+      "loss": 1.6663,
+      "step": 7789
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5108310626702998e-05,
+      "loss": 1.7588,
+      "step": 7790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5106267029972752e-05,
+      "loss": 1.7079,
+      "step": 7791
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5104223433242507e-05,
+      "loss": 1.6908,
+      "step": 7792
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5102179836512261e-05,
+      "loss": 1.6375,
+      "step": 7793
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5100136239782015e-05,
+      "loss": 1.7629,
+      "step": 7794
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5098092643051772e-05,
+      "loss": 1.815,
+      "step": 7795
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5096049046321528e-05,
+      "loss": 1.6414,
+      "step": 7796
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5094005449591282e-05,
+      "loss": 1.7414,
+      "step": 7797
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5091961852861036e-05,
+      "loss": 1.7681,
+      "step": 7798
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.508991825613079e-05,
+      "loss": 1.8229,
+      "step": 7799
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5087874659400544e-05,
+      "loss": 1.7321,
+      "step": 7800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.50858310626703e-05,
+      "loss": 1.6832,
+      "step": 7801
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5083787465940057e-05,
+      "loss": 1.9359,
+      "step": 7802
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.508174386920981e-05,
+      "loss": 1.9541,
+      "step": 7803
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5079700272479565e-05,
+      "loss": 1.8284,
+      "step": 7804
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.507765667574932e-05,
+      "loss": 1.8539,
+      "step": 7805
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5075613079019074e-05,
+      "loss": 1.73,
+      "step": 7806
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5073569482288828e-05,
+      "loss": 1.76,
+      "step": 7807
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5071525885558582e-05,
+      "loss": 1.8306,
+      "step": 7808
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5069482288828336e-05,
+      "loss": 1.6437,
+      "step": 7809
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5067438692098095e-05,
+      "loss": 1.7848,
+      "step": 7810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5065395095367849e-05,
+      "loss": 1.7137,
+      "step": 7811
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5063351498637603e-05,
+      "loss": 1.7213,
+      "step": 7812
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5061307901907357e-05,
+      "loss": 1.6976,
+      "step": 7813
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5059264305177112e-05,
+      "loss": 1.7354,
+      "step": 7814
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5057220708446866e-05,
+      "loss": 1.7823,
+      "step": 7815
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.505517711171662e-05,
+      "loss": 1.6884,
+      "step": 7816
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5053133514986378e-05,
+      "loss": 1.6839,
+      "step": 7817
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5051089918256132e-05,
+      "loss": 1.8595,
+      "step": 7818
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5049046321525887e-05,
+      "loss": 1.6409,
+      "step": 7819
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5047002724795641e-05,
+      "loss": 1.6612,
+      "step": 7820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5044959128065395e-05,
+      "loss": 1.7191,
+      "step": 7821
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5042915531335149e-05,
+      "loss": 1.8477,
+      "step": 7822
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5040871934604905e-05,
+      "loss": 1.6654,
+      "step": 7823
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5038828337874659e-05,
+      "loss": 1.6571,
+      "step": 7824
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5036784741144416e-05,
+      "loss": 1.6567,
+      "step": 7825
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.503474114441417e-05,
+      "loss": 1.6559,
+      "step": 7826
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5032697547683924e-05,
+      "loss": 1.6702,
+      "step": 7827
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.503065395095368e-05,
+      "loss": 1.6308,
+      "step": 7828
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5028610354223433e-05,
+      "loss": 1.8297,
+      "step": 7829
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5026566757493187e-05,
+      "loss": 1.7628,
+      "step": 7830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5024523160762941e-05,
+      "loss": 1.7121,
+      "step": 7831
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.50224795640327e-05,
+      "loss": 1.7346,
+      "step": 7832
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5020435967302454e-05,
+      "loss": 1.8002,
+      "step": 7833
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5018392370572208e-05,
+      "loss": 1.7645,
+      "step": 7834
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5016348773841962e-05,
+      "loss": 1.84,
+      "step": 7835
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5014305177111716e-05,
+      "loss": 1.7416,
+      "step": 7836
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5012261580381472e-05,
+      "loss": 1.6816,
+      "step": 7837
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5010217983651226e-05,
+      "loss": 1.675,
+      "step": 7838
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.500817438692098e-05,
+      "loss": 1.6638,
+      "step": 7839
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5006130790190737e-05,
+      "loss": 1.9543,
+      "step": 7840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5004087193460493e-05,
+      "loss": 1.7774,
+      "step": 7841
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5002043596730246e-05,
+      "loss": 1.7599,
+      "step": 7842
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5e-05,
+      "loss": 1.7954,
+      "step": 7843
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4997956403269754e-05,
+      "loss": 1.6312,
+      "step": 7844
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.499591280653951e-05,
+      "loss": 1.7334,
+      "step": 7845
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4993869209809266e-05,
+      "loss": 1.6922,
+      "step": 7846
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.499182561307902e-05,
+      "loss": 1.7559,
+      "step": 7847
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4989782016348773e-05,
+      "loss": 1.5989,
+      "step": 7848
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4987738419618529e-05,
+      "loss": 1.7138,
+      "step": 7849
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4985694822888285e-05,
+      "loss": 1.7279,
+      "step": 7850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4983651226158039e-05,
+      "loss": 1.6393,
+      "step": 7851
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4981607629427793e-05,
+      "loss": 1.5143,
+      "step": 7852
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4979564032697548e-05,
+      "loss": 1.7404,
+      "step": 7853
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4977520435967302e-05,
+      "loss": 1.5499,
+      "step": 7854
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4975476839237058e-05,
+      "loss": 1.8188,
+      "step": 7855
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4973433242506812e-05,
+      "loss": 1.679,
+      "step": 7856
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4971389645776567e-05,
+      "loss": 1.7396,
+      "step": 7857
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4969346049046321e-05,
+      "loss": 1.5808,
+      "step": 7858
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4967302452316077e-05,
+      "loss": 1.8154,
+      "step": 7859
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4965258855585833e-05,
+      "loss": 1.5655,
+      "step": 7860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4963215258855587e-05,
+      "loss": 1.5683,
+      "step": 7861
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.496117166212534e-05,
+      "loss": 1.6719,
+      "step": 7862
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4959128065395094e-05,
+      "loss": 1.6471,
+      "step": 7863
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4957084468664852e-05,
+      "loss": 1.846,
+      "step": 7864
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4955040871934606e-05,
+      "loss": 1.7107,
+      "step": 7865
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.495299727520436e-05,
+      "loss": 1.6278,
+      "step": 7866
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4950953678474114e-05,
+      "loss": 1.5945,
+      "step": 7867
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4948910081743871e-05,
+      "loss": 1.7226,
+      "step": 7868
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4946866485013625e-05,
+      "loss": 1.6129,
+      "step": 7869
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4944822888283379e-05,
+      "loss": 1.8168,
+      "step": 7870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4942779291553134e-05,
+      "loss": 1.6883,
+      "step": 7871
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4940735694822888e-05,
+      "loss": 1.8031,
+      "step": 7872
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4938692098092644e-05,
+      "loss": 1.5947,
+      "step": 7873
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4936648501362398e-05,
+      "loss": 1.6927,
+      "step": 7874
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4934604904632154e-05,
+      "loss": 1.8546,
+      "step": 7875
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4932561307901907e-05,
+      "loss": 1.7081,
+      "step": 7876
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4930517711171663e-05,
+      "loss": 1.6371,
+      "step": 7877
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4928474114441417e-05,
+      "loss": 1.574,
+      "step": 7878
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4926430517711173e-05,
+      "loss": 1.599,
+      "step": 7879
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4924386920980927e-05,
+      "loss": 1.6809,
+      "step": 7880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.492234332425068e-05,
+      "loss": 1.6967,
+      "step": 7881
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4920299727520436e-05,
+      "loss": 1.6724,
+      "step": 7882
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4918256130790192e-05,
+      "loss": 1.7102,
+      "step": 7883
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4916212534059946e-05,
+      "loss": 1.6643,
+      "step": 7884
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.49141689373297e-05,
+      "loss": 1.741,
+      "step": 7885
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4912125340599457e-05,
+      "loss": 1.7042,
+      "step": 7886
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4910081743869211e-05,
+      "loss": 1.7297,
+      "step": 7887
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4908038147138965e-05,
+      "loss": 1.7823,
+      "step": 7888
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4905994550408719e-05,
+      "loss": 1.8382,
+      "step": 7889
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4903950953678474e-05,
+      "loss": 1.7897,
+      "step": 7890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.490190735694823e-05,
+      "loss": 1.7514,
+      "step": 7891
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4899863760217984e-05,
+      "loss": 1.6581,
+      "step": 7892
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4897820163487738e-05,
+      "loss": 1.6481,
+      "step": 7893
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4895776566757494e-05,
+      "loss": 1.6118,
+      "step": 7894
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.489373297002725e-05,
+      "loss": 1.6725,
+      "step": 7895
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4891689373297003e-05,
+      "loss": 1.8177,
+      "step": 7896
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4889645776566757e-05,
+      "loss": 1.7466,
+      "step": 7897
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4887602179836513e-05,
+      "loss": 1.7016,
+      "step": 7898
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4885558583106267e-05,
+      "loss": 1.7211,
+      "step": 7899
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4883514986376022e-05,
+      "loss": 1.7382,
+      "step": 7900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4881471389645778e-05,
+      "loss": 1.4889,
+      "step": 7901
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4879427792915532e-05,
+      "loss": 1.74,
+      "step": 7902
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4877384196185286e-05,
+      "loss": 1.8159,
+      "step": 7903
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4875340599455041e-05,
+      "loss": 1.8586,
+      "step": 7904
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4873297002724797e-05,
+      "loss": 1.6056,
+      "step": 7905
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4871253405994551e-05,
+      "loss": 1.6954,
+      "step": 7906
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4869209809264305e-05,
+      "loss": 1.698,
+      "step": 7907
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4867166212534059e-05,
+      "loss": 1.6754,
+      "step": 7908
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4865122615803816e-05,
+      "loss": 1.7325,
+      "step": 7909
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.486307901907357e-05,
+      "loss": 1.7976,
+      "step": 7910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4861035422343324e-05,
+      "loss": 1.8175,
+      "step": 7911
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.485899182561308e-05,
+      "loss": 1.5737,
+      "step": 7912
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4856948228882835e-05,
+      "loss": 1.6539,
+      "step": 7913
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.485490463215259e-05,
+      "loss": 1.6195,
+      "step": 7914
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4852861035422343e-05,
+      "loss": 1.5426,
+      "step": 7915
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4850817438692099e-05,
+      "loss": 1.6394,
+      "step": 7916
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4848773841961853e-05,
+      "loss": 1.8201,
+      "step": 7917
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4846730245231608e-05,
+      "loss": 1.7153,
+      "step": 7918
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4844686648501362e-05,
+      "loss": 1.6477,
+      "step": 7919
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4842643051771118e-05,
+      "loss": 1.6687,
+      "step": 7920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4840599455040872e-05,
+      "loss": 1.6353,
+      "step": 7921
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4838555858310628e-05,
+      "loss": 1.7597,
+      "step": 7922
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4836512261580382e-05,
+      "loss": 1.6895,
+      "step": 7923
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4834468664850137e-05,
+      "loss": 1.6614,
+      "step": 7924
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4832425068119891e-05,
+      "loss": 1.7673,
+      "step": 7925
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4830381471389645e-05,
+      "loss": 1.5924,
+      "step": 7926
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4828337874659402e-05,
+      "loss": 1.6946,
+      "step": 7927
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4826294277929156e-05,
+      "loss": 1.7635,
+      "step": 7928
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.482425068119891e-05,
+      "loss": 1.6309,
+      "step": 7929
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4822207084468664e-05,
+      "loss": 1.7779,
+      "step": 7930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.482016348773842e-05,
+      "loss": 1.9684,
+      "step": 7931
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4818119891008175e-05,
+      "loss": 1.7779,
+      "step": 7932
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.481607629427793e-05,
+      "loss": 1.7251,
+      "step": 7933
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4814032697547683e-05,
+      "loss": 1.4682,
+      "step": 7934
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4811989100817439e-05,
+      "loss": 1.6046,
+      "step": 7935
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4809945504087195e-05,
+      "loss": 1.6476,
+      "step": 7936
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4807901907356949e-05,
+      "loss": 1.788,
+      "step": 7937
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4805858310626702e-05,
+      "loss": 1.8031,
+      "step": 7938
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4803814713896458e-05,
+      "loss": 1.6439,
+      "step": 7939
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4801771117166214e-05,
+      "loss": 1.7234,
+      "step": 7940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4799727520435968e-05,
+      "loss": 1.738,
+      "step": 7941
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4797683923705723e-05,
+      "loss": 1.7351,
+      "step": 7942
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4795640326975477e-05,
+      "loss": 1.6861,
+      "step": 7943
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4793596730245231e-05,
+      "loss": 1.7034,
+      "step": 7944
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4791553133514987e-05,
+      "loss": 1.736,
+      "step": 7945
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4789509536784742e-05,
+      "loss": 1.636,
+      "step": 7946
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4787465940054496e-05,
+      "loss": 1.6747,
+      "step": 7947
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.478542234332425e-05,
+      "loss": 1.6476,
+      "step": 7948
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4783378746594006e-05,
+      "loss": 1.6556,
+      "step": 7949
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4781335149863762e-05,
+      "loss": 1.6963,
+      "step": 7950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4779291553133516e-05,
+      "loss": 1.7179,
+      "step": 7951
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.477724795640327e-05,
+      "loss": 1.7191,
+      "step": 7952
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4775204359673023e-05,
+      "loss": 1.8658,
+      "step": 7953
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.477316076294278e-05,
+      "loss": 1.6843,
+      "step": 7954
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4771117166212535e-05,
+      "loss": 1.8066,
+      "step": 7955
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4769073569482289e-05,
+      "loss": 1.7902,
+      "step": 7956
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4767029972752044e-05,
+      "loss": 1.6282,
+      "step": 7957
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4764986376021798e-05,
+      "loss": 1.672,
+      "step": 7958
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4762942779291554e-05,
+      "loss": 1.9023,
+      "step": 7959
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4760899182561308e-05,
+      "loss": 1.5634,
+      "step": 7960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4758855585831063e-05,
+      "loss": 1.6986,
+      "step": 7961
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4756811989100817e-05,
+      "loss": 1.7622,
+      "step": 7962
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4754768392370573e-05,
+      "loss": 1.7409,
+      "step": 7963
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4752724795640327e-05,
+      "loss": 1.7971,
+      "step": 7964
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4750681198910083e-05,
+      "loss": 1.698,
+      "step": 7965
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4748637602179836e-05,
+      "loss": 1.6474,
+      "step": 7966
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.474659400544959e-05,
+      "loss": 1.8995,
+      "step": 7967
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4744550408719348e-05,
+      "loss": 1.7123,
+      "step": 7968
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4742506811989102e-05,
+      "loss": 1.698,
+      "step": 7969
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4740463215258856e-05,
+      "loss": 1.7018,
+      "step": 7970
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.473841961852861e-05,
+      "loss": 1.5779,
+      "step": 7971
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4736376021798367e-05,
+      "loss": 1.7584,
+      "step": 7972
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.473433242506812e-05,
+      "loss": 1.7379,
+      "step": 7973
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4732288828337875e-05,
+      "loss": 1.761,
+      "step": 7974
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4730245231607629e-05,
+      "loss": 1.662,
+      "step": 7975
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4728201634877384e-05,
+      "loss": 1.6959,
+      "step": 7976
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.472615803814714e-05,
+      "loss": 1.6046,
+      "step": 7977
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4724114441416894e-05,
+      "loss": 1.8243,
+      "step": 7978
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4722070844686648e-05,
+      "loss": 1.6977,
+      "step": 7979
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4720027247956403e-05,
+      "loss": 1.5991,
+      "step": 7980
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4717983651226159e-05,
+      "loss": 1.7361,
+      "step": 7981
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4715940054495913e-05,
+      "loss": 1.6564,
+      "step": 7982
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4713896457765669e-05,
+      "loss": 1.9004,
+      "step": 7983
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4711852861035423e-05,
+      "loss": 1.691,
+      "step": 7984
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4709809264305177e-05,
+      "loss": 1.7643,
+      "step": 7985
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4707765667574932e-05,
+      "loss": 1.6276,
+      "step": 7986
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4705722070844688e-05,
+      "loss": 1.554,
+      "step": 7987
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4703678474114442e-05,
+      "loss": 1.6242,
+      "step": 7988
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4701634877384196e-05,
+      "loss": 1.8774,
+      "step": 7989
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4699591280653951e-05,
+      "loss": 1.6247,
+      "step": 7990
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4697547683923707e-05,
+      "loss": 1.6601,
+      "step": 7991
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4695504087193461e-05,
+      "loss": 1.8134,
+      "step": 7992
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4693460490463215e-05,
+      "loss": 1.697,
+      "step": 7993
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4691416893732969e-05,
+      "loss": 1.7288,
+      "step": 7994
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4689373297002726e-05,
+      "loss": 1.6016,
+      "step": 7995
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.468732970027248e-05,
+      "loss": 1.7217,
+      "step": 7996
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4685286103542234e-05,
+      "loss": 1.7724,
+      "step": 7997
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.468324250681199e-05,
+      "loss": 1.7421,
+      "step": 7998
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4681198910081745e-05,
+      "loss": 1.7264,
+      "step": 7999
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.46791553133515e-05,
+      "loss": 1.806,
+      "step": 8000
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 1.5377315282821655,
+      "eval_runtime": 721.2674,
+      "eval_samples_per_second": 5.754,
+      "eval_steps_per_second": 1.439,
+      "step": 8000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4677111716621253e-05,
+      "loss": 1.8313,
+      "step": 8001
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4675068119891009e-05,
+      "loss": 1.7691,
+      "step": 8002
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4673024523160763e-05,
+      "loss": 1.8084,
+      "step": 8003
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4670980926430518e-05,
+      "loss": 1.6779,
+      "step": 8004
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4668937329700272e-05,
+      "loss": 1.7661,
+      "step": 8005
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4666893732970028e-05,
+      "loss": 1.747,
+      "step": 8006
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4664850136239782e-05,
+      "loss": 1.6945,
+      "step": 8007
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4662806539509537e-05,
+      "loss": 1.8891,
+      "step": 8008
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4660762942779293e-05,
+      "loss": 1.745,
+      "step": 8009
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4660762942779293e-05,
+      "loss": 1.8254,
+      "step": 8010
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4658719346049047e-05,
+      "loss": 1.6737,
+      "step": 8011
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4656675749318801e-05,
+      "loss": 1.6699,
+      "step": 8012
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4654632152588555e-05,
+      "loss": 1.6396,
+      "step": 8013
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4652588555858312e-05,
+      "loss": 1.6821,
+      "step": 8014
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4650544959128066e-05,
+      "loss": 1.8347,
+      "step": 8015
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.464850136239782e-05,
+      "loss": 1.6668,
+      "step": 8016
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4646457765667574e-05,
+      "loss": 1.6304,
+      "step": 8017
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4644414168937331e-05,
+      "loss": 1.6942,
+      "step": 8018
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4642370572207085e-05,
+      "loss": 1.5964,
+      "step": 8019
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.464032697547684e-05,
+      "loss": 1.5793,
+      "step": 8020
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4638283378746593e-05,
+      "loss": 1.6812,
+      "step": 8021
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4636239782016349e-05,
+      "loss": 1.708,
+      "step": 8022
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4634196185286104e-05,
+      "loss": 1.7571,
+      "step": 8023
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4632152588555858e-05,
+      "loss": 1.7672,
+      "step": 8024
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4630108991825614e-05,
+      "loss": 1.7768,
+      "step": 8025
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4628065395095368e-05,
+      "loss": 1.6607,
+      "step": 8026
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4626021798365124e-05,
+      "loss": 1.6309,
+      "step": 8027
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4623978201634878e-05,
+      "loss": 1.9112,
+      "step": 8028
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4621934604904633e-05,
+      "loss": 1.8924,
+      "step": 8029
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4619891008174387e-05,
+      "loss": 1.7947,
+      "step": 8030
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4617847411444141e-05,
+      "loss": 1.6591,
+      "step": 8031
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4615803814713897e-05,
+      "loss": 1.6762,
+      "step": 8032
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4613760217983652e-05,
+      "loss": 1.7123,
+      "step": 8033
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4611716621253406e-05,
+      "loss": 1.9027,
+      "step": 8034
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.460967302452316e-05,
+      "loss": 1.6074,
+      "step": 8035
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4607629427792916e-05,
+      "loss": 1.6599,
+      "step": 8036
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4605585831062671e-05,
+      "loss": 1.6212,
+      "step": 8037
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4603542234332425e-05,
+      "loss": 1.6104,
+      "step": 8038
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.460149863760218e-05,
+      "loss": 1.7199,
+      "step": 8039
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4599455040871935e-05,
+      "loss": 1.649,
+      "step": 8040
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.459741144414169e-05,
+      "loss": 1.6906,
+      "step": 8041
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4595367847411445e-05,
+      "loss": 1.5435,
+      "step": 8042
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4593324250681198e-05,
+      "loss": 1.8131,
+      "step": 8043
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4591280653950954e-05,
+      "loss": 1.7833,
+      "step": 8044
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.458923705722071e-05,
+      "loss": 1.6301,
+      "step": 8045
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4587193460490464e-05,
+      "loss": 1.6697,
+      "step": 8046
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4585149863760218e-05,
+      "loss": 1.6421,
+      "step": 8047
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4583106267029973e-05,
+      "loss": 1.6637,
+      "step": 8048
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4581062670299727e-05,
+      "loss": 1.6662,
+      "step": 8049
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4579019073569483e-05,
+      "loss": 1.736,
+      "step": 8050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4576975476839238e-05,
+      "loss": 1.6754,
+      "step": 8051
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4574931880108992e-05,
+      "loss": 1.7168,
+      "step": 8052
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4572888283378746e-05,
+      "loss": 1.7323,
+      "step": 8053
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4570844686648502e-05,
+      "loss": 1.7364,
+      "step": 8054
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4568801089918258e-05,
+      "loss": 1.724,
+      "step": 8055
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4566757493188012e-05,
+      "loss": 1.6053,
+      "step": 8056
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4564713896457765e-05,
+      "loss": 1.7973,
+      "step": 8057
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.456267029972752e-05,
+      "loss": 1.7438,
+      "step": 8058
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4560626702997277e-05,
+      "loss": 1.6178,
+      "step": 8059
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.455858310626703e-05,
+      "loss": 1.758,
+      "step": 8060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4556539509536785e-05,
+      "loss": 1.6674,
+      "step": 8061
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4554495912806539e-05,
+      "loss": 1.6119,
+      "step": 8062
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4552452316076296e-05,
+      "loss": 1.6418,
+      "step": 8063
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.455040871934605e-05,
+      "loss": 1.7198,
+      "step": 8064
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4548365122615804e-05,
+      "loss": 1.7122,
+      "step": 8065
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.454632152588556e-05,
+      "loss": 1.6848,
+      "step": 8066
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4544277929155313e-05,
+      "loss": 1.7127,
+      "step": 8067
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4542234332425069e-05,
+      "loss": 1.7507,
+      "step": 8068
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4540190735694823e-05,
+      "loss": 1.6961,
+      "step": 8069
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4538147138964579e-05,
+      "loss": 1.6505,
+      "step": 8070
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4536103542234332e-05,
+      "loss": 1.6179,
+      "step": 8071
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4534059945504088e-05,
+      "loss": 1.7215,
+      "step": 8072
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4532016348773842e-05,
+      "loss": 1.6804,
+      "step": 8073
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4529972752043598e-05,
+      "loss": 1.6871,
+      "step": 8074
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4527929155313352e-05,
+      "loss": 1.7006,
+      "step": 8075
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4525885558583106e-05,
+      "loss": 1.6048,
+      "step": 8076
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4523841961852861e-05,
+      "loss": 1.7403,
+      "step": 8077
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4521798365122617e-05,
+      "loss": 1.724,
+      "step": 8078
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.451975476839237e-05,
+      "loss": 1.7517,
+      "step": 8079
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4517711171662125e-05,
+      "loss": 1.6307,
+      "step": 8080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4515667574931882e-05,
+      "loss": 1.7142,
+      "step": 8081
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4513623978201636e-05,
+      "loss": 1.6222,
+      "step": 8082
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.451158038147139e-05,
+      "loss": 1.7284,
+      "step": 8083
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4509536784741144e-05,
+      "loss": 1.7683,
+      "step": 8084
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.45074931880109e-05,
+      "loss": 1.6019,
+      "step": 8085
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4505449591280655e-05,
+      "loss": 1.6994,
+      "step": 8086
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4503405994550409e-05,
+      "loss": 1.4732,
+      "step": 8087
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4501362397820163e-05,
+      "loss": 1.6665,
+      "step": 8088
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4499318801089919e-05,
+      "loss": 1.7097,
+      "step": 8089
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4497275204359674e-05,
+      "loss": 1.6893,
+      "step": 8090
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4495231607629428e-05,
+      "loss": 1.6563,
+      "step": 8091
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4493188010899182e-05,
+      "loss": 1.7067,
+      "step": 8092
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4491144414168938e-05,
+      "loss": 1.7672,
+      "step": 8093
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4489100817438692e-05,
+      "loss": 1.76,
+      "step": 8094
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4487057220708447e-05,
+      "loss": 1.7317,
+      "step": 8095
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4485013623978203e-05,
+      "loss": 1.6727,
+      "step": 8096
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4482970027247957e-05,
+      "loss": 1.6964,
+      "step": 8097
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4480926430517711e-05,
+      "loss": 1.6234,
+      "step": 8098
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4478882833787466e-05,
+      "loss": 1.7403,
+      "step": 8099
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4476839237057222e-05,
+      "loss": 1.7209,
+      "step": 8100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4474795640326976e-05,
+      "loss": 1.7546,
+      "step": 8101
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.447275204359673e-05,
+      "loss": 1.8565,
+      "step": 8102
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4470708446866484e-05,
+      "loss": 1.6869,
+      "step": 8103
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4468664850136241e-05,
+      "loss": 1.9088,
+      "step": 8104
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4466621253405995e-05,
+      "loss": 1.7243,
+      "step": 8105
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4464577656675749e-05,
+      "loss": 1.8327,
+      "step": 8106
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4462534059945505e-05,
+      "loss": 1.642,
+      "step": 8107
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.446049046321526e-05,
+      "loss": 1.9108,
+      "step": 8108
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4458446866485014e-05,
+      "loss": 1.772,
+      "step": 8109
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4456403269754768e-05,
+      "loss": 1.7525,
+      "step": 8110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4454359673024524e-05,
+      "loss": 1.5723,
+      "step": 8111
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4452316076294278e-05,
+      "loss": 1.6845,
+      "step": 8112
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4450272479564033e-05,
+      "loss": 1.6965,
+      "step": 8113
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4448228882833787e-05,
+      "loss": 1.6456,
+      "step": 8114
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4446185286103543e-05,
+      "loss": 1.6849,
+      "step": 8115
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4444141689373297e-05,
+      "loss": 1.5437,
+      "step": 8116
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4442098092643053e-05,
+      "loss": 1.6766,
+      "step": 8117
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4440054495912807e-05,
+      "loss": 1.6614,
+      "step": 8118
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4438010899182562e-05,
+      "loss": 1.7657,
+      "step": 8119
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4435967302452316e-05,
+      "loss": 1.7073,
+      "step": 8120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.443392370572207e-05,
+      "loss": 1.6882,
+      "step": 8121
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4431880108991827e-05,
+      "loss": 1.8021,
+      "step": 8122
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4429836512261581e-05,
+      "loss": 1.7835,
+      "step": 8123
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4427792915531335e-05,
+      "loss": 1.7004,
+      "step": 8124
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.442574931880109e-05,
+      "loss": 1.6925,
+      "step": 8125
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4423705722070847e-05,
+      "loss": 1.7347,
+      "step": 8126
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.44216621253406e-05,
+      "loss": 1.6829,
+      "step": 8127
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4419618528610354e-05,
+      "loss": 1.6335,
+      "step": 8128
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4417574931880108e-05,
+      "loss": 1.6684,
+      "step": 8129
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4415531335149864e-05,
+      "loss": 1.755,
+      "step": 8130
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.441348773841962e-05,
+      "loss": 1.6602,
+      "step": 8131
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4411444141689374e-05,
+      "loss": 1.6505,
+      "step": 8132
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4409400544959128e-05,
+      "loss": 1.7281,
+      "step": 8133
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4407356948228883e-05,
+      "loss": 1.6629,
+      "step": 8134
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4405313351498639e-05,
+      "loss": 1.819,
+      "step": 8135
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4403269754768393e-05,
+      "loss": 1.7143,
+      "step": 8136
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4401226158038148e-05,
+      "loss": 1.6434,
+      "step": 8137
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4399182561307902e-05,
+      "loss": 1.6854,
+      "step": 8138
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4397138964577656e-05,
+      "loss": 1.667,
+      "step": 8139
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4395095367847412e-05,
+      "loss": 1.6585,
+      "step": 8140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4393051771117167e-05,
+      "loss": 1.6275,
+      "step": 8141
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4391008174386921e-05,
+      "loss": 1.5841,
+      "step": 8142
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4388964577656675e-05,
+      "loss": 1.7808,
+      "step": 8143
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4386920980926431e-05,
+      "loss": 1.6662,
+      "step": 8144
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4384877384196187e-05,
+      "loss": 1.7165,
+      "step": 8145
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.438283378746594e-05,
+      "loss": 1.7511,
+      "step": 8146
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4380790190735695e-05,
+      "loss": 1.773,
+      "step": 8147
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.437874659400545e-05,
+      "loss": 1.7419,
+      "step": 8148
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4376702997275206e-05,
+      "loss": 1.8582,
+      "step": 8149
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.437465940054496e-05,
+      "loss": 1.7149,
+      "step": 8150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4372615803814714e-05,
+      "loss": 1.7259,
+      "step": 8151
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.437057220708447e-05,
+      "loss": 1.5806,
+      "step": 8152
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4368528610354223e-05,
+      "loss": 1.8681,
+      "step": 8153
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4366485013623979e-05,
+      "loss": 1.6901,
+      "step": 8154
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4364441416893733e-05,
+      "loss": 1.7961,
+      "step": 8155
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4362397820163488e-05,
+      "loss": 1.793,
+      "step": 8156
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4360354223433242e-05,
+      "loss": 1.7306,
+      "step": 8157
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4358310626702998e-05,
+      "loss": 1.7316,
+      "step": 8158
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4356267029972752e-05,
+      "loss": 1.6551,
+      "step": 8159
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4354223433242508e-05,
+      "loss": 1.63,
+      "step": 8160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4352179836512262e-05,
+      "loss": 1.6943,
+      "step": 8161
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4350136239782015e-05,
+      "loss": 1.6867,
+      "step": 8162
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4348092643051773e-05,
+      "loss": 1.6664,
+      "step": 8163
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4346049046321527e-05,
+      "loss": 1.5381,
+      "step": 8164
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.434400544959128e-05,
+      "loss": 1.6986,
+      "step": 8165
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4341961852861035e-05,
+      "loss": 1.6133,
+      "step": 8166
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4339918256130792e-05,
+      "loss": 1.5939,
+      "step": 8167
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4337874659400546e-05,
+      "loss": 1.5978,
+      "step": 8168
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.43358310626703e-05,
+      "loss": 1.7452,
+      "step": 8169
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4333787465940054e-05,
+      "loss": 1.7664,
+      "step": 8170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.433174386920981e-05,
+      "loss": 1.7254,
+      "step": 8171
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.4329700272479565e-05,
+      "loss": 1.6701,
+      "step": 8172
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4327656675749319e-05,
+      "loss": 1.7295,
+      "step": 8173
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4325613079019073e-05,
+      "loss": 1.669,
+      "step": 8174
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4323569482288829e-05,
+      "loss": 1.7395,
+      "step": 8175
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4321525885558584e-05,
+      "loss": 1.6764,
+      "step": 8176
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4319482288828338e-05,
+      "loss": 1.8954,
+      "step": 8177
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4317438692098094e-05,
+      "loss": 1.5839,
+      "step": 8178
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4315395095367848e-05,
+      "loss": 1.6805,
+      "step": 8179
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4313351498637602e-05,
+      "loss": 1.6082,
+      "step": 8180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4311307901907357e-05,
+      "loss": 1.7119,
+      "step": 8181
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4309264305177113e-05,
+      "loss": 1.6869,
+      "step": 8182
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4307220708446867e-05,
+      "loss": 1.7116,
+      "step": 8183
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.430517711171662e-05,
+      "loss": 1.8116,
+      "step": 8184
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4303133514986376e-05,
+      "loss": 1.6712,
+      "step": 8185
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4301089918256132e-05,
+      "loss": 1.7127,
+      "step": 8186
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4299046321525886e-05,
+      "loss": 1.6955,
+      "step": 8187
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.429700272479564e-05,
+      "loss": 1.6761,
+      "step": 8188
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4294959128065396e-05,
+      "loss": 1.694,
+      "step": 8189
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4292915531335151e-05,
+      "loss": 1.6392,
+      "step": 8190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4290871934604905e-05,
+      "loss": 1.6989,
+      "step": 8191
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4288828337874659e-05,
+      "loss": 1.6586,
+      "step": 8192
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4286784741144415e-05,
+      "loss": 1.7317,
+      "step": 8193
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.428474114441417e-05,
+      "loss": 1.759,
+      "step": 8194
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4282697547683924e-05,
+      "loss": 1.6842,
+      "step": 8195
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4280653950953678e-05,
+      "loss": 1.6755,
+      "step": 8196
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4278610354223434e-05,
+      "loss": 1.6885,
+      "step": 8197
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4276566757493188e-05,
+      "loss": 1.5699,
+      "step": 8198
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4274523160762943e-05,
+      "loss": 1.5302,
+      "step": 8199
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4272479564032697e-05,
+      "loss": 1.6451,
+      "step": 8200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4270435967302453e-05,
+      "loss": 1.6501,
+      "step": 8201
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4268392370572207e-05,
+      "loss": 1.6497,
+      "step": 8202
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4266348773841962e-05,
+      "loss": 1.6427,
+      "step": 8203
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4264305177111718e-05,
+      "loss": 1.7672,
+      "step": 8204
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4262261580381472e-05,
+      "loss": 1.7961,
+      "step": 8205
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4260217983651226e-05,
+      "loss": 1.6259,
+      "step": 8206
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.425817438692098e-05,
+      "loss": 1.6463,
+      "step": 8207
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4256130790190737e-05,
+      "loss": 1.6826,
+      "step": 8208
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4254087193460491e-05,
+      "loss": 1.6666,
+      "step": 8209
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4252043596730245e-05,
+      "loss": 1.5875,
+      "step": 8210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 1.6557,
+      "step": 8211
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4247956403269756e-05,
+      "loss": 1.7414,
+      "step": 8212
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.424591280653951e-05,
+      "loss": 1.7685,
+      "step": 8213
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4243869209809264e-05,
+      "loss": 1.9106,
+      "step": 8214
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4241825613079018e-05,
+      "loss": 1.6142,
+      "step": 8215
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4239782016348774e-05,
+      "loss": 1.6274,
+      "step": 8216
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.423773841961853e-05,
+      "loss": 1.6497,
+      "step": 8217
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4235694822888283e-05,
+      "loss": 1.9536,
+      "step": 8218
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4233651226158039e-05,
+      "loss": 1.6897,
+      "step": 8219
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4231607629427793e-05,
+      "loss": 1.6608,
+      "step": 8220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4229564032697549e-05,
+      "loss": 1.8365,
+      "step": 8221
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.4227520435967303e-05,
+      "loss": 1.8086,
+      "step": 8222
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4225476839237058e-05,
+      "loss": 1.6387,
+      "step": 8223
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4223433242506812e-05,
+      "loss": 1.7557,
+      "step": 8224
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4221389645776566e-05,
+      "loss": 1.7337,
+      "step": 8225
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4219346049046322e-05,
+      "loss": 1.7507,
+      "step": 8226
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4217302452316077e-05,
+      "loss": 1.7191,
+      "step": 8227
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4215258855585831e-05,
+      "loss": 1.6642,
+      "step": 8228
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4213215258855585e-05,
+      "loss": 1.6753,
+      "step": 8229
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4211171662125341e-05,
+      "loss": 1.6719,
+      "step": 8230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4209128065395096e-05,
+      "loss": 1.681,
+      "step": 8231
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.420708446866485e-05,
+      "loss": 1.664,
+      "step": 8232
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4205040871934604e-05,
+      "loss": 1.7993,
+      "step": 8233
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.420299727520436e-05,
+      "loss": 1.886,
+      "step": 8234
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4200953678474116e-05,
+      "loss": 1.6673,
+      "step": 8235
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.419891008174387e-05,
+      "loss": 1.6198,
+      "step": 8236
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4196866485013624e-05,
+      "loss": 1.7924,
+      "step": 8237
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4194822888283379e-05,
+      "loss": 1.7374,
+      "step": 8238
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4192779291553135e-05,
+      "loss": 1.8145,
+      "step": 8239
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4190735694822889e-05,
+      "loss": 1.7576,
+      "step": 8240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4188692098092643e-05,
+      "loss": 1.6889,
+      "step": 8241
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4186648501362398e-05,
+      "loss": 1.6417,
+      "step": 8242
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4184604904632152e-05,
+      "loss": 1.6496,
+      "step": 8243
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4182561307901908e-05,
+      "loss": 1.6807,
+      "step": 8244
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4180517711171663e-05,
+      "loss": 1.7683,
+      "step": 8245
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4178474114441417e-05,
+      "loss": 1.726,
+      "step": 8246
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4176430517711171e-05,
+      "loss": 1.6248,
+      "step": 8247
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4174386920980927e-05,
+      "loss": 1.7376,
+      "step": 8248
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4172343324250683e-05,
+      "loss": 1.6827,
+      "step": 8249
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4170299727520437e-05,
+      "loss": 1.617,
+      "step": 8250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.416825613079019e-05,
+      "loss": 1.5474,
+      "step": 8251
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4166212534059944e-05,
+      "loss": 1.7955,
+      "step": 8252
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4164168937329702e-05,
+      "loss": 1.7202,
+      "step": 8253
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4162125340599456e-05,
+      "loss": 1.6166,
+      "step": 8254
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.416008174386921e-05,
+      "loss": 1.617,
+      "step": 8255
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4158038147138964e-05,
+      "loss": 1.6444,
+      "step": 8256
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4155994550408721e-05,
+      "loss": 1.6637,
+      "step": 8257
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4153950953678475e-05,
+      "loss": 1.6466,
+      "step": 8258
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4151907356948229e-05,
+      "loss": 1.6949,
+      "step": 8259
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4149863760217984e-05,
+      "loss": 1.6367,
+      "step": 8260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4147820163487738e-05,
+      "loss": 1.5786,
+      "step": 8261
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4145776566757494e-05,
+      "loss": 1.6449,
+      "step": 8262
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4143732970027248e-05,
+      "loss": 1.7896,
+      "step": 8263
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4141689373297004e-05,
+      "loss": 1.7475,
+      "step": 8264
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4139645776566758e-05,
+      "loss": 1.6993,
+      "step": 8265
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4137602179836513e-05,
+      "loss": 1.6525,
+      "step": 8266
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4135558583106267e-05,
+      "loss": 1.7053,
+      "step": 8267
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4133514986376023e-05,
+      "loss": 1.6521,
+      "step": 8268
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4131471389645777e-05,
+      "loss": 1.59,
+      "step": 8269
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.412942779291553e-05,
+      "loss": 1.6554,
+      "step": 8270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4127384196185286e-05,
+      "loss": 1.8115,
+      "step": 8271
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4125340599455042e-05,
+      "loss": 1.854,
+      "step": 8272
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.4123297002724796e-05,
+      "loss": 1.719,
+      "step": 8273
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.412125340599455e-05,
+      "loss": 1.615,
+      "step": 8274
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4119209809264307e-05,
+      "loss": 1.7878,
+      "step": 8275
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4117166212534061e-05,
+      "loss": 1.7331,
+      "step": 8276
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4115122615803815e-05,
+      "loss": 1.7096,
+      "step": 8277
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4113079019073569e-05,
+      "loss": 1.6488,
+      "step": 8278
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4111035422343325e-05,
+      "loss": 1.6982,
+      "step": 8279
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.410899182561308e-05,
+      "loss": 1.7479,
+      "step": 8280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4106948228882834e-05,
+      "loss": 1.7005,
+      "step": 8281
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4104904632152588e-05,
+      "loss": 1.7568,
+      "step": 8282
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4102861035422344e-05,
+      "loss": 1.6343,
+      "step": 8283
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.41008174386921e-05,
+      "loss": 1.6208,
+      "step": 8284
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4098773841961853e-05,
+      "loss": 1.7785,
+      "step": 8285
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4096730245231609e-05,
+      "loss": 1.8053,
+      "step": 8286
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4094686648501363e-05,
+      "loss": 1.7598,
+      "step": 8287
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4092643051771117e-05,
+      "loss": 1.7749,
+      "step": 8288
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4090599455040872e-05,
+      "loss": 1.5887,
+      "step": 8289
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4088555858310628e-05,
+      "loss": 1.6747,
+      "step": 8290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4086512261580382e-05,
+      "loss": 1.6446,
+      "step": 8291
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4084468664850136e-05,
+      "loss": 1.7114,
+      "step": 8292
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4082425068119892e-05,
+      "loss": 1.6841,
+      "step": 8293
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4080381471389647e-05,
+      "loss": 1.7137,
+      "step": 8294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4078337874659401e-05,
+      "loss": 1.7256,
+      "step": 8295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4076294277929155e-05,
+      "loss": 1.61,
+      "step": 8296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4074250681198909e-05,
+      "loss": 1.5684,
+      "step": 8297
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4072207084468666e-05,
+      "loss": 1.7549,
+      "step": 8298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.407016348773842e-05,
+      "loss": 1.8336,
+      "step": 8299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4068119891008174e-05,
+      "loss": 1.7969,
+      "step": 8300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.406607629427793e-05,
+      "loss": 1.8961,
+      "step": 8301
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4064032697547685e-05,
+      "loss": 1.8666,
+      "step": 8302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.406198910081744e-05,
+      "loss": 1.7264,
+      "step": 8303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4059945504087193e-05,
+      "loss": 1.6102,
+      "step": 8304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4057901907356949e-05,
+      "loss": 1.7192,
+      "step": 8305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4055858310626703e-05,
+      "loss": 1.8844,
+      "step": 8306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4053814713896459e-05,
+      "loss": 1.7312,
+      "step": 8307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4051771117166212e-05,
+      "loss": 1.715,
+      "step": 8308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4049727520435968e-05,
+      "loss": 1.7459,
+      "step": 8309
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4047683923705722e-05,
+      "loss": 1.4537,
+      "step": 8310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4045640326975478e-05,
+      "loss": 1.6509,
+      "step": 8311
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4043596730245232e-05,
+      "loss": 1.7087,
+      "step": 8312
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4041553133514987e-05,
+      "loss": 1.8158,
+      "step": 8313
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4039509536784741e-05,
+      "loss": 1.7277,
+      "step": 8314
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4037465940054495e-05,
+      "loss": 1.5916,
+      "step": 8315
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4035422343324252e-05,
+      "loss": 1.822,
+      "step": 8316
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4033378746594006e-05,
+      "loss": 1.7116,
+      "step": 8317
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.403133514986376e-05,
+      "loss": 1.7455,
+      "step": 8318
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4029291553133514e-05,
+      "loss": 1.646,
+      "step": 8319
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4027247956403272e-05,
+      "loss": 1.7664,
+      "step": 8320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4025204359673026e-05,
+      "loss": 1.6171,
+      "step": 8321
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.402316076294278e-05,
+      "loss": 1.7226,
+      "step": 8322
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4021117166212533e-05,
+      "loss": 1.738,
+      "step": 8323
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.4019073569482289e-05,
+      "loss": 1.6449,
+      "step": 8324
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4017029972752045e-05,
+      "loss": 1.7998,
+      "step": 8325
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4014986376021799e-05,
+      "loss": 1.5575,
+      "step": 8326
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4012942779291554e-05,
+      "loss": 1.7107,
+      "step": 8327
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4010899182561308e-05,
+      "loss": 1.5873,
+      "step": 8328
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4008855585831064e-05,
+      "loss": 1.8641,
+      "step": 8329
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4006811989100818e-05,
+      "loss": 1.7439,
+      "step": 8330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4004768392370573e-05,
+      "loss": 1.7539,
+      "step": 8331
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4002724795640327e-05,
+      "loss": 1.7249,
+      "step": 8332
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4000681198910081e-05,
+      "loss": 1.6779,
+      "step": 8333
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3998637602179837e-05,
+      "loss": 1.6601,
+      "step": 8334
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3996594005449593e-05,
+      "loss": 1.7304,
+      "step": 8335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3994550408719346e-05,
+      "loss": 1.7024,
+      "step": 8336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.39925068119891e-05,
+      "loss": 1.7437,
+      "step": 8337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3990463215258856e-05,
+      "loss": 1.7415,
+      "step": 8338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3988419618528612e-05,
+      "loss": 1.6634,
+      "step": 8339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3986376021798366e-05,
+      "loss": 1.7222,
+      "step": 8340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.398433242506812e-05,
+      "loss": 1.8365,
+      "step": 8341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3982288828337875e-05,
+      "loss": 1.7706,
+      "step": 8342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.398024523160763e-05,
+      "loss": 1.6916,
+      "step": 8343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3978201634877385e-05,
+      "loss": 1.7825,
+      "step": 8344
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3976158038147139e-05,
+      "loss": 1.6859,
+      "step": 8345
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3974114441416894e-05,
+      "loss": 1.5973,
+      "step": 8346
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3972070844686648e-05,
+      "loss": 1.6542,
+      "step": 8347
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3970027247956404e-05,
+      "loss": 1.5927,
+      "step": 8348
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3967983651226158e-05,
+      "loss": 1.7057,
+      "step": 8349
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3965940054495913e-05,
+      "loss": 1.6666,
+      "step": 8350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3963896457765667e-05,
+      "loss": 1.7436,
+      "step": 8351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3961852861035423e-05,
+      "loss": 1.6798,
+      "step": 8352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3959809264305177e-05,
+      "loss": 1.8087,
+      "step": 8353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3957765667574933e-05,
+      "loss": 1.6787,
+      "step": 8354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3955722070844687e-05,
+      "loss": 1.867,
+      "step": 8355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.395367847411444e-05,
+      "loss": 1.7499,
+      "step": 8356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3951634877384198e-05,
+      "loss": 1.754,
+      "step": 8357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3949591280653952e-05,
+      "loss": 1.5198,
+      "step": 8358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3947547683923706e-05,
+      "loss": 1.8469,
+      "step": 8359
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.394550408719346e-05,
+      "loss": 1.649,
+      "step": 8360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3943460490463217e-05,
+      "loss": 1.7353,
+      "step": 8361
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3941416893732971e-05,
+      "loss": 1.6142,
+      "step": 8362
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3939373297002725e-05,
+      "loss": 1.7422,
+      "step": 8363
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3937329700272479e-05,
+      "loss": 1.7015,
+      "step": 8364
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3935286103542234e-05,
+      "loss": 1.9222,
+      "step": 8365
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.393324250681199e-05,
+      "loss": 1.733,
+      "step": 8366
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3931198910081744e-05,
+      "loss": 1.6583,
+      "step": 8367
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3929155313351498e-05,
+      "loss": 1.6698,
+      "step": 8368
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3927111716621254e-05,
+      "loss": 1.8316,
+      "step": 8369
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.392506811989101e-05,
+      "loss": 1.689,
+      "step": 8370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3923024523160763e-05,
+      "loss": 1.7871,
+      "step": 8371
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3920980926430519e-05,
+      "loss": 1.6842,
+      "step": 8372
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3918937329700273e-05,
+      "loss": 1.6672,
+      "step": 8373
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3916893732970027e-05,
+      "loss": 1.6771,
+      "step": 8374
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3914850136239782e-05,
+      "loss": 1.8317,
+      "step": 8375
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3912806539509538e-05,
+      "loss": 1.5353,
+      "step": 8376
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3910762942779292e-05,
+      "loss": 1.7621,
+      "step": 8377
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3908719346049046e-05,
+      "loss": 1.734,
+      "step": 8378
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3906675749318801e-05,
+      "loss": 1.7744,
+      "step": 8379
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3904632152588557e-05,
+      "loss": 1.6583,
+      "step": 8380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3902588555858311e-05,
+      "loss": 1.6469,
+      "step": 8381
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3900544959128065e-05,
+      "loss": 1.6951,
+      "step": 8382
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.389850136239782e-05,
+      "loss": 1.6551,
+      "step": 8383
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3896457765667576e-05,
+      "loss": 1.7531,
+      "step": 8384
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.389441416893733e-05,
+      "loss": 1.6354,
+      "step": 8385
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3892370572207084e-05,
+      "loss": 1.7746,
+      "step": 8386
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.389032697547684e-05,
+      "loss": 1.7769,
+      "step": 8387
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3888283378746595e-05,
+      "loss": 1.6546,
+      "step": 8388
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.388623978201635e-05,
+      "loss": 1.6386,
+      "step": 8389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3884196185286103e-05,
+      "loss": 1.6661,
+      "step": 8390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3882152588555859e-05,
+      "loss": 1.6072,
+      "step": 8391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3880108991825613e-05,
+      "loss": 1.7734,
+      "step": 8392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3878065395095368e-05,
+      "loss": 1.7176,
+      "step": 8393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3876021798365122e-05,
+      "loss": 1.781,
+      "step": 8394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3873978201634878e-05,
+      "loss": 1.7655,
+      "step": 8395
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3871934604904632e-05,
+      "loss": 1.7597,
+      "step": 8396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3869891008174388e-05,
+      "loss": 1.6879,
+      "step": 8397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3867847411444143e-05,
+      "loss": 1.675,
+      "step": 8398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3865803814713897e-05,
+      "loss": 1.8753,
+      "step": 8399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3863760217983651e-05,
+      "loss": 1.7316,
+      "step": 8400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3861716621253405e-05,
+      "loss": 1.7724,
+      "step": 8401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3859673024523162e-05,
+      "loss": 1.6663,
+      "step": 8402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3857629427792916e-05,
+      "loss": 1.5093,
+      "step": 8403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.385558583106267e-05,
+      "loss": 1.7129,
+      "step": 8404
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3853542234332424e-05,
+      "loss": 1.584,
+      "step": 8405
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3851498637602181e-05,
+      "loss": 1.6591,
+      "step": 8406
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3849455040871935e-05,
+      "loss": 1.5697,
+      "step": 8407
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.384741144414169e-05,
+      "loss": 1.7702,
+      "step": 8408
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3845367847411443e-05,
+      "loss": 1.6403,
+      "step": 8409
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3843324250681199e-05,
+      "loss": 1.6651,
+      "step": 8410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3841280653950955e-05,
+      "loss": 1.8179,
+      "step": 8411
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3839237057220708e-05,
+      "loss": 1.6858,
+      "step": 8412
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3837193460490464e-05,
+      "loss": 1.6074,
+      "step": 8413
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3835149863760218e-05,
+      "loss": 1.7808,
+      "step": 8414
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3833106267029974e-05,
+      "loss": 1.7841,
+      "step": 8415
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3831062670299728e-05,
+      "loss": 1.7436,
+      "step": 8416
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3829019073569483e-05,
+      "loss": 1.5605,
+      "step": 8417
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3826975476839237e-05,
+      "loss": 1.6769,
+      "step": 8418
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3824931880108991e-05,
+      "loss": 1.7202,
+      "step": 8419
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3822888283378747e-05,
+      "loss": 1.5251,
+      "step": 8420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3820844686648502e-05,
+      "loss": 1.5842,
+      "step": 8421
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3818801089918256e-05,
+      "loss": 1.6738,
+      "step": 8422
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.381675749318801e-05,
+      "loss": 1.7694,
+      "step": 8423
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3814713896457768e-05,
+      "loss": 1.7997,
+      "step": 8424
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3812670299727522e-05,
+      "loss": 1.6928,
+      "step": 8425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3810626702997275e-05,
+      "loss": 1.725,
+      "step": 8426
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.380858310626703e-05,
+      "loss": 1.8231,
+      "step": 8427
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3806539509536785e-05,
+      "loss": 1.7582,
+      "step": 8428
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.380449591280654e-05,
+      "loss": 1.7847,
+      "step": 8429
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3802452316076295e-05,
+      "loss": 1.7435,
+      "step": 8430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3800408719346049e-05,
+      "loss": 1.8739,
+      "step": 8431
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3798365122615804e-05,
+      "loss": 1.6914,
+      "step": 8432
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.379632152588556e-05,
+      "loss": 1.6005,
+      "step": 8433
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3794277929155314e-05,
+      "loss": 1.6262,
+      "step": 8434
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3792234332425068e-05,
+      "loss": 1.6723,
+      "step": 8435
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3790190735694823e-05,
+      "loss": 1.7054,
+      "step": 8436
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3788147138964577e-05,
+      "loss": 1.7643,
+      "step": 8437
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3786103542234333e-05,
+      "loss": 1.7052,
+      "step": 8438
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3784059945504089e-05,
+      "loss": 1.6445,
+      "step": 8439
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3782016348773842e-05,
+      "loss": 1.7264,
+      "step": 8440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3779972752043596e-05,
+      "loss": 1.7375,
+      "step": 8441
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3777929155313352e-05,
+      "loss": 1.6093,
+      "step": 8442
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3775885558583108e-05,
+      "loss": 1.6633,
+      "step": 8443
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3773841961852862e-05,
+      "loss": 1.761,
+      "step": 8444
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3771798365122616e-05,
+      "loss": 1.5781,
+      "step": 8445
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.376975476839237e-05,
+      "loss": 1.6272,
+      "step": 8446
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3767711171662127e-05,
+      "loss": 1.6261,
+      "step": 8447
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.376566757493188e-05,
+      "loss": 1.6755,
+      "step": 8448
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3763623978201635e-05,
+      "loss": 1.7351,
+      "step": 8449
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3761580381471389e-05,
+      "loss": 1.8325,
+      "step": 8450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3759536784741146e-05,
+      "loss": 1.7883,
+      "step": 8451
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.37574931880109e-05,
+      "loss": 1.7423,
+      "step": 8452
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3755449591280654e-05,
+      "loss": 1.7084,
+      "step": 8453
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.375340599455041e-05,
+      "loss": 1.9765,
+      "step": 8454
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3751362397820163e-05,
+      "loss": 1.7006,
+      "step": 8455
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3749318801089919e-05,
+      "loss": 1.7256,
+      "step": 8456
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3747275204359673e-05,
+      "loss": 1.6891,
+      "step": 8457
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3745231607629429e-05,
+      "loss": 1.6686,
+      "step": 8458
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3743188010899183e-05,
+      "loss": 1.8007,
+      "step": 8459
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3741144414168938e-05,
+      "loss": 1.6996,
+      "step": 8460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3739100817438692e-05,
+      "loss": 1.8149,
+      "step": 8461
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3737057220708448e-05,
+      "loss": 1.6626,
+      "step": 8462
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3735013623978202e-05,
+      "loss": 1.6224,
+      "step": 8463
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3732970027247956e-05,
+      "loss": 1.6252,
+      "step": 8464
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3730926430517713e-05,
+      "loss": 1.6119,
+      "step": 8465
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3728882833787467e-05,
+      "loss": 1.7679,
+      "step": 8466
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.372683923705722e-05,
+      "loss": 1.6471,
+      "step": 8467
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3724795640326975e-05,
+      "loss": 1.6823,
+      "step": 8468
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3722752043596732e-05,
+      "loss": 1.7135,
+      "step": 8469
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3720708446866486e-05,
+      "loss": 1.6819,
+      "step": 8470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.371866485013624e-05,
+      "loss": 1.5502,
+      "step": 8471
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3716621253405994e-05,
+      "loss": 1.6828,
+      "step": 8472
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.371457765667575e-05,
+      "loss": 1.8359,
+      "step": 8473
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3712534059945505e-05,
+      "loss": 1.6503,
+      "step": 8474
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3710490463215259e-05,
+      "loss": 1.6659,
+      "step": 8475
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3708446866485013e-05,
+      "loss": 1.9352,
+      "step": 8476
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3706403269754769e-05,
+      "loss": 1.736,
+      "step": 8477
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3704359673024524e-05,
+      "loss": 1.8998,
+      "step": 8478
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3702316076294278e-05,
+      "loss": 1.6796,
+      "step": 8479
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3700272479564034e-05,
+      "loss": 1.7066,
+      "step": 8480
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3698228882833788e-05,
+      "loss": 1.7542,
+      "step": 8481
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3696185286103542e-05,
+      "loss": 1.7188,
+      "step": 8482
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3694141689373297e-05,
+      "loss": 1.7125,
+      "step": 8483
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3692098092643053e-05,
+      "loss": 1.6987,
+      "step": 8484
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3690054495912807e-05,
+      "loss": 1.6842,
+      "step": 8485
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3688010899182561e-05,
+      "loss": 1.8712,
+      "step": 8486
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3685967302452317e-05,
+      "loss": 1.6236,
+      "step": 8487
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3683923705722072e-05,
+      "loss": 1.7363,
+      "step": 8488
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3681880108991826e-05,
+      "loss": 1.5742,
+      "step": 8489
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.367983651226158e-05,
+      "loss": 1.7711,
+      "step": 8490
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3677792915531334e-05,
+      "loss": 1.7142,
+      "step": 8491
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3675749318801091e-05,
+      "loss": 1.7989,
+      "step": 8492
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3673705722070845e-05,
+      "loss": 1.7386,
+      "step": 8493
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.36716621253406e-05,
+      "loss": 1.66,
+      "step": 8494
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3669618528610355e-05,
+      "loss": 1.5884,
+      "step": 8495
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.366757493188011e-05,
+      "loss": 1.7255,
+      "step": 8496
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3665531335149864e-05,
+      "loss": 1.597,
+      "step": 8497
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3663487738419618e-05,
+      "loss": 1.8101,
+      "step": 8498
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3661444141689374e-05,
+      "loss": 1.7287,
+      "step": 8499
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3659400544959128e-05,
+      "loss": 1.8407,
+      "step": 8500
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 1.5326870679855347,
+      "eval_runtime": 721.2859,
+      "eval_samples_per_second": 5.754,
+      "eval_steps_per_second": 1.439,
+      "step": 8500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3657356948228884e-05,
+      "loss": 1.784,
+      "step": 8501
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3655313351498637e-05,
+      "loss": 1.8304,
+      "step": 8502
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3653269754768393e-05,
+      "loss": 1.6242,
+      "step": 8503
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3651226158038147e-05,
+      "loss": 1.7162,
+      "step": 8504
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3649182561307903e-05,
+      "loss": 1.7215,
+      "step": 8505
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3647138964577657e-05,
+      "loss": 1.8058,
+      "step": 8506
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3645095367847412e-05,
+      "loss": 1.6394,
+      "step": 8507
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3643051771117166e-05,
+      "loss": 1.8311,
+      "step": 8508
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.364100817438692e-05,
+      "loss": 1.7199,
+      "step": 8509
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3638964577656677e-05,
+      "loss": 1.7859,
+      "step": 8510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3636920980926431e-05,
+      "loss": 1.7154,
+      "step": 8511
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3634877384196185e-05,
+      "loss": 1.6183,
+      "step": 8512
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.363283378746594e-05,
+      "loss": 1.6108,
+      "step": 8513
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3630790190735697e-05,
+      "loss": 1.8561,
+      "step": 8514
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.362874659400545e-05,
+      "loss": 1.5893,
+      "step": 8515
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3626702997275204e-05,
+      "loss": 1.7025,
+      "step": 8516
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3624659400544958e-05,
+      "loss": 1.69,
+      "step": 8517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3622615803814714e-05,
+      "loss": 1.7451,
+      "step": 8518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.362057220708447e-05,
+      "loss": 1.6752,
+      "step": 8519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3618528610354224e-05,
+      "loss": 1.6647,
+      "step": 8520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.361648501362398e-05,
+      "loss": 1.6264,
+      "step": 8521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3614441416893733e-05,
+      "loss": 1.7874,
+      "step": 8522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3612397820163489e-05,
+      "loss": 1.866,
+      "step": 8523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3610354223433243e-05,
+      "loss": 1.6848,
+      "step": 8524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3608310626702998e-05,
+      "loss": 1.927,
+      "step": 8525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3606267029972752e-05,
+      "loss": 1.6999,
+      "step": 8526
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3604223433242506e-05,
+      "loss": 1.7352,
+      "step": 8527
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3602179836512262e-05,
+      "loss": 1.7824,
+      "step": 8528
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3600136239782018e-05,
+      "loss": 1.6135,
+      "step": 8529
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3598092643051771e-05,
+      "loss": 1.8326,
+      "step": 8530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3596049046321525e-05,
+      "loss": 1.7358,
+      "step": 8531
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3594005449591281e-05,
+      "loss": 1.7325,
+      "step": 8532
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3591961852861037e-05,
+      "loss": 1.7533,
+      "step": 8533
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.358991825613079e-05,
+      "loss": 1.7716,
+      "step": 8534
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3587874659400545e-05,
+      "loss": 1.7696,
+      "step": 8535
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.35858310626703e-05,
+      "loss": 1.6227,
+      "step": 8536
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3583787465940056e-05,
+      "loss": 1.7698,
+      "step": 8537
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.358174386920981e-05,
+      "loss": 1.7683,
+      "step": 8538
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3579700272479564e-05,
+      "loss": 1.641,
+      "step": 8539
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.357765667574932e-05,
+      "loss": 1.8372,
+      "step": 8540
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3575613079019073e-05,
+      "loss": 1.8146,
+      "step": 8541
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3573569482288829e-05,
+      "loss": 1.5272,
+      "step": 8542
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3571525885558583e-05,
+      "loss": 1.5765,
+      "step": 8543
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3569482288828338e-05,
+      "loss": 1.7709,
+      "step": 8544
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3567438692098092e-05,
+      "loss": 1.7013,
+      "step": 8545
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3565395095367848e-05,
+      "loss": 1.6861,
+      "step": 8546
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3563351498637602e-05,
+      "loss": 1.5362,
+      "step": 8547
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3561307901907358e-05,
+      "loss": 1.7497,
+      "step": 8548
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3559264305177112e-05,
+      "loss": 1.7362,
+      "step": 8549
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3557220708446865e-05,
+      "loss": 1.6451,
+      "step": 8550
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3555177111716623e-05,
+      "loss": 1.7178,
+      "step": 8551
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3553133514986377e-05,
+      "loss": 1.7375,
+      "step": 8552
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.355108991825613e-05,
+      "loss": 1.5335,
+      "step": 8553
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3549046321525885e-05,
+      "loss": 1.7577,
+      "step": 8554
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3547002724795642e-05,
+      "loss": 1.7758,
+      "step": 8555
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3544959128065396e-05,
+      "loss": 1.7806,
+      "step": 8556
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.354291553133515e-05,
+      "loss": 1.8265,
+      "step": 8557
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3540871934604904e-05,
+      "loss": 1.628,
+      "step": 8558
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.353882833787466e-05,
+      "loss": 1.6704,
+      "step": 8559
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3536784741144415e-05,
+      "loss": 1.8507,
+      "step": 8560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3534741144414169e-05,
+      "loss": 1.7697,
+      "step": 8561
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3532697547683925e-05,
+      "loss": 1.5732,
+      "step": 8562
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3530653950953679e-05,
+      "loss": 1.6669,
+      "step": 8563
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3528610354223434e-05,
+      "loss": 1.8755,
+      "step": 8564
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3526566757493188e-05,
+      "loss": 1.7344,
+      "step": 8565
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3524523160762944e-05,
+      "loss": 1.8305,
+      "step": 8566
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3522479564032698e-05,
+      "loss": 1.7585,
+      "step": 8567
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3520435967302452e-05,
+      "loss": 1.8261,
+      "step": 8568
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3518392370572207e-05,
+      "loss": 1.727,
+      "step": 8569
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3516348773841963e-05,
+      "loss": 1.7973,
+      "step": 8570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3514305177111717e-05,
+      "loss": 1.7398,
+      "step": 8571
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.351226158038147e-05,
+      "loss": 1.6106,
+      "step": 8572
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3510217983651226e-05,
+      "loss": 1.7421,
+      "step": 8573
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3508174386920982e-05,
+      "loss": 1.7844,
+      "step": 8574
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3506130790190736e-05,
+      "loss": 1.636,
+      "step": 8575
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.350408719346049e-05,
+      "loss": 1.5161,
+      "step": 8576
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3502043596730246e-05,
+      "loss": 1.6947,
+      "step": 8577
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.6662,
+      "step": 8578
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3497956403269755e-05,
+      "loss": 1.6311,
+      "step": 8579
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3495912806539509e-05,
+      "loss": 1.5541,
+      "step": 8580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3493869209809265e-05,
+      "loss": 1.7417,
+      "step": 8581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.349182561307902e-05,
+      "loss": 1.6914,
+      "step": 8582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3489782016348774e-05,
+      "loss": 1.7263,
+      "step": 8583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3487738419618528e-05,
+      "loss": 1.6955,
+      "step": 8584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3485694822888284e-05,
+      "loss": 1.7306,
+      "step": 8585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3483651226158038e-05,
+      "loss": 1.678,
+      "step": 8586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3481607629427793e-05,
+      "loss": 1.6885,
+      "step": 8587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3479564032697547e-05,
+      "loss": 1.7426,
+      "step": 8588
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3477520435967303e-05,
+      "loss": 1.6534,
+      "step": 8589
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3475476839237057e-05,
+      "loss": 1.6701,
+      "step": 8590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3473433242506813e-05,
+      "loss": 1.5787,
+      "step": 8591
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3471389645776568e-05,
+      "loss": 1.8261,
+      "step": 8592
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3469346049046322e-05,
+      "loss": 1.7358,
+      "step": 8593
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3467302452316076e-05,
+      "loss": 1.6921,
+      "step": 8594
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.346525885558583e-05,
+      "loss": 1.6277,
+      "step": 8595
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3463215258855587e-05,
+      "loss": 1.6327,
+      "step": 8596
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3461171662125341e-05,
+      "loss": 1.7357,
+      "step": 8597
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3459128065395095e-05,
+      "loss": 1.6204,
+      "step": 8598
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3457084468664849e-05,
+      "loss": 1.763,
+      "step": 8599
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3455040871934606e-05,
+      "loss": 1.7394,
+      "step": 8600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.345299727520436e-05,
+      "loss": 1.7749,
+      "step": 8601
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3450953678474114e-05,
+      "loss": 1.61,
+      "step": 8602
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.344891008174387e-05,
+      "loss": 1.726,
+      "step": 8603
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3446866485013624e-05,
+      "loss": 1.8984,
+      "step": 8604
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.344482288828338e-05,
+      "loss": 1.6646,
+      "step": 8605
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3442779291553133e-05,
+      "loss": 1.7548,
+      "step": 8606
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3440735694822889e-05,
+      "loss": 1.7805,
+      "step": 8607
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3438692098092643e-05,
+      "loss": 1.6246,
+      "step": 8608
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3436648501362399e-05,
+      "loss": 1.5994,
+      "step": 8609
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3434604904632153e-05,
+      "loss": 1.7161,
+      "step": 8610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3432561307901908e-05,
+      "loss": 1.669,
+      "step": 8611
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3430517711171662e-05,
+      "loss": 1.6296,
+      "step": 8612
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3428474114441416e-05,
+      "loss": 1.7231,
+      "step": 8613
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3426430517711172e-05,
+      "loss": 1.745,
+      "step": 8614
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3424386920980927e-05,
+      "loss": 1.7389,
+      "step": 8615
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3422343324250681e-05,
+      "loss": 1.7408,
+      "step": 8616
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3420299727520435e-05,
+      "loss": 1.6673,
+      "step": 8617
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3418256130790193e-05,
+      "loss": 1.683,
+      "step": 8618
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3416212534059947e-05,
+      "loss": 1.7539,
+      "step": 8619
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.34141689373297e-05,
+      "loss": 1.6638,
+      "step": 8620
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3412125340599454e-05,
+      "loss": 1.8068,
+      "step": 8621
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.341008174386921e-05,
+      "loss": 1.7409,
+      "step": 8622
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3408038147138966e-05,
+      "loss": 1.6825,
+      "step": 8623
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.340599455040872e-05,
+      "loss": 1.7753,
+      "step": 8624
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3403950953678474e-05,
+      "loss": 1.7393,
+      "step": 8625
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.340190735694823e-05,
+      "loss": 1.6619,
+      "step": 8626
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3399863760217985e-05,
+      "loss": 1.6179,
+      "step": 8627
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3397820163487739e-05,
+      "loss": 1.7141,
+      "step": 8628
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3395776566757493e-05,
+      "loss": 1.6099,
+      "step": 8629
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3393732970027248e-05,
+      "loss": 1.6954,
+      "step": 8630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3391689373297002e-05,
+      "loss": 1.719,
+      "step": 8631
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3389645776566758e-05,
+      "loss": 1.7548,
+      "step": 8632
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3387602179836514e-05,
+      "loss": 1.6245,
+      "step": 8633
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3385558583106267e-05,
+      "loss": 1.6134,
+      "step": 8634
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3383514986376021e-05,
+      "loss": 1.5855,
+      "step": 8635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3381471389645777e-05,
+      "loss": 1.6112,
+      "step": 8636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3379427792915533e-05,
+      "loss": 1.8044,
+      "step": 8637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3377384196185287e-05,
+      "loss": 1.5067,
+      "step": 8638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.337534059945504e-05,
+      "loss": 1.7382,
+      "step": 8639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3373297002724795e-05,
+      "loss": 1.6063,
+      "step": 8640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3371253405994552e-05,
+      "loss": 1.7967,
+      "step": 8641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3369209809264306e-05,
+      "loss": 1.597,
+      "step": 8642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.336716621253406e-05,
+      "loss": 1.5484,
+      "step": 8643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3365122615803814e-05,
+      "loss": 1.7197,
+      "step": 8644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3363079019073571e-05,
+      "loss": 1.6399,
+      "step": 8645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3361035422343325e-05,
+      "loss": 1.677,
+      "step": 8646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3358991825613079e-05,
+      "loss": 1.6855,
+      "step": 8647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3356948228882834e-05,
+      "loss": 1.711,
+      "step": 8648
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3354904632152588e-05,
+      "loss": 1.7315,
+      "step": 8649
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3352861035422344e-05,
+      "loss": 1.5784,
+      "step": 8650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3350817438692098e-05,
+      "loss": 1.7498,
+      "step": 8651
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3348773841961854e-05,
+      "loss": 1.5851,
+      "step": 8652
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3346730245231608e-05,
+      "loss": 1.7744,
+      "step": 8653
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3344686648501363e-05,
+      "loss": 1.6594,
+      "step": 8654
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3342643051771117e-05,
+      "loss": 1.6723,
+      "step": 8655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3340599455040873e-05,
+      "loss": 1.6238,
+      "step": 8656
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3338555858310627e-05,
+      "loss": 1.7172,
+      "step": 8657
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.333651226158038e-05,
+      "loss": 1.7373,
+      "step": 8658
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3334468664850138e-05,
+      "loss": 1.7121,
+      "step": 8659
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3332425068119892e-05,
+      "loss": 1.8418,
+      "step": 8660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3330381471389646e-05,
+      "loss": 1.7511,
+      "step": 8661
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.33283378746594e-05,
+      "loss": 1.7503,
+      "step": 8662
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3326294277929157e-05,
+      "loss": 1.8269,
+      "step": 8663
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3324250681198911e-05,
+      "loss": 1.7189,
+      "step": 8664
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3322207084468665e-05,
+      "loss": 1.7746,
+      "step": 8665
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3320163487738419e-05,
+      "loss": 1.7644,
+      "step": 8666
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3318119891008175e-05,
+      "loss": 1.7425,
+      "step": 8667
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.331607629427793e-05,
+      "loss": 1.7099,
+      "step": 8668
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3314032697547684e-05,
+      "loss": 1.6665,
+      "step": 8669
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3311989100817438e-05,
+      "loss": 1.7335,
+      "step": 8670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3309945504087194e-05,
+      "loss": 1.764,
+      "step": 8671
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.330790190735695e-05,
+      "loss": 1.6999,
+      "step": 8672
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3305858310626703e-05,
+      "loss": 1.7068,
+      "step": 8673
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3303814713896459e-05,
+      "loss": 1.6873,
+      "step": 8674
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3301771117166213e-05,
+      "loss": 1.6103,
+      "step": 8675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3299727520435967e-05,
+      "loss": 1.7219,
+      "step": 8676
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3297683923705722e-05,
+      "loss": 1.6149,
+      "step": 8677
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3295640326975478e-05,
+      "loss": 1.6875,
+      "step": 8678
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3293596730245232e-05,
+      "loss": 1.7645,
+      "step": 8679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3291553133514986e-05,
+      "loss": 1.7836,
+      "step": 8680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3289509536784742e-05,
+      "loss": 1.6668,
+      "step": 8681
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3287465940054497e-05,
+      "loss": 1.5842,
+      "step": 8682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3285422343324251e-05,
+      "loss": 1.7438,
+      "step": 8683
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3283378746594005e-05,
+      "loss": 1.7917,
+      "step": 8684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3281335149863759e-05,
+      "loss": 1.7538,
+      "step": 8685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3279291553133516e-05,
+      "loss": 1.6688,
+      "step": 8686
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.327724795640327e-05,
+      "loss": 1.6929,
+      "step": 8687
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3275204359673024e-05,
+      "loss": 1.7615,
+      "step": 8688
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.327316076294278e-05,
+      "loss": 1.7817,
+      "step": 8689
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3271117166212535e-05,
+      "loss": 1.6407,
+      "step": 8690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.326907356948229e-05,
+      "loss": 1.7439,
+      "step": 8691
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3267029972752043e-05,
+      "loss": 1.6551,
+      "step": 8692
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3264986376021799e-05,
+      "loss": 1.6464,
+      "step": 8693
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3262942779291553e-05,
+      "loss": 1.6506,
+      "step": 8694
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3260899182561309e-05,
+      "loss": 1.6842,
+      "step": 8695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3258855585831063e-05,
+      "loss": 1.6383,
+      "step": 8696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3256811989100818e-05,
+      "loss": 1.7394,
+      "step": 8697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3254768392370572e-05,
+      "loss": 1.6888,
+      "step": 8698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3252724795640328e-05,
+      "loss": 1.5436,
+      "step": 8699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3250681198910083e-05,
+      "loss": 1.7037,
+      "step": 8700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3248637602179837e-05,
+      "loss": 1.661,
+      "step": 8701
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3246594005449591e-05,
+      "loss": 1.7917,
+      "step": 8702
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3244550408719345e-05,
+      "loss": 1.5981,
+      "step": 8703
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3242506811989102e-05,
+      "loss": 1.6522,
+      "step": 8704
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3240463215258856e-05,
+      "loss": 1.7186,
+      "step": 8705
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.323841961852861e-05,
+      "loss": 1.6922,
+      "step": 8706
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3236376021798364e-05,
+      "loss": 1.6088,
+      "step": 8707
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3234332425068122e-05,
+      "loss": 1.6456,
+      "step": 8708
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3232288828337876e-05,
+      "loss": 1.696,
+      "step": 8709
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.323024523160763e-05,
+      "loss": 1.6836,
+      "step": 8710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3228201634877383e-05,
+      "loss": 1.7023,
+      "step": 8711
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3226158038147139e-05,
+      "loss": 1.7733,
+      "step": 8712
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3224114441416895e-05,
+      "loss": 1.6355,
+      "step": 8713
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3222070844686649e-05,
+      "loss": 1.7248,
+      "step": 8714
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3220027247956404e-05,
+      "loss": 1.6851,
+      "step": 8715
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3217983651226158e-05,
+      "loss": 1.851,
+      "step": 8716
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3215940054495914e-05,
+      "loss": 1.5338,
+      "step": 8717
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3213896457765668e-05,
+      "loss": 1.542,
+      "step": 8718
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3211852861035423e-05,
+      "loss": 1.7426,
+      "step": 8719
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3209809264305177e-05,
+      "loss": 1.5976,
+      "step": 8720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3207765667574931e-05,
+      "loss": 1.6337,
+      "step": 8721
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3205722070844687e-05,
+      "loss": 1.7159,
+      "step": 8722
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3203678474114443e-05,
+      "loss": 1.7707,
+      "step": 8723
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3201634877384197e-05,
+      "loss": 1.7331,
+      "step": 8724
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.319959128065395e-05,
+      "loss": 1.6939,
+      "step": 8725
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3197547683923706e-05,
+      "loss": 1.6966,
+      "step": 8726
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3195504087193462e-05,
+      "loss": 1.6537,
+      "step": 8727
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3193460490463216e-05,
+      "loss": 1.7284,
+      "step": 8728
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.319141689373297e-05,
+      "loss": 1.7568,
+      "step": 8729
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3189373297002725e-05,
+      "loss": 1.775,
+      "step": 8730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3187329700272481e-05,
+      "loss": 1.6197,
+      "step": 8731
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3185286103542235e-05,
+      "loss": 1.5987,
+      "step": 8732
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3183242506811989e-05,
+      "loss": 1.6443,
+      "step": 8733
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3181198910081744e-05,
+      "loss": 1.6906,
+      "step": 8734
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3179155313351498e-05,
+      "loss": 1.6767,
+      "step": 8735
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3177111716621254e-05,
+      "loss": 1.6821,
+      "step": 8736
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3175068119891008e-05,
+      "loss": 1.7231,
+      "step": 8737
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3173024523160764e-05,
+      "loss": 1.8917,
+      "step": 8738
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3170980926430517e-05,
+      "loss": 1.7328,
+      "step": 8739
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3168937329700273e-05,
+      "loss": 1.7828,
+      "step": 8740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3166893732970029e-05,
+      "loss": 1.7051,
+      "step": 8741
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3164850136239783e-05,
+      "loss": 1.673,
+      "step": 8742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3162806539509537e-05,
+      "loss": 1.6853,
+      "step": 8743
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3160762942779292e-05,
+      "loss": 1.7542,
+      "step": 8744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3158719346049048e-05,
+      "loss": 1.792,
+      "step": 8745
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3156675749318802e-05,
+      "loss": 1.7255,
+      "step": 8746
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3154632152588556e-05,
+      "loss": 1.6964,
+      "step": 8747
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.315258855585831e-05,
+      "loss": 1.7139,
+      "step": 8748
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3150544959128067e-05,
+      "loss": 1.6844,
+      "step": 8749
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3148501362397821e-05,
+      "loss": 1.7224,
+      "step": 8750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3146457765667575e-05,
+      "loss": 1.7151,
+      "step": 8751
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3144414168937329e-05,
+      "loss": 1.7506,
+      "step": 8752
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3142370572207084e-05,
+      "loss": 1.5725,
+      "step": 8753
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.314032697547684e-05,
+      "loss": 1.8652,
+      "step": 8754
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3138283378746594e-05,
+      "loss": 1.6513,
+      "step": 8755
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.313623978201635e-05,
+      "loss": 1.7569,
+      "step": 8756
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3134196185286104e-05,
+      "loss": 1.7229,
+      "step": 8757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.313215258855586e-05,
+      "loss": 1.885,
+      "step": 8758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3130108991825613e-05,
+      "loss": 1.5934,
+      "step": 8759
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3128065395095369e-05,
+      "loss": 1.5943,
+      "step": 8760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3126021798365123e-05,
+      "loss": 1.5814,
+      "step": 8761
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3123978201634877e-05,
+      "loss": 1.7692,
+      "step": 8762
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3121934604904632e-05,
+      "loss": 1.7463,
+      "step": 8763
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3119891008174388e-05,
+      "loss": 1.7588,
+      "step": 8764
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3117847411444142e-05,
+      "loss": 1.6519,
+      "step": 8765
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3115803814713896e-05,
+      "loss": 1.6862,
+      "step": 8766
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3113760217983651e-05,
+      "loss": 1.8864,
+      "step": 8767
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3111716621253407e-05,
+      "loss": 1.72,
+      "step": 8768
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3109673024523161e-05,
+      "loss": 1.6751,
+      "step": 8769
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3107629427792915e-05,
+      "loss": 1.6556,
+      "step": 8770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.310558583106267e-05,
+      "loss": 1.5423,
+      "step": 8771
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3103542234332426e-05,
+      "loss": 1.6741,
+      "step": 8772
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.310149863760218e-05,
+      "loss": 1.7056,
+      "step": 8773
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3099455040871934e-05,
+      "loss": 1.6678,
+      "step": 8774
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.309741144414169e-05,
+      "loss": 1.6576,
+      "step": 8775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3095367847411445e-05,
+      "loss": 1.6873,
+      "step": 8776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.30933242506812e-05,
+      "loss": 1.6722,
+      "step": 8777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3091280653950953e-05,
+      "loss": 1.6412,
+      "step": 8778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3089237057220709e-05,
+      "loss": 1.6475,
+      "step": 8779
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3087193460490463e-05,
+      "loss": 1.6999,
+      "step": 8780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3085149863760218e-05,
+      "loss": 1.6525,
+      "step": 8781
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3083106267029972e-05,
+      "loss": 1.6931,
+      "step": 8782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3081062670299728e-05,
+      "loss": 1.7361,
+      "step": 8783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3079019073569482e-05,
+      "loss": 1.6194,
+      "step": 8784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3076975476839238e-05,
+      "loss": 1.6575,
+      "step": 8785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3074931880108993e-05,
+      "loss": 1.7429,
+      "step": 8786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3072888283378747e-05,
+      "loss": 1.6958,
+      "step": 8787
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3070844686648501e-05,
+      "loss": 1.8227,
+      "step": 8788
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3068801089918255e-05,
+      "loss": 1.5585,
+      "step": 8789
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3066757493188012e-05,
+      "loss": 1.5508,
+      "step": 8790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3064713896457766e-05,
+      "loss": 1.6369,
+      "step": 8791
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.306267029972752e-05,
+      "loss": 1.6669,
+      "step": 8792
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3060626702997274e-05,
+      "loss": 1.7641,
+      "step": 8793
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3058583106267032e-05,
+      "loss": 1.6478,
+      "step": 8794
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3056539509536785e-05,
+      "loss": 1.6541,
+      "step": 8795
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.305449591280654e-05,
+      "loss": 1.5909,
+      "step": 8796
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3052452316076295e-05,
+      "loss": 1.6469,
+      "step": 8797
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3050408719346049e-05,
+      "loss": 1.6335,
+      "step": 8798
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3048365122615805e-05,
+      "loss": 1.7861,
+      "step": 8799
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3046321525885559e-05,
+      "loss": 1.8118,
+      "step": 8800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3044277929155314e-05,
+      "loss": 1.7689,
+      "step": 8801
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3042234332425068e-05,
+      "loss": 1.783,
+      "step": 8802
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3040190735694824e-05,
+      "loss": 1.8246,
+      "step": 8803
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3038147138964578e-05,
+      "loss": 1.8296,
+      "step": 8804
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3036103542234333e-05,
+      "loss": 1.7571,
+      "step": 8805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3034059945504087e-05,
+      "loss": 1.6685,
+      "step": 8806
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3032016348773841e-05,
+      "loss": 1.7564,
+      "step": 8807
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3029972752043597e-05,
+      "loss": 1.8286,
+      "step": 8808
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3027929155313352e-05,
+      "loss": 1.5645,
+      "step": 8809
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3025885558583106e-05,
+      "loss": 1.7429,
+      "step": 8810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.302384196185286e-05,
+      "loss": 1.6056,
+      "step": 8811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3021798365122618e-05,
+      "loss": 1.5918,
+      "step": 8812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3019754768392372e-05,
+      "loss": 1.7785,
+      "step": 8813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3017711171662126e-05,
+      "loss": 1.68,
+      "step": 8814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.301566757493188e-05,
+      "loss": 1.6955,
+      "step": 8815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3013623978201635e-05,
+      "loss": 1.6469,
+      "step": 8816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.301158038147139e-05,
+      "loss": 1.6228,
+      "step": 8817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3009536784741145e-05,
+      "loss": 1.7179,
+      "step": 8818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3007493188010899e-05,
+      "loss": 1.7879,
+      "step": 8819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3005449591280654e-05,
+      "loss": 1.7138,
+      "step": 8820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.300340599455041e-05,
+      "loss": 1.5557,
+      "step": 8821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3001362397820164e-05,
+      "loss": 1.8419,
+      "step": 8822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2999318801089918e-05,
+      "loss": 1.7241,
+      "step": 8823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2997275204359673e-05,
+      "loss": 1.6059,
+      "step": 8824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2995231607629427e-05,
+      "loss": 1.933,
+      "step": 8825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2993188010899183e-05,
+      "loss": 1.707,
+      "step": 8826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2991144414168939e-05,
+      "loss": 1.6017,
+      "step": 8827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2989100817438693e-05,
+      "loss": 1.6578,
+      "step": 8828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2987057220708446e-05,
+      "loss": 1.761,
+      "step": 8829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2985013623978202e-05,
+      "loss": 1.7042,
+      "step": 8830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2982970027247958e-05,
+      "loss": 1.7097,
+      "step": 8831
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2980926430517712e-05,
+      "loss": 1.727,
+      "step": 8832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2978882833787466e-05,
+      "loss": 1.7037,
+      "step": 8833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.297683923705722e-05,
+      "loss": 1.7006,
+      "step": 8834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2974795640326977e-05,
+      "loss": 1.7318,
+      "step": 8835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.297275204359673e-05,
+      "loss": 1.6759,
+      "step": 8836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2970708446866485e-05,
+      "loss": 1.5869,
+      "step": 8837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.296866485013624e-05,
+      "loss": 1.7391,
+      "step": 8838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2966621253405996e-05,
+      "loss": 1.6752,
+      "step": 8839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.296457765667575e-05,
+      "loss": 1.7439,
+      "step": 8840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2962534059945504e-05,
+      "loss": 1.7311,
+      "step": 8841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.296049046321526e-05,
+      "loss": 1.7367,
+      "step": 8842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2958446866485013e-05,
+      "loss": 1.7044,
+      "step": 8843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2956403269754769e-05,
+      "loss": 1.7136,
+      "step": 8844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2954359673024523e-05,
+      "loss": 1.8304,
+      "step": 8845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2952316076294279e-05,
+      "loss": 1.773,
+      "step": 8846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2950272479564033e-05,
+      "loss": 1.5242,
+      "step": 8847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2948228882833788e-05,
+      "loss": 1.6354,
+      "step": 8848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2946185286103542e-05,
+      "loss": 1.7321,
+      "step": 8849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2944141689373298e-05,
+      "loss": 1.5892,
+      "step": 8850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2942098092643052e-05,
+      "loss": 1.7746,
+      "step": 8851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2940054495912806e-05,
+      "loss": 1.7407,
+      "step": 8852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2938010899182563e-05,
+      "loss": 1.6971,
+      "step": 8853
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2935967302452317e-05,
+      "loss": 1.6899,
+      "step": 8854
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2933923705722071e-05,
+      "loss": 1.569,
+      "step": 8855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2931880108991825e-05,
+      "loss": 1.653,
+      "step": 8856
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2929836512261582e-05,
+      "loss": 1.7181,
+      "step": 8857
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2927792915531336e-05,
+      "loss": 1.6403,
+      "step": 8858
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.292574931880109e-05,
+      "loss": 1.7608,
+      "step": 8859
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2923705722070844e-05,
+      "loss": 1.6847,
+      "step": 8860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.29216621253406e-05,
+      "loss": 1.7541,
+      "step": 8861
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2919618528610355e-05,
+      "loss": 1.6739,
+      "step": 8862
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.291757493188011e-05,
+      "loss": 1.7961,
+      "step": 8863
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2915531335149863e-05,
+      "loss": 1.7732,
+      "step": 8864
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2913487738419619e-05,
+      "loss": 1.9178,
+      "step": 8865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2911444141689374e-05,
+      "loss": 1.7071,
+      "step": 8866
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2909400544959128e-05,
+      "loss": 1.6869,
+      "step": 8867
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2907356948228884e-05,
+      "loss": 1.7728,
+      "step": 8868
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2905313351498638e-05,
+      "loss": 1.6633,
+      "step": 8869
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2903269754768392e-05,
+      "loss": 1.7517,
+      "step": 8870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2901226158038147e-05,
+      "loss": 1.6433,
+      "step": 8871
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2899182561307903e-05,
+      "loss": 1.6046,
+      "step": 8872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2897138964577657e-05,
+      "loss": 1.7696,
+      "step": 8873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2895095367847411e-05,
+      "loss": 1.6371,
+      "step": 8874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2893051771117167e-05,
+      "loss": 1.6201,
+      "step": 8875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2891008174386922e-05,
+      "loss": 1.6721,
+      "step": 8876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2888964577656676e-05,
+      "loss": 1.7116,
+      "step": 8877
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.288692098092643e-05,
+      "loss": 1.7793,
+      "step": 8878
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2884877384196186e-05,
+      "loss": 1.7546,
+      "step": 8879
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2882833787465941e-05,
+      "loss": 1.7374,
+      "step": 8880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2880790190735695e-05,
+      "loss": 1.7075,
+      "step": 8881
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.287874659400545e-05,
+      "loss": 1.6225,
+      "step": 8882
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2876702997275205e-05,
+      "loss": 1.7421,
+      "step": 8883
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.287465940054496e-05,
+      "loss": 1.6958,
+      "step": 8884
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2872615803814714e-05,
+      "loss": 1.7773,
+      "step": 8885
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2870572207084468e-05,
+      "loss": 1.6898,
+      "step": 8886
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2868528610354224e-05,
+      "loss": 1.7416,
+      "step": 8887
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2866485013623978e-05,
+      "loss": 1.7416,
+      "step": 8888
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2864441416893734e-05,
+      "loss": 1.6745,
+      "step": 8889
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2862397820163488e-05,
+      "loss": 1.6252,
+      "step": 8890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2860354223433243e-05,
+      "loss": 1.7814,
+      "step": 8891
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2858310626702997e-05,
+      "loss": 1.4453,
+      "step": 8892
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2856267029972753e-05,
+      "loss": 1.6362,
+      "step": 8893
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2854223433242508e-05,
+      "loss": 1.6815,
+      "step": 8894
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2852179836512262e-05,
+      "loss": 1.7982,
+      "step": 8895
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2850136239782016e-05,
+      "loss": 1.7619,
+      "step": 8896
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.284809264305177e-05,
+      "loss": 1.7256,
+      "step": 8897
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2846049046321528e-05,
+      "loss": 1.6215,
+      "step": 8898
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2844005449591281e-05,
+      "loss": 1.6782,
+      "step": 8899
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2841961852861035e-05,
+      "loss": 1.6671,
+      "step": 8900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.283991825613079e-05,
+      "loss": 1.7255,
+      "step": 8901
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2837874659400547e-05,
+      "loss": 1.7509,
+      "step": 8902
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.28358310626703e-05,
+      "loss": 1.7129,
+      "step": 8903
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2833787465940055e-05,
+      "loss": 1.7491,
+      "step": 8904
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2831743869209808e-05,
+      "loss": 1.6207,
+      "step": 8905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2829700272479564e-05,
+      "loss": 1.7285,
+      "step": 8906
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.282765667574932e-05,
+      "loss": 1.7046,
+      "step": 8907
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2825613079019074e-05,
+      "loss": 1.6375,
+      "step": 8908
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.282356948228883e-05,
+      "loss": 1.7572,
+      "step": 8909
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2821525885558583e-05,
+      "loss": 1.7013,
+      "step": 8910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2819482288828339e-05,
+      "loss": 1.8306,
+      "step": 8911
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2817438692098093e-05,
+      "loss": 1.7355,
+      "step": 8912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2815395095367848e-05,
+      "loss": 1.7054,
+      "step": 8913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2813351498637602e-05,
+      "loss": 1.5828,
+      "step": 8914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2811307901907356e-05,
+      "loss": 1.7077,
+      "step": 8915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2809264305177112e-05,
+      "loss": 1.6254,
+      "step": 8916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2807220708446868e-05,
+      "loss": 1.7073,
+      "step": 8917
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2805177111716622e-05,
+      "loss": 1.6639,
+      "step": 8918
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2803133514986375e-05,
+      "loss": 1.639,
+      "step": 8919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2801089918256131e-05,
+      "loss": 1.7541,
+      "step": 8920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2799046321525887e-05,
+      "loss": 1.6578,
+      "step": 8921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.279700272479564e-05,
+      "loss": 1.6504,
+      "step": 8922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2794959128065395e-05,
+      "loss": 1.9641,
+      "step": 8923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.279291553133515e-05,
+      "loss": 1.7583,
+      "step": 8924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2790871934604906e-05,
+      "loss": 1.7355,
+      "step": 8925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.278882833787466e-05,
+      "loss": 1.6949,
+      "step": 8926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2786784741144414e-05,
+      "loss": 1.7523,
+      "step": 8927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.278474114441417e-05,
+      "loss": 1.7765,
+      "step": 8928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2782697547683925e-05,
+      "loss": 1.5864,
+      "step": 8929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2780653950953679e-05,
+      "loss": 1.6159,
+      "step": 8930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2778610354223433e-05,
+      "loss": 1.7448,
+      "step": 8931
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2776566757493189e-05,
+      "loss": 1.655,
+      "step": 8932
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2774523160762942e-05,
+      "loss": 1.6432,
+      "step": 8933
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2772479564032698e-05,
+      "loss": 1.7189,
+      "step": 8934
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2770435967302454e-05,
+      "loss": 1.6968,
+      "step": 8935
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2768392370572208e-05,
+      "loss": 1.7358,
+      "step": 8936
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2766348773841962e-05,
+      "loss": 1.7315,
+      "step": 8937
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2764305177111717e-05,
+      "loss": 1.7023,
+      "step": 8938
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2762261580381473e-05,
+      "loss": 1.7036,
+      "step": 8939
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2760217983651227e-05,
+      "loss": 1.6716,
+      "step": 8940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.275817438692098e-05,
+      "loss": 1.7166,
+      "step": 8941
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2756130790190735e-05,
+      "loss": 1.6584,
+      "step": 8942
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2754087193460492e-05,
+      "loss": 1.7348,
+      "step": 8943
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2752043596730246e-05,
+      "loss": 1.6825,
+      "step": 8944
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.275e-05,
+      "loss": 1.7334,
+      "step": 8945
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2747956403269754e-05,
+      "loss": 1.7158,
+      "step": 8946
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.274591280653951e-05,
+      "loss": 1.7909,
+      "step": 8947
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2743869209809265e-05,
+      "loss": 1.722,
+      "step": 8948
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2741825613079019e-05,
+      "loss": 1.6999,
+      "step": 8949
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2739782016348775e-05,
+      "loss": 1.7256,
+      "step": 8950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2737738419618529e-05,
+      "loss": 1.6837,
+      "step": 8951
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2735694822888284e-05,
+      "loss": 1.7471,
+      "step": 8952
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2733651226158038e-05,
+      "loss": 1.6169,
+      "step": 8953
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2731607629427794e-05,
+      "loss": 1.7709,
+      "step": 8954
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2729564032697548e-05,
+      "loss": 1.7422,
+      "step": 8955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2727520435967302e-05,
+      "loss": 1.7828,
+      "step": 8956
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2725476839237057e-05,
+      "loss": 1.7157,
+      "step": 8957
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2723433242506813e-05,
+      "loss": 1.5109,
+      "step": 8958
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2721389645776567e-05,
+      "loss": 1.6891,
+      "step": 8959
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.271934604904632e-05,
+      "loss": 1.7164,
+      "step": 8960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2717302452316076e-05,
+      "loss": 1.7081,
+      "step": 8961
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2715258855585832e-05,
+      "loss": 1.5992,
+      "step": 8962
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2713215258855586e-05,
+      "loss": 1.6457,
+      "step": 8963
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.271117166212534e-05,
+      "loss": 1.6342,
+      "step": 8964
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2709128065395096e-05,
+      "loss": 1.8076,
+      "step": 8965
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2707084468664851e-05,
+      "loss": 1.576,
+      "step": 8966
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2705040871934605e-05,
+      "loss": 1.761,
+      "step": 8967
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2702997275204359e-05,
+      "loss": 1.609,
+      "step": 8968
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2700953678474115e-05,
+      "loss": 1.8815,
+      "step": 8969
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.269891008174387e-05,
+      "loss": 1.7309,
+      "step": 8970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2696866485013624e-05,
+      "loss": 1.6623,
+      "step": 8971
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2694822888283378e-05,
+      "loss": 1.7371,
+      "step": 8972
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2692779291553134e-05,
+      "loss": 1.6083,
+      "step": 8973
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2690735694822888e-05,
+      "loss": 1.802,
+      "step": 8974
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2688692098092643e-05,
+      "loss": 1.6809,
+      "step": 8975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2686648501362399e-05,
+      "loss": 1.5716,
+      "step": 8976
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2684604904632153e-05,
+      "loss": 1.6055,
+      "step": 8977
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2682561307901907e-05,
+      "loss": 1.6527,
+      "step": 8978
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2680517711171663e-05,
+      "loss": 1.6312,
+      "step": 8979
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2678474114441418e-05,
+      "loss": 1.769,
+      "step": 8980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2676430517711172e-05,
+      "loss": 1.6905,
+      "step": 8981
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2674386920980926e-05,
+      "loss": 1.6321,
+      "step": 8982
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.267234332425068e-05,
+      "loss": 1.6764,
+      "step": 8983
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2670299727520437e-05,
+      "loss": 1.6038,
+      "step": 8984
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2668256130790191e-05,
+      "loss": 1.757,
+      "step": 8985
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2666212534059945e-05,
+      "loss": 1.6821,
+      "step": 8986
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.26641689373297e-05,
+      "loss": 1.6661,
+      "step": 8987
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2662125340599457e-05,
+      "loss": 1.713,
+      "step": 8988
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.266008174386921e-05,
+      "loss": 1.6739,
+      "step": 8989
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2658038147138964e-05,
+      "loss": 1.6587,
+      "step": 8990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.265599455040872e-05,
+      "loss": 1.568,
+      "step": 8991
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2653950953678474e-05,
+      "loss": 1.6277,
+      "step": 8992
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.265190735694823e-05,
+      "loss": 1.7369,
+      "step": 8993
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2649863760217984e-05,
+      "loss": 1.7308,
+      "step": 8994
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.264782016348774e-05,
+      "loss": 1.5414,
+      "step": 8995
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2645776566757493e-05,
+      "loss": 1.489,
+      "step": 8996
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2643732970027249e-05,
+      "loss": 1.6762,
+      "step": 8997
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2641689373297003e-05,
+      "loss": 1.6256,
+      "step": 8998
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2639645776566758e-05,
+      "loss": 1.9374,
+      "step": 8999
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2637602179836512e-05,
+      "loss": 1.6371,
+      "step": 9000
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 1.527760624885559,
+      "eval_runtime": 721.8093,
+      "eval_samples_per_second": 5.749,
+      "eval_steps_per_second": 1.438,
+      "step": 9000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2635558583106266e-05,
+      "loss": 1.8707,
+      "step": 9001
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2633514986376022e-05,
+      "loss": 1.7271,
+      "step": 9002
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2631471389645777e-05,
+      "loss": 1.6275,
+      "step": 9003
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2629427792915531e-05,
+      "loss": 1.6293,
+      "step": 9004
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2627384196185285e-05,
+      "loss": 1.7242,
+      "step": 9005
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2625340599455043e-05,
+      "loss": 1.5771,
+      "step": 9006
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2623297002724797e-05,
+      "loss": 1.6887,
+      "step": 9007
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.262125340599455e-05,
+      "loss": 1.5939,
+      "step": 9008
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2619209809264304e-05,
+      "loss": 1.6801,
+      "step": 9009
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.261716621253406e-05,
+      "loss": 1.5446,
+      "step": 9010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2615122615803816e-05,
+      "loss": 1.7235,
+      "step": 9011
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.261307901907357e-05,
+      "loss": 1.6667,
+      "step": 9012
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2611035422343324e-05,
+      "loss": 1.653,
+      "step": 9013
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.260899182561308e-05,
+      "loss": 1.8727,
+      "step": 9014
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2606948228882835e-05,
+      "loss": 1.7056,
+      "step": 9015
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2604904632152589e-05,
+      "loss": 1.6347,
+      "step": 9016
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2602861035422344e-05,
+      "loss": 1.6719,
+      "step": 9017
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2600817438692098e-05,
+      "loss": 1.6411,
+      "step": 9018
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2598773841961852e-05,
+      "loss": 1.7479,
+      "step": 9019
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2596730245231608e-05,
+      "loss": 1.6096,
+      "step": 9020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2594686648501364e-05,
+      "loss": 1.7963,
+      "step": 9021
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2592643051771118e-05,
+      "loss": 1.7668,
+      "step": 9022
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2590599455040871e-05,
+      "loss": 1.6934,
+      "step": 9023
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2588555858310627e-05,
+      "loss": 1.7748,
+      "step": 9024
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2586512261580383e-05,
+      "loss": 1.5718,
+      "step": 9025
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2584468664850137e-05,
+      "loss": 1.6497,
+      "step": 9026
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.258242506811989e-05,
+      "loss": 1.5961,
+      "step": 9027
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2580381471389645e-05,
+      "loss": 1.7857,
+      "step": 9028
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2578337874659402e-05,
+      "loss": 1.6517,
+      "step": 9029
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2576294277929156e-05,
+      "loss": 1.7582,
+      "step": 9030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.257425068119891e-05,
+      "loss": 1.6303,
+      "step": 9031
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2572207084468665e-05,
+      "loss": 1.8132,
+      "step": 9032
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2570163487738421e-05,
+      "loss": 1.6947,
+      "step": 9033
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2568119891008175e-05,
+      "loss": 1.589,
+      "step": 9034
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2566076294277929e-05,
+      "loss": 1.6868,
+      "step": 9035
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2564032697547685e-05,
+      "loss": 1.7597,
+      "step": 9036
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2561989100817438e-05,
+      "loss": 1.7359,
+      "step": 9037
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2559945504087194e-05,
+      "loss": 1.7371,
+      "step": 9038
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2557901907356948e-05,
+      "loss": 1.5738,
+      "step": 9039
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2555858310626704e-05,
+      "loss": 1.7706,
+      "step": 9040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2553814713896458e-05,
+      "loss": 1.6729,
+      "step": 9041
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2551771117166213e-05,
+      "loss": 1.6084,
+      "step": 9042
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2549727520435967e-05,
+      "loss": 1.6517,
+      "step": 9043
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2547683923705723e-05,
+      "loss": 1.6003,
+      "step": 9044
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2545640326975477e-05,
+      "loss": 1.6194,
+      "step": 9045
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.254359673024523e-05,
+      "loss": 1.9328,
+      "step": 9046
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2541553133514988e-05,
+      "loss": 1.737,
+      "step": 9047
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2539509536784742e-05,
+      "loss": 1.5667,
+      "step": 9048
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2537465940054496e-05,
+      "loss": 1.5698,
+      "step": 9049
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.253542234332425e-05,
+      "loss": 1.5643,
+      "step": 9050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2533378746594007e-05,
+      "loss": 1.815,
+      "step": 9051
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2531335149863761e-05,
+      "loss": 1.6924,
+      "step": 9052
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2529291553133515e-05,
+      "loss": 1.6327,
+      "step": 9053
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2527247956403269e-05,
+      "loss": 1.7076,
+      "step": 9054
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2525204359673025e-05,
+      "loss": 1.6745,
+      "step": 9055
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.252316076294278e-05,
+      "loss": 1.776,
+      "step": 9056
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2521117166212534e-05,
+      "loss": 1.7599,
+      "step": 9057
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2519073569482288e-05,
+      "loss": 1.6396,
+      "step": 9058
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2517029972752044e-05,
+      "loss": 1.7569,
+      "step": 9059
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.25149863760218e-05,
+      "loss": 1.6945,
+      "step": 9060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2512942779291553e-05,
+      "loss": 1.6863,
+      "step": 9061
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2510899182561309e-05,
+      "loss": 1.7822,
+      "step": 9062
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2508855585831063e-05,
+      "loss": 1.8156,
+      "step": 9063
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2506811989100817e-05,
+      "loss": 1.7599,
+      "step": 9064
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2504768392370572e-05,
+      "loss": 1.6553,
+      "step": 9065
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2502724795640328e-05,
+      "loss": 1.7667,
+      "step": 9066
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2500681198910082e-05,
+      "loss": 1.6319,
+      "step": 9067
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2498637602179836e-05,
+      "loss": 1.7052,
+      "step": 9068
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2496594005449592e-05,
+      "loss": 1.6877,
+      "step": 9069
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2494550408719347e-05,
+      "loss": 1.7751,
+      "step": 9070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2492506811989101e-05,
+      "loss": 1.7059,
+      "step": 9071
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2490463215258855e-05,
+      "loss": 1.6319,
+      "step": 9072
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.248841961852861e-05,
+      "loss": 1.6509,
+      "step": 9073
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2486376021798366e-05,
+      "loss": 1.5708,
+      "step": 9074
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.248433242506812e-05,
+      "loss": 1.7355,
+      "step": 9075
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2482288828337874e-05,
+      "loss": 1.6685,
+      "step": 9076
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.248024523160763e-05,
+      "loss": 1.6571,
+      "step": 9077
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2478201634877386e-05,
+      "loss": 1.6649,
+      "step": 9078
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.247615803814714e-05,
+      "loss": 1.6819,
+      "step": 9079
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2474114441416893e-05,
+      "loss": 1.5981,
+      "step": 9080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2472070844686649e-05,
+      "loss": 1.7419,
+      "step": 9081
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2470027247956403e-05,
+      "loss": 1.7821,
+      "step": 9082
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2467983651226159e-05,
+      "loss": 1.6991,
+      "step": 9083
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2465940054495913e-05,
+      "loss": 1.6405,
+      "step": 9084
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2463896457765668e-05,
+      "loss": 1.6518,
+      "step": 9085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2461852861035422e-05,
+      "loss": 1.8484,
+      "step": 9086
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2459809264305178e-05,
+      "loss": 1.7208,
+      "step": 9087
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2457765667574933e-05,
+      "loss": 1.759,
+      "step": 9088
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2455722070844687e-05,
+      "loss": 1.6635,
+      "step": 9089
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2453678474114441e-05,
+      "loss": 1.6924,
+      "step": 9090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2451634877384195e-05,
+      "loss": 1.6476,
+      "step": 9091
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2449591280653953e-05,
+      "loss": 1.5977,
+      "step": 9092
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2447547683923706e-05,
+      "loss": 1.6548,
+      "step": 9093
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.244550408719346e-05,
+      "loss": 1.7455,
+      "step": 9094
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2443460490463214e-05,
+      "loss": 1.7858,
+      "step": 9095
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2441416893732972e-05,
+      "loss": 1.611,
+      "step": 9096
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2439373297002726e-05,
+      "loss": 1.6106,
+      "step": 9097
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.243732970027248e-05,
+      "loss": 1.5685,
+      "step": 9098
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2435286103542234e-05,
+      "loss": 1.9115,
+      "step": 9099
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2433242506811989e-05,
+      "loss": 1.5989,
+      "step": 9100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2431198910081745e-05,
+      "loss": 1.741,
+      "step": 9101
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2429155313351499e-05,
+      "loss": 1.767,
+      "step": 9102
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2427111716621254e-05,
+      "loss": 1.7735,
+      "step": 9103
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2425068119891008e-05,
+      "loss": 1.753,
+      "step": 9104
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2423024523160764e-05,
+      "loss": 1.6944,
+      "step": 9105
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2420980926430518e-05,
+      "loss": 1.7721,
+      "step": 9106
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2418937329700273e-05,
+      "loss": 1.6946,
+      "step": 9107
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2416893732970027e-05,
+      "loss": 1.7097,
+      "step": 9108
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2414850136239781e-05,
+      "loss": 1.6881,
+      "step": 9109
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2412806539509537e-05,
+      "loss": 1.5767,
+      "step": 9110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2410762942779293e-05,
+      "loss": 1.6886,
+      "step": 9111
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2408719346049047e-05,
+      "loss": 1.671,
+      "step": 9112
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.24066757493188e-05,
+      "loss": 1.6419,
+      "step": 9113
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2404632152588558e-05,
+      "loss": 1.5828,
+      "step": 9114
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2402588555858312e-05,
+      "loss": 1.5491,
+      "step": 9115
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2400544959128066e-05,
+      "loss": 1.6516,
+      "step": 9116
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.239850136239782e-05,
+      "loss": 1.6586,
+      "step": 9117
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2396457765667575e-05,
+      "loss": 1.7471,
+      "step": 9118
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2394414168937331e-05,
+      "loss": 1.718,
+      "step": 9119
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2392370572207085e-05,
+      "loss": 1.7079,
+      "step": 9120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2390326975476839e-05,
+      "loss": 1.691,
+      "step": 9121
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2388283378746594e-05,
+      "loss": 1.7245,
+      "step": 9122
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.238623978201635e-05,
+      "loss": 1.711,
+      "step": 9123
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2384196185286104e-05,
+      "loss": 1.656,
+      "step": 9124
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2382152588555858e-05,
+      "loss": 1.7277,
+      "step": 9125
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2380108991825614e-05,
+      "loss": 1.7178,
+      "step": 9126
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2378065395095367e-05,
+      "loss": 1.7025,
+      "step": 9127
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2376021798365123e-05,
+      "loss": 1.6791,
+      "step": 9128
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2373978201634879e-05,
+      "loss": 1.6841,
+      "step": 9129
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2371934604904633e-05,
+      "loss": 1.6934,
+      "step": 9130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2369891008174387e-05,
+      "loss": 1.6896,
+      "step": 9131
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2367847411444142e-05,
+      "loss": 1.6721,
+      "step": 9132
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2365803814713898e-05,
+      "loss": 1.751,
+      "step": 9133
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2363760217983652e-05,
+      "loss": 1.6352,
+      "step": 9134
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2361716621253406e-05,
+      "loss": 1.6925,
+      "step": 9135
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.235967302452316e-05,
+      "loss": 1.7426,
+      "step": 9136
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2357629427792917e-05,
+      "loss": 1.6966,
+      "step": 9137
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2355585831062671e-05,
+      "loss": 1.5916,
+      "step": 9138
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2353542234332425e-05,
+      "loss": 1.7544,
+      "step": 9139
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2351498637602179e-05,
+      "loss": 1.5566,
+      "step": 9140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2349455040871934e-05,
+      "loss": 1.6938,
+      "step": 9141
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.234741144414169e-05,
+      "loss": 1.7641,
+      "step": 9142
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2345367847411444e-05,
+      "loss": 1.8979,
+      "step": 9143
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.23433242506812e-05,
+      "loss": 1.6892,
+      "step": 9144
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2341280653950954e-05,
+      "loss": 1.6902,
+      "step": 9145
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.233923705722071e-05,
+      "loss": 1.658,
+      "step": 9146
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2337193460490463e-05,
+      "loss": 1.7404,
+      "step": 9147
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2335149863760219e-05,
+      "loss": 1.7641,
+      "step": 9148
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2333106267029973e-05,
+      "loss": 1.6652,
+      "step": 9149
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2331062670299727e-05,
+      "loss": 1.7486,
+      "step": 9150
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2329019073569482e-05,
+      "loss": 1.6475,
+      "step": 9151
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2326975476839238e-05,
+      "loss": 1.709,
+      "step": 9152
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2324931880108992e-05,
+      "loss": 1.7394,
+      "step": 9153
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2322888283378746e-05,
+      "loss": 1.7595,
+      "step": 9154
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2320844686648503e-05,
+      "loss": 1.7193,
+      "step": 9155
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2318801089918257e-05,
+      "loss": 1.7782,
+      "step": 9156
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2316757493188011e-05,
+      "loss": 1.7084,
+      "step": 9157
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2314713896457765e-05,
+      "loss": 1.8001,
+      "step": 9158
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.231267029972752e-05,
+      "loss": 1.6663,
+      "step": 9159
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2310626702997276e-05,
+      "loss": 1.6978,
+      "step": 9160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.230858310626703e-05,
+      "loss": 1.7549,
+      "step": 9161
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2306539509536784e-05,
+      "loss": 1.7504,
+      "step": 9162
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.230449591280654e-05,
+      "loss": 1.8273,
+      "step": 9163
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2302452316076295e-05,
+      "loss": 1.8073,
+      "step": 9164
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.230040871934605e-05,
+      "loss": 1.7461,
+      "step": 9165
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2298365122615803e-05,
+      "loss": 1.8123,
+      "step": 9166
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2296321525885559e-05,
+      "loss": 1.7098,
+      "step": 9167
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2294277929155313e-05,
+      "loss": 1.6987,
+      "step": 9168
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2292234332425068e-05,
+      "loss": 1.6503,
+      "step": 9169
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2290190735694824e-05,
+      "loss": 1.6837,
+      "step": 9170
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2288147138964578e-05,
+      "loss": 1.7966,
+      "step": 9171
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2286103542234332e-05,
+      "loss": 1.838,
+      "step": 9172
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2284059945504088e-05,
+      "loss": 1.7088,
+      "step": 9173
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2282016348773843e-05,
+      "loss": 1.7223,
+      "step": 9174
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2279972752043597e-05,
+      "loss": 1.8003,
+      "step": 9175
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2277929155313351e-05,
+      "loss": 1.6147,
+      "step": 9176
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2275885558583105e-05,
+      "loss": 1.757,
+      "step": 9177
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2273841961852862e-05,
+      "loss": 1.6696,
+      "step": 9178
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2271798365122616e-05,
+      "loss": 1.7753,
+      "step": 9179
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.226975476839237e-05,
+      "loss": 1.6222,
+      "step": 9180
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2267711171662124e-05,
+      "loss": 1.5243,
+      "step": 9181
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2265667574931882e-05,
+      "loss": 1.6189,
+      "step": 9182
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2263623978201635e-05,
+      "loss": 1.6622,
+      "step": 9183
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.226158038147139e-05,
+      "loss": 1.7491,
+      "step": 9184
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2259536784741145e-05,
+      "loss": 1.7494,
+      "step": 9185
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2257493188010899e-05,
+      "loss": 1.6826,
+      "step": 9186
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2255449591280655e-05,
+      "loss": 1.7929,
+      "step": 9187
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2253405994550409e-05,
+      "loss": 1.6643,
+      "step": 9188
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2251362397820164e-05,
+      "loss": 1.7832,
+      "step": 9189
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2249318801089918e-05,
+      "loss": 1.7275,
+      "step": 9190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2247275204359674e-05,
+      "loss": 1.5844,
+      "step": 9191
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2245231607629428e-05,
+      "loss": 1.5441,
+      "step": 9192
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2243188010899183e-05,
+      "loss": 1.488,
+      "step": 9193
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2241144414168937e-05,
+      "loss": 1.6954,
+      "step": 9194
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2239100817438691e-05,
+      "loss": 1.8412,
+      "step": 9195
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2237057220708447e-05,
+      "loss": 1.7769,
+      "step": 9196
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2235013623978202e-05,
+      "loss": 1.7737,
+      "step": 9197
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2232970027247956e-05,
+      "loss": 1.7077,
+      "step": 9198
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.223092643051771e-05,
+      "loss": 1.6487,
+      "step": 9199
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2228882833787468e-05,
+      "loss": 1.7588,
+      "step": 9200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2226839237057222e-05,
+      "loss": 1.7284,
+      "step": 9201
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2224795640326976e-05,
+      "loss": 1.6313,
+      "step": 9202
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.222275204359673e-05,
+      "loss": 1.7509,
+      "step": 9203
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2220708446866485e-05,
+      "loss": 1.6308,
+      "step": 9204
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.221866485013624e-05,
+      "loss": 1.7015,
+      "step": 9205
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2216621253405995e-05,
+      "loss": 1.7851,
+      "step": 9206
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2214577656675749e-05,
+      "loss": 1.7332,
+      "step": 9207
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2212534059945504e-05,
+      "loss": 1.7433,
+      "step": 9208
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.221049046321526e-05,
+      "loss": 1.6951,
+      "step": 9209
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2208446866485014e-05,
+      "loss": 1.7321,
+      "step": 9210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.220640326975477e-05,
+      "loss": 1.6386,
+      "step": 9211
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2204359673024523e-05,
+      "loss": 1.6211,
+      "step": 9212
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2202316076294277e-05,
+      "loss": 1.726,
+      "step": 9213
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2200272479564033e-05,
+      "loss": 1.8111,
+      "step": 9214
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2198228882833789e-05,
+      "loss": 1.6493,
+      "step": 9215
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2196185286103543e-05,
+      "loss": 1.7265,
+      "step": 9216
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2194141689373297e-05,
+      "loss": 1.7831,
+      "step": 9217
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2192098092643052e-05,
+      "loss": 1.7139,
+      "step": 9218
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2190054495912808e-05,
+      "loss": 1.6811,
+      "step": 9219
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2188010899182562e-05,
+      "loss": 1.7273,
+      "step": 9220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2185967302452316e-05,
+      "loss": 1.7637,
+      "step": 9221
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.218392370572207e-05,
+      "loss": 1.6101,
+      "step": 9222
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2181880108991827e-05,
+      "loss": 1.6507,
+      "step": 9223
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2179836512261581e-05,
+      "loss": 1.6791,
+      "step": 9224
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2177792915531335e-05,
+      "loss": 1.6706,
+      "step": 9225
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.217574931880109e-05,
+      "loss": 1.7002,
+      "step": 9226
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2173705722070846e-05,
+      "loss": 1.597,
+      "step": 9227
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.21716621253406e-05,
+      "loss": 1.6643,
+      "step": 9228
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2169618528610354e-05,
+      "loss": 1.7106,
+      "step": 9229
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.216757493188011e-05,
+      "loss": 1.7856,
+      "step": 9230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2165531335149864e-05,
+      "loss": 1.6162,
+      "step": 9231
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2163487738419619e-05,
+      "loss": 1.5426,
+      "step": 9232
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2161444141689373e-05,
+      "loss": 1.6192,
+      "step": 9233
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2159400544959129e-05,
+      "loss": 1.8113,
+      "step": 9234
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2157356948228883e-05,
+      "loss": 1.7058,
+      "step": 9235
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2155313351498638e-05,
+      "loss": 1.7304,
+      "step": 9236
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2153269754768392e-05,
+      "loss": 1.6996,
+      "step": 9237
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2151226158038148e-05,
+      "loss": 1.6152,
+      "step": 9238
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2149182561307902e-05,
+      "loss": 1.7156,
+      "step": 9239
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2147138964577656e-05,
+      "loss": 1.7645,
+      "step": 9240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2145095367847413e-05,
+      "loss": 1.8511,
+      "step": 9241
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2143051771117167e-05,
+      "loss": 1.6398,
+      "step": 9242
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2141008174386921e-05,
+      "loss": 1.6787,
+      "step": 9243
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2138964577656675e-05,
+      "loss": 1.6569,
+      "step": 9244
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2136920980926432e-05,
+      "loss": 1.6381,
+      "step": 9245
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2134877384196186e-05,
+      "loss": 1.6421,
+      "step": 9246
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.213283378746594e-05,
+      "loss": 1.694,
+      "step": 9247
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2130790190735694e-05,
+      "loss": 1.6225,
+      "step": 9248
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.212874659400545e-05,
+      "loss": 1.4993,
+      "step": 9249
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2126702997275205e-05,
+      "loss": 1.75,
+      "step": 9250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.212465940054496e-05,
+      "loss": 1.787,
+      "step": 9251
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2122615803814715e-05,
+      "loss": 1.6491,
+      "step": 9252
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2120572207084469e-05,
+      "loss": 1.5788,
+      "step": 9253
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2118528610354224e-05,
+      "loss": 1.6909,
+      "step": 9254
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2116485013623978e-05,
+      "loss": 1.6327,
+      "step": 9255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2114441416893734e-05,
+      "loss": 1.7907,
+      "step": 9256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2112397820163488e-05,
+      "loss": 1.8309,
+      "step": 9257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2110354223433242e-05,
+      "loss": 1.6724,
+      "step": 9258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2108310626702998e-05,
+      "loss": 1.657,
+      "step": 9259
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2106267029972753e-05,
+      "loss": 1.7975,
+      "step": 9260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2104223433242507e-05,
+      "loss": 1.7212,
+      "step": 9261
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2102179836512261e-05,
+      "loss": 1.6987,
+      "step": 9262
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2100136239782017e-05,
+      "loss": 1.6353,
+      "step": 9263
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2098092643051772e-05,
+      "loss": 1.6893,
+      "step": 9264
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2096049046321526e-05,
+      "loss": 1.6856,
+      "step": 9265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.209400544959128e-05,
+      "loss": 1.6049,
+      "step": 9266
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2091961852861036e-05,
+      "loss": 1.7326,
+      "step": 9267
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2089918256130791e-05,
+      "loss": 1.6712,
+      "step": 9268
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2087874659400545e-05,
+      "loss": 1.761,
+      "step": 9269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.20858310626703e-05,
+      "loss": 1.6473,
+      "step": 9270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2083787465940055e-05,
+      "loss": 1.6645,
+      "step": 9271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.208174386920981e-05,
+      "loss": 1.6677,
+      "step": 9272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2079700272479565e-05,
+      "loss": 1.6473,
+      "step": 9273
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2077656675749318e-05,
+      "loss": 1.7103,
+      "step": 9274
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2075613079019074e-05,
+      "loss": 1.691,
+      "step": 9275
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2073569482288828e-05,
+      "loss": 1.5635,
+      "step": 9276
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2071525885558584e-05,
+      "loss": 1.6521,
+      "step": 9277
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2069482288828338e-05,
+      "loss": 1.7192,
+      "step": 9278
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2067438692098093e-05,
+      "loss": 1.6814,
+      "step": 9279
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2065395095367847e-05,
+      "loss": 1.7364,
+      "step": 9280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2063351498637603e-05,
+      "loss": 1.6345,
+      "step": 9281
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2061307901907358e-05,
+      "loss": 1.8417,
+      "step": 9282
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2059264305177112e-05,
+      "loss": 1.6729,
+      "step": 9283
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2057220708446866e-05,
+      "loss": 1.6093,
+      "step": 9284
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.205517711171662e-05,
+      "loss": 1.7235,
+      "step": 9285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2053133514986378e-05,
+      "loss": 1.5911,
+      "step": 9286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2051089918256132e-05,
+      "loss": 1.7231,
+      "step": 9287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2049046321525885e-05,
+      "loss": 1.7482,
+      "step": 9288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.204700272479564e-05,
+      "loss": 1.8032,
+      "step": 9289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2044959128065397e-05,
+      "loss": 1.8561,
+      "step": 9290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.204291553133515e-05,
+      "loss": 1.6465,
+      "step": 9291
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2040871934604905e-05,
+      "loss": 1.6967,
+      "step": 9292
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.203882833787466e-05,
+      "loss": 1.6513,
+      "step": 9293
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2036784741144414e-05,
+      "loss": 1.7251,
+      "step": 9294
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.203474114441417e-05,
+      "loss": 1.6916,
+      "step": 9295
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2032697547683924e-05,
+      "loss": 1.6533,
+      "step": 9296
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.203065395095368e-05,
+      "loss": 1.6631,
+      "step": 9297
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2028610354223433e-05,
+      "loss": 1.6281,
+      "step": 9298
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2026566757493189e-05,
+      "loss": 1.6468,
+      "step": 9299
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2024523160762943e-05,
+      "loss": 1.6743,
+      "step": 9300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2022479564032699e-05,
+      "loss": 1.6541,
+      "step": 9301
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2020435967302452e-05,
+      "loss": 1.6543,
+      "step": 9302
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2018392370572206e-05,
+      "loss": 1.5039,
+      "step": 9303
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2016348773841962e-05,
+      "loss": 1.6901,
+      "step": 9304
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2014305177111718e-05,
+      "loss": 1.5775,
+      "step": 9305
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2012261580381472e-05,
+      "loss": 1.7296,
+      "step": 9306
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2010217983651226e-05,
+      "loss": 1.7373,
+      "step": 9307
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2008174386920983e-05,
+      "loss": 1.7187,
+      "step": 9308
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2006130790190737e-05,
+      "loss": 1.5658,
+      "step": 9309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.200408719346049e-05,
+      "loss": 1.7201,
+      "step": 9310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2002043596730245e-05,
+      "loss": 1.6786,
+      "step": 9311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2e-05,
+      "loss": 1.7494,
+      "step": 9312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1997956403269756e-05,
+      "loss": 1.618,
+      "step": 9313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.199591280653951e-05,
+      "loss": 1.7609,
+      "step": 9314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1993869209809264e-05,
+      "loss": 1.6207,
+      "step": 9315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.199182561307902e-05,
+      "loss": 1.5684,
+      "step": 9316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1989782016348775e-05,
+      "loss": 1.6786,
+      "step": 9317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1987738419618529e-05,
+      "loss": 1.6771,
+      "step": 9318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1985694822888283e-05,
+      "loss": 1.7473,
+      "step": 9319
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1983651226158039e-05,
+      "loss": 1.8477,
+      "step": 9320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1981607629427793e-05,
+      "loss": 1.6987,
+      "step": 9321
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1979564032697548e-05,
+      "loss": 1.6936,
+      "step": 9322
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1977520435967304e-05,
+      "loss": 1.6887,
+      "step": 9323
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1975476839237058e-05,
+      "loss": 1.6539,
+      "step": 9324
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1973433242506812e-05,
+      "loss": 1.6201,
+      "step": 9325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1971389645776567e-05,
+      "loss": 1.5513,
+      "step": 9326
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1969346049046323e-05,
+      "loss": 1.7968,
+      "step": 9327
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1967302452316077e-05,
+      "loss": 1.7375,
+      "step": 9328
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.196525885558583e-05,
+      "loss": 1.642,
+      "step": 9329
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1963215258855585e-05,
+      "loss": 1.6684,
+      "step": 9330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1961171662125342e-05,
+      "loss": 1.8275,
+      "step": 9331
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1959128065395096e-05,
+      "loss": 1.5103,
+      "step": 9332
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.195708446866485e-05,
+      "loss": 1.7607,
+      "step": 9333
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1955040871934604e-05,
+      "loss": 1.7079,
+      "step": 9334
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.195299727520436e-05,
+      "loss": 1.7689,
+      "step": 9335
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1950953678474115e-05,
+      "loss": 1.8242,
+      "step": 9336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1948910081743869e-05,
+      "loss": 1.6445,
+      "step": 9337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1946866485013625e-05,
+      "loss": 1.8496,
+      "step": 9338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1944822888283379e-05,
+      "loss": 1.5535,
+      "step": 9339
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1942779291553134e-05,
+      "loss": 1.6923,
+      "step": 9340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1940735694822888e-05,
+      "loss": 1.6528,
+      "step": 9341
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1938692098092644e-05,
+      "loss": 1.6085,
+      "step": 9342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1936648501362398e-05,
+      "loss": 1.7147,
+      "step": 9343
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1934604904632152e-05,
+      "loss": 1.6347,
+      "step": 9344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1932561307901907e-05,
+      "loss": 1.5824,
+      "step": 9345
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1930517711171663e-05,
+      "loss": 1.5615,
+      "step": 9346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1928474114441417e-05,
+      "loss": 1.7584,
+      "step": 9347
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1926430517711171e-05,
+      "loss": 1.6779,
+      "step": 9348
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1924386920980928e-05,
+      "loss": 1.6738,
+      "step": 9349
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1922343324250682e-05,
+      "loss": 1.7261,
+      "step": 9350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1920299727520436e-05,
+      "loss": 1.7273,
+      "step": 9351
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.191825613079019e-05,
+      "loss": 1.9472,
+      "step": 9352
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1916212534059946e-05,
+      "loss": 1.7059,
+      "step": 9353
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1914168937329701e-05,
+      "loss": 1.5499,
+      "step": 9354
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1912125340599455e-05,
+      "loss": 1.5813,
+      "step": 9355
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.191008174386921e-05,
+      "loss": 1.6722,
+      "step": 9356
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1908038147138965e-05,
+      "loss": 1.8548,
+      "step": 9357
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.190599455040872e-05,
+      "loss": 1.7299,
+      "step": 9358
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1903950953678474e-05,
+      "loss": 1.77,
+      "step": 9359
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1901907356948228e-05,
+      "loss": 1.6977,
+      "step": 9360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1899863760217984e-05,
+      "loss": 1.6101,
+      "step": 9361
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1897820163487738e-05,
+      "loss": 1.7928,
+      "step": 9362
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1895776566757494e-05,
+      "loss": 1.7508,
+      "step": 9363
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.189373297002725e-05,
+      "loss": 1.4521,
+      "step": 9364
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1891689373297003e-05,
+      "loss": 1.9639,
+      "step": 9365
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1889645776566757e-05,
+      "loss": 1.6836,
+      "step": 9366
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1887602179836513e-05,
+      "loss": 1.7159,
+      "step": 9367
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1885558583106268e-05,
+      "loss": 1.6477,
+      "step": 9368
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1883514986376022e-05,
+      "loss": 1.742,
+      "step": 9369
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1881471389645776e-05,
+      "loss": 1.7159,
+      "step": 9370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.187942779291553e-05,
+      "loss": 1.6292,
+      "step": 9371
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1877384196185287e-05,
+      "loss": 1.6315,
+      "step": 9372
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1875340599455041e-05,
+      "loss": 1.6222,
+      "step": 9373
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1873297002724795e-05,
+      "loss": 1.7304,
+      "step": 9374
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.187125340599455e-05,
+      "loss": 1.6691,
+      "step": 9375
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1869209809264307e-05,
+      "loss": 1.6579,
+      "step": 9376
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.186716621253406e-05,
+      "loss": 1.6365,
+      "step": 9377
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1865122615803814e-05,
+      "loss": 1.6849,
+      "step": 9378
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.186307901907357e-05,
+      "loss": 1.8679,
+      "step": 9379
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1861035422343324e-05,
+      "loss": 1.6456,
+      "step": 9380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.185899182561308e-05,
+      "loss": 1.6687,
+      "step": 9381
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1856948228882834e-05,
+      "loss": 1.6708,
+      "step": 9382
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.185490463215259e-05,
+      "loss": 1.6927,
+      "step": 9383
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1852861035422343e-05,
+      "loss": 1.6949,
+      "step": 9384
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1850817438692099e-05,
+      "loss": 1.6065,
+      "step": 9385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1848773841961853e-05,
+      "loss": 1.6433,
+      "step": 9386
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1846730245231608e-05,
+      "loss": 1.5685,
+      "step": 9387
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1844686648501362e-05,
+      "loss": 1.6396,
+      "step": 9388
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1842643051771116e-05,
+      "loss": 1.7563,
+      "step": 9389
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1840599455040874e-05,
+      "loss": 1.6602,
+      "step": 9390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1838555858310628e-05,
+      "loss": 1.7564,
+      "step": 9391
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1836512261580381e-05,
+      "loss": 1.6465,
+      "step": 9392
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1834468664850135e-05,
+      "loss": 1.7746,
+      "step": 9393
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1832425068119893e-05,
+      "loss": 1.7843,
+      "step": 9394
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1830381471389647e-05,
+      "loss": 1.6383,
+      "step": 9395
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.18283378746594e-05,
+      "loss": 1.7077,
+      "step": 9396
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1826294277929155e-05,
+      "loss": 1.5895,
+      "step": 9397
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.182425068119891e-05,
+      "loss": 1.6984,
+      "step": 9398
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1822207084468666e-05,
+      "loss": 1.6907,
+      "step": 9399
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.182016348773842e-05,
+      "loss": 1.6472,
+      "step": 9400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1818119891008174e-05,
+      "loss": 1.6973,
+      "step": 9401
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.181607629427793e-05,
+      "loss": 1.6903,
+      "step": 9402
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1814032697547685e-05,
+      "loss": 1.5571,
+      "step": 9403
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1811989100817439e-05,
+      "loss": 1.667,
+      "step": 9404
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1809945504087195e-05,
+      "loss": 1.6871,
+      "step": 9405
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1807901907356948e-05,
+      "loss": 1.674,
+      "step": 9406
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1805858310626702e-05,
+      "loss": 1.6638,
+      "step": 9407
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1803814713896458e-05,
+      "loss": 1.8081,
+      "step": 9408
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1801771117166214e-05,
+      "loss": 1.8588,
+      "step": 9409
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1799727520435968e-05,
+      "loss": 1.5852,
+      "step": 9410
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1797683923705722e-05,
+      "loss": 1.647,
+      "step": 9411
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1795640326975477e-05,
+      "loss": 1.7661,
+      "step": 9412
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1793596730245233e-05,
+      "loss": 1.6771,
+      "step": 9413
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1791553133514987e-05,
+      "loss": 1.6795,
+      "step": 9414
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.178950953678474e-05,
+      "loss": 1.7022,
+      "step": 9415
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1787465940054495e-05,
+      "loss": 1.63,
+      "step": 9416
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1785422343324252e-05,
+      "loss": 1.7084,
+      "step": 9417
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1783378746594006e-05,
+      "loss": 1.757,
+      "step": 9418
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.178133514986376e-05,
+      "loss": 1.7141,
+      "step": 9419
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1779291553133515e-05,
+      "loss": 1.814,
+      "step": 9420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1777247956403271e-05,
+      "loss": 1.698,
+      "step": 9421
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1775204359673025e-05,
+      "loss": 1.5992,
+      "step": 9422
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1773160762942779e-05,
+      "loss": 1.7541,
+      "step": 9423
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1771117166212535e-05,
+      "loss": 1.6907,
+      "step": 9424
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1769073569482289e-05,
+      "loss": 1.6205,
+      "step": 9425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1767029972752044e-05,
+      "loss": 1.6786,
+      "step": 9426
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1764986376021798e-05,
+      "loss": 1.7362,
+      "step": 9427
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1762942779291554e-05,
+      "loss": 1.6519,
+      "step": 9428
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1760899182561308e-05,
+      "loss": 1.6679,
+      "step": 9429
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1758855585831063e-05,
+      "loss": 1.7922,
+      "step": 9430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1756811989100819e-05,
+      "loss": 1.7018,
+      "step": 9431
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1754768392370573e-05,
+      "loss": 1.7048,
+      "step": 9432
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1752724795640327e-05,
+      "loss": 1.5962,
+      "step": 9433
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.175068119891008e-05,
+      "loss": 1.6932,
+      "step": 9434
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1748637602179838e-05,
+      "loss": 1.7773,
+      "step": 9435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1746594005449592e-05,
+      "loss": 1.7216,
+      "step": 9436
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1744550408719346e-05,
+      "loss": 1.6368,
+      "step": 9437
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.17425068119891e-05,
+      "loss": 1.7628,
+      "step": 9438
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1740463215258857e-05,
+      "loss": 1.6713,
+      "step": 9439
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1738419618528611e-05,
+      "loss": 1.8032,
+      "step": 9440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1736376021798365e-05,
+      "loss": 1.5991,
+      "step": 9441
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1734332425068119e-05,
+      "loss": 1.6775,
+      "step": 9442
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1732288828337875e-05,
+      "loss": 1.7316,
+      "step": 9443
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.173024523160763e-05,
+      "loss": 1.7676,
+      "step": 9444
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1728201634877384e-05,
+      "loss": 1.5796,
+      "step": 9445
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.172615803814714e-05,
+      "loss": 1.7058,
+      "step": 9446
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1724114441416894e-05,
+      "loss": 1.6184,
+      "step": 9447
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.172207084468665e-05,
+      "loss": 1.6618,
+      "step": 9448
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1720027247956403e-05,
+      "loss": 1.6772,
+      "step": 9449
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1717983651226159e-05,
+      "loss": 1.6997,
+      "step": 9450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1715940054495913e-05,
+      "loss": 1.6703,
+      "step": 9451
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1713896457765667e-05,
+      "loss": 1.6519,
+      "step": 9452
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1711852861035423e-05,
+      "loss": 1.7916,
+      "step": 9453
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1709809264305178e-05,
+      "loss": 1.7264,
+      "step": 9454
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1707765667574932e-05,
+      "loss": 1.8329,
+      "step": 9455
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1705722070844686e-05,
+      "loss": 1.6698,
+      "step": 9456
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1703678474114442e-05,
+      "loss": 1.6338,
+      "step": 9457
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1701634877384197e-05,
+      "loss": 1.6426,
+      "step": 9458
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1699591280653951e-05,
+      "loss": 1.7114,
+      "step": 9459
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1697547683923705e-05,
+      "loss": 1.6433,
+      "step": 9460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.169550408719346e-05,
+      "loss": 1.6317,
+      "step": 9461
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1693460490463216e-05,
+      "loss": 1.5832,
+      "step": 9462
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.169141689373297e-05,
+      "loss": 1.6263,
+      "step": 9463
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1689373297002724e-05,
+      "loss": 1.5911,
+      "step": 9464
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.168732970027248e-05,
+      "loss": 1.6476,
+      "step": 9465
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1685286103542236e-05,
+      "loss": 1.755,
+      "step": 9466
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.168324250681199e-05,
+      "loss": 1.7325,
+      "step": 9467
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1681198910081743e-05,
+      "loss": 1.6306,
+      "step": 9468
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1679155313351499e-05,
+      "loss": 1.6493,
+      "step": 9469
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1677111716621253e-05,
+      "loss": 1.6798,
+      "step": 9470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1675068119891009e-05,
+      "loss": 1.6998,
+      "step": 9471
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1673024523160763e-05,
+      "loss": 1.6947,
+      "step": 9472
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1670980926430518e-05,
+      "loss": 1.8269,
+      "step": 9473
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1668937329700272e-05,
+      "loss": 1.652,
+      "step": 9474
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1666893732970028e-05,
+      "loss": 1.7159,
+      "step": 9475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1664850136239783e-05,
+      "loss": 1.6245,
+      "step": 9476
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1662806539509537e-05,
+      "loss": 1.6958,
+      "step": 9477
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1660762942779291e-05,
+      "loss": 1.7257,
+      "step": 9478
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1658719346049045e-05,
+      "loss": 1.6858,
+      "step": 9479
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1656675749318803e-05,
+      "loss": 1.7301,
+      "step": 9480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1654632152588557e-05,
+      "loss": 1.6985,
+      "step": 9481
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.165258855585831e-05,
+      "loss": 1.6063,
+      "step": 9482
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1650544959128064e-05,
+      "loss": 1.7668,
+      "step": 9483
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1648501362397822e-05,
+      "loss": 1.7499,
+      "step": 9484
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1646457765667576e-05,
+      "loss": 1.7012,
+      "step": 9485
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.164441416893733e-05,
+      "loss": 1.7788,
+      "step": 9486
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1642370572207085e-05,
+      "loss": 1.7363,
+      "step": 9487
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.164032697547684e-05,
+      "loss": 1.619,
+      "step": 9488
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1638283378746595e-05,
+      "loss": 1.7485,
+      "step": 9489
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1636239782016349e-05,
+      "loss": 1.8122,
+      "step": 9490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1634196185286104e-05,
+      "loss": 1.7077,
+      "step": 9491
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1632152588555858e-05,
+      "loss": 1.7144,
+      "step": 9492
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1630108991825614e-05,
+      "loss": 1.6372,
+      "step": 9493
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1628065395095368e-05,
+      "loss": 1.6959,
+      "step": 9494
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1626021798365124e-05,
+      "loss": 1.6848,
+      "step": 9495
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1623978201634877e-05,
+      "loss": 1.7686,
+      "step": 9496
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1621934604904631e-05,
+      "loss": 1.6899,
+      "step": 9497
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1619891008174387e-05,
+      "loss": 1.6856,
+      "step": 9498
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1617847411444143e-05,
+      "loss": 1.6577,
+      "step": 9499
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1615803814713897e-05,
+      "loss": 1.6453,
+      "step": 9500
+    },
+    {
+      "epoch": 1.88,
+      "eval_loss": 1.5230976343154907,
+      "eval_runtime": 721.121,
+      "eval_samples_per_second": 5.755,
+      "eval_steps_per_second": 1.439,
+      "step": 9500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.161376021798365e-05,
+      "loss": 1.6769,
+      "step": 9501
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1611716621253408e-05,
+      "loss": 1.7239,
+      "step": 9502
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1609673024523162e-05,
+      "loss": 1.5828,
+      "step": 9503
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1607629427792916e-05,
+      "loss": 1.7394,
+      "step": 9504
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.160558583106267e-05,
+      "loss": 1.8058,
+      "step": 9505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1603542234332425e-05,
+      "loss": 1.7117,
+      "step": 9506
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1601498637602181e-05,
+      "loss": 1.7931,
+      "step": 9507
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1599455040871935e-05,
+      "loss": 1.676,
+      "step": 9508
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1597411444141689e-05,
+      "loss": 1.6998,
+      "step": 9509
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1595367847411444e-05,
+      "loss": 1.8116,
+      "step": 9510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.15933242506812e-05,
+      "loss": 1.5204,
+      "step": 9511
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1591280653950954e-05,
+      "loss": 1.5845,
+      "step": 9512
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1589237057220708e-05,
+      "loss": 1.7063,
+      "step": 9513
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1587193460490464e-05,
+      "loss": 1.6959,
+      "step": 9514
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1585149863760218e-05,
+      "loss": 1.7304,
+      "step": 9515
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1583106267029973e-05,
+      "loss": 1.6716,
+      "step": 9516
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1581062670299729e-05,
+      "loss": 1.687,
+      "step": 9517
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1579019073569483e-05,
+      "loss": 1.8051,
+      "step": 9518
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1576975476839237e-05,
+      "loss": 1.699,
+      "step": 9519
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1574931880108992e-05,
+      "loss": 1.7332,
+      "step": 9520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1572888283378748e-05,
+      "loss": 1.9135,
+      "step": 9521
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1570844686648502e-05,
+      "loss": 1.69,
+      "step": 9522
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1568801089918256e-05,
+      "loss": 1.7743,
+      "step": 9523
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.156675749318801e-05,
+      "loss": 1.6298,
+      "step": 9524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1564713896457767e-05,
+      "loss": 1.7543,
+      "step": 9525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1562670299727521e-05,
+      "loss": 1.6956,
+      "step": 9526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1560626702997275e-05,
+      "loss": 1.6115,
+      "step": 9527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.155858310626703e-05,
+      "loss": 1.6323,
+      "step": 9528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1556539509536785e-05,
+      "loss": 1.6608,
+      "step": 9529
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.155449591280654e-05,
+      "loss": 1.7747,
+      "step": 9530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1552452316076294e-05,
+      "loss": 1.695,
+      "step": 9531
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.155040871934605e-05,
+      "loss": 1.8004,
+      "step": 9532
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1548365122615804e-05,
+      "loss": 1.7733,
+      "step": 9533
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.154632152588556e-05,
+      "loss": 1.8073,
+      "step": 9534
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1544277929155313e-05,
+      "loss": 1.6701,
+      "step": 9535
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1542234332425069e-05,
+      "loss": 1.7508,
+      "step": 9536
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1540190735694823e-05,
+      "loss": 1.7419,
+      "step": 9537
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1538147138964577e-05,
+      "loss": 1.5168,
+      "step": 9538
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1536103542234332e-05,
+      "loss": 1.631,
+      "step": 9539
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1534059945504088e-05,
+      "loss": 1.7883,
+      "step": 9540
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1532016348773842e-05,
+      "loss": 1.7045,
+      "step": 9541
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1529972752043596e-05,
+      "loss": 1.7078,
+      "step": 9542
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1527929155313353e-05,
+      "loss": 1.6964,
+      "step": 9543
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1525885558583107e-05,
+      "loss": 1.7368,
+      "step": 9544
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1523841961852861e-05,
+      "loss": 1.6589,
+      "step": 9545
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1521798365122615e-05,
+      "loss": 1.6994,
+      "step": 9546
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.151975476839237e-05,
+      "loss": 1.6603,
+      "step": 9547
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1517711171662126e-05,
+      "loss": 1.6016,
+      "step": 9548
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.151566757493188e-05,
+      "loss": 1.6945,
+      "step": 9549
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1513623978201634e-05,
+      "loss": 1.7782,
+      "step": 9550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.151158038147139e-05,
+      "loss": 1.6049,
+      "step": 9551
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1509536784741145e-05,
+      "loss": 1.6015,
+      "step": 9552
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.15074931880109e-05,
+      "loss": 1.6986,
+      "step": 9553
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1505449591280653e-05,
+      "loss": 1.6597,
+      "step": 9554
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1503405994550409e-05,
+      "loss": 1.6938,
+      "step": 9555
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1501362397820163e-05,
+      "loss": 1.8515,
+      "step": 9556
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1499318801089919e-05,
+      "loss": 1.5903,
+      "step": 9557
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1497275204359674e-05,
+      "loss": 1.5522,
+      "step": 9558
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1495231607629428e-05,
+      "loss": 1.7947,
+      "step": 9559
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1493188010899182e-05,
+      "loss": 1.5788,
+      "step": 9560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1491144414168938e-05,
+      "loss": 1.682,
+      "step": 9561
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1489100817438693e-05,
+      "loss": 1.6705,
+      "step": 9562
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1487057220708447e-05,
+      "loss": 1.5797,
+      "step": 9563
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1485013623978201e-05,
+      "loss": 1.7376,
+      "step": 9564
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1482970027247955e-05,
+      "loss": 1.681,
+      "step": 9565
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1480926430517712e-05,
+      "loss": 1.5981,
+      "step": 9566
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1478882833787466e-05,
+      "loss": 1.821,
+      "step": 9567
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.147683923705722e-05,
+      "loss": 1.6251,
+      "step": 9568
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1474795640326976e-05,
+      "loss": 1.6438,
+      "step": 9569
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1472752043596732e-05,
+      "loss": 1.7063,
+      "step": 9570
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1470708446866486e-05,
+      "loss": 1.6808,
+      "step": 9571
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.146866485013624e-05,
+      "loss": 1.7999,
+      "step": 9572
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1466621253405995e-05,
+      "loss": 1.6229,
+      "step": 9573
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1464577656675749e-05,
+      "loss": 1.6231,
+      "step": 9574
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1462534059945505e-05,
+      "loss": 1.6937,
+      "step": 9575
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1460490463215259e-05,
+      "loss": 1.5739,
+      "step": 9576
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1458446866485014e-05,
+      "loss": 1.6486,
+      "step": 9577
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1456403269754768e-05,
+      "loss": 1.6477,
+      "step": 9578
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1454359673024524e-05,
+      "loss": 1.6239,
+      "step": 9579
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1452316076294278e-05,
+      "loss": 1.6436,
+      "step": 9580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1450272479564033e-05,
+      "loss": 1.5842,
+      "step": 9581
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1448228882833787e-05,
+      "loss": 1.8246,
+      "step": 9582
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1446185286103541e-05,
+      "loss": 1.7381,
+      "step": 9583
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1444141689373299e-05,
+      "loss": 1.6604,
+      "step": 9584
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1442098092643053e-05,
+      "loss": 1.7208,
+      "step": 9585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1440054495912806e-05,
+      "loss": 1.551,
+      "step": 9586
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.143801089918256e-05,
+      "loss": 1.6514,
+      "step": 9587
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1435967302452318e-05,
+      "loss": 1.8023,
+      "step": 9588
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1433923705722072e-05,
+      "loss": 1.6356,
+      "step": 9589
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1431880108991826e-05,
+      "loss": 1.6441,
+      "step": 9590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.142983651226158e-05,
+      "loss": 1.755,
+      "step": 9591
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1427792915531335e-05,
+      "loss": 1.6586,
+      "step": 9592
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.142574931880109e-05,
+      "loss": 1.7795,
+      "step": 9593
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1423705722070845e-05,
+      "loss": 1.6221,
+      "step": 9594
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1421662125340599e-05,
+      "loss": 1.7223,
+      "step": 9595
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1419618528610354e-05,
+      "loss": 1.7508,
+      "step": 9596
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.141757493188011e-05,
+      "loss": 1.7288,
+      "step": 9597
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1415531335149864e-05,
+      "loss": 1.8195,
+      "step": 9598
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.141348773841962e-05,
+      "loss": 1.6683,
+      "step": 9599
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1411444141689373e-05,
+      "loss": 1.7515,
+      "step": 9600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1409400544959127e-05,
+      "loss": 1.6249,
+      "step": 9601
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1407356948228883e-05,
+      "loss": 1.7807,
+      "step": 9602
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1405313351498639e-05,
+      "loss": 1.6206,
+      "step": 9603
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1403269754768393e-05,
+      "loss": 1.7284,
+      "step": 9604
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1401226158038147e-05,
+      "loss": 1.7131,
+      "step": 9605
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1399182561307902e-05,
+      "loss": 1.7061,
+      "step": 9606
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1397138964577658e-05,
+      "loss": 1.7838,
+      "step": 9607
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1395095367847412e-05,
+      "loss": 1.5951,
+      "step": 9608
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1393051771117166e-05,
+      "loss": 1.764,
+      "step": 9609
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.139100817438692e-05,
+      "loss": 1.6832,
+      "step": 9610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1388964577656677e-05,
+      "loss": 1.6448,
+      "step": 9611
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1386920980926431e-05,
+      "loss": 1.618,
+      "step": 9612
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1384877384196185e-05,
+      "loss": 1.8853,
+      "step": 9613
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.138283378746594e-05,
+      "loss": 1.6696,
+      "step": 9614
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1380790190735696e-05,
+      "loss": 1.6685,
+      "step": 9615
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.137874659400545e-05,
+      "loss": 1.7376,
+      "step": 9616
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1376702997275204e-05,
+      "loss": 1.6416,
+      "step": 9617
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.137465940054496e-05,
+      "loss": 1.6897,
+      "step": 9618
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1372615803814714e-05,
+      "loss": 1.7543,
+      "step": 9619
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.137057220708447e-05,
+      "loss": 1.7438,
+      "step": 9620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1368528610354223e-05,
+      "loss": 1.6204,
+      "step": 9621
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1366485013623979e-05,
+      "loss": 1.7341,
+      "step": 9622
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1364441416893733e-05,
+      "loss": 1.5848,
+      "step": 9623
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1362397820163488e-05,
+      "loss": 1.5755,
+      "step": 9624
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1360354223433244e-05,
+      "loss": 1.6642,
+      "step": 9625
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1358310626702998e-05,
+      "loss": 1.6687,
+      "step": 9626
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1356267029972752e-05,
+      "loss": 1.6866,
+      "step": 9627
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1354223433242506e-05,
+      "loss": 1.5833,
+      "step": 9628
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1352179836512263e-05,
+      "loss": 1.5767,
+      "step": 9629
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1350136239782017e-05,
+      "loss": 1.7339,
+      "step": 9630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1348092643051771e-05,
+      "loss": 1.5882,
+      "step": 9631
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1346049046321525e-05,
+      "loss": 1.6898,
+      "step": 9632
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1344005449591282e-05,
+      "loss": 1.7667,
+      "step": 9633
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1341961852861036e-05,
+      "loss": 1.6735,
+      "step": 9634
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.133991825613079e-05,
+      "loss": 1.8898,
+      "step": 9635
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1337874659400544e-05,
+      "loss": 1.6418,
+      "step": 9636
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.13358310626703e-05,
+      "loss": 1.6535,
+      "step": 9637
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1333787465940055e-05,
+      "loss": 1.7209,
+      "step": 9638
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.133174386920981e-05,
+      "loss": 1.7182,
+      "step": 9639
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1329700272479565e-05,
+      "loss": 1.6883,
+      "step": 9640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1327656675749319e-05,
+      "loss": 1.7532,
+      "step": 9641
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1325613079019074e-05,
+      "loss": 1.6416,
+      "step": 9642
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1323569482288828e-05,
+      "loss": 1.7615,
+      "step": 9643
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1321525885558584e-05,
+      "loss": 1.6639,
+      "step": 9644
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1319482288828338e-05,
+      "loss": 1.5739,
+      "step": 9645
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1317438692098092e-05,
+      "loss": 1.6038,
+      "step": 9646
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1315395095367848e-05,
+      "loss": 1.7686,
+      "step": 9647
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1313351498637603e-05,
+      "loss": 1.6651,
+      "step": 9648
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1311307901907357e-05,
+      "loss": 1.6763,
+      "step": 9649
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1309264305177111e-05,
+      "loss": 1.5934,
+      "step": 9650
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1307220708446867e-05,
+      "loss": 1.6274,
+      "step": 9651
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1305177111716622e-05,
+      "loss": 1.6708,
+      "step": 9652
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1303133514986376e-05,
+      "loss": 1.639,
+      "step": 9653
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.130108991825613e-05,
+      "loss": 1.7398,
+      "step": 9654
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1299046321525886e-05,
+      "loss": 1.9091,
+      "step": 9655
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1297002724795641e-05,
+      "loss": 1.5536,
+      "step": 9656
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1294959128065395e-05,
+      "loss": 1.6343,
+      "step": 9657
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.129291553133515e-05,
+      "loss": 1.7903,
+      "step": 9658
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1290871934604905e-05,
+      "loss": 1.736,
+      "step": 9659
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.128882833787466e-05,
+      "loss": 1.7627,
+      "step": 9660
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1286784741144415e-05,
+      "loss": 1.7322,
+      "step": 9661
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1284741144414169e-05,
+      "loss": 1.624,
+      "step": 9662
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1282697547683924e-05,
+      "loss": 1.6779,
+      "step": 9663
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1280653950953678e-05,
+      "loss": 1.7487,
+      "step": 9664
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1278610354223434e-05,
+      "loss": 1.6555,
+      "step": 9665
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.127656675749319e-05,
+      "loss": 1.6988,
+      "step": 9666
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1274523160762943e-05,
+      "loss": 1.6635,
+      "step": 9667
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1272479564032697e-05,
+      "loss": 1.7144,
+      "step": 9668
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1270435967302453e-05,
+      "loss": 1.6771,
+      "step": 9669
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1268392370572208e-05,
+      "loss": 1.7519,
+      "step": 9670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1266348773841962e-05,
+      "loss": 1.7859,
+      "step": 9671
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1264305177111716e-05,
+      "loss": 1.7546,
+      "step": 9672
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.126226158038147e-05,
+      "loss": 1.6977,
+      "step": 9673
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1260217983651228e-05,
+      "loss": 1.658,
+      "step": 9674
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1258174386920982e-05,
+      "loss": 1.7275,
+      "step": 9675
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1256130790190736e-05,
+      "loss": 1.7869,
+      "step": 9676
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.125408719346049e-05,
+      "loss": 1.6931,
+      "step": 9677
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1252043596730247e-05,
+      "loss": 1.6062,
+      "step": 9678
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.125e-05,
+      "loss": 1.8289,
+      "step": 9679
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1247956403269755e-05,
+      "loss": 1.6585,
+      "step": 9680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.124591280653951e-05,
+      "loss": 1.7745,
+      "step": 9681
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1243869209809264e-05,
+      "loss": 1.6249,
+      "step": 9682
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.124182561307902e-05,
+      "loss": 1.6008,
+      "step": 9683
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1239782016348774e-05,
+      "loss": 1.693,
+      "step": 9684
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.123773841961853e-05,
+      "loss": 1.7464,
+      "step": 9685
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1235694822888283e-05,
+      "loss": 1.755,
+      "step": 9686
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1233651226158039e-05,
+      "loss": 1.646,
+      "step": 9687
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1231607629427793e-05,
+      "loss": 1.6902,
+      "step": 9688
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1229564032697549e-05,
+      "loss": 1.7344,
+      "step": 9689
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1227520435967303e-05,
+      "loss": 1.5794,
+      "step": 9690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1225476839237056e-05,
+      "loss": 1.7215,
+      "step": 9691
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1223433242506812e-05,
+      "loss": 1.6715,
+      "step": 9692
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1221389645776568e-05,
+      "loss": 1.5179,
+      "step": 9693
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1219346049046322e-05,
+      "loss": 1.7399,
+      "step": 9694
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1217302452316076e-05,
+      "loss": 1.6833,
+      "step": 9695
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1215258855585833e-05,
+      "loss": 1.8025,
+      "step": 9696
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1213215258855587e-05,
+      "loss": 1.6574,
+      "step": 9697
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.121117166212534e-05,
+      "loss": 1.6845,
+      "step": 9698
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1209128065395095e-05,
+      "loss": 1.709,
+      "step": 9699
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.120708446866485e-05,
+      "loss": 1.6641,
+      "step": 9700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1205040871934606e-05,
+      "loss": 1.8161,
+      "step": 9701
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.120299727520436e-05,
+      "loss": 1.7581,
+      "step": 9702
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1200953678474114e-05,
+      "loss": 1.6578,
+      "step": 9703
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.119891008174387e-05,
+      "loss": 1.7064,
+      "step": 9704
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1196866485013625e-05,
+      "loss": 1.5804,
+      "step": 9705
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1194822888283379e-05,
+      "loss": 1.5504,
+      "step": 9706
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1192779291553135e-05,
+      "loss": 1.8258,
+      "step": 9707
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1190735694822889e-05,
+      "loss": 1.7545,
+      "step": 9708
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1188692098092643e-05,
+      "loss": 1.6824,
+      "step": 9709
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1186648501362398e-05,
+      "loss": 1.7187,
+      "step": 9710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1184604904632154e-05,
+      "loss": 1.6217,
+      "step": 9711
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1182561307901908e-05,
+      "loss": 1.6727,
+      "step": 9712
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1180517711171662e-05,
+      "loss": 1.7536,
+      "step": 9713
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1178474114441417e-05,
+      "loss": 1.5795,
+      "step": 9714
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1176430517711173e-05,
+      "loss": 1.6545,
+      "step": 9715
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1174386920980927e-05,
+      "loss": 1.6861,
+      "step": 9716
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1172343324250681e-05,
+      "loss": 1.5997,
+      "step": 9717
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1170299727520435e-05,
+      "loss": 1.7989,
+      "step": 9718
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1168256130790192e-05,
+      "loss": 1.7944,
+      "step": 9719
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1166212534059946e-05,
+      "loss": 1.6326,
+      "step": 9720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.11641689373297e-05,
+      "loss": 1.6831,
+      "step": 9721
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1162125340599456e-05,
+      "loss": 1.7187,
+      "step": 9722
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.116008174386921e-05,
+      "loss": 1.7702,
+      "step": 9723
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1158038147138965e-05,
+      "loss": 1.7028,
+      "step": 9724
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.115599455040872e-05,
+      "loss": 1.8141,
+      "step": 9725
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1153950953678475e-05,
+      "loss": 1.6067,
+      "step": 9726
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1151907356948229e-05,
+      "loss": 1.602,
+      "step": 9727
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1149863760217984e-05,
+      "loss": 1.7062,
+      "step": 9728
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1147820163487738e-05,
+      "loss": 1.7153,
+      "step": 9729
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1145776566757494e-05,
+      "loss": 1.6854,
+      "step": 9730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1143732970027248e-05,
+      "loss": 1.7096,
+      "step": 9731
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1141689373297004e-05,
+      "loss": 1.7231,
+      "step": 9732
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1139645776566757e-05,
+      "loss": 1.7323,
+      "step": 9733
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1137602179836513e-05,
+      "loss": 1.6367,
+      "step": 9734
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1135558583106267e-05,
+      "loss": 1.6198,
+      "step": 9735
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1133514986376021e-05,
+      "loss": 1.6416,
+      "step": 9736
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1131471389645778e-05,
+      "loss": 1.5488,
+      "step": 9737
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1129427792915532e-05,
+      "loss": 1.7043,
+      "step": 9738
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1127384196185286e-05,
+      "loss": 1.7799,
+      "step": 9739
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.112534059945504e-05,
+      "loss": 1.6768,
+      "step": 9740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1123297002724796e-05,
+      "loss": 1.7648,
+      "step": 9741
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1121253405994551e-05,
+      "loss": 1.5729,
+      "step": 9742
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1119209809264305e-05,
+      "loss": 1.6608,
+      "step": 9743
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.111716621253406e-05,
+      "loss": 1.7997,
+      "step": 9744
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1115122615803815e-05,
+      "loss": 1.7409,
+      "step": 9745
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.111307901907357e-05,
+      "loss": 1.6557,
+      "step": 9746
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1111035422343324e-05,
+      "loss": 1.5736,
+      "step": 9747
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.110899182561308e-05,
+      "loss": 1.7723,
+      "step": 9748
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1106948228882834e-05,
+      "loss": 1.7147,
+      "step": 9749
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1104904632152588e-05,
+      "loss": 1.7305,
+      "step": 9750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1102861035422344e-05,
+      "loss": 1.8004,
+      "step": 9751
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.11008174386921e-05,
+      "loss": 1.6765,
+      "step": 9752
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1098773841961853e-05,
+      "loss": 1.6749,
+      "step": 9753
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1096730245231607e-05,
+      "loss": 1.6514,
+      "step": 9754
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1094686648501363e-05,
+      "loss": 1.7272,
+      "step": 9755
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1092643051771118e-05,
+      "loss": 1.5489,
+      "step": 9756
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1090599455040872e-05,
+      "loss": 1.6608,
+      "step": 9757
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1088555858310626e-05,
+      "loss": 1.8683,
+      "step": 9758
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.108651226158038e-05,
+      "loss": 1.7164,
+      "step": 9759
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1084468664850137e-05,
+      "loss": 1.6515,
+      "step": 9760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1082425068119891e-05,
+      "loss": 1.7953,
+      "step": 9761
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1080381471389645e-05,
+      "loss": 1.7166,
+      "step": 9762
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1078337874659401e-05,
+      "loss": 1.7927,
+      "step": 9763
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1076294277929157e-05,
+      "loss": 1.7113,
+      "step": 9764
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.107425068119891e-05,
+      "loss": 1.6159,
+      "step": 9765
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1072207084468665e-05,
+      "loss": 1.5699,
+      "step": 9766
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.107016348773842e-05,
+      "loss": 1.749,
+      "step": 9767
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1068119891008174e-05,
+      "loss": 1.6327,
+      "step": 9768
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.106607629427793e-05,
+      "loss": 1.7054,
+      "step": 9769
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1064032697547684e-05,
+      "loss": 1.7945,
+      "step": 9770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.106198910081744e-05,
+      "loss": 1.6828,
+      "step": 9771
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1059945504087193e-05,
+      "loss": 1.6775,
+      "step": 9772
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1057901907356949e-05,
+      "loss": 1.6759,
+      "step": 9773
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1055858310626703e-05,
+      "loss": 1.6493,
+      "step": 9774
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1053814713896458e-05,
+      "loss": 1.829,
+      "step": 9775
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1051771117166212e-05,
+      "loss": 1.7724,
+      "step": 9776
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1049727520435966e-05,
+      "loss": 1.6762,
+      "step": 9777
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1047683923705724e-05,
+      "loss": 1.7725,
+      "step": 9778
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1045640326975478e-05,
+      "loss": 1.7027,
+      "step": 9779
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1043596730245232e-05,
+      "loss": 1.7013,
+      "step": 9780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1041553133514985e-05,
+      "loss": 1.648,
+      "step": 9781
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1039509536784743e-05,
+      "loss": 1.7465,
+      "step": 9782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1037465940054497e-05,
+      "loss": 1.5202,
+      "step": 9783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.103542234332425e-05,
+      "loss": 1.6177,
+      "step": 9784
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1033378746594005e-05,
+      "loss": 1.7216,
+      "step": 9785
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.103133514986376e-05,
+      "loss": 1.7467,
+      "step": 9786
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1029291553133516e-05,
+      "loss": 1.7558,
+      "step": 9787
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.102724795640327e-05,
+      "loss": 1.683,
+      "step": 9788
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1025204359673024e-05,
+      "loss": 1.6126,
+      "step": 9789
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.102316076294278e-05,
+      "loss": 1.693,
+      "step": 9790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1021117166212535e-05,
+      "loss": 1.5799,
+      "step": 9791
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1019073569482289e-05,
+      "loss": 1.6605,
+      "step": 9792
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1017029972752045e-05,
+      "loss": 1.6982,
+      "step": 9793
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1014986376021799e-05,
+      "loss": 1.6978,
+      "step": 9794
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1012942779291552e-05,
+      "loss": 1.6995,
+      "step": 9795
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1010899182561308e-05,
+      "loss": 1.6994,
+      "step": 9796
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1008855585831064e-05,
+      "loss": 1.7285,
+      "step": 9797
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1006811989100818e-05,
+      "loss": 1.7343,
+      "step": 9798
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1004768392370572e-05,
+      "loss": 1.6495,
+      "step": 9799
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1002724795640327e-05,
+      "loss": 1.7354,
+      "step": 9800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1000681198910083e-05,
+      "loss": 1.8198,
+      "step": 9801
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0998637602179837e-05,
+      "loss": 1.7274,
+      "step": 9802
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.099659400544959e-05,
+      "loss": 1.7469,
+      "step": 9803
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0994550408719346e-05,
+      "loss": 1.8505,
+      "step": 9804
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0992506811989102e-05,
+      "loss": 1.7663,
+      "step": 9805
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0990463215258856e-05,
+      "loss": 1.5561,
+      "step": 9806
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.098841961852861e-05,
+      "loss": 1.545,
+      "step": 9807
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0986376021798366e-05,
+      "loss": 1.803,
+      "step": 9808
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0984332425068121e-05,
+      "loss": 1.643,
+      "step": 9809
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0982288828337875e-05,
+      "loss": 1.7106,
+      "step": 9810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0980245231607629e-05,
+      "loss": 1.8373,
+      "step": 9811
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0978201634877385e-05,
+      "loss": 1.5263,
+      "step": 9812
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0976158038147139e-05,
+      "loss": 1.8485,
+      "step": 9813
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0974114441416894e-05,
+      "loss": 1.6579,
+      "step": 9814
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0972070844686648e-05,
+      "loss": 1.6846,
+      "step": 9815
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0970027247956404e-05,
+      "loss": 1.6322,
+      "step": 9816
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0967983651226158e-05,
+      "loss": 1.7246,
+      "step": 9817
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0965940054495913e-05,
+      "loss": 1.66,
+      "step": 9818
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0963896457765669e-05,
+      "loss": 1.7576,
+      "step": 9819
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0961852861035423e-05,
+      "loss": 1.7179,
+      "step": 9820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0959809264305177e-05,
+      "loss": 1.705,
+      "step": 9821
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.095776566757493e-05,
+      "loss": 1.6754,
+      "step": 9822
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0955722070844688e-05,
+      "loss": 1.6575,
+      "step": 9823
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0953678474114442e-05,
+      "loss": 1.664,
+      "step": 9824
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0951634877384196e-05,
+      "loss": 1.8338,
+      "step": 9825
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.094959128065395e-05,
+      "loss": 1.6415,
+      "step": 9826
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0947547683923707e-05,
+      "loss": 1.5964,
+      "step": 9827
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0945504087193461e-05,
+      "loss": 1.64,
+      "step": 9828
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0943460490463215e-05,
+      "loss": 1.7228,
+      "step": 9829
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0941416893732969e-05,
+      "loss": 1.5326,
+      "step": 9830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0939373297002725e-05,
+      "loss": 1.7408,
+      "step": 9831
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.093732970027248e-05,
+      "loss": 1.6338,
+      "step": 9832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0935286103542234e-05,
+      "loss": 1.6385,
+      "step": 9833
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.093324250681199e-05,
+      "loss": 1.7445,
+      "step": 9834
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0931198910081744e-05,
+      "loss": 1.6881,
+      "step": 9835
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.09291553133515e-05,
+      "loss": 1.7279,
+      "step": 9836
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0927111716621253e-05,
+      "loss": 1.7255,
+      "step": 9837
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0925068119891009e-05,
+      "loss": 1.6426,
+      "step": 9838
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0923024523160763e-05,
+      "loss": 1.7051,
+      "step": 9839
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0920980926430517e-05,
+      "loss": 1.7155,
+      "step": 9840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0918937329700273e-05,
+      "loss": 1.6464,
+      "step": 9841
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0916893732970028e-05,
+      "loss": 1.6195,
+      "step": 9842
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0914850136239782e-05,
+      "loss": 1.697,
+      "step": 9843
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0912806539509536e-05,
+      "loss": 1.6885,
+      "step": 9844
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0910762942779293e-05,
+      "loss": 1.7948,
+      "step": 9845
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0908719346049047e-05,
+      "loss": 1.6531,
+      "step": 9846
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0906675749318801e-05,
+      "loss": 1.6617,
+      "step": 9847
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0904632152588555e-05,
+      "loss": 1.675,
+      "step": 9848
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0902588555858311e-05,
+      "loss": 1.7185,
+      "step": 9849
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0900544959128067e-05,
+      "loss": 1.6842,
+      "step": 9850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.089850136239782e-05,
+      "loss": 1.6591,
+      "step": 9851
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0896457765667574e-05,
+      "loss": 1.722,
+      "step": 9852
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.089441416893733e-05,
+      "loss": 1.7495,
+      "step": 9853
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0892370572207086e-05,
+      "loss": 1.7682,
+      "step": 9854
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.089032697547684e-05,
+      "loss": 1.7013,
+      "step": 9855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0888283378746594e-05,
+      "loss": 1.7505,
+      "step": 9856
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.088623978201635e-05,
+      "loss": 1.6833,
+      "step": 9857
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0884196185286103e-05,
+      "loss": 1.5915,
+      "step": 9858
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0882152588555859e-05,
+      "loss": 1.7032,
+      "step": 9859
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0880108991825614e-05,
+      "loss": 1.6583,
+      "step": 9860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0878065395095368e-05,
+      "loss": 1.797,
+      "step": 9861
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0876021798365122e-05,
+      "loss": 1.7648,
+      "step": 9862
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0873978201634878e-05,
+      "loss": 1.5693,
+      "step": 9863
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0871934604904634e-05,
+      "loss": 1.7019,
+      "step": 9864
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0869891008174387e-05,
+      "loss": 1.7348,
+      "step": 9865
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0867847411444141e-05,
+      "loss": 1.7094,
+      "step": 9866
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0865803814713895e-05,
+      "loss": 1.6951,
+      "step": 9867
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0863760217983653e-05,
+      "loss": 1.6445,
+      "step": 9868
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0861716621253407e-05,
+      "loss": 1.6839,
+      "step": 9869
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.085967302452316e-05,
+      "loss": 1.6563,
+      "step": 9870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0857629427792914e-05,
+      "loss": 1.6951,
+      "step": 9871
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0855585831062672e-05,
+      "loss": 1.6856,
+      "step": 9872
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0853542234332426e-05,
+      "loss": 1.7,
+      "step": 9873
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.085149863760218e-05,
+      "loss": 1.6825,
+      "step": 9874
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0849455040871935e-05,
+      "loss": 1.6508,
+      "step": 9875
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.084741144414169e-05,
+      "loss": 1.7534,
+      "step": 9876
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0845367847411445e-05,
+      "loss": 1.6637,
+      "step": 9877
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0843324250681199e-05,
+      "loss": 1.6945,
+      "step": 9878
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0841280653950954e-05,
+      "loss": 1.7241,
+      "step": 9879
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0839237057220708e-05,
+      "loss": 1.8371,
+      "step": 9880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0837193460490464e-05,
+      "loss": 1.7552,
+      "step": 9881
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0835149863760218e-05,
+      "loss": 1.7277,
+      "step": 9882
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0833106267029974e-05,
+      "loss": 1.6384,
+      "step": 9883
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0831062670299728e-05,
+      "loss": 1.8144,
+      "step": 9884
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0829019073569481e-05,
+      "loss": 1.636,
+      "step": 9885
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0826975476839239e-05,
+      "loss": 1.6265,
+      "step": 9886
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0824931880108993e-05,
+      "loss": 1.6813,
+      "step": 9887
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0822888283378747e-05,
+      "loss": 1.7281,
+      "step": 9888
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.08208446866485e-05,
+      "loss": 1.7485,
+      "step": 9889
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0818801089918258e-05,
+      "loss": 1.6185,
+      "step": 9890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0816757493188012e-05,
+      "loss": 1.7475,
+      "step": 9891
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0814713896457766e-05,
+      "loss": 1.6683,
+      "step": 9892
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.081267029972752e-05,
+      "loss": 1.5136,
+      "step": 9893
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0810626702997275e-05,
+      "loss": 1.678,
+      "step": 9894
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0808583106267031e-05,
+      "loss": 1.7516,
+      "step": 9895
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0806539509536785e-05,
+      "loss": 1.7354,
+      "step": 9896
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0804495912806539e-05,
+      "loss": 1.7332,
+      "step": 9897
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0802452316076295e-05,
+      "loss": 1.7059,
+      "step": 9898
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.080040871934605e-05,
+      "loss": 1.7705,
+      "step": 9899
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0798365122615804e-05,
+      "loss": 1.7427,
+      "step": 9900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.079632152588556e-05,
+      "loss": 1.753,
+      "step": 9901
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0794277929155314e-05,
+      "loss": 1.6523,
+      "step": 9902
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0792234332425068e-05,
+      "loss": 1.5708,
+      "step": 9903
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0790190735694823e-05,
+      "loss": 1.7847,
+      "step": 9904
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0788147138964579e-05,
+      "loss": 1.63,
+      "step": 9905
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0786103542234333e-05,
+      "loss": 1.6025,
+      "step": 9906
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0784059945504087e-05,
+      "loss": 1.5184,
+      "step": 9907
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0782016348773842e-05,
+      "loss": 1.7005,
+      "step": 9908
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0779972752043598e-05,
+      "loss": 1.5869,
+      "step": 9909
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0777929155313352e-05,
+      "loss": 1.6207,
+      "step": 9910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0775885558583106e-05,
+      "loss": 1.5732,
+      "step": 9911
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.077384196185286e-05,
+      "loss": 1.7038,
+      "step": 9912
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0771798365122617e-05,
+      "loss": 1.6347,
+      "step": 9913
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0769754768392371e-05,
+      "loss": 1.6718,
+      "step": 9914
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0767711171662125e-05,
+      "loss": 1.6788,
+      "step": 9915
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.076566757493188e-05,
+      "loss": 1.7417,
+      "step": 9916
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0763623978201636e-05,
+      "loss": 1.7678,
+      "step": 9917
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.076158038147139e-05,
+      "loss": 1.742,
+      "step": 9918
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0759536784741144e-05,
+      "loss": 1.7628,
+      "step": 9919
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.07574931880109e-05,
+      "loss": 1.6357,
+      "step": 9920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0755449591280654e-05,
+      "loss": 1.5768,
+      "step": 9921
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.075340599455041e-05,
+      "loss": 1.7143,
+      "step": 9922
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0751362397820163e-05,
+      "loss": 1.7977,
+      "step": 9923
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0749318801089919e-05,
+      "loss": 1.5259,
+      "step": 9924
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0747275204359673e-05,
+      "loss": 1.5837,
+      "step": 9925
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0745231607629429e-05,
+      "loss": 1.7865,
+      "step": 9926
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0743188010899182e-05,
+      "loss": 1.6669,
+      "step": 9927
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0741144414168938e-05,
+      "loss": 1.5966,
+      "step": 9928
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0739100817438692e-05,
+      "loss": 1.7058,
+      "step": 9929
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0737057220708446e-05,
+      "loss": 1.7026,
+      "step": 9930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0735013623978203e-05,
+      "loss": 1.6783,
+      "step": 9931
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0732970027247957e-05,
+      "loss": 1.697,
+      "step": 9932
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0730926430517711e-05,
+      "loss": 1.6544,
+      "step": 9933
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0728882833787465e-05,
+      "loss": 1.5772,
+      "step": 9934
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.072683923705722e-05,
+      "loss": 1.7326,
+      "step": 9935
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0724795640326976e-05,
+      "loss": 1.7626,
+      "step": 9936
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.072275204359673e-05,
+      "loss": 1.7494,
+      "step": 9937
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0720708446866484e-05,
+      "loss": 1.7946,
+      "step": 9938
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.071866485013624e-05,
+      "loss": 1.6848,
+      "step": 9939
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0716621253405996e-05,
+      "loss": 1.6658,
+      "step": 9940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.071457765667575e-05,
+      "loss": 1.6812,
+      "step": 9941
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0712534059945505e-05,
+      "loss": 1.5377,
+      "step": 9942
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0710490463215259e-05,
+      "loss": 1.7603,
+      "step": 9943
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0708446866485013e-05,
+      "loss": 1.6325,
+      "step": 9944
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0706403269754769e-05,
+      "loss": 1.6891,
+      "step": 9945
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0704359673024524e-05,
+      "loss": 1.6028,
+      "step": 9946
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0702316076294278e-05,
+      "loss": 1.5668,
+      "step": 9947
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0700272479564032e-05,
+      "loss": 1.7301,
+      "step": 9948
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0698228882833788e-05,
+      "loss": 1.5672,
+      "step": 9949
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0696185286103543e-05,
+      "loss": 1.705,
+      "step": 9950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0694141689373297e-05,
+      "loss": 1.7748,
+      "step": 9951
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0692098092643051e-05,
+      "loss": 1.6055,
+      "step": 9952
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0690054495912805e-05,
+      "loss": 1.7957,
+      "step": 9953
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0688010899182563e-05,
+      "loss": 1.6465,
+      "step": 9954
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0685967302452316e-05,
+      "loss": 1.6424,
+      "step": 9955
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.068392370572207e-05,
+      "loss": 1.7048,
+      "step": 9956
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0681880108991826e-05,
+      "loss": 1.7252,
+      "step": 9957
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0679836512261582e-05,
+      "loss": 1.7948,
+      "step": 9958
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0677792915531336e-05,
+      "loss": 1.8139,
+      "step": 9959
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.067574931880109e-05,
+      "loss": 1.6407,
+      "step": 9960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0673705722070845e-05,
+      "loss": 1.7745,
+      "step": 9961
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0671662125340599e-05,
+      "loss": 1.7493,
+      "step": 9962
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0669618528610355e-05,
+      "loss": 1.6364,
+      "step": 9963
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0667574931880109e-05,
+      "loss": 1.6081,
+      "step": 9964
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0665531335149864e-05,
+      "loss": 1.6714,
+      "step": 9965
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0663487738419618e-05,
+      "loss": 1.6249,
+      "step": 9966
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0661444141689374e-05,
+      "loss": 1.6567,
+      "step": 9967
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0659400544959128e-05,
+      "loss": 1.6701,
+      "step": 9968
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0657356948228883e-05,
+      "loss": 1.6702,
+      "step": 9969
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0655313351498637e-05,
+      "loss": 1.7343,
+      "step": 9970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0653269754768391e-05,
+      "loss": 1.7059,
+      "step": 9971
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0651226158038149e-05,
+      "loss": 1.7718,
+      "step": 9972
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0649182561307903e-05,
+      "loss": 1.9325,
+      "step": 9973
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0647138964577657e-05,
+      "loss": 1.5781,
+      "step": 9974
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.064509536784741e-05,
+      "loss": 1.6702,
+      "step": 9975
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0643051771117168e-05,
+      "loss": 1.689,
+      "step": 9976
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0641008174386922e-05,
+      "loss": 1.6826,
+      "step": 9977
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0638964577656676e-05,
+      "loss": 1.7731,
+      "step": 9978
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.063692098092643e-05,
+      "loss": 1.6583,
+      "step": 9979
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0634877384196185e-05,
+      "loss": 1.6849,
+      "step": 9980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0632833787465941e-05,
+      "loss": 1.6612,
+      "step": 9981
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0630790190735695e-05,
+      "loss": 1.7264,
+      "step": 9982
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.062874659400545e-05,
+      "loss": 1.6508,
+      "step": 9983
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0626702997275204e-05,
+      "loss": 1.6394,
+      "step": 9984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.062465940054496e-05,
+      "loss": 1.7272,
+      "step": 9985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0622615803814714e-05,
+      "loss": 1.7931,
+      "step": 9986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.062057220708447e-05,
+      "loss": 1.5354,
+      "step": 9987
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0618528610354224e-05,
+      "loss": 1.7164,
+      "step": 9988
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0616485013623977e-05,
+      "loss": 1.6179,
+      "step": 9989
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0614441416893733e-05,
+      "loss": 1.7696,
+      "step": 9990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0612397820163489e-05,
+      "loss": 1.7965,
+      "step": 9991
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0610354223433243e-05,
+      "loss": 1.5782,
+      "step": 9992
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0608310626702997e-05,
+      "loss": 1.7001,
+      "step": 9993
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0606267029972752e-05,
+      "loss": 1.7978,
+      "step": 9994
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0604223433242508e-05,
+      "loss": 1.7618,
+      "step": 9995
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0602179836512262e-05,
+      "loss": 1.5899,
+      "step": 9996
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0600136239782016e-05,
+      "loss": 1.6447,
+      "step": 9997
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0598092643051771e-05,
+      "loss": 1.8294,
+      "step": 9998
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0596049046321527e-05,
+      "loss": 1.7219,
+      "step": 9999
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0594005449591281e-05,
+      "loss": 1.7754,
+      "step": 10000
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 1.5213923454284668,
+      "eval_runtime": 721.2548,
+      "eval_samples_per_second": 5.754,
+      "eval_steps_per_second": 1.439,
+      "step": 10000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0591961852861035e-05,
+      "loss": 1.5853,
+      "step": 10001
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.058991825613079e-05,
+      "loss": 1.7234,
+      "step": 10002
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0587874659400546e-05,
+      "loss": 1.5199,
+      "step": 10003
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.05858310626703e-05,
+      "loss": 1.7205,
+      "step": 10004
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0583787465940054e-05,
+      "loss": 1.5621,
+      "step": 10005
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.058174386920981e-05,
+      "loss": 1.5714,
+      "step": 10006
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0579700272479564e-05,
+      "loss": 1.6481,
+      "step": 10007
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.057765667574932e-05,
+      "loss": 1.6564,
+      "step": 10008
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0575613079019073e-05,
+      "loss": 1.7139,
+      "step": 10009
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0573569482288829e-05,
+      "loss": 1.6602,
+      "step": 10010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0571525885558583e-05,
+      "loss": 1.6446,
+      "step": 10011
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0569482288828338e-05,
+      "loss": 1.7746,
+      "step": 10012
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0567438692098094e-05,
+      "loss": 1.7319,
+      "step": 10013
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0565395095367848e-05,
+      "loss": 1.7324,
+      "step": 10014
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0563351498637602e-05,
+      "loss": 1.807,
+      "step": 10015
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0561307901907356e-05,
+      "loss": 1.7048,
+      "step": 10016
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0559264305177113e-05,
+      "loss": 1.7189,
+      "step": 10017
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0557220708446867e-05,
+      "loss": 1.6644,
+      "step": 10018
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0555177111716621e-05,
+      "loss": 1.7065,
+      "step": 10019
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0553133514986375e-05,
+      "loss": 1.6923,
+      "step": 10020
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0551089918256132e-05,
+      "loss": 1.761,
+      "step": 10021
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0549046321525886e-05,
+      "loss": 1.8069,
+      "step": 10022
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.054700272479564e-05,
+      "loss": 1.6484,
+      "step": 10023
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0544959128065396e-05,
+      "loss": 1.6353,
+      "step": 10024
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.054291553133515e-05,
+      "loss": 1.5963,
+      "step": 10025
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0540871934604905e-05,
+      "loss": 1.6736,
+      "step": 10026
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.053882833787466e-05,
+      "loss": 1.6711,
+      "step": 10027
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0536784741144415e-05,
+      "loss": 1.736,
+      "step": 10028
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0534741144414169e-05,
+      "loss": 1.8316,
+      "step": 10029
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0532697547683925e-05,
+      "loss": 1.6389,
+      "step": 10030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0530653950953678e-05,
+      "loss": 1.6671,
+      "step": 10031
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0528610354223434e-05,
+      "loss": 1.7064,
+      "step": 10032
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0526566757493188e-05,
+      "loss": 1.7073,
+      "step": 10033
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0524523160762942e-05,
+      "loss": 1.6652,
+      "step": 10034
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0522479564032698e-05,
+      "loss": 1.6756,
+      "step": 10035
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0520435967302453e-05,
+      "loss": 1.7215,
+      "step": 10036
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0518392370572207e-05,
+      "loss": 1.7157,
+      "step": 10037
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0516348773841961e-05,
+      "loss": 1.6739,
+      "step": 10038
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0514305177111718e-05,
+      "loss": 1.7501,
+      "step": 10039
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0512261580381472e-05,
+      "loss": 1.6953,
+      "step": 10040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0510217983651226e-05,
+      "loss": 1.6539,
+      "step": 10041
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.050817438692098e-05,
+      "loss": 1.6909,
+      "step": 10042
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0506130790190736e-05,
+      "loss": 1.702,
+      "step": 10043
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0504087193460492e-05,
+      "loss": 1.6751,
+      "step": 10044
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0502043596730245e-05,
+      "loss": 1.5801,
+      "step": 10045
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.05e-05,
+      "loss": 1.7168,
+      "step": 10046
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0497956403269755e-05,
+      "loss": 1.5905,
+      "step": 10047
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.049591280653951e-05,
+      "loss": 1.7029,
+      "step": 10048
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0493869209809265e-05,
+      "loss": 1.6243,
+      "step": 10049
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0491825613079019e-05,
+      "loss": 1.7102,
+      "step": 10050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0489782016348774e-05,
+      "loss": 1.5855,
+      "step": 10051
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0487738419618528e-05,
+      "loss": 1.5727,
+      "step": 10052
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0485694822888284e-05,
+      "loss": 1.6763,
+      "step": 10053
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.048365122615804e-05,
+      "loss": 1.6392,
+      "step": 10054
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0481607629427793e-05,
+      "loss": 1.6779,
+      "step": 10055
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0479564032697547e-05,
+      "loss": 1.7415,
+      "step": 10056
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0477520435967303e-05,
+      "loss": 1.7114,
+      "step": 10057
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0475476839237059e-05,
+      "loss": 1.8415,
+      "step": 10058
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0473433242506812e-05,
+      "loss": 1.7041,
+      "step": 10059
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0471389645776566e-05,
+      "loss": 1.6708,
+      "step": 10060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.046934604904632e-05,
+      "loss": 1.6629,
+      "step": 10061
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0467302452316078e-05,
+      "loss": 1.6836,
+      "step": 10062
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0465258855585832e-05,
+      "loss": 1.5707,
+      "step": 10063
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0463215258855586e-05,
+      "loss": 1.7679,
+      "step": 10064
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.046117166212534e-05,
+      "loss": 1.5833,
+      "step": 10065
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0459128065395097e-05,
+      "loss": 1.688,
+      "step": 10066
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.045708446866485e-05,
+      "loss": 1.652,
+      "step": 10067
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0455040871934605e-05,
+      "loss": 1.8334,
+      "step": 10068
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.045299727520436e-05,
+      "loss": 1.745,
+      "step": 10069
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0450953678474114e-05,
+      "loss": 1.6723,
+      "step": 10070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.044891008174387e-05,
+      "loss": 1.7223,
+      "step": 10071
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0446866485013624e-05,
+      "loss": 1.5992,
+      "step": 10072
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.044482288828338e-05,
+      "loss": 1.605,
+      "step": 10073
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0442779291553133e-05,
+      "loss": 1.7141,
+      "step": 10074
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0440735694822889e-05,
+      "loss": 1.6116,
+      "step": 10075
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0438692098092643e-05,
+      "loss": 1.6718,
+      "step": 10076
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0436648501362399e-05,
+      "loss": 1.6888,
+      "step": 10077
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0434604904632153e-05,
+      "loss": 1.6354,
+      "step": 10078
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0432561307901906e-05,
+      "loss": 1.6264,
+      "step": 10079
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0430517711171664e-05,
+      "loss": 1.7722,
+      "step": 10080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0428474114441418e-05,
+      "loss": 1.5663,
+      "step": 10081
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0426430517711172e-05,
+      "loss": 1.6532,
+      "step": 10082
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0424386920980926e-05,
+      "loss": 1.6682,
+      "step": 10083
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0422343324250683e-05,
+      "loss": 1.6552,
+      "step": 10084
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0420299727520437e-05,
+      "loss": 1.7764,
+      "step": 10085
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.041825613079019e-05,
+      "loss": 1.5863,
+      "step": 10086
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0416212534059945e-05,
+      "loss": 1.686,
+      "step": 10087
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.04141689373297e-05,
+      "loss": 1.5602,
+      "step": 10088
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0412125340599456e-05,
+      "loss": 1.7004,
+      "step": 10089
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.041008174386921e-05,
+      "loss": 1.6205,
+      "step": 10090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0408038147138964e-05,
+      "loss": 1.6812,
+      "step": 10091
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.040599455040872e-05,
+      "loss": 1.7467,
+      "step": 10092
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0403950953678475e-05,
+      "loss": 1.6769,
+      "step": 10093
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0401907356948229e-05,
+      "loss": 1.5628,
+      "step": 10094
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0399863760217985e-05,
+      "loss": 1.6585,
+      "step": 10095
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0397820163487739e-05,
+      "loss": 1.6186,
+      "step": 10096
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0395776566757493e-05,
+      "loss": 1.6677,
+      "step": 10097
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0393732970027248e-05,
+      "loss": 1.6985,
+      "step": 10098
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0391689373297004e-05,
+      "loss": 1.7792,
+      "step": 10099
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0389645776566758e-05,
+      "loss": 1.7744,
+      "step": 10100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0387602179836512e-05,
+      "loss": 1.679,
+      "step": 10101
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0385558583106267e-05,
+      "loss": 1.6687,
+      "step": 10102
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0383514986376023e-05,
+      "loss": 1.6419,
+      "step": 10103
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0381471389645777e-05,
+      "loss": 1.7086,
+      "step": 10104
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0379427792915531e-05,
+      "loss": 1.5174,
+      "step": 10105
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0377384196185285e-05,
+      "loss": 1.6119,
+      "step": 10106
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0375340599455042e-05,
+      "loss": 1.7297,
+      "step": 10107
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0373297002724796e-05,
+      "loss": 1.7469,
+      "step": 10108
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.037125340599455e-05,
+      "loss": 1.6111,
+      "step": 10109
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0369209809264306e-05,
+      "loss": 1.6284,
+      "step": 10110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0367166212534061e-05,
+      "loss": 1.6736,
+      "step": 10111
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0365122615803815e-05,
+      "loss": 1.7138,
+      "step": 10112
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.036307901907357e-05,
+      "loss": 1.7175,
+      "step": 10113
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0361035422343325e-05,
+      "loss": 1.7829,
+      "step": 10114
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0358991825613079e-05,
+      "loss": 1.6339,
+      "step": 10115
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0356948228882834e-05,
+      "loss": 1.7,
+      "step": 10116
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0354904632152588e-05,
+      "loss": 1.8024,
+      "step": 10117
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0352861035422344e-05,
+      "loss": 1.8305,
+      "step": 10118
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0350817438692098e-05,
+      "loss": 1.678,
+      "step": 10119
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0348773841961854e-05,
+      "loss": 1.6037,
+      "step": 10120
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.034673024523161e-05,
+      "loss": 2.9743,
+      "step": 10121
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0344686648501363e-05,
+      "loss": 1.6778,
+      "step": 10122
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0342643051771117e-05,
+      "loss": 1.723,
+      "step": 10123
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0340599455040871e-05,
+      "loss": 1.7702,
+      "step": 10124
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0338555858310628e-05,
+      "loss": 1.6271,
+      "step": 10125
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0336512261580382e-05,
+      "loss": 1.523,
+      "step": 10126
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0334468664850136e-05,
+      "loss": 1.6066,
+      "step": 10127
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.033242506811989e-05,
+      "loss": 1.6917,
+      "step": 10128
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0330381471389646e-05,
+      "loss": 1.6524,
+      "step": 10129
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0328337874659401e-05,
+      "loss": 1.7399,
+      "step": 10130
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0326294277929155e-05,
+      "loss": 1.8229,
+      "step": 10131
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.032425068119891e-05,
+      "loss": 1.6109,
+      "step": 10132
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0322207084468665e-05,
+      "loss": 1.7033,
+      "step": 10133
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.032016348773842e-05,
+      "loss": 1.6067,
+      "step": 10134
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0318119891008174e-05,
+      "loss": 1.6644,
+      "step": 10135
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.031607629427793e-05,
+      "loss": 1.7437,
+      "step": 10136
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0314032697547684e-05,
+      "loss": 1.8786,
+      "step": 10137
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0311989100817438e-05,
+      "loss": 1.6922,
+      "step": 10138
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0309945504087194e-05,
+      "loss": 1.7574,
+      "step": 10139
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.030790190735695e-05,
+      "loss": 1.6485,
+      "step": 10140
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0305858310626703e-05,
+      "loss": 1.7024,
+      "step": 10141
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0303814713896457e-05,
+      "loss": 1.6607,
+      "step": 10142
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0301771117166213e-05,
+      "loss": 1.6205,
+      "step": 10143
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0299727520435968e-05,
+      "loss": 1.6515,
+      "step": 10144
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0297683923705722e-05,
+      "loss": 1.7006,
+      "step": 10145
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0295640326975476e-05,
+      "loss": 1.6661,
+      "step": 10146
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.029359673024523e-05,
+      "loss": 1.6239,
+      "step": 10147
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0291553133514988e-05,
+      "loss": 1.6142,
+      "step": 10148
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0289509536784741e-05,
+      "loss": 1.5958,
+      "step": 10149
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0287465940054495e-05,
+      "loss": 1.9298,
+      "step": 10150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0285422343324251e-05,
+      "loss": 1.6203,
+      "step": 10151
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0283378746594007e-05,
+      "loss": 1.7278,
+      "step": 10152
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.028133514986376e-05,
+      "loss": 1.5688,
+      "step": 10153
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0279291553133515e-05,
+      "loss": 1.6203,
+      "step": 10154
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.027724795640327e-05,
+      "loss": 1.69,
+      "step": 10155
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0275204359673024e-05,
+      "loss": 1.6939,
+      "step": 10156
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.027316076294278e-05,
+      "loss": 1.7786,
+      "step": 10157
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0271117166212534e-05,
+      "loss": 1.5509,
+      "step": 10158
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.026907356948229e-05,
+      "loss": 1.6391,
+      "step": 10159
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0267029972752043e-05,
+      "loss": 1.7173,
+      "step": 10160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0264986376021799e-05,
+      "loss": 1.6407,
+      "step": 10161
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0262942779291555e-05,
+      "loss": 1.6566,
+      "step": 10162
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0260899182561308e-05,
+      "loss": 1.5953,
+      "step": 10163
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0258855585831062e-05,
+      "loss": 1.86,
+      "step": 10164
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0256811989100816e-05,
+      "loss": 1.7927,
+      "step": 10165
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0254768392370574e-05,
+      "loss": 1.6761,
+      "step": 10166
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0252724795640328e-05,
+      "loss": 1.7237,
+      "step": 10167
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0250681198910082e-05,
+      "loss": 1.7535,
+      "step": 10168
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0248637602179836e-05,
+      "loss": 1.7197,
+      "step": 10169
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0246594005449593e-05,
+      "loss": 1.7091,
+      "step": 10170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0244550408719347e-05,
+      "loss": 1.6609,
+      "step": 10171
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.02425068119891e-05,
+      "loss": 1.5889,
+      "step": 10172
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0240463215258855e-05,
+      "loss": 1.6404,
+      "step": 10173
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.023841961852861e-05,
+      "loss": 1.7503,
+      "step": 10174
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0236376021798366e-05,
+      "loss": 1.7339,
+      "step": 10175
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.023433242506812e-05,
+      "loss": 1.8189,
+      "step": 10176
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0232288828337875e-05,
+      "loss": 1.8087,
+      "step": 10177
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.023024523160763e-05,
+      "loss": 1.7049,
+      "step": 10178
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0228201634877385e-05,
+      "loss": 1.7353,
+      "step": 10179
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0226158038147139e-05,
+      "loss": 1.5843,
+      "step": 10180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0224114441416895e-05,
+      "loss": 1.6823,
+      "step": 10181
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0222070844686649e-05,
+      "loss": 1.6995,
+      "step": 10182
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0220027247956403e-05,
+      "loss": 1.7195,
+      "step": 10183
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0217983651226158e-05,
+      "loss": 1.7526,
+      "step": 10184
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0215940054495914e-05,
+      "loss": 1.5972,
+      "step": 10185
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0213896457765668e-05,
+      "loss": 1.7301,
+      "step": 10186
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0211852861035422e-05,
+      "loss": 1.6703,
+      "step": 10187
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0209809264305177e-05,
+      "loss": 1.5755,
+      "step": 10188
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0207765667574933e-05,
+      "loss": 1.758,
+      "step": 10189
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0205722070844687e-05,
+      "loss": 1.6939,
+      "step": 10190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.020367847411444e-05,
+      "loss": 1.7235,
+      "step": 10191
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0201634877384196e-05,
+      "loss": 1.6356,
+      "step": 10192
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0199591280653952e-05,
+      "loss": 1.7226,
+      "step": 10193
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0197547683923706e-05,
+      "loss": 1.6539,
+      "step": 10194
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.019550408719346e-05,
+      "loss": 1.7246,
+      "step": 10195
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0193460490463216e-05,
+      "loss": 1.6131,
+      "step": 10196
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0191416893732971e-05,
+      "loss": 1.6952,
+      "step": 10197
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0189373297002725e-05,
+      "loss": 1.6753,
+      "step": 10198
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0187329700272479e-05,
+      "loss": 1.8015,
+      "step": 10199
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0185286103542235e-05,
+      "loss": 1.7313,
+      "step": 10200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0183242506811989e-05,
+      "loss": 1.7697,
+      "step": 10201
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0181198910081744e-05,
+      "loss": 1.6669,
+      "step": 10202
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0179155313351498e-05,
+      "loss": 1.7619,
+      "step": 10203
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0177111716621254e-05,
+      "loss": 1.5719,
+      "step": 10204
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0175068119891008e-05,
+      "loss": 1.7436,
+      "step": 10205
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0173024523160763e-05,
+      "loss": 1.5744,
+      "step": 10206
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0170980926430519e-05,
+      "loss": 1.8061,
+      "step": 10207
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0168937329700273e-05,
+      "loss": 1.666,
+      "step": 10208
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0166893732970027e-05,
+      "loss": 1.5827,
+      "step": 10209
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0164850136239781e-05,
+      "loss": 1.7106,
+      "step": 10210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0162806539509538e-05,
+      "loss": 1.6521,
+      "step": 10211
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0160762942779292e-05,
+      "loss": 1.7132,
+      "step": 10212
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0158719346049046e-05,
+      "loss": 1.8039,
+      "step": 10213
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.01566757493188e-05,
+      "loss": 1.5907,
+      "step": 10214
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0154632152588557e-05,
+      "loss": 1.6748,
+      "step": 10215
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0152588555858311e-05,
+      "loss": 1.5451,
+      "step": 10216
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0150544959128065e-05,
+      "loss": 1.7232,
+      "step": 10217
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0148501362397821e-05,
+      "loss": 1.7519,
+      "step": 10218
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0146457765667575e-05,
+      "loss": 1.8036,
+      "step": 10219
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.014441416893733e-05,
+      "loss": 1.7372,
+      "step": 10220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0142370572207084e-05,
+      "loss": 1.8017,
+      "step": 10221
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.014032697547684e-05,
+      "loss": 1.7256,
+      "step": 10222
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0138283378746594e-05,
+      "loss": 1.7801,
+      "step": 10223
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.013623978201635e-05,
+      "loss": 1.7126,
+      "step": 10224
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0134196185286104e-05,
+      "loss": 1.6127,
+      "step": 10225
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0132152588555859e-05,
+      "loss": 1.7334,
+      "step": 10226
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0130108991825613e-05,
+      "loss": 1.7484,
+      "step": 10227
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0128065395095367e-05,
+      "loss": 1.5297,
+      "step": 10228
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0126021798365123e-05,
+      "loss": 1.6911,
+      "step": 10229
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0123978201634878e-05,
+      "loss": 1.6944,
+      "step": 10230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0121934604904632e-05,
+      "loss": 1.5811,
+      "step": 10231
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0119891008174386e-05,
+      "loss": 1.6637,
+      "step": 10232
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0117847411444143e-05,
+      "loss": 1.6969,
+      "step": 10233
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0115803814713897e-05,
+      "loss": 1.7021,
+      "step": 10234
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0113760217983651e-05,
+      "loss": 1.6629,
+      "step": 10235
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0111716621253405e-05,
+      "loss": 1.6821,
+      "step": 10236
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0109673024523161e-05,
+      "loss": 1.6666,
+      "step": 10237
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0107629427792917e-05,
+      "loss": 1.5858,
+      "step": 10238
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.010558583106267e-05,
+      "loss": 1.689,
+      "step": 10239
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0103542234332424e-05,
+      "loss": 1.6156,
+      "step": 10240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.010149863760218e-05,
+      "loss": 1.7181,
+      "step": 10241
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0099455040871936e-05,
+      "loss": 1.7509,
+      "step": 10242
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.009741144414169e-05,
+      "loss": 1.7134,
+      "step": 10243
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0095367847411444e-05,
+      "loss": 1.7859,
+      "step": 10244
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.00933242506812e-05,
+      "loss": 1.7329,
+      "step": 10245
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0091280653950953e-05,
+      "loss": 1.7228,
+      "step": 10246
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0089237057220709e-05,
+      "loss": 1.7558,
+      "step": 10247
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0087193460490464e-05,
+      "loss": 1.6817,
+      "step": 10248
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0085149863760218e-05,
+      "loss": 1.6269,
+      "step": 10249
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0083106267029972e-05,
+      "loss": 1.7713,
+      "step": 10250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0081062670299728e-05,
+      "loss": 1.7507,
+      "step": 10251
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0079019073569484e-05,
+      "loss": 1.6362,
+      "step": 10252
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0076975476839238e-05,
+      "loss": 1.6673,
+      "step": 10253
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0074931880108991e-05,
+      "loss": 1.8397,
+      "step": 10254
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0072888283378745e-05,
+      "loss": 1.6482,
+      "step": 10255
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0070844686648503e-05,
+      "loss": 1.7421,
+      "step": 10256
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0068801089918257e-05,
+      "loss": 1.7071,
+      "step": 10257
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.006675749318801e-05,
+      "loss": 1.753,
+      "step": 10258
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0064713896457766e-05,
+      "loss": 1.6374,
+      "step": 10259
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0062670299727522e-05,
+      "loss": 1.5519,
+      "step": 10260
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0060626702997276e-05,
+      "loss": 1.7176,
+      "step": 10261
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.005858310626703e-05,
+      "loss": 1.6729,
+      "step": 10262
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0056539509536785e-05,
+      "loss": 1.6199,
+      "step": 10263
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.005449591280654e-05,
+      "loss": 1.7135,
+      "step": 10264
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0052452316076295e-05,
+      "loss": 1.6348,
+      "step": 10265
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0050408719346049e-05,
+      "loss": 1.7212,
+      "step": 10266
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0048365122615805e-05,
+      "loss": 1.8007,
+      "step": 10267
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0046321525885558e-05,
+      "loss": 1.6626,
+      "step": 10268
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0044277929155314e-05,
+      "loss": 1.6847,
+      "step": 10269
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0042234332425068e-05,
+      "loss": 1.7091,
+      "step": 10270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0040190735694824e-05,
+      "loss": 1.7145,
+      "step": 10271
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0038147138964578e-05,
+      "loss": 1.8182,
+      "step": 10272
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0036103542234332e-05,
+      "loss": 1.5931,
+      "step": 10273
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0034059945504089e-05,
+      "loss": 1.6058,
+      "step": 10274
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0032016348773843e-05,
+      "loss": 1.6845,
+      "step": 10275
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0029972752043597e-05,
+      "loss": 1.8422,
+      "step": 10276
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.002792915531335e-05,
+      "loss": 1.6483,
+      "step": 10277
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0025885558583108e-05,
+      "loss": 1.5805,
+      "step": 10278
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0023841961852862e-05,
+      "loss": 1.6601,
+      "step": 10279
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0021798365122616e-05,
+      "loss": 1.6366,
+      "step": 10280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.001975476839237e-05,
+      "loss": 1.5898,
+      "step": 10281
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0017711171662125e-05,
+      "loss": 1.7374,
+      "step": 10282
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0015667574931881e-05,
+      "loss": 1.6767,
+      "step": 10283
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0013623978201635e-05,
+      "loss": 1.785,
+      "step": 10284
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0011580381471389e-05,
+      "loss": 1.5514,
+      "step": 10285
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0009536784741145e-05,
+      "loss": 1.6973,
+      "step": 10286
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.00074931880109e-05,
+      "loss": 1.8316,
+      "step": 10287
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0005449591280654e-05,
+      "loss": 1.657,
+      "step": 10288
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.000340599455041e-05,
+      "loss": 1.7004,
+      "step": 10289
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0001362397820164e-05,
+      "loss": 1.5991,
+      "step": 10290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.999318801089918e-06,
+      "loss": 1.7011,
+      "step": 10291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.997275204359673e-06,
+      "loss": 1.6386,
+      "step": 10292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.995231607629429e-06,
+      "loss": 1.7393,
+      "step": 10293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.993188010899183e-06,
+      "loss": 1.7349,
+      "step": 10294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.991144414168937e-06,
+      "loss": 1.6023,
+      "step": 10295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.989100817438692e-06,
+      "loss": 1.7206,
+      "step": 10296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.987057220708448e-06,
+      "loss": 1.7485,
+      "step": 10297
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.985013623978202e-06,
+      "loss": 1.6268,
+      "step": 10298
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.982970027247956e-06,
+      "loss": 1.6958,
+      "step": 10299
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.980926430517712e-06,
+      "loss": 1.6948,
+      "step": 10300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.978882833787467e-06,
+      "loss": 1.771,
+      "step": 10301
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.976839237057221e-06,
+      "loss": 1.5707,
+      "step": 10302
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.974795640326975e-06,
+      "loss": 1.5948,
+      "step": 10303
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.97275204359673e-06,
+      "loss": 1.6586,
+      "step": 10304
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.970708446866486e-06,
+      "loss": 1.6258,
+      "step": 10305
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.96866485013624e-06,
+      "loss": 1.6785,
+      "step": 10306
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.966621253405994e-06,
+      "loss": 1.7038,
+      "step": 10307
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.96457765667575e-06,
+      "loss": 1.578,
+      "step": 10308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.962534059945504e-06,
+      "loss": 1.6497,
+      "step": 10309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.96049046321526e-06,
+      "loss": 1.7241,
+      "step": 10310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.958446866485013e-06,
+      "loss": 1.5854,
+      "step": 10311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.956403269754769e-06,
+      "loss": 1.5979,
+      "step": 10312
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.954359673024523e-06,
+      "loss": 1.6748,
+      "step": 10313
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.952316076294279e-06,
+      "loss": 1.5889,
+      "step": 10314
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.950272479564034e-06,
+      "loss": 1.698,
+      "step": 10315
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.948228882833788e-06,
+      "loss": 1.6525,
+      "step": 10316
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.946185286103542e-06,
+      "loss": 1.5856,
+      "step": 10317
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.944141689373296e-06,
+      "loss": 1.7381,
+      "step": 10318
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.942098092643053e-06,
+      "loss": 1.698,
+      "step": 10319
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.940054495912807e-06,
+      "loss": 1.5343,
+      "step": 10320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.938010899182561e-06,
+      "loss": 1.6081,
+      "step": 10321
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.935967302452315e-06,
+      "loss": 1.6685,
+      "step": 10322
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.93392370572207e-06,
+      "loss": 1.7023,
+      "step": 10323
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.931880108991826e-06,
+      "loss": 1.7092,
+      "step": 10324
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.92983651226158e-06,
+      "loss": 1.6607,
+      "step": 10325
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.927792915531334e-06,
+      "loss": 1.5729,
+      "step": 10326
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.92574931880109e-06,
+      "loss": 1.5996,
+      "step": 10327
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.923705722070846e-06,
+      "loss": 1.6705,
+      "step": 10328
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.9216621253406e-06,
+      "loss": 1.5891,
+      "step": 10329
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.919618528610355e-06,
+      "loss": 1.6377,
+      "step": 10330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.917574931880109e-06,
+      "loss": 1.7346,
+      "step": 10331
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.915531335149863e-06,
+      "loss": 1.7063,
+      "step": 10332
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.913487738419619e-06,
+      "loss": 1.59,
+      "step": 10333
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.911444141689374e-06,
+      "loss": 1.6045,
+      "step": 10334
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.909400544959128e-06,
+      "loss": 1.7018,
+      "step": 10335
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.907356948228882e-06,
+      "loss": 1.6403,
+      "step": 10336
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.905313351498638e-06,
+      "loss": 1.8106,
+      "step": 10337
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.903269754768393e-06,
+      "loss": 1.7389,
+      "step": 10338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.901226158038147e-06,
+      "loss": 1.6667,
+      "step": 10339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.899182561307901e-06,
+      "loss": 1.6792,
+      "step": 10340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.897138964577655e-06,
+      "loss": 1.6564,
+      "step": 10341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.895095367847413e-06,
+      "loss": 1.589,
+      "step": 10342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.893051771117167e-06,
+      "loss": 1.6689,
+      "step": 10343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.89100817438692e-06,
+      "loss": 1.719,
+      "step": 10344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.888964577656676e-06,
+      "loss": 1.6957,
+      "step": 10345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.886920980926432e-06,
+      "loss": 1.6163,
+      "step": 10346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.884877384196186e-06,
+      "loss": 1.8076,
+      "step": 10347
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.88283378746594e-06,
+      "loss": 1.6782,
+      "step": 10348
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.880790190735695e-06,
+      "loss": 1.7599,
+      "step": 10349
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.87874659400545e-06,
+      "loss": 1.5591,
+      "step": 10350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.876702997275205e-06,
+      "loss": 1.7402,
+      "step": 10351
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.874659400544959e-06,
+      "loss": 1.7179,
+      "step": 10352
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.872615803814714e-06,
+      "loss": 1.6767,
+      "step": 10353
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.870572207084468e-06,
+      "loss": 1.7073,
+      "step": 10354
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.868528610354224e-06,
+      "loss": 1.6711,
+      "step": 10355
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.86648501362398e-06,
+      "loss": 1.6203,
+      "step": 10356
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.864441416893734e-06,
+      "loss": 1.6423,
+      "step": 10357
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.862397820163487e-06,
+      "loss": 1.7281,
+      "step": 10358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.860354223433241e-06,
+      "loss": 1.7561,
+      "step": 10359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.858310626702999e-06,
+      "loss": 1.6063,
+      "step": 10360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.856267029972753e-06,
+      "loss": 1.7296,
+      "step": 10361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.854223433242507e-06,
+      "loss": 1.6067,
+      "step": 10362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.85217983651226e-06,
+      "loss": 1.7574,
+      "step": 10363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.850136239782018e-06,
+      "loss": 1.6598,
+      "step": 10364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.848092643051772e-06,
+      "loss": 1.7219,
+      "step": 10365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.846049046321526e-06,
+      "loss": 1.6997,
+      "step": 10366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.84400544959128e-06,
+      "loss": 1.6513,
+      "step": 10367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.841961852861035e-06,
+      "loss": 1.6401,
+      "step": 10368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.839918256130791e-06,
+      "loss": 1.6772,
+      "step": 10369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.837874659400545e-06,
+      "loss": 1.7002,
+      "step": 10370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.8358310626703e-06,
+      "loss": 1.7164,
+      "step": 10371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.833787465940054e-06,
+      "loss": 1.6439,
+      "step": 10372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.83174386920981e-06,
+      "loss": 1.6679,
+      "step": 10373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.829700272479564e-06,
+      "loss": 1.7029,
+      "step": 10374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.82765667574932e-06,
+      "loss": 1.5437,
+      "step": 10375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.825613079019074e-06,
+      "loss": 1.7362,
+      "step": 10376
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.823569482288828e-06,
+      "loss": 1.6944,
+      "step": 10377
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.821525885558583e-06,
+      "loss": 1.6889,
+      "step": 10378
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.819482288828339e-06,
+      "loss": 1.6763,
+      "step": 10379
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.817438692098093e-06,
+      "loss": 1.8065,
+      "step": 10380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.815395095367847e-06,
+      "loss": 1.6211,
+      "step": 10381
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.813351498637602e-06,
+      "loss": 1.6934,
+      "step": 10382
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.811307901907358e-06,
+      "loss": 1.6269,
+      "step": 10383
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.809264305177112e-06,
+      "loss": 1.7387,
+      "step": 10384
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.807220708446866e-06,
+      "loss": 1.6069,
+      "step": 10385
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.805177111716621e-06,
+      "loss": 1.7128,
+      "step": 10386
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.803133514986377e-06,
+      "loss": 1.6119,
+      "step": 10387
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.801089918256131e-06,
+      "loss": 1.7799,
+      "step": 10388
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.799046321525885e-06,
+      "loss": 1.8482,
+      "step": 10389
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.79700272479564e-06,
+      "loss": 1.704,
+      "step": 10390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.794959128065396e-06,
+      "loss": 1.6685,
+      "step": 10391
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.79291553133515e-06,
+      "loss": 1.617,
+      "step": 10392
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.790871934604904e-06,
+      "loss": 1.6773,
+      "step": 10393
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.78882833787466e-06,
+      "loss": 1.6676,
+      "step": 10394
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.786784741144414e-06,
+      "loss": 1.7161,
+      "step": 10395
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.78474114441417e-06,
+      "loss": 1.6012,
+      "step": 10396
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.782697547683925e-06,
+      "loss": 1.7796,
+      "step": 10397
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.780653950953679e-06,
+      "loss": 1.796,
+      "step": 10398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.778610354223433e-06,
+      "loss": 1.7879,
+      "step": 10399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.776566757493188e-06,
+      "loss": 1.7128,
+      "step": 10400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.774523160762944e-06,
+      "loss": 1.843,
+      "step": 10401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.772479564032698e-06,
+      "loss": 1.7211,
+      "step": 10402
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.770435967302452e-06,
+      "loss": 1.5381,
+      "step": 10403
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.768392370572206e-06,
+      "loss": 1.6545,
+      "step": 10404
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.766348773841963e-06,
+      "loss": 1.7185,
+      "step": 10405
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.764305177111717e-06,
+      "loss": 1.6877,
+      "step": 10406
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.762261580381471e-06,
+      "loss": 1.5449,
+      "step": 10407
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.760217983651225e-06,
+      "loss": 1.668,
+      "step": 10408
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.758174386920982e-06,
+      "loss": 1.7771,
+      "step": 10409
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.756130790190736e-06,
+      "loss": 1.7938,
+      "step": 10410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.75408719346049e-06,
+      "loss": 1.6842,
+      "step": 10411
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.752043596730246e-06,
+      "loss": 1.7193,
+      "step": 10412
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.75e-06,
+      "loss": 1.5851,
+      "step": 10413
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.747956403269755e-06,
+      "loss": 1.7874,
+      "step": 10414
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.74591280653951e-06,
+      "loss": 1.6745,
+      "step": 10415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.743869209809265e-06,
+      "loss": 1.6954,
+      "step": 10416
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.741825613079019e-06,
+      "loss": 1.6672,
+      "step": 10417
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.739782016348775e-06,
+      "loss": 1.6098,
+      "step": 10418
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.737738419618529e-06,
+      "loss": 1.6333,
+      "step": 10419
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.735694822888284e-06,
+      "loss": 1.6927,
+      "step": 10420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.733651226158038e-06,
+      "loss": 1.6321,
+      "step": 10421
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.731607629427792e-06,
+      "loss": 1.7129,
+      "step": 10422
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.729564032697548e-06,
+      "loss": 1.5479,
+      "step": 10423
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.727520435967303e-06,
+      "loss": 1.6801,
+      "step": 10424
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.725476839237057e-06,
+      "loss": 1.5587,
+      "step": 10425
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.723433242506811e-06,
+      "loss": 1.7326,
+      "step": 10426
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.721389645776569e-06,
+      "loss": 1.6528,
+      "step": 10427
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.719346049046322e-06,
+      "loss": 1.6633,
+      "step": 10428
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.717302452316076e-06,
+      "loss": 1.5184,
+      "step": 10429
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.71525885558583e-06,
+      "loss": 1.6229,
+      "step": 10430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.713215258855586e-06,
+      "loss": 1.5825,
+      "step": 10431
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.711171662125342e-06,
+      "loss": 1.8045,
+      "step": 10432
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.709128065395096e-06,
+      "loss": 1.6347,
+      "step": 10433
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.70708446866485e-06,
+      "loss": 1.5859,
+      "step": 10434
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.705040871934605e-06,
+      "loss": 1.7219,
+      "step": 10435
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.70299727520436e-06,
+      "loss": 1.6662,
+      "step": 10436
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.700953678474115e-06,
+      "loss": 1.5339,
+      "step": 10437
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.69891008174387e-06,
+      "loss": 1.7875,
+      "step": 10438
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.696866485013624e-06,
+      "loss": 1.6763,
+      "step": 10439
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.694822888283378e-06,
+      "loss": 1.674,
+      "step": 10440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.692779291553134e-06,
+      "loss": 1.774,
+      "step": 10441
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.69073569482289e-06,
+      "loss": 1.8186,
+      "step": 10442
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.688692098092643e-06,
+      "loss": 1.7073,
+      "step": 10443
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.686648501362397e-06,
+      "loss": 1.8332,
+      "step": 10444
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.684604904632153e-06,
+      "loss": 1.6351,
+      "step": 10445
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.682561307901909e-06,
+      "loss": 1.5664,
+      "step": 10446
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.680517711171663e-06,
+      "loss": 1.6074,
+      "step": 10447
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.678474114441416e-06,
+      "loss": 1.645,
+      "step": 10448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.67643051771117e-06,
+      "loss": 1.7348,
+      "step": 10449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.674386920980928e-06,
+      "loss": 1.7721,
+      "step": 10450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.672343324250682e-06,
+      "loss": 1.6238,
+      "step": 10451
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.670299727520436e-06,
+      "loss": 1.597,
+      "step": 10452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.668256130790191e-06,
+      "loss": 1.6889,
+      "step": 10453
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.666212534059947e-06,
+      "loss": 1.5936,
+      "step": 10454
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.6641689373297e-06,
+      "loss": 1.7089,
+      "step": 10455
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.662125340599455e-06,
+      "loss": 1.743,
+      "step": 10456
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.66008174386921e-06,
+      "loss": 1.7737,
+      "step": 10457
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.658038147138964e-06,
+      "loss": 1.6761,
+      "step": 10458
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.65599455040872e-06,
+      "loss": 1.5259,
+      "step": 10459
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.653950953678474e-06,
+      "loss": 1.6889,
+      "step": 10460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.65190735694823e-06,
+      "loss": 1.8198,
+      "step": 10461
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.649863760217983e-06,
+      "loss": 1.6129,
+      "step": 10462
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.647820163487739e-06,
+      "loss": 1.6106,
+      "step": 10463
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.645776566757493e-06,
+      "loss": 1.6242,
+      "step": 10464
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.643732970027249e-06,
+      "loss": 1.7514,
+      "step": 10465
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.641689373297003e-06,
+      "loss": 1.6899,
+      "step": 10466
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.639645776566757e-06,
+      "loss": 1.5529,
+      "step": 10467
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.637602179836514e-06,
+      "loss": 1.7304,
+      "step": 10468
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.635558583106268e-06,
+      "loss": 1.738,
+      "step": 10469
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.633514986376022e-06,
+      "loss": 1.5912,
+      "step": 10470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.631471389645776e-06,
+      "loss": 1.5608,
+      "step": 10471
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.629427792915533e-06,
+      "loss": 1.6825,
+      "step": 10472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.627384196185287e-06,
+      "loss": 1.7136,
+      "step": 10473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.625340599455041e-06,
+      "loss": 1.653,
+      "step": 10474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.623297002724795e-06,
+      "loss": 1.6368,
+      "step": 10475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.62125340599455e-06,
+      "loss": 1.7409,
+      "step": 10476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.619209809264306e-06,
+      "loss": 1.6584,
+      "step": 10477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.61716621253406e-06,
+      "loss": 1.7804,
+      "step": 10478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.615122615803814e-06,
+      "loss": 1.6709,
+      "step": 10479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.61307901907357e-06,
+      "loss": 1.756,
+      "step": 10480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.611035422343325e-06,
+      "loss": 1.5808,
+      "step": 10481
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.60899182561308e-06,
+      "loss": 1.7793,
+      "step": 10482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.606948228882835e-06,
+      "loss": 1.7278,
+      "step": 10483
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.604904632152589e-06,
+      "loss": 1.6537,
+      "step": 10484
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.602861035422343e-06,
+      "loss": 1.6922,
+      "step": 10485
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.600817438692098e-06,
+      "loss": 1.5756,
+      "step": 10486
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.598773841961854e-06,
+      "loss": 1.6785,
+      "step": 10487
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.596730245231608e-06,
+      "loss": 1.782,
+      "step": 10488
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.594686648501362e-06,
+      "loss": 1.8299,
+      "step": 10489
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.592643051771117e-06,
+      "loss": 1.6871,
+      "step": 10490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.590599455040873e-06,
+      "loss": 1.6712,
+      "step": 10491
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.588555858310627e-06,
+      "loss": 1.7562,
+      "step": 10492
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586512261580381e-06,
+      "loss": 1.6126,
+      "step": 10493
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.584468664850137e-06,
+      "loss": 1.6295,
+      "step": 10494
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.582425068119892e-06,
+      "loss": 1.6372,
+      "step": 10495
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.580381471389646e-06,
+      "loss": 1.5748,
+      "step": 10496
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.5783378746594e-06,
+      "loss": 1.7398,
+      "step": 10497
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.576294277929156e-06,
+      "loss": 1.6826,
+      "step": 10498
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.574250681198911e-06,
+      "loss": 1.7121,
+      "step": 10499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.572207084468665e-06,
+      "loss": 1.7695,
+      "step": 10500
+    },
+    {
+      "epoch": 2.08,
+      "eval_loss": 1.5164508819580078,
+      "eval_runtime": 721.1727,
+      "eval_samples_per_second": 5.755,
+      "eval_steps_per_second": 1.439,
+      "step": 10500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.57016348773842e-06,
+      "loss": 1.7071,
+      "step": 10501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.568119891008175e-06,
+      "loss": 1.6336,
+      "step": 10502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.566076294277929e-06,
+      "loss": 1.6169,
+      "step": 10503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.564032697547684e-06,
+      "loss": 1.6864,
+      "step": 10504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.561989100817438e-06,
+      "loss": 1.6481,
+      "step": 10505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.559945504087194e-06,
+      "loss": 1.6856,
+      "step": 10506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.557901907356948e-06,
+      "loss": 1.7043,
+      "step": 10507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.555858310626704e-06,
+      "loss": 1.7133,
+      "step": 10508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.55381471389646e-06,
+      "loss": 1.7098,
+      "step": 10509
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.551771117166213e-06,
+      "loss": 1.727,
+      "step": 10510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.549727520435967e-06,
+      "loss": 1.6093,
+      "step": 10511
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.547683923705721e-06,
+      "loss": 1.6909,
+      "step": 10512
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.545640326975478e-06,
+      "loss": 1.5134,
+      "step": 10513
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.543596730245232e-06,
+      "loss": 1.6941,
+      "step": 10514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.541553133514986e-06,
+      "loss": 1.6505,
+      "step": 10515
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.53950953678474e-06,
+      "loss": 1.6512,
+      "step": 10516
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.537465940054496e-06,
+      "loss": 1.6721,
+      "step": 10517
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.535422343324251e-06,
+      "loss": 1.6233,
+      "step": 10518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.533378746594005e-06,
+      "loss": 1.7277,
+      "step": 10519
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.53133514986376e-06,
+      "loss": 1.5539,
+      "step": 10520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.529291553133515e-06,
+      "loss": 1.6784,
+      "step": 10521
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.52724795640327e-06,
+      "loss": 1.7035,
+      "step": 10522
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.525204359673025e-06,
+      "loss": 1.7024,
+      "step": 10523
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.52316076294278e-06,
+      "loss": 1.7204,
+      "step": 10524
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.521117166212534e-06,
+      "loss": 1.7283,
+      "step": 10525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.519073569482288e-06,
+      "loss": 1.7798,
+      "step": 10526
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.517029972752044e-06,
+      "loss": 1.5978,
+      "step": 10527
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.5149863760218e-06,
+      "loss": 1.7492,
+      "step": 10528
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.512942779291553e-06,
+      "loss": 1.6089,
+      "step": 10529
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.510899182561307e-06,
+      "loss": 1.7238,
+      "step": 10530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.508855585831063e-06,
+      "loss": 1.7169,
+      "step": 10531
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.506811989100818e-06,
+      "loss": 1.7219,
+      "step": 10532
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.504768392370572e-06,
+      "loss": 1.6833,
+      "step": 10533
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.502724795640326e-06,
+      "loss": 1.6509,
+      "step": 10534
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.500681198910082e-06,
+      "loss": 1.7133,
+      "step": 10535
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.498637602179838e-06,
+      "loss": 1.5718,
+      "step": 10536
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.496594005449592e-06,
+      "loss": 1.4954,
+      "step": 10537
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.494550408719345e-06,
+      "loss": 1.6848,
+      "step": 10538
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.492506811989101e-06,
+      "loss": 1.5297,
+      "step": 10539
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.490463215258857e-06,
+      "loss": 1.6823,
+      "step": 10540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.48841961852861e-06,
+      "loss": 1.7608,
+      "step": 10541
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.486376021798365e-06,
+      "loss": 1.7752,
+      "step": 10542
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.48433242506812e-06,
+      "loss": 1.6179,
+      "step": 10543
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.482288828337874e-06,
+      "loss": 1.8591,
+      "step": 10544
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.48024523160763e-06,
+      "loss": 1.5887,
+      "step": 10545
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.478201634877384e-06,
+      "loss": 1.8061,
+      "step": 10546
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.47615803814714e-06,
+      "loss": 1.6621,
+      "step": 10547
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.474114441416893e-06,
+      "loss": 1.7101,
+      "step": 10548
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.472070844686649e-06,
+      "loss": 1.7103,
+      "step": 10549
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.470027247956405e-06,
+      "loss": 1.8355,
+      "step": 10550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.467983651226159e-06,
+      "loss": 1.6437,
+      "step": 10551
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.465940054495912e-06,
+      "loss": 1.7139,
+      "step": 10552
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.463896457765666e-06,
+      "loss": 1.7051,
+      "step": 10553
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.461852861035424e-06,
+      "loss": 1.7256,
+      "step": 10554
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.459809264305178e-06,
+      "loss": 1.6234,
+      "step": 10555
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.457765667574932e-06,
+      "loss": 1.7082,
+      "step": 10556
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.455722070844686e-06,
+      "loss": 1.641,
+      "step": 10557
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.453678474114443e-06,
+      "loss": 1.6813,
+      "step": 10558
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.451634877384197e-06,
+      "loss": 1.7041,
+      "step": 10559
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.44959128065395e-06,
+      "loss": 1.823,
+      "step": 10560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.447547683923705e-06,
+      "loss": 1.7954,
+      "step": 10561
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.44550408719346e-06,
+      "loss": 1.8296,
+      "step": 10562
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.443460490463216e-06,
+      "loss": 1.6998,
+      "step": 10563
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.44141689373297e-06,
+      "loss": 1.5731,
+      "step": 10564
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.439373297002726e-06,
+      "loss": 1.6284,
+      "step": 10565
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.43732970027248e-06,
+      "loss": 1.6445,
+      "step": 10566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.435286103542235e-06,
+      "loss": 1.8591,
+      "step": 10567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.433242506811989e-06,
+      "loss": 1.6411,
+      "step": 10568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.431198910081745e-06,
+      "loss": 1.5765,
+      "step": 10569
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.429155313351499e-06,
+      "loss": 1.6691,
+      "step": 10570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.427111716621253e-06,
+      "loss": 1.6043,
+      "step": 10571
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.425068119891008e-06,
+      "loss": 1.5753,
+      "step": 10572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.423024523160764e-06,
+      "loss": 1.6796,
+      "step": 10573
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.420980926430518e-06,
+      "loss": 1.6393,
+      "step": 10574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.418937329700272e-06,
+      "loss": 1.6724,
+      "step": 10575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.416893732970029e-06,
+      "loss": 1.6607,
+      "step": 10576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.414850136239783e-06,
+      "loss": 1.6182,
+      "step": 10577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.412806539509537e-06,
+      "loss": 1.6326,
+      "step": 10578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.410762942779291e-06,
+      "loss": 1.8313,
+      "step": 10579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.408719346049046e-06,
+      "loss": 1.7427,
+      "step": 10580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.406675749318802e-06,
+      "loss": 1.8216,
+      "step": 10581
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.404632152588556e-06,
+      "loss": 1.7278,
+      "step": 10582
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.40258855585831e-06,
+      "loss": 1.6482,
+      "step": 10583
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.400544959128066e-06,
+      "loss": 1.5886,
+      "step": 10584
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.398501362397821e-06,
+      "loss": 1.6242,
+      "step": 10585
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.396457765667575e-06,
+      "loss": 1.6679,
+      "step": 10586
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.394414168937329e-06,
+      "loss": 1.688,
+      "step": 10587
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.392370572207085e-06,
+      "loss": 1.6967,
+      "step": 10588
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.390326975476839e-06,
+      "loss": 1.6012,
+      "step": 10589
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.388283378746594e-06,
+      "loss": 1.7507,
+      "step": 10590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.38623978201635e-06,
+      "loss": 1.4823,
+      "step": 10591
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.384196185286104e-06,
+      "loss": 1.6732,
+      "step": 10592
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.382152588555858e-06,
+      "loss": 1.7136,
+      "step": 10593
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.380108991825613e-06,
+      "loss": 1.7388,
+      "step": 10594
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.378065395095369e-06,
+      "loss": 1.7108,
+      "step": 10595
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.376021798365123e-06,
+      "loss": 1.7126,
+      "step": 10596
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.373978201634877e-06,
+      "loss": 1.6391,
+      "step": 10597
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.371934604904631e-06,
+      "loss": 1.5743,
+      "step": 10598
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.369891008174388e-06,
+      "loss": 1.6374,
+      "step": 10599
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.367847411444142e-06,
+      "loss": 1.7026,
+      "step": 10600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.365803814713896e-06,
+      "loss": 1.6566,
+      "step": 10601
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.36376021798365e-06,
+      "loss": 1.6327,
+      "step": 10602
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.361716621253407e-06,
+      "loss": 1.583,
+      "step": 10603
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.359673024523161e-06,
+      "loss": 1.6931,
+      "step": 10604
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.357629427792915e-06,
+      "loss": 1.6621,
+      "step": 10605
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.355585831062671e-06,
+      "loss": 1.738,
+      "step": 10606
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.353542234332425e-06,
+      "loss": 1.5253,
+      "step": 10607
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.35149863760218e-06,
+      "loss": 1.7671,
+      "step": 10608
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.349455040871934e-06,
+      "loss": 1.5945,
+      "step": 10609
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.34741144414169e-06,
+      "loss": 1.8179,
+      "step": 10610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.345367847411444e-06,
+      "loss": 1.8376,
+      "step": 10611
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.3433242506812e-06,
+      "loss": 1.6692,
+      "step": 10612
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.341280653950954e-06,
+      "loss": 1.6632,
+      "step": 10613
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.33923705722071e-06,
+      "loss": 1.7593,
+      "step": 10614
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.337193460490463e-06,
+      "loss": 1.6249,
+      "step": 10615
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.335149863760217e-06,
+      "loss": 1.7064,
+      "step": 10616
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.333106267029973e-06,
+      "loss": 1.634,
+      "step": 10617
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.331062670299728e-06,
+      "loss": 1.6011,
+      "step": 10618
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.329019073569482e-06,
+      "loss": 1.7693,
+      "step": 10619
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.326975476839236e-06,
+      "loss": 1.7856,
+      "step": 10620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.324931880108994e-06,
+      "loss": 1.5824,
+      "step": 10621
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.322888283378747e-06,
+      "loss": 1.6991,
+      "step": 10622
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.320844686648501e-06,
+      "loss": 1.572,
+      "step": 10623
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.318801089918255e-06,
+      "loss": 1.7121,
+      "step": 10624
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.316757493188011e-06,
+      "loss": 1.8945,
+      "step": 10625
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.314713896457767e-06,
+      "loss": 1.8133,
+      "step": 10626
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.31267029972752e-06,
+      "loss": 1.7401,
+      "step": 10627
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.310626702997275e-06,
+      "loss": 1.5565,
+      "step": 10628
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.30858310626703e-06,
+      "loss": 1.6452,
+      "step": 10629
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.306539509536786e-06,
+      "loss": 1.5308,
+      "step": 10630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.30449591280654e-06,
+      "loss": 1.7294,
+      "step": 10631
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.302452316076295e-06,
+      "loss": 1.587,
+      "step": 10632
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.30040871934605e-06,
+      "loss": 1.7569,
+      "step": 10633
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.298365122615803e-06,
+      "loss": 1.7297,
+      "step": 10634
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.296321525885559e-06,
+      "loss": 1.584,
+      "step": 10635
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.294277929155314e-06,
+      "loss": 1.5708,
+      "step": 10636
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.292234332425068e-06,
+      "loss": 1.7404,
+      "step": 10637
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.290190735694822e-06,
+      "loss": 1.7071,
+      "step": 10638
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.288147138964578e-06,
+      "loss": 1.7467,
+      "step": 10639
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.286103542234334e-06,
+      "loss": 1.643,
+      "step": 10640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.284059945504088e-06,
+      "loss": 1.6519,
+      "step": 10641
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.282016348773842e-06,
+      "loss": 1.7443,
+      "step": 10642
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.279972752043595e-06,
+      "loss": 1.6739,
+      "step": 10643
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.277929155313353e-06,
+      "loss": 1.6284,
+      "step": 10644
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.275885558583107e-06,
+      "loss": 1.7182,
+      "step": 10645
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.27384196185286e-06,
+      "loss": 1.6403,
+      "step": 10646
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.271798365122616e-06,
+      "loss": 1.7,
+      "step": 10647
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.269754768392372e-06,
+      "loss": 1.6771,
+      "step": 10648
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.267711171662126e-06,
+      "loss": 1.6685,
+      "step": 10649
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.26566757493188e-06,
+      "loss": 1.6824,
+      "step": 10650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.263623978201635e-06,
+      "loss": 1.518,
+      "step": 10651
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.26158038147139e-06,
+      "loss": 1.7448,
+      "step": 10652
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.259536784741145e-06,
+      "loss": 1.488,
+      "step": 10653
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.257493188010899e-06,
+      "loss": 1.6875,
+      "step": 10654
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.255449591280655e-06,
+      "loss": 1.6622,
+      "step": 10655
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.253405994550409e-06,
+      "loss": 1.8086,
+      "step": 10656
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.251362397820164e-06,
+      "loss": 1.6229,
+      "step": 10657
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.249318801089918e-06,
+      "loss": 1.6567,
+      "step": 10658
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.247275204359674e-06,
+      "loss": 1.7334,
+      "step": 10659
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.245231607629428e-06,
+      "loss": 1.6618,
+      "step": 10660
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.243188010899182e-06,
+      "loss": 1.7385,
+      "step": 10661
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.241144414168939e-06,
+      "loss": 1.6475,
+      "step": 10662
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.239100817438693e-06,
+      "loss": 1.6097,
+      "step": 10663
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.237057220708447e-06,
+      "loss": 1.613,
+      "step": 10664
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.2350136239782e-06,
+      "loss": 1.4971,
+      "step": 10665
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.232970027247958e-06,
+      "loss": 1.7904,
+      "step": 10666
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.230926430517712e-06,
+      "loss": 1.6659,
+      "step": 10667
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.228882833787466e-06,
+      "loss": 1.7014,
+      "step": 10668
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.22683923705722e-06,
+      "loss": 1.6417,
+      "step": 10669
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.224795640326976e-06,
+      "loss": 1.5987,
+      "step": 10670
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.222752043596731e-06,
+      "loss": 1.7725,
+      "step": 10671
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.220708446866485e-06,
+      "loss": 1.6714,
+      "step": 10672
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.21866485013624e-06,
+      "loss": 1.6813,
+      "step": 10673
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.216621253405995e-06,
+      "loss": 1.7296,
+      "step": 10674
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.21457765667575e-06,
+      "loss": 1.7072,
+      "step": 10675
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.212534059945504e-06,
+      "loss": 1.6489,
+      "step": 10676
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.21049046321526e-06,
+      "loss": 1.6213,
+      "step": 10677
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.208446866485014e-06,
+      "loss": 1.7654,
+      "step": 10678
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.206403269754768e-06,
+      "loss": 1.6157,
+      "step": 10679
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.204359673024523e-06,
+      "loss": 1.71,
+      "step": 10680
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.202316076294279e-06,
+      "loss": 1.5126,
+      "step": 10681
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.200272479564033e-06,
+      "loss": 1.6588,
+      "step": 10682
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.198228882833787e-06,
+      "loss": 1.4747,
+      "step": 10683
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.196185286103542e-06,
+      "loss": 1.874,
+      "step": 10684
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.194141689373298e-06,
+      "loss": 1.7322,
+      "step": 10685
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.192098092643052e-06,
+      "loss": 1.8468,
+      "step": 10686
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.190054495912806e-06,
+      "loss": 1.6686,
+      "step": 10687
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.188010899182562e-06,
+      "loss": 1.7429,
+      "step": 10688
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.185967302452317e-06,
+      "loss": 1.6781,
+      "step": 10689
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.183923705722071e-06,
+      "loss": 1.6851,
+      "step": 10690
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.181880108991825e-06,
+      "loss": 1.6274,
+      "step": 10691
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.17983651226158e-06,
+      "loss": 1.793,
+      "step": 10692
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.177792915531336e-06,
+      "loss": 1.6989,
+      "step": 10693
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.17574931880109e-06,
+      "loss": 1.683,
+      "step": 10694
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.173705722070844e-06,
+      "loss": 1.7266,
+      "step": 10695
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.1716621253406e-06,
+      "loss": 1.6033,
+      "step": 10696
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.169618528610354e-06,
+      "loss": 1.6564,
+      "step": 10697
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.16757493188011e-06,
+      "loss": 1.7495,
+      "step": 10698
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.165531335149863e-06,
+      "loss": 1.6591,
+      "step": 10699
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.163487738419619e-06,
+      "loss": 1.6545,
+      "step": 10700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.161444141689373e-06,
+      "loss": 1.7336,
+      "step": 10701
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.159400544959129e-06,
+      "loss": 1.6725,
+      "step": 10702
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.157356948228884e-06,
+      "loss": 1.6832,
+      "step": 10703
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.155313351498638e-06,
+      "loss": 1.5498,
+      "step": 10704
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.153269754768392e-06,
+      "loss": 1.6677,
+      "step": 10705
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.151226158038146e-06,
+      "loss": 1.7219,
+      "step": 10706
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.149182561307903e-06,
+      "loss": 1.6516,
+      "step": 10707
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.147138964577657e-06,
+      "loss": 1.7382,
+      "step": 10708
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.145095367847411e-06,
+      "loss": 1.6694,
+      "step": 10709
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.143051771117165e-06,
+      "loss": 1.7543,
+      "step": 10710
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.141008174386923e-06,
+      "loss": 1.784,
+      "step": 10711
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.138964577656676e-06,
+      "loss": 1.7676,
+      "step": 10712
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.13692098092643e-06,
+      "loss": 1.7535,
+      "step": 10713
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.134877384196186e-06,
+      "loss": 1.789,
+      "step": 10714
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.13283378746594e-06,
+      "loss": 1.7329,
+      "step": 10715
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.130790190735696e-06,
+      "loss": 1.795,
+      "step": 10716
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.12874659400545e-06,
+      "loss": 1.6501,
+      "step": 10717
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.126702997275205e-06,
+      "loss": 1.7843,
+      "step": 10718
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.124659400544959e-06,
+      "loss": 1.6756,
+      "step": 10719
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.122615803814715e-06,
+      "loss": 1.7071,
+      "step": 10720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.120572207084469e-06,
+      "loss": 1.7837,
+      "step": 10721
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.118528610354224e-06,
+      "loss": 1.699,
+      "step": 10722
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.116485013623978e-06,
+      "loss": 1.6527,
+      "step": 10723
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.114441416893732e-06,
+      "loss": 1.725,
+      "step": 10724
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.112397820163488e-06,
+      "loss": 1.688,
+      "step": 10725
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.110354223433243e-06,
+      "loss": 1.8422,
+      "step": 10726
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.108310626702997e-06,
+      "loss": 1.7671,
+      "step": 10727
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.106267029972751e-06,
+      "loss": 1.602,
+      "step": 10728
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.104223433242507e-06,
+      "loss": 1.4786,
+      "step": 10729
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.102179836512263e-06,
+      "loss": 1.6326,
+      "step": 10730
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.100136239782017e-06,
+      "loss": 1.6747,
+      "step": 10731
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.09809264305177e-06,
+      "loss": 1.65,
+      "step": 10732
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.096049046321526e-06,
+      "loss": 1.6896,
+      "step": 10733
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.094005449591282e-06,
+      "loss": 1.7017,
+      "step": 10734
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.091961852861036e-06,
+      "loss": 1.6631,
+      "step": 10735
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.08991825613079e-06,
+      "loss": 1.5066,
+      "step": 10736
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.087874659400545e-06,
+      "loss": 1.7038,
+      "step": 10737
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.0858310626703e-06,
+      "loss": 1.745,
+      "step": 10738
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.083787465940055e-06,
+      "loss": 1.6746,
+      "step": 10739
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.081743869209809e-06,
+      "loss": 1.5954,
+      "step": 10740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.079700272479564e-06,
+      "loss": 1.6844,
+      "step": 10741
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.077656675749318e-06,
+      "loss": 1.7129,
+      "step": 10742
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.075613079019074e-06,
+      "loss": 1.6341,
+      "step": 10743
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.07356948228883e-06,
+      "loss": 1.7211,
+      "step": 10744
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.071525885558584e-06,
+      "loss": 1.6284,
+      "step": 10745
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.069482288828338e-06,
+      "loss": 1.7606,
+      "step": 10746
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.067438692098091e-06,
+      "loss": 1.6825,
+      "step": 10747
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.065395095367849e-06,
+      "loss": 1.6637,
+      "step": 10748
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.063351498637603e-06,
+      "loss": 1.6778,
+      "step": 10749
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.061307901907357e-06,
+      "loss": 1.8141,
+      "step": 10750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.05926430517711e-06,
+      "loss": 1.6609,
+      "step": 10751
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.057220708446868e-06,
+      "loss": 1.5081,
+      "step": 10752
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.055177111716622e-06,
+      "loss": 1.6694,
+      "step": 10753
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.053133514986376e-06,
+      "loss": 1.6512,
+      "step": 10754
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.05108991825613e-06,
+      "loss": 1.6604,
+      "step": 10755
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.049046321525885e-06,
+      "loss": 1.6315,
+      "step": 10756
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.047002724795641e-06,
+      "loss": 1.615,
+      "step": 10757
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.044959128065395e-06,
+      "loss": 1.6697,
+      "step": 10758
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.04291553133515e-06,
+      "loss": 1.6515,
+      "step": 10759
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.040871934604905e-06,
+      "loss": 1.6403,
+      "step": 10760
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.03882833787466e-06,
+      "loss": 1.6576,
+      "step": 10761
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.036784741144414e-06,
+      "loss": 1.6112,
+      "step": 10762
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.03474114441417e-06,
+      "loss": 1.6566,
+      "step": 10763
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.032697547683924e-06,
+      "loss": 1.6676,
+      "step": 10764
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.030653950953678e-06,
+      "loss": 1.7501,
+      "step": 10765
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.028610354223433e-06,
+      "loss": 1.6766,
+      "step": 10766
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.026566757493189e-06,
+      "loss": 1.7438,
+      "step": 10767
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.024523160762943e-06,
+      "loss": 1.628,
+      "step": 10768
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.022479564032697e-06,
+      "loss": 1.7137,
+      "step": 10769
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.020435967302454e-06,
+      "loss": 1.7411,
+      "step": 10770
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.018392370572208e-06,
+      "loss": 1.6552,
+      "step": 10771
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.016348773841962e-06,
+      "loss": 1.8263,
+      "step": 10772
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.014305177111716e-06,
+      "loss": 1.6874,
+      "step": 10773
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.012261580381472e-06,
+      "loss": 1.6276,
+      "step": 10774
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.010217983651227e-06,
+      "loss": 1.6892,
+      "step": 10775
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.008174386920981e-06,
+      "loss": 1.6264,
+      "step": 10776
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.006130790190735e-06,
+      "loss": 1.5926,
+      "step": 10777
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.00408719346049e-06,
+      "loss": 1.6922,
+      "step": 10778
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.002043596730246e-06,
+      "loss": 1.6902,
+      "step": 10779
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9e-06,
+      "loss": 1.776,
+      "step": 10780
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.997956403269754e-06,
+      "loss": 1.6908,
+      "step": 10781
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.99591280653951e-06,
+      "loss": 1.6745,
+      "step": 10782
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.993869209809264e-06,
+      "loss": 1.5574,
+      "step": 10783
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.99182561307902e-06,
+      "loss": 1.7661,
+      "step": 10784
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.989782016348775e-06,
+      "loss": 1.6008,
+      "step": 10785
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.987738419618529e-06,
+      "loss": 1.7659,
+      "step": 10786
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.985694822888283e-06,
+      "loss": 1.6717,
+      "step": 10787
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.983651226158039e-06,
+      "loss": 1.7314,
+      "step": 10788
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.981607629427794e-06,
+      "loss": 1.5917,
+      "step": 10789
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.979564032697548e-06,
+      "loss": 1.6278,
+      "step": 10790
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.977520435967302e-06,
+      "loss": 1.708,
+      "step": 10791
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.975476839237056e-06,
+      "loss": 1.5902,
+      "step": 10792
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.973433242506813e-06,
+      "loss": 1.6935,
+      "step": 10793
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.971389645776567e-06,
+      "loss": 1.7203,
+      "step": 10794
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.969346049046321e-06,
+      "loss": 1.7468,
+      "step": 10795
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.967302452316075e-06,
+      "loss": 1.7262,
+      "step": 10796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.965258855585832e-06,
+      "loss": 1.6033,
+      "step": 10797
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.963215258855586e-06,
+      "loss": 1.6333,
+      "step": 10798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.96117166212534e-06,
+      "loss": 1.673,
+      "step": 10799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.959128065395096e-06,
+      "loss": 1.6186,
+      "step": 10800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.95708446866485e-06,
+      "loss": 1.6168,
+      "step": 10801
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.955040871934606e-06,
+      "loss": 1.709,
+      "step": 10802
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.95299727520436e-06,
+      "loss": 1.6491,
+      "step": 10803
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.950953678474115e-06,
+      "loss": 1.7256,
+      "step": 10804
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.948910081743869e-06,
+      "loss": 1.6338,
+      "step": 10805
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.946866485013625e-06,
+      "loss": 1.7497,
+      "step": 10806
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.944822888283379e-06,
+      "loss": 1.5776,
+      "step": 10807
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.942779291553134e-06,
+      "loss": 1.6543,
+      "step": 10808
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.940735694822888e-06,
+      "loss": 1.5768,
+      "step": 10809
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.938692098092642e-06,
+      "loss": 1.8146,
+      "step": 10810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.9366485013624e-06,
+      "loss": 1.8502,
+      "step": 10811
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.934604904632153e-06,
+      "loss": 1.7034,
+      "step": 10812
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.932561307901907e-06,
+      "loss": 1.71,
+      "step": 10813
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.930517711171661e-06,
+      "loss": 1.8323,
+      "step": 10814
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.928474114441419e-06,
+      "loss": 1.7057,
+      "step": 10815
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.926430517711173e-06,
+      "loss": 1.652,
+      "step": 10816
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.924386920980926e-06,
+      "loss": 1.6821,
+      "step": 10817
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.92234332425068e-06,
+      "loss": 1.6986,
+      "step": 10818
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.920299727520436e-06,
+      "loss": 1.8505,
+      "step": 10819
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.918256130790192e-06,
+      "loss": 1.7337,
+      "step": 10820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.916212534059946e-06,
+      "loss": 1.8209,
+      "step": 10821
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.9141689373297e-06,
+      "loss": 1.7072,
+      "step": 10822
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.912125340599455e-06,
+      "loss": 1.631,
+      "step": 10823
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.91008174386921e-06,
+      "loss": 1.822,
+      "step": 10824
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.908038147138965e-06,
+      "loss": 1.694,
+      "step": 10825
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.90599455040872e-06,
+      "loss": 1.7122,
+      "step": 10826
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.903950953678474e-06,
+      "loss": 1.6739,
+      "step": 10827
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.901907356948228e-06,
+      "loss": 1.7077,
+      "step": 10828
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.899863760217984e-06,
+      "loss": 1.667,
+      "step": 10829
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.89782016348774e-06,
+      "loss": 1.5789,
+      "step": 10830
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.895776566757493e-06,
+      "loss": 1.7883,
+      "step": 10831
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.893732970027247e-06,
+      "loss": 1.6595,
+      "step": 10832
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.891689373297003e-06,
+      "loss": 1.6463,
+      "step": 10833
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.889645776566759e-06,
+      "loss": 1.5512,
+      "step": 10834
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.887602179836513e-06,
+      "loss": 1.6078,
+      "step": 10835
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.885558583106267e-06,
+      "loss": 1.8311,
+      "step": 10836
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.88351498637602e-06,
+      "loss": 1.7739,
+      "step": 10837
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.881471389645778e-06,
+      "loss": 1.7049,
+      "step": 10838
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.879427792915532e-06,
+      "loss": 1.8669,
+      "step": 10839
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.877384196185286e-06,
+      "loss": 1.5783,
+      "step": 10840
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.875340599455041e-06,
+      "loss": 1.6345,
+      "step": 10841
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.873297002724797e-06,
+      "loss": 1.6438,
+      "step": 10842
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.871253405994551e-06,
+      "loss": 1.6012,
+      "step": 10843
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.869209809264305e-06,
+      "loss": 1.5821,
+      "step": 10844
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.86716621253406e-06,
+      "loss": 1.6601,
+      "step": 10845
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.865122615803814e-06,
+      "loss": 1.7449,
+      "step": 10846
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.86307901907357e-06,
+      "loss": 1.7021,
+      "step": 10847
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.861035422343324e-06,
+      "loss": 1.6026,
+      "step": 10848
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.85899182561308e-06,
+      "loss": 1.7495,
+      "step": 10849
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.856948228882834e-06,
+      "loss": 1.7223,
+      "step": 10850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.85490463215259e-06,
+      "loss": 1.6679,
+      "step": 10851
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.852861035422345e-06,
+      "loss": 1.7501,
+      "step": 10852
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.850817438692099e-06,
+      "loss": 1.6177,
+      "step": 10853
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.848773841961853e-06,
+      "loss": 1.6759,
+      "step": 10854
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.846730245231607e-06,
+      "loss": 1.7387,
+      "step": 10855
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.844686648501364e-06,
+      "loss": 1.6362,
+      "step": 10856
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.842643051771118e-06,
+      "loss": 1.6978,
+      "step": 10857
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.840599455040872e-06,
+      "loss": 1.7065,
+      "step": 10858
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.838555858310626e-06,
+      "loss": 1.6622,
+      "step": 10859
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.836512261580383e-06,
+      "loss": 1.6615,
+      "step": 10860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.834468664850137e-06,
+      "loss": 1.7322,
+      "step": 10861
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.832425068119891e-06,
+      "loss": 1.757,
+      "step": 10862
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.830381471389645e-06,
+      "loss": 1.6395,
+      "step": 10863
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.8283378746594e-06,
+      "loss": 1.5786,
+      "step": 10864
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.826294277929156e-06,
+      "loss": 1.7294,
+      "step": 10865
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.82425068119891e-06,
+      "loss": 1.5868,
+      "step": 10866
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.822207084468666e-06,
+      "loss": 1.6234,
+      "step": 10867
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.82016348773842e-06,
+      "loss": 1.6377,
+      "step": 10868
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.818119891008175e-06,
+      "loss": 1.6822,
+      "step": 10869
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.81607629427793e-06,
+      "loss": 1.6017,
+      "step": 10870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.814032697547685e-06,
+      "loss": 1.6647,
+      "step": 10871
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.811989100817439e-06,
+      "loss": 1.6755,
+      "step": 10872
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.809945504087193e-06,
+      "loss": 1.6807,
+      "step": 10873
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.807901907356948e-06,
+      "loss": 1.6056,
+      "step": 10874
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.805858310626704e-06,
+      "loss": 1.6719,
+      "step": 10875
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.803814713896458e-06,
+      "loss": 1.8043,
+      "step": 10876
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.801771117166212e-06,
+      "loss": 1.6787,
+      "step": 10877
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.799727520435968e-06,
+      "loss": 1.7371,
+      "step": 10878
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.797683923705723e-06,
+      "loss": 1.7567,
+      "step": 10879
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.795640326975477e-06,
+      "loss": 1.637,
+      "step": 10880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.793596730245231e-06,
+      "loss": 1.8096,
+      "step": 10881
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.791553133514987e-06,
+      "loss": 1.6748,
+      "step": 10882
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.789509536784742e-06,
+      "loss": 1.6511,
+      "step": 10883
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.787465940054496e-06,
+      "loss": 1.7274,
+      "step": 10884
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.78542234332425e-06,
+      "loss": 1.6937,
+      "step": 10885
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.783378746594006e-06,
+      "loss": 1.6798,
+      "step": 10886
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.781335149863761e-06,
+      "loss": 1.7558,
+      "step": 10887
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.779291553133515e-06,
+      "loss": 1.7813,
+      "step": 10888
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.77724795640327e-06,
+      "loss": 1.8382,
+      "step": 10889
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.775204359673025e-06,
+      "loss": 1.698,
+      "step": 10890
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.773160762942779e-06,
+      "loss": 1.7067,
+      "step": 10891
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.771117166212535e-06,
+      "loss": 1.6841,
+      "step": 10892
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.769073569482288e-06,
+      "loss": 1.7067,
+      "step": 10893
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.767029972752044e-06,
+      "loss": 1.5693,
+      "step": 10894
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.764986376021798e-06,
+      "loss": 1.6185,
+      "step": 10895
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.762942779291554e-06,
+      "loss": 1.6953,
+      "step": 10896
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.76089918256131e-06,
+      "loss": 1.6165,
+      "step": 10897
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.758855585831063e-06,
+      "loss": 1.6211,
+      "step": 10898
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.756811989100817e-06,
+      "loss": 1.6044,
+      "step": 10899
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.754768392370571e-06,
+      "loss": 1.6791,
+      "step": 10900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.752724795640328e-06,
+      "loss": 1.7596,
+      "step": 10901
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.750681198910082e-06,
+      "loss": 1.6648,
+      "step": 10902
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.748637602179836e-06,
+      "loss": 1.6422,
+      "step": 10903
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.74659400544959e-06,
+      "loss": 1.7393,
+      "step": 10904
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.744550408719348e-06,
+      "loss": 1.7298,
+      "step": 10905
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.742506811989102e-06,
+      "loss": 1.597,
+      "step": 10906
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.740463215258855e-06,
+      "loss": 1.6788,
+      "step": 10907
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.738419618528611e-06,
+      "loss": 1.6082,
+      "step": 10908
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.736376021798365e-06,
+      "loss": 1.6818,
+      "step": 10909
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.73433242506812e-06,
+      "loss": 1.7088,
+      "step": 10910
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.732288828337875e-06,
+      "loss": 1.6282,
+      "step": 10911
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.73024523160763e-06,
+      "loss": 1.6641,
+      "step": 10912
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.728201634877384e-06,
+      "loss": 1.6921,
+      "step": 10913
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.72615803814714e-06,
+      "loss": 1.737,
+      "step": 10914
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.724114441416894e-06,
+      "loss": 1.5486,
+      "step": 10915
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.72207084468665e-06,
+      "loss": 1.6527,
+      "step": 10916
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.720027247956403e-06,
+      "loss": 1.7094,
+      "step": 10917
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.717983651226157e-06,
+      "loss": 1.6113,
+      "step": 10918
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.715940054495913e-06,
+      "loss": 1.6209,
+      "step": 10919
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.713896457765669e-06,
+      "loss": 1.6933,
+      "step": 10920
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.711852861035422e-06,
+      "loss": 1.7034,
+      "step": 10921
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.709809264305176e-06,
+      "loss": 1.812,
+      "step": 10922
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.707765667574932e-06,
+      "loss": 1.7765,
+      "step": 10923
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.705722070844688e-06,
+      "loss": 1.563,
+      "step": 10924
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.703678474114442e-06,
+      "loss": 1.6764,
+      "step": 10925
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.701634877384196e-06,
+      "loss": 1.7811,
+      "step": 10926
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.699591280653951e-06,
+      "loss": 1.6864,
+      "step": 10927
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.697547683923707e-06,
+      "loss": 1.5323,
+      "step": 10928
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.69550408719346e-06,
+      "loss": 1.6149,
+      "step": 10929
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.693460490463215e-06,
+      "loss": 1.6155,
+      "step": 10930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.69141689373297e-06,
+      "loss": 1.706,
+      "step": 10931
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.689373297002724e-06,
+      "loss": 1.6389,
+      "step": 10932
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.68732970027248e-06,
+      "loss": 1.7133,
+      "step": 10933
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.685286103542234e-06,
+      "loss": 1.6275,
+      "step": 10934
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.68324250681199e-06,
+      "loss": 1.6671,
+      "step": 10935
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.681198910081743e-06,
+      "loss": 1.8289,
+      "step": 10936
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.679155313351499e-06,
+      "loss": 1.6304,
+      "step": 10937
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.677111716621255e-06,
+      "loss": 1.6524,
+      "step": 10938
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.675068119891009e-06,
+      "loss": 1.6956,
+      "step": 10939
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.673024523160763e-06,
+      "loss": 1.8373,
+      "step": 10940
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.670980926430516e-06,
+      "loss": 1.7454,
+      "step": 10941
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.668937329700274e-06,
+      "loss": 1.5724,
+      "step": 10942
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.666893732970028e-06,
+      "loss": 1.6956,
+      "step": 10943
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.664850136239782e-06,
+      "loss": 1.5704,
+      "step": 10944
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.662806539509536e-06,
+      "loss": 1.6867,
+      "step": 10945
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.660762942779293e-06,
+      "loss": 1.6627,
+      "step": 10946
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.658719346049047e-06,
+      "loss": 1.7484,
+      "step": 10947
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.6566757493188e-06,
+      "loss": 1.6617,
+      "step": 10948
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.654632152588556e-06,
+      "loss": 1.6304,
+      "step": 10949
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.65258855585831e-06,
+      "loss": 1.5958,
+      "step": 10950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.650544959128066e-06,
+      "loss": 1.6405,
+      "step": 10951
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.64850136239782e-06,
+      "loss": 1.5856,
+      "step": 10952
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.646457765667576e-06,
+      "loss": 1.6335,
+      "step": 10953
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.64441416893733e-06,
+      "loss": 1.6146,
+      "step": 10954
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.642370572207085e-06,
+      "loss": 1.6261,
+      "step": 10955
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.640326975476839e-06,
+      "loss": 1.8684,
+      "step": 10956
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.638283378746595e-06,
+      "loss": 1.7652,
+      "step": 10957
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.636239782016349e-06,
+      "loss": 1.5936,
+      "step": 10958
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.634196185286103e-06,
+      "loss": 1.6753,
+      "step": 10959
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.632152588555858e-06,
+      "loss": 1.7466,
+      "step": 10960
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.630108991825614e-06,
+      "loss": 1.6716,
+      "step": 10961
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.628065395095368e-06,
+      "loss": 1.7415,
+      "step": 10962
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.626021798365122e-06,
+      "loss": 1.5312,
+      "step": 10963
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.623978201634879e-06,
+      "loss": 1.6687,
+      "step": 10964
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.621934604904633e-06,
+      "loss": 1.7093,
+      "step": 10965
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.619891008174387e-06,
+      "loss": 1.5944,
+      "step": 10966
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.617847411444141e-06,
+      "loss": 1.6663,
+      "step": 10967
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.615803814713897e-06,
+      "loss": 1.7386,
+      "step": 10968
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.613760217983652e-06,
+      "loss": 1.6887,
+      "step": 10969
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.611716621253406e-06,
+      "loss": 1.5739,
+      "step": 10970
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.60967302452316e-06,
+      "loss": 1.6639,
+      "step": 10971
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.607629427792916e-06,
+      "loss": 1.6764,
+      "step": 10972
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.605585831062671e-06,
+      "loss": 1.618,
+      "step": 10973
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.603542234332425e-06,
+      "loss": 1.6535,
+      "step": 10974
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.60149863760218e-06,
+      "loss": 1.7692,
+      "step": 10975
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.599455040871935e-06,
+      "loss": 1.5366,
+      "step": 10976
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.597411444141689e-06,
+      "loss": 1.7689,
+      "step": 10977
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.595367847411444e-06,
+      "loss": 1.7106,
+      "step": 10978
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.5933242506812e-06,
+      "loss": 1.7104,
+      "step": 10979
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.591280653950954e-06,
+      "loss": 1.6501,
+      "step": 10980
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.589237057220708e-06,
+      "loss": 1.586,
+      "step": 10981
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.587193460490464e-06,
+      "loss": 1.5857,
+      "step": 10982
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.58514986376022e-06,
+      "loss": 1.6616,
+      "step": 10983
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.583106267029973e-06,
+      "loss": 1.6071,
+      "step": 10984
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.581062670299727e-06,
+      "loss": 1.7882,
+      "step": 10985
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.579019073569481e-06,
+      "loss": 1.6154,
+      "step": 10986
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.576975476839238e-06,
+      "loss": 1.7173,
+      "step": 10987
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.574931880108992e-06,
+      "loss": 1.6775,
+      "step": 10988
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.572888283378746e-06,
+      "loss": 1.6954,
+      "step": 10989
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.570844686648502e-06,
+      "loss": 1.6321,
+      "step": 10990
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.568801089918257e-06,
+      "loss": 1.7358,
+      "step": 10991
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.566757493188011e-06,
+      "loss": 1.7762,
+      "step": 10992
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.564713896457765e-06,
+      "loss": 1.6454,
+      "step": 10993
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.562670299727521e-06,
+      "loss": 1.7231,
+      "step": 10994
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.560626702997275e-06,
+      "loss": 1.6145,
+      "step": 10995
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.55858310626703e-06,
+      "loss": 1.7805,
+      "step": 10996
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.556539509536784e-06,
+      "loss": 1.7083,
+      "step": 10997
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.55449591280654e-06,
+      "loss": 1.6837,
+      "step": 10998
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.552452316076294e-06,
+      "loss": 1.5719,
+      "step": 10999
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.55040871934605e-06,
+      "loss": 1.7109,
+      "step": 11000
+    },
+    {
+      "epoch": 2.17,
+      "eval_loss": 1.5138320922851562,
+      "eval_runtime": 721.9157,
+      "eval_samples_per_second": 5.749,
+      "eval_steps_per_second": 1.438,
+      "step": 11000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.548365122615804e-06,
+      "loss": 1.7165,
+      "step": 11001
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.54632152588556e-06,
+      "loss": 1.6957,
+      "step": 11002
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.544277929155313e-06,
+      "loss": 1.7364,
+      "step": 11003
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.542234332425067e-06,
+      "loss": 1.6489,
+      "step": 11004
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.540190735694824e-06,
+      "loss": 1.581,
+      "step": 11005
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.538147138964578e-06,
+      "loss": 1.6738,
+      "step": 11006
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.536103542234332e-06,
+      "loss": 1.6014,
+      "step": 11007
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.534059945504086e-06,
+      "loss": 1.5914,
+      "step": 11008
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.532016348773844e-06,
+      "loss": 1.7367,
+      "step": 11009
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.529972752043598e-06,
+      "loss": 1.4561,
+      "step": 11010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.527929155313351e-06,
+      "loss": 1.6471,
+      "step": 11011
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.525885558583105e-06,
+      "loss": 1.8621,
+      "step": 11012
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.523841961852861e-06,
+      "loss": 1.7134,
+      "step": 11013
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.521798365122617e-06,
+      "loss": 1.7375,
+      "step": 11014
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.51975476839237e-06,
+      "loss": 1.6332,
+      "step": 11015
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.517711171662125e-06,
+      "loss": 1.836,
+      "step": 11016
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.51566757493188e-06,
+      "loss": 1.6532,
+      "step": 11017
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.513623978201636e-06,
+      "loss": 1.5584,
+      "step": 11018
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.51158038147139e-06,
+      "loss": 1.5895,
+      "step": 11019
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.509536784741145e-06,
+      "loss": 1.602,
+      "step": 11020
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.5074931880109e-06,
+      "loss": 1.67,
+      "step": 11021
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.505449591280653e-06,
+      "loss": 1.7871,
+      "step": 11022
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.503405994550409e-06,
+      "loss": 1.7439,
+      "step": 11023
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.501362397820165e-06,
+      "loss": 1.693,
+      "step": 11024
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.499318801089918e-06,
+      "loss": 1.6343,
+      "step": 11025
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.497275204359672e-06,
+      "loss": 1.8612,
+      "step": 11026
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.495231607629428e-06,
+      "loss": 1.5813,
+      "step": 11027
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.493188010899184e-06,
+      "loss": 1.6685,
+      "step": 11028
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.491144414168938e-06,
+      "loss": 1.6017,
+      "step": 11029
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.489100817438692e-06,
+      "loss": 1.6812,
+      "step": 11030
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.487057220708445e-06,
+      "loss": 1.8595,
+      "step": 11031
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.485013623978203e-06,
+      "loss": 1.7645,
+      "step": 11032
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.482970027247957e-06,
+      "loss": 1.6242,
+      "step": 11033
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.48092643051771e-06,
+      "loss": 1.6668,
+      "step": 11034
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.478882833787466e-06,
+      "loss": 1.948,
+      "step": 11035
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.476839237057222e-06,
+      "loss": 1.6883,
+      "step": 11036
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.474795640326976e-06,
+      "loss": 1.9253,
+      "step": 11037
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.47275204359673e-06,
+      "loss": 1.6519,
+      "step": 11038
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.470708446866485e-06,
+      "loss": 1.7346,
+      "step": 11039
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.46866485013624e-06,
+      "loss": 1.7193,
+      "step": 11040
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.466621253405995e-06,
+      "loss": 1.7674,
+      "step": 11041
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.464577656675749e-06,
+      "loss": 1.7002,
+      "step": 11042
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.462534059945505e-06,
+      "loss": 1.7133,
+      "step": 11043
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.460490463215259e-06,
+      "loss": 1.5763,
+      "step": 11044
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.458446866485014e-06,
+      "loss": 1.6763,
+      "step": 11045
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.45640326975477e-06,
+      "loss": 1.6222,
+      "step": 11046
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.454359673024524e-06,
+      "loss": 1.642,
+      "step": 11047
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.452316076294278e-06,
+      "loss": 1.6911,
+      "step": 11048
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.450272479564032e-06,
+      "loss": 1.6796,
+      "step": 11049
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.448228882833789e-06,
+      "loss": 1.6325,
+      "step": 11050
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.446185286103543e-06,
+      "loss": 1.7829,
+      "step": 11051
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.444141689373297e-06,
+      "loss": 1.7189,
+      "step": 11052
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.44209809264305e-06,
+      "loss": 1.6569,
+      "step": 11053
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.440054495912808e-06,
+      "loss": 1.7049,
+      "step": 11054
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.438010899182562e-06,
+      "loss": 1.6312,
+      "step": 11055
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.435967302452316e-06,
+      "loss": 1.6799,
+      "step": 11056
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.43392370572207e-06,
+      "loss": 1.5243,
+      "step": 11057
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.431880108991826e-06,
+      "loss": 1.759,
+      "step": 11058
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.429836512261581e-06,
+      "loss": 1.768,
+      "step": 11059
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.427792915531335e-06,
+      "loss": 1.7145,
+      "step": 11060
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.42574931880109e-06,
+      "loss": 1.7613,
+      "step": 11061
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.423705722070845e-06,
+      "loss": 1.7949,
+      "step": 11062
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.4216621253406e-06,
+      "loss": 1.6881,
+      "step": 11063
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.419618528610354e-06,
+      "loss": 1.6181,
+      "step": 11064
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.41757493188011e-06,
+      "loss": 1.6399,
+      "step": 11065
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.415531335149864e-06,
+      "loss": 1.6315,
+      "step": 11066
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.413487738419618e-06,
+      "loss": 1.744,
+      "step": 11067
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.411444141689373e-06,
+      "loss": 1.7625,
+      "step": 11068
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.409400544959129e-06,
+      "loss": 1.6899,
+      "step": 11069
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.407356948228883e-06,
+      "loss": 1.6753,
+      "step": 11070
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.405313351498637e-06,
+      "loss": 1.7363,
+      "step": 11071
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.403269754768393e-06,
+      "loss": 1.6047,
+      "step": 11072
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.401226158038148e-06,
+      "loss": 1.6807,
+      "step": 11073
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.399182561307902e-06,
+      "loss": 1.7936,
+      "step": 11074
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.397138964577656e-06,
+      "loss": 1.6587,
+      "step": 11075
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.395095367847412e-06,
+      "loss": 1.6327,
+      "step": 11076
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.393051771117167e-06,
+      "loss": 1.6644,
+      "step": 11077
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.391008174386921e-06,
+      "loss": 1.8386,
+      "step": 11078
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.388964577656675e-06,
+      "loss": 1.7333,
+      "step": 11079
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.38692098092643e-06,
+      "loss": 1.8219,
+      "step": 11080
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.384877384196186e-06,
+      "loss": 1.6617,
+      "step": 11081
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.38283378746594e-06,
+      "loss": 1.6294,
+      "step": 11082
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.380790190735694e-06,
+      "loss": 1.6828,
+      "step": 11083
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.37874659400545e-06,
+      "loss": 1.7403,
+      "step": 11084
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.376702997275204e-06,
+      "loss": 1.6265,
+      "step": 11085
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.37465940054496e-06,
+      "loss": 1.7532,
+      "step": 11086
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.372615803814715e-06,
+      "loss": 1.7587,
+      "step": 11087
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.370572207084469e-06,
+      "loss": 1.441,
+      "step": 11088
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.368528610354223e-06,
+      "loss": 1.6658,
+      "step": 11089
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.366485013623979e-06,
+      "loss": 1.519,
+      "step": 11090
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.364441416893734e-06,
+      "loss": 1.7116,
+      "step": 11091
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.362397820163488e-06,
+      "loss": 1.648,
+      "step": 11092
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.360354223433242e-06,
+      "loss": 1.5201,
+      "step": 11093
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.358310626702996e-06,
+      "loss": 1.6406,
+      "step": 11094
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.356267029972753e-06,
+      "loss": 1.7495,
+      "step": 11095
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.354223433242507e-06,
+      "loss": 1.6308,
+      "step": 11096
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.352179836512261e-06,
+      "loss": 1.659,
+      "step": 11097
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.350136239782015e-06,
+      "loss": 1.6711,
+      "step": 11098
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.348092643051773e-06,
+      "loss": 1.6853,
+      "step": 11099
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.346049046321527e-06,
+      "loss": 1.6927,
+      "step": 11100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.34400544959128e-06,
+      "loss": 1.7199,
+      "step": 11101
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.341961852861036e-06,
+      "loss": 1.7175,
+      "step": 11102
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.33991825613079e-06,
+      "loss": 1.6539,
+      "step": 11103
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.337874659400546e-06,
+      "loss": 1.7033,
+      "step": 11104
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.3358310626703e-06,
+      "loss": 1.7843,
+      "step": 11105
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.333787465940055e-06,
+      "loss": 1.8009,
+      "step": 11106
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.33174386920981e-06,
+      "loss": 1.8111,
+      "step": 11107
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.329700272479565e-06,
+      "loss": 1.542,
+      "step": 11108
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.327656675749319e-06,
+      "loss": 1.8151,
+      "step": 11109
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.325613079019074e-06,
+      "loss": 1.598,
+      "step": 11110
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.323569482288828e-06,
+      "loss": 1.7376,
+      "step": 11111
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.321525885558582e-06,
+      "loss": 1.5621,
+      "step": 11112
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.319482288828338e-06,
+      "loss": 1.5161,
+      "step": 11113
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.317438692098094e-06,
+      "loss": 1.4921,
+      "step": 11114
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.315395095367847e-06,
+      "loss": 1.7766,
+      "step": 11115
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.313351498637601e-06,
+      "loss": 1.6211,
+      "step": 11116
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.311307901907357e-06,
+      "loss": 1.6693,
+      "step": 11117
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.309264305177113e-06,
+      "loss": 1.7302,
+      "step": 11118
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.307220708446867e-06,
+      "loss": 1.8257,
+      "step": 11119
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.30517711171662e-06,
+      "loss": 1.7434,
+      "step": 11120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.303133514986376e-06,
+      "loss": 1.6744,
+      "step": 11121
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.301089918256132e-06,
+      "loss": 1.7152,
+      "step": 11122
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.299046321525886e-06,
+      "loss": 1.6657,
+      "step": 11123
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.29700272479564e-06,
+      "loss": 1.5426,
+      "step": 11124
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.294959128065395e-06,
+      "loss": 1.6599,
+      "step": 11125
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.29291553133515e-06,
+      "loss": 1.663,
+      "step": 11126
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.290871934604905e-06,
+      "loss": 1.7417,
+      "step": 11127
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.28882833787466e-06,
+      "loss": 1.7126,
+      "step": 11128
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.286784741144414e-06,
+      "loss": 1.598,
+      "step": 11129
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.284741144414168e-06,
+      "loss": 1.652,
+      "step": 11130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.282697547683924e-06,
+      "loss": 1.6531,
+      "step": 11131
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.28065395095368e-06,
+      "loss": 1.8617,
+      "step": 11132
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.278610354223434e-06,
+      "loss": 1.749,
+      "step": 11133
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.276566757493188e-06,
+      "loss": 1.6814,
+      "step": 11134
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.274523160762942e-06,
+      "loss": 1.8086,
+      "step": 11135
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.272479564032699e-06,
+      "loss": 1.7077,
+      "step": 11136
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.270435967302453e-06,
+      "loss": 1.6637,
+      "step": 11137
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.268392370572207e-06,
+      "loss": 1.8479,
+      "step": 11138
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.26634877384196e-06,
+      "loss": 1.7134,
+      "step": 11139
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.264305177111718e-06,
+      "loss": 1.6464,
+      "step": 11140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.262261580381472e-06,
+      "loss": 1.635,
+      "step": 11141
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.260217983651226e-06,
+      "loss": 1.584,
+      "step": 11142
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.258174386920981e-06,
+      "loss": 1.6135,
+      "step": 11143
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.256130790190735e-06,
+      "loss": 1.6462,
+      "step": 11144
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.254087193460491e-06,
+      "loss": 1.6321,
+      "step": 11145
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.252043596730245e-06,
+      "loss": 1.618,
+      "step": 11146
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.25e-06,
+      "loss": 1.6564,
+      "step": 11147
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.247956403269755e-06,
+      "loss": 1.6639,
+      "step": 11148
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.24591280653951e-06,
+      "loss": 1.6392,
+      "step": 11149
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.243869209809264e-06,
+      "loss": 1.6128,
+      "step": 11150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.24182561307902e-06,
+      "loss": 1.6645,
+      "step": 11151
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.239782016348774e-06,
+      "loss": 1.6819,
+      "step": 11152
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.237738419618528e-06,
+      "loss": 1.6415,
+      "step": 11153
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.235694822888283e-06,
+      "loss": 1.5356,
+      "step": 11154
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.233651226158039e-06,
+      "loss": 1.5966,
+      "step": 11155
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.231607629427793e-06,
+      "loss": 1.7262,
+      "step": 11156
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.229564032697547e-06,
+      "loss": 1.6831,
+      "step": 11157
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.227520435967304e-06,
+      "loss": 1.6882,
+      "step": 11158
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.225476839237058e-06,
+      "loss": 1.6678,
+      "step": 11159
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.223433242506812e-06,
+      "loss": 1.7641,
+      "step": 11160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.221389645776566e-06,
+      "loss": 1.7006,
+      "step": 11161
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.219346049046322e-06,
+      "loss": 1.684,
+      "step": 11162
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.217302452316077e-06,
+      "loss": 1.7884,
+      "step": 11163
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.215258855585831e-06,
+      "loss": 1.5831,
+      "step": 11164
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.213215258855585e-06,
+      "loss": 1.7055,
+      "step": 11165
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.21117166212534e-06,
+      "loss": 1.5582,
+      "step": 11166
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.209128065395096e-06,
+      "loss": 1.8151,
+      "step": 11167
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.20708446866485e-06,
+      "loss": 1.6348,
+      "step": 11168
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.205040871934604e-06,
+      "loss": 1.4963,
+      "step": 11169
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.20299727520436e-06,
+      "loss": 1.5736,
+      "step": 11170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.200953678474114e-06,
+      "loss": 1.7472,
+      "step": 11171
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.19891008174387e-06,
+      "loss": 1.7697,
+      "step": 11172
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.196866485013625e-06,
+      "loss": 1.6358,
+      "step": 11173
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.194822888283379e-06,
+      "loss": 1.7,
+      "step": 11174
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.192779291553133e-06,
+      "loss": 1.7047,
+      "step": 11175
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.190735694822889e-06,
+      "loss": 1.6055,
+      "step": 11176
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.188692098092644e-06,
+      "loss": 1.6246,
+      "step": 11177
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.186648501362398e-06,
+      "loss": 1.5695,
+      "step": 11178
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.184604904632152e-06,
+      "loss": 1.7436,
+      "step": 11179
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.182561307901906e-06,
+      "loss": 1.7301,
+      "step": 11180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.180517711171663e-06,
+      "loss": 1.5455,
+      "step": 11181
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.178474114441417e-06,
+      "loss": 1.8166,
+      "step": 11182
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.176430517711171e-06,
+      "loss": 1.5496,
+      "step": 11183
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.174386920980927e-06,
+      "loss": 1.6536,
+      "step": 11184
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.172343324250682e-06,
+      "loss": 1.5982,
+      "step": 11185
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.170299727520436e-06,
+      "loss": 1.6482,
+      "step": 11186
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.16825613079019e-06,
+      "loss": 1.6974,
+      "step": 11187
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.166212534059946e-06,
+      "loss": 1.7278,
+      "step": 11188
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.1641689373297e-06,
+      "loss": 1.5828,
+      "step": 11189
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.162125340599456e-06,
+      "loss": 1.7066,
+      "step": 11190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.16008174386921e-06,
+      "loss": 1.563,
+      "step": 11191
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.158038147138965e-06,
+      "loss": 1.7437,
+      "step": 11192
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.155994550408719e-06,
+      "loss": 1.6271,
+      "step": 11193
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.153950953678475e-06,
+      "loss": 1.6691,
+      "step": 11194
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.151907356948229e-06,
+      "loss": 1.5178,
+      "step": 11195
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.149863760217984e-06,
+      "loss": 1.5902,
+      "step": 11196
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.147820163487738e-06,
+      "loss": 1.6724,
+      "step": 11197
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.145776566757492e-06,
+      "loss": 1.8474,
+      "step": 11198
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.14373297002725e-06,
+      "loss": 1.7783,
+      "step": 11199
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.141689373297003e-06,
+      "loss": 1.6182,
+      "step": 11200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.139645776566757e-06,
+      "loss": 1.631,
+      "step": 11201
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.137602179836511e-06,
+      "loss": 1.601,
+      "step": 11202
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.135558583106269e-06,
+      "loss": 1.6624,
+      "step": 11203
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.133514986376023e-06,
+      "loss": 1.6643,
+      "step": 11204
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.131471389645777e-06,
+      "loss": 1.652,
+      "step": 11205
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.12942779291553e-06,
+      "loss": 1.7619,
+      "step": 11206
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.127384196185286e-06,
+      "loss": 1.7062,
+      "step": 11207
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.125340599455042e-06,
+      "loss": 1.6599,
+      "step": 11208
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.123297002724796e-06,
+      "loss": 1.7006,
+      "step": 11209
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.12125340599455e-06,
+      "loss": 1.6336,
+      "step": 11210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.119209809264305e-06,
+      "loss": 1.5536,
+      "step": 11211
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.117166212534061e-06,
+      "loss": 1.6404,
+      "step": 11212
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.115122615803815e-06,
+      "loss": 1.781,
+      "step": 11213
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.11307901907357e-06,
+      "loss": 1.6969,
+      "step": 11214
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.111035422343324e-06,
+      "loss": 1.7534,
+      "step": 11215
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.108991825613078e-06,
+      "loss": 1.6973,
+      "step": 11216
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.106948228882834e-06,
+      "loss": 1.7316,
+      "step": 11217
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.10490463215259e-06,
+      "loss": 1.5664,
+      "step": 11218
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.102861035422344e-06,
+      "loss": 1.6229,
+      "step": 11219
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.100817438692097e-06,
+      "loss": 1.5203,
+      "step": 11220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.098773841961853e-06,
+      "loss": 1.7355,
+      "step": 11221
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.096730245231609e-06,
+      "loss": 1.7119,
+      "step": 11222
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.094686648501363e-06,
+      "loss": 1.7027,
+      "step": 11223
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.092643051771117e-06,
+      "loss": 1.6538,
+      "step": 11224
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.090599455040872e-06,
+      "loss": 1.713,
+      "step": 11225
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.088555858310628e-06,
+      "loss": 1.6025,
+      "step": 11226
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.086512261580382e-06,
+      "loss": 1.7226,
+      "step": 11227
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.084468664850136e-06,
+      "loss": 1.6356,
+      "step": 11228
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.082425068119891e-06,
+      "loss": 1.6406,
+      "step": 11229
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.080381471389647e-06,
+      "loss": 1.6196,
+      "step": 11230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.078337874659401e-06,
+      "loss": 1.6484,
+      "step": 11231
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.076294277929155e-06,
+      "loss": 1.783,
+      "step": 11232
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.07425068119891e-06,
+      "loss": 1.6467,
+      "step": 11233
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.072207084468664e-06,
+      "loss": 1.6193,
+      "step": 11234
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.07016348773842e-06,
+      "loss": 1.6273,
+      "step": 11235
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.068119891008174e-06,
+      "loss": 1.5407,
+      "step": 11236
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.06607629427793e-06,
+      "loss": 1.6716,
+      "step": 11237
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.064032697547684e-06,
+      "loss": 1.6278,
+      "step": 11238
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.06198910081744e-06,
+      "loss": 1.6113,
+      "step": 11239
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.059945504087195e-06,
+      "loss": 1.6173,
+      "step": 11240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.057901907356949e-06,
+      "loss": 1.7245,
+      "step": 11241
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.055858310626703e-06,
+      "loss": 1.7233,
+      "step": 11242
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.053814713896457e-06,
+      "loss": 1.6309,
+      "step": 11243
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.051771117166214e-06,
+      "loss": 1.6006,
+      "step": 11244
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.049727520435968e-06,
+      "loss": 1.6786,
+      "step": 11245
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.047683923705722e-06,
+      "loss": 1.6744,
+      "step": 11246
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.045640326975476e-06,
+      "loss": 1.6548,
+      "step": 11247
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.043596730245233e-06,
+      "loss": 1.6693,
+      "step": 11248
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.041553133514987e-06,
+      "loss": 1.6378,
+      "step": 11249
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.039509536784741e-06,
+      "loss": 1.7865,
+      "step": 11250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.037465940054495e-06,
+      "loss": 1.807,
+      "step": 11251
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.03542234332425e-06,
+      "loss": 1.6531,
+      "step": 11252
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.033378746594006e-06,
+      "loss": 1.613,
+      "step": 11253
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.03133514986376e-06,
+      "loss": 1.6743,
+      "step": 11254
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.029291553133516e-06,
+      "loss": 1.7296,
+      "step": 11255
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.02724795640327e-06,
+      "loss": 1.6331,
+      "step": 11256
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.025204359673025e-06,
+      "loss": 1.7071,
+      "step": 11257
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.02316076294278e-06,
+      "loss": 1.6567,
+      "step": 11258
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.021117166212535e-06,
+      "loss": 1.5825,
+      "step": 11259
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.019073569482289e-06,
+      "loss": 1.5694,
+      "step": 11260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.017029972752043e-06,
+      "loss": 1.7568,
+      "step": 11261
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.014986376021798e-06,
+      "loss": 1.7671,
+      "step": 11262
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.012942779291554e-06,
+      "loss": 1.7546,
+      "step": 11263
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.010899182561308e-06,
+      "loss": 1.6673,
+      "step": 11264
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.008855585831062e-06,
+      "loss": 1.6823,
+      "step": 11265
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.00681198910082e-06,
+      "loss": 1.6166,
+      "step": 11266
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.004768392370573e-06,
+      "loss": 1.62,
+      "step": 11267
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.002724795640327e-06,
+      "loss": 1.7001,
+      "step": 11268
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.000681198910081e-06,
+      "loss": 1.7134,
+      "step": 11269
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.998637602179837e-06,
+      "loss": 1.7666,
+      "step": 11270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.996594005449592e-06,
+      "loss": 1.6579,
+      "step": 11271
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.994550408719346e-06,
+      "loss": 1.7943,
+      "step": 11272
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.9925068119891e-06,
+      "loss": 1.7748,
+      "step": 11273
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.990463215258856e-06,
+      "loss": 1.6581,
+      "step": 11274
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.988419618528612e-06,
+      "loss": 1.6169,
+      "step": 11275
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.986376021798365e-06,
+      "loss": 1.6822,
+      "step": 11276
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.98433242506812e-06,
+      "loss": 1.6795,
+      "step": 11277
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.982288828337875e-06,
+      "loss": 1.7334,
+      "step": 11278
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.980245231607629e-06,
+      "loss": 1.6377,
+      "step": 11279
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.978201634877385e-06,
+      "loss": 1.7319,
+      "step": 11280
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.97615803814714e-06,
+      "loss": 1.6615,
+      "step": 11281
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.974114441416894e-06,
+      "loss": 1.5939,
+      "step": 11282
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.972070844686648e-06,
+      "loss": 1.6503,
+      "step": 11283
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.970027247956404e-06,
+      "loss": 1.8336,
+      "step": 11284
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.96798365122616e-06,
+      "loss": 1.6878,
+      "step": 11285
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.965940054495913e-06,
+      "loss": 1.6156,
+      "step": 11286
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.963896457765667e-06,
+      "loss": 1.6038,
+      "step": 11287
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.961852861035421e-06,
+      "loss": 1.7577,
+      "step": 11288
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.959809264305179e-06,
+      "loss": 1.6824,
+      "step": 11289
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.957765667574932e-06,
+      "loss": 1.6023,
+      "step": 11290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.955722070844686e-06,
+      "loss": 1.629,
+      "step": 11291
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.95367847411444e-06,
+      "loss": 1.6502,
+      "step": 11292
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.951634877384198e-06,
+      "loss": 1.5924,
+      "step": 11293
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.949591280653952e-06,
+      "loss": 1.5442,
+      "step": 11294
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.947547683923706e-06,
+      "loss": 1.6597,
+      "step": 11295
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.945504087193461e-06,
+      "loss": 1.641,
+      "step": 11296
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.943460490463215e-06,
+      "loss": 1.6403,
+      "step": 11297
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.94141689373297e-06,
+      "loss": 1.6539,
+      "step": 11298
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.939373297002725e-06,
+      "loss": 1.6718,
+      "step": 11299
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.93732970027248e-06,
+      "loss": 1.6638,
+      "step": 11300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.935286103542234e-06,
+      "loss": 1.7064,
+      "step": 11301
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.93324250681199e-06,
+      "loss": 1.7228,
+      "step": 11302
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.931198910081744e-06,
+      "loss": 1.6639,
+      "step": 11303
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.9291553133515e-06,
+      "loss": 1.6287,
+      "step": 11304
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.927111716621253e-06,
+      "loss": 1.7492,
+      "step": 11305
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.925068119891007e-06,
+      "loss": 1.6359,
+      "step": 11306
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.923024523160763e-06,
+      "loss": 1.6933,
+      "step": 11307
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.920980926430519e-06,
+      "loss": 1.6825,
+      "step": 11308
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.918937329700273e-06,
+      "loss": 1.6772,
+      "step": 11309
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.916893732970026e-06,
+      "loss": 1.5787,
+      "step": 11310
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.914850136239782e-06,
+      "loss": 1.6931,
+      "step": 11311
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.912806539509538e-06,
+      "loss": 1.5438,
+      "step": 11312
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.910762942779292e-06,
+      "loss": 1.6112,
+      "step": 11313
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.908719346049046e-06,
+      "loss": 1.7954,
+      "step": 11314
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.906675749318801e-06,
+      "loss": 1.9344,
+      "step": 11315
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.904632152588557e-06,
+      "loss": 1.6352,
+      "step": 11316
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.90258855585831e-06,
+      "loss": 1.7154,
+      "step": 11317
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.900544959128065e-06,
+      "loss": 1.7598,
+      "step": 11318
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.89850136239782e-06,
+      "loss": 1.7077,
+      "step": 11319
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.896457765667574e-06,
+      "loss": 1.7393,
+      "step": 11320
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.89441416893733e-06,
+      "loss": 1.6302,
+      "step": 11321
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.892370572207086e-06,
+      "loss": 1.5987,
+      "step": 11322
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.89032697547684e-06,
+      "loss": 1.6543,
+      "step": 11323
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.888283378746593e-06,
+      "loss": 1.6423,
+      "step": 11324
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.886239782016349e-06,
+      "loss": 1.5976,
+      "step": 11325
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.884196185286105e-06,
+      "loss": 1.7043,
+      "step": 11326
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.882152588555859e-06,
+      "loss": 1.6973,
+      "step": 11327
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.880108991825613e-06,
+      "loss": 1.6734,
+      "step": 11328
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.878065395095367e-06,
+      "loss": 1.7484,
+      "step": 11329
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.876021798365124e-06,
+      "loss": 1.7911,
+      "step": 11330
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.873978201634878e-06,
+      "loss": 1.6502,
+      "step": 11331
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.871934604904632e-06,
+      "loss": 1.8639,
+      "step": 11332
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.869891008174386e-06,
+      "loss": 1.5179,
+      "step": 11333
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.867847411444143e-06,
+      "loss": 1.5979,
+      "step": 11334
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.865803814713897e-06,
+      "loss": 1.7088,
+      "step": 11335
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.863760217983651e-06,
+      "loss": 1.8282,
+      "step": 11336
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.861716621253407e-06,
+      "loss": 1.5521,
+      "step": 11337
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.85967302452316e-06,
+      "loss": 1.7006,
+      "step": 11338
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.857629427792916e-06,
+      "loss": 1.5389,
+      "step": 11339
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.85558583106267e-06,
+      "loss": 1.5867,
+      "step": 11340
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.853542234332426e-06,
+      "loss": 1.5369,
+      "step": 11341
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.85149863760218e-06,
+      "loss": 1.831,
+      "step": 11342
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.849455040871935e-06,
+      "loss": 1.6319,
+      "step": 11343
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.84741144414169e-06,
+      "loss": 1.5778,
+      "step": 11344
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.845367847411445e-06,
+      "loss": 1.7229,
+      "step": 11345
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.843324250681199e-06,
+      "loss": 1.8037,
+      "step": 11346
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.841280653950953e-06,
+      "loss": 1.6563,
+      "step": 11347
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.839237057220708e-06,
+      "loss": 1.6602,
+      "step": 11348
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.837193460490464e-06,
+      "loss": 1.7664,
+      "step": 11349
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.835149863760218e-06,
+      "loss": 1.4961,
+      "step": 11350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.833106267029972e-06,
+      "loss": 1.8337,
+      "step": 11351
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.831062670299729e-06,
+      "loss": 1.6008,
+      "step": 11352
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.829019073569483e-06,
+      "loss": 1.7605,
+      "step": 11353
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.826975476839237e-06,
+      "loss": 1.6946,
+      "step": 11354
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.824931880108991e-06,
+      "loss": 1.7501,
+      "step": 11355
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.822888283378747e-06,
+      "loss": 1.693,
+      "step": 11356
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.820844686648502e-06,
+      "loss": 1.7452,
+      "step": 11357
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.818801089918256e-06,
+      "loss": 1.6972,
+      "step": 11358
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.81675749318801e-06,
+      "loss": 1.5684,
+      "step": 11359
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.814713896457766e-06,
+      "loss": 1.8736,
+      "step": 11360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.812670299727521e-06,
+      "loss": 1.6105,
+      "step": 11361
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.810626702997275e-06,
+      "loss": 1.6578,
+      "step": 11362
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.808583106267031e-06,
+      "loss": 1.6831,
+      "step": 11363
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.806539509536785e-06,
+      "loss": 1.6046,
+      "step": 11364
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.804495912806539e-06,
+      "loss": 1.8085,
+      "step": 11365
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.802452316076294e-06,
+      "loss": 1.6398,
+      "step": 11366
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.80040871934605e-06,
+      "loss": 1.6184,
+      "step": 11367
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.798365122615804e-06,
+      "loss": 1.7338,
+      "step": 11368
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.796321525885558e-06,
+      "loss": 1.6345,
+      "step": 11369
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.794277929155314e-06,
+      "loss": 1.6847,
+      "step": 11370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.79223433242507e-06,
+      "loss": 1.5698,
+      "step": 11371
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.790190735694823e-06,
+      "loss": 1.7151,
+      "step": 11372
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.788147138964577e-06,
+      "loss": 1.759,
+      "step": 11373
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.786103542234331e-06,
+      "loss": 1.7343,
+      "step": 11374
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.784059945504088e-06,
+      "loss": 1.6221,
+      "step": 11375
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.782016348773842e-06,
+      "loss": 1.7138,
+      "step": 11376
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.779972752043596e-06,
+      "loss": 1.674,
+      "step": 11377
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.777929155313352e-06,
+      "loss": 1.6802,
+      "step": 11378
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.775885558583108e-06,
+      "loss": 1.5739,
+      "step": 11379
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.773841961852861e-06,
+      "loss": 1.7288,
+      "step": 11380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.771798365122615e-06,
+      "loss": 1.9164,
+      "step": 11381
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.769754768392371e-06,
+      "loss": 1.6961,
+      "step": 11382
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.767711171662125e-06,
+      "loss": 1.6012,
+      "step": 11383
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.76566757493188e-06,
+      "loss": 1.5383,
+      "step": 11384
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.763623978201635e-06,
+      "loss": 1.7094,
+      "step": 11385
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.76158038147139e-06,
+      "loss": 1.5869,
+      "step": 11386
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.759536784741144e-06,
+      "loss": 1.5859,
+      "step": 11387
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.7574931880109e-06,
+      "loss": 1.713,
+      "step": 11388
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.755449591280654e-06,
+      "loss": 1.7576,
+      "step": 11389
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.75340599455041e-06,
+      "loss": 1.5876,
+      "step": 11390
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.751362397820163e-06,
+      "loss": 1.8237,
+      "step": 11391
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.749318801089917e-06,
+      "loss": 1.6006,
+      "step": 11392
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.747275204359675e-06,
+      "loss": 1.6614,
+      "step": 11393
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.745231607629428e-06,
+      "loss": 1.6939,
+      "step": 11394
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.743188010899182e-06,
+      "loss": 1.7257,
+      "step": 11395
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.741144414168936e-06,
+      "loss": 1.6548,
+      "step": 11396
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.739100817438694e-06,
+      "loss": 1.665,
+      "step": 11397
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.737057220708448e-06,
+      "loss": 1.5595,
+      "step": 11398
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.735013623978202e-06,
+      "loss": 1.6419,
+      "step": 11399
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.732970027247955e-06,
+      "loss": 1.7592,
+      "step": 11400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.730926430517711e-06,
+      "loss": 1.615,
+      "step": 11401
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.728882833787467e-06,
+      "loss": 1.7381,
+      "step": 11402
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.72683923705722e-06,
+      "loss": 1.5567,
+      "step": 11403
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.724795640326976e-06,
+      "loss": 1.6923,
+      "step": 11404
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.72275204359673e-06,
+      "loss": 1.645,
+      "step": 11405
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.720708446866486e-06,
+      "loss": 1.6423,
+      "step": 11406
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.71866485013624e-06,
+      "loss": 1.6114,
+      "step": 11407
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.716621253405995e-06,
+      "loss": 1.6488,
+      "step": 11408
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.71457765667575e-06,
+      "loss": 1.5836,
+      "step": 11409
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.712534059945503e-06,
+      "loss": 1.5868,
+      "step": 11410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.710490463215259e-06,
+      "loss": 1.6606,
+      "step": 11411
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.708446866485015e-06,
+      "loss": 1.6777,
+      "step": 11412
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.706403269754769e-06,
+      "loss": 1.5734,
+      "step": 11413
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.704359673024522e-06,
+      "loss": 1.5846,
+      "step": 11414
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.702316076294278e-06,
+      "loss": 1.6639,
+      "step": 11415
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.700272479564034e-06,
+      "loss": 1.7336,
+      "step": 11416
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.698228882833788e-06,
+      "loss": 1.673,
+      "step": 11417
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.696185286103542e-06,
+      "loss": 1.7591,
+      "step": 11418
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.694141689373297e-06,
+      "loss": 1.6102,
+      "step": 11419
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.692098092643053e-06,
+      "loss": 1.6704,
+      "step": 11420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.690054495912807e-06,
+      "loss": 1.6398,
+      "step": 11421
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.68801089918256e-06,
+      "loss": 1.6513,
+      "step": 11422
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.685967302452316e-06,
+      "loss": 1.5875,
+      "step": 11423
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.683923705722072e-06,
+      "loss": 1.6567,
+      "step": 11424
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.681880108991826e-06,
+      "loss": 1.5887,
+      "step": 11425
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.67983651226158e-06,
+      "loss": 1.7083,
+      "step": 11426
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.677792915531336e-06,
+      "loss": 1.7654,
+      "step": 11427
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.67574931880109e-06,
+      "loss": 1.6595,
+      "step": 11428
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.673705722070845e-06,
+      "loss": 1.5916,
+      "step": 11429
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.671662125340599e-06,
+      "loss": 1.629,
+      "step": 11430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.669618528610355e-06,
+      "loss": 1.6369,
+      "step": 11431
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.667574931880109e-06,
+      "loss": 1.7322,
+      "step": 11432
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.665531335149864e-06,
+      "loss": 1.6391,
+      "step": 11433
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.66348773841962e-06,
+      "loss": 1.6778,
+      "step": 11434
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.661444141689374e-06,
+      "loss": 1.6591,
+      "step": 11435
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.659400544959128e-06,
+      "loss": 1.5715,
+      "step": 11436
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.657356948228882e-06,
+      "loss": 1.6709,
+      "step": 11437
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.655313351498639e-06,
+      "loss": 1.8331,
+      "step": 11438
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.653269754768393e-06,
+      "loss": 1.6547,
+      "step": 11439
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.651226158038147e-06,
+      "loss": 1.6988,
+      "step": 11440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.6491825613079e-06,
+      "loss": 1.7187,
+      "step": 11441
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.647138964577658e-06,
+      "loss": 1.7026,
+      "step": 11442
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.645095367847412e-06,
+      "loss": 1.6892,
+      "step": 11443
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.643051771117166e-06,
+      "loss": 1.6098,
+      "step": 11444
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.64100817438692e-06,
+      "loss": 1.8005,
+      "step": 11445
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.638964577656676e-06,
+      "loss": 1.7033,
+      "step": 11446
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.636920980926431e-06,
+      "loss": 1.7128,
+      "step": 11447
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.634877384196185e-06,
+      "loss": 1.6076,
+      "step": 11448
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.63283378746594e-06,
+      "loss": 1.6709,
+      "step": 11449
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.630790190735695e-06,
+      "loss": 1.7147,
+      "step": 11450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.6287465940054495e-06,
+      "loss": 1.8408,
+      "step": 11451
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.626702997275204e-06,
+      "loss": 1.6846,
+      "step": 11452
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.62465940054496e-06,
+      "loss": 1.6976,
+      "step": 11453
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.622615803814714e-06,
+      "loss": 1.6061,
+      "step": 11454
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.620572207084469e-06,
+      "loss": 1.7208,
+      "step": 11455
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.618528610354223e-06,
+      "loss": 1.6333,
+      "step": 11456
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.616485013623979e-06,
+      "loss": 1.5621,
+      "step": 11457
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.614441416893733e-06,
+      "loss": 1.589,
+      "step": 11458
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.612397820163488e-06,
+      "loss": 1.7217,
+      "step": 11459
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.6103542234332435e-06,
+      "loss": 1.5647,
+      "step": 11460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.608310626702997e-06,
+      "loss": 1.6365,
+      "step": 11461
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.606267029972752e-06,
+      "loss": 1.6299,
+      "step": 11462
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.604223433242506e-06,
+      "loss": 1.6549,
+      "step": 11463
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.602179836512263e-06,
+      "loss": 1.7416,
+      "step": 11464
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.6001362397820165e-06,
+      "loss": 1.7784,
+      "step": 11465
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.598092643051771e-06,
+      "loss": 1.5667,
+      "step": 11466
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.596049046321525e-06,
+      "loss": 1.7272,
+      "step": 11467
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.594005449591282e-06,
+      "loss": 1.627,
+      "step": 11468
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.591961852861036e-06,
+      "loss": 1.8278,
+      "step": 11469
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.5899182561307905e-06,
+      "loss": 1.6011,
+      "step": 11470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.587874659400544e-06,
+      "loss": 1.6902,
+      "step": 11471
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.5858310626703e-06,
+      "loss": 1.5787,
+      "step": 11472
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.583787465940055e-06,
+      "loss": 1.5639,
+      "step": 11473
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.581743869209809e-06,
+      "loss": 1.7345,
+      "step": 11474
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.579700272479565e-06,
+      "loss": 1.6161,
+      "step": 11475
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.577656675749319e-06,
+      "loss": 1.7236,
+      "step": 11476
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.575613079019074e-06,
+      "loss": 1.6407,
+      "step": 11477
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.573569482288828e-06,
+      "loss": 1.728,
+      "step": 11478
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.5715258855585835e-06,
+      "loss": 1.6813,
+      "step": 11479
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.569482288828338e-06,
+      "loss": 1.7758,
+      "step": 11480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.567438692098092e-06,
+      "loss": 1.6861,
+      "step": 11481
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.565395095367847e-06,
+      "loss": 1.7384,
+      "step": 11482
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.563351498637603e-06,
+      "loss": 1.6035,
+      "step": 11483
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.5613079019073575e-06,
+      "loss": 1.6843,
+      "step": 11484
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.559264305177111e-06,
+      "loss": 1.7142,
+      "step": 11485
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.557220708446866e-06,
+      "loss": 1.6671,
+      "step": 11486
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.555177111716622e-06,
+      "loss": 1.6315,
+      "step": 11487
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.553133514986376e-06,
+      "loss": 1.6301,
+      "step": 11488
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.5510899182561305e-06,
+      "loss": 1.6719,
+      "step": 11489
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.549046321525886e-06,
+      "loss": 1.7225,
+      "step": 11490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.547002724795641e-06,
+      "loss": 1.6329,
+      "step": 11491
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.544959128065395e-06,
+      "loss": 1.5627,
+      "step": 11492
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.54291553133515e-06,
+      "loss": 1.6735,
+      "step": 11493
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.540871934604905e-06,
+      "loss": 1.7134,
+      "step": 11494
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.53882833787466e-06,
+      "loss": 1.6887,
+      "step": 11495
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.536784741144414e-06,
+      "loss": 1.5135,
+      "step": 11496
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.534741144414168e-06,
+      "loss": 1.5882,
+      "step": 11497
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.5326975476839245e-06,
+      "loss": 1.4734,
+      "step": 11498
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.530653950953678e-06,
+      "loss": 1.7008,
+      "step": 11499
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.528610354223433e-06,
+      "loss": 1.6992,
+      "step": 11500
+    },
+    {
+      "epoch": 2.27,
+      "eval_loss": 1.5107401609420776,
+      "eval_runtime": 720.5105,
+      "eval_samples_per_second": 5.76,
+      "eval_steps_per_second": 1.441,
+      "step": 11500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.526566757493189e-06,
+      "loss": 1.6525,
+      "step": 11501
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.524523160762944e-06,
+      "loss": 1.6431,
+      "step": 11502
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.5224795640326975e-06,
+      "loss": 1.5353,
+      "step": 11503
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.520435967302452e-06,
+      "loss": 1.6426,
+      "step": 11504
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.518392370572208e-06,
+      "loss": 1.6647,
+      "step": 11505
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.516348773841962e-06,
+      "loss": 1.7119,
+      "step": 11506
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.514305177111717e-06,
+      "loss": 1.5825,
+      "step": 11507
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.512261580381471e-06,
+      "loss": 1.7713,
+      "step": 11508
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.510217983651227e-06,
+      "loss": 1.6634,
+      "step": 11509
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.508174386920981e-06,
+      "loss": 1.6822,
+      "step": 11510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.506130790190736e-06,
+      "loss": 1.7169,
+      "step": 11511
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.50408719346049e-06,
+      "loss": 1.6276,
+      "step": 11512
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.502043596730246e-06,
+      "loss": 1.6648,
+      "step": 11513
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.5e-06,
+      "loss": 1.7048,
+      "step": 11514
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.497956403269755e-06,
+      "loss": 1.6186,
+      "step": 11515
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.49591280653951e-06,
+      "loss": 1.7344,
+      "step": 11516
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4938692098092645e-06,
+      "loss": 1.6523,
+      "step": 11517
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.491825613079019e-06,
+      "loss": 1.6736,
+      "step": 11518
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.489782016348774e-06,
+      "loss": 1.6411,
+      "step": 11519
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.487738419618529e-06,
+      "loss": 1.6485,
+      "step": 11520
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.485694822888284e-06,
+      "loss": 1.6036,
+      "step": 11521
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4836512261580385e-06,
+      "loss": 1.7625,
+      "step": 11522
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.481607629427793e-06,
+      "loss": 1.7258,
+      "step": 11523
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.479564032697547e-06,
+      "loss": 1.7682,
+      "step": 11524
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.477520435967303e-06,
+      "loss": 1.6721,
+      "step": 11525
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.475476839237057e-06,
+      "loss": 1.7736,
+      "step": 11526
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.473433242506812e-06,
+      "loss": 1.6722,
+      "step": 11527
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.471389645776567e-06,
+      "loss": 1.773,
+      "step": 11528
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.469346049046322e-06,
+      "loss": 1.6618,
+      "step": 11529
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.467302452316077e-06,
+      "loss": 1.6366,
+      "step": 11530
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4652588555858315e-06,
+      "loss": 1.6441,
+      "step": 11531
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.463215258855586e-06,
+      "loss": 1.687,
+      "step": 11532
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.46117166212534e-06,
+      "loss": 1.6275,
+      "step": 11533
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.459128065395096e-06,
+      "loss": 1.7398,
+      "step": 11534
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.45708446866485e-06,
+      "loss": 1.749,
+      "step": 11535
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4550408719346055e-06,
+      "loss": 1.6799,
+      "step": 11536
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.452997275204359e-06,
+      "loss": 1.741,
+      "step": 11537
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.450953678474115e-06,
+      "loss": 1.6748,
+      "step": 11538
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.448910081743869e-06,
+      "loss": 1.603,
+      "step": 11539
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.446866485013625e-06,
+      "loss": 1.7842,
+      "step": 11540
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4448228882833786e-06,
+      "loss": 1.6952,
+      "step": 11541
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.442779291553133e-06,
+      "loss": 1.7386,
+      "step": 11542
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.440735694822889e-06,
+      "loss": 1.527,
+      "step": 11543
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.438692098092643e-06,
+      "loss": 1.7374,
+      "step": 11544
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4366485013623985e-06,
+      "loss": 1.7088,
+      "step": 11545
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4346049046321525e-06,
+      "loss": 1.7169,
+      "step": 11546
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.432561307901908e-06,
+      "loss": 1.6469,
+      "step": 11547
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.430517711171662e-06,
+      "loss": 1.6346,
+      "step": 11548
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.428474114441418e-06,
+      "loss": 1.7763,
+      "step": 11549
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.426430517711172e-06,
+      "loss": 1.7122,
+      "step": 11550
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.424386920980926e-06,
+      "loss": 1.5968,
+      "step": 11551
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.422343324250681e-06,
+      "loss": 1.7023,
+      "step": 11552
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.420299727520436e-06,
+      "loss": 1.8548,
+      "step": 11553
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.418256130790191e-06,
+      "loss": 1.7168,
+      "step": 11554
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4162125340599456e-06,
+      "loss": 1.6369,
+      "step": 11555
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.414168937329701e-06,
+      "loss": 1.5737,
+      "step": 11556
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.412125340599455e-06,
+      "loss": 1.7907,
+      "step": 11557
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.41008174386921e-06,
+      "loss": 1.6233,
+      "step": 11558
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.408038147138965e-06,
+      "loss": 1.6726,
+      "step": 11559
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.4059945504087195e-06,
+      "loss": 1.5815,
+      "step": 11560
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.403950953678474e-06,
+      "loss": 1.5234,
+      "step": 11561
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.401907356948229e-06,
+      "loss": 1.7095,
+      "step": 11562
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.399863760217984e-06,
+      "loss": 1.5548,
+      "step": 11563
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.397820163487739e-06,
+      "loss": 1.6769,
+      "step": 11564
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.395776566757493e-06,
+      "loss": 1.6775,
+      "step": 11565
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.393732970027248e-06,
+      "loss": 1.537,
+      "step": 11566
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.391689373297003e-06,
+      "loss": 1.8234,
+      "step": 11567
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.389645776566758e-06,
+      "loss": 1.6718,
+      "step": 11568
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.387602179836512e-06,
+      "loss": 1.6885,
+      "step": 11569
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.385558583106267e-06,
+      "loss": 1.7636,
+      "step": 11570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.383514986376022e-06,
+      "loss": 1.5625,
+      "step": 11571
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.381471389645777e-06,
+      "loss": 1.6432,
+      "step": 11572
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.379427792915532e-06,
+      "loss": 1.7955,
+      "step": 11573
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.3773841961852865e-06,
+      "loss": 1.5705,
+      "step": 11574
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.375340599455041e-06,
+      "loss": 1.7436,
+      "step": 11575
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.373297002724795e-06,
+      "loss": 1.6785,
+      "step": 11576
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.371253405994551e-06,
+      "loss": 1.6493,
+      "step": 11577
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.369209809264305e-06,
+      "loss": 1.6629,
+      "step": 11578
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.36716621253406e-06,
+      "loss": 1.6373,
+      "step": 11579
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.365122615803814e-06,
+      "loss": 1.7,
+      "step": 11580
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.36307901907357e-06,
+      "loss": 1.71,
+      "step": 11581
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.361035422343324e-06,
+      "loss": 1.7657,
+      "step": 11582
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.3589918256130796e-06,
+      "loss": 1.6996,
+      "step": 11583
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.356948228882834e-06,
+      "loss": 1.8124,
+      "step": 11584
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.354904632152588e-06,
+      "loss": 1.7571,
+      "step": 11585
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.352861035422344e-06,
+      "loss": 1.5991,
+      "step": 11586
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.350817438692098e-06,
+      "loss": 1.6859,
+      "step": 11587
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.3487738419618535e-06,
+      "loss": 1.7598,
+      "step": 11588
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.346730245231607e-06,
+      "loss": 1.6497,
+      "step": 11589
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.344686648501363e-06,
+      "loss": 1.755,
+      "step": 11590
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.342643051771117e-06,
+      "loss": 1.7976,
+      "step": 11591
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.340599455040873e-06,
+      "loss": 1.7064,
+      "step": 11592
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.3385558583106266e-06,
+      "loss": 1.675,
+      "step": 11593
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.336512261580381e-06,
+      "loss": 1.5236,
+      "step": 11594
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.334468664850136e-06,
+      "loss": 1.6203,
+      "step": 11595
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.332425068119891e-06,
+      "loss": 1.6551,
+      "step": 11596
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.3303814713896466e-06,
+      "loss": 1.7635,
+      "step": 11597
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.3283378746594005e-06,
+      "loss": 1.6194,
+      "step": 11598
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.326294277929156e-06,
+      "loss": 1.5778,
+      "step": 11599
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.32425068119891e-06,
+      "loss": 1.6361,
+      "step": 11600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.322207084468666e-06,
+      "loss": 1.6413,
+      "step": 11601
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.32016348773842e-06,
+      "loss": 1.7859,
+      "step": 11602
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.318119891008174e-06,
+      "loss": 1.6903,
+      "step": 11603
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.316076294277929e-06,
+      "loss": 1.7164,
+      "step": 11604
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.314032697547684e-06,
+      "loss": 1.7544,
+      "step": 11605
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.311989100817439e-06,
+      "loss": 1.6566,
+      "step": 11606
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.3099455040871936e-06,
+      "loss": 1.6392,
+      "step": 11607
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.307901907356948e-06,
+      "loss": 1.6091,
+      "step": 11608
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.305858310626703e-06,
+      "loss": 1.5621,
+      "step": 11609
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.303814713896458e-06,
+      "loss": 1.646,
+      "step": 11610
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.301771117166213e-06,
+      "loss": 1.6227,
+      "step": 11611
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.2997275204359675e-06,
+      "loss": 1.6624,
+      "step": 11612
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.297683923705722e-06,
+      "loss": 1.4328,
+      "step": 11613
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.295640326975477e-06,
+      "loss": 1.6586,
+      "step": 11614
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.293596730245232e-06,
+      "loss": 1.6822,
+      "step": 11615
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.291553133514987e-06,
+      "loss": 1.7582,
+      "step": 11616
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.289509536784741e-06,
+      "loss": 1.5649,
+      "step": 11617
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.287465940054496e-06,
+      "loss": 1.7783,
+      "step": 11618
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.285422343324251e-06,
+      "loss": 1.5009,
+      "step": 11619
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.283378746594006e-06,
+      "loss": 1.7957,
+      "step": 11620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.28133514986376e-06,
+      "loss": 1.7086,
+      "step": 11621
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.279291553133515e-06,
+      "loss": 1.5881,
+      "step": 11622
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.277247956403269e-06,
+      "loss": 1.6303,
+      "step": 11623
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.275204359673025e-06,
+      "loss": 1.6283,
+      "step": 11624
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.27316076294278e-06,
+      "loss": 1.6979,
+      "step": 11625
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2711171662125345e-06,
+      "loss": 1.5912,
+      "step": 11626
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.269073569482289e-06,
+      "loss": 1.7076,
+      "step": 11627
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.267029972752044e-06,
+      "loss": 1.5813,
+      "step": 11628
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.264986376021799e-06,
+      "loss": 1.7022,
+      "step": 11629
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.262942779291553e-06,
+      "loss": 1.5808,
+      "step": 11630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.260899182561308e-06,
+      "loss": 1.626,
+      "step": 11631
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.258855585831062e-06,
+      "loss": 1.6623,
+      "step": 11632
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.256811989100818e-06,
+      "loss": 1.7249,
+      "step": 11633
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.254768392370572e-06,
+      "loss": 1.6735,
+      "step": 11634
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2527247956403276e-06,
+      "loss": 1.6692,
+      "step": 11635
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2506811989100815e-06,
+      "loss": 1.6247,
+      "step": 11636
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.248637602179837e-06,
+      "loss": 1.6232,
+      "step": 11637
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.246594005449591e-06,
+      "loss": 1.7019,
+      "step": 11638
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.244550408719346e-06,
+      "loss": 1.7522,
+      "step": 11639
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2425068119891015e-06,
+      "loss": 1.6503,
+      "step": 11640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2404632152588554e-06,
+      "loss": 1.6312,
+      "step": 11641
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.238419618528611e-06,
+      "loss": 1.6435,
+      "step": 11642
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.236376021798365e-06,
+      "loss": 1.5169,
+      "step": 11643
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.234332425068121e-06,
+      "loss": 1.6356,
+      "step": 11644
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2322888283378746e-06,
+      "loss": 1.8138,
+      "step": 11645
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.23024523160763e-06,
+      "loss": 1.5581,
+      "step": 11646
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.228201634877384e-06,
+      "loss": 1.7486,
+      "step": 11647
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.226158038147139e-06,
+      "loss": 1.7059,
+      "step": 11648
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.224114441416894e-06,
+      "loss": 1.6044,
+      "step": 11649
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2220708446866485e-06,
+      "loss": 1.704,
+      "step": 11650
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.220027247956403e-06,
+      "loss": 1.5511,
+      "step": 11651
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.217983651226158e-06,
+      "loss": 1.6049,
+      "step": 11652
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.215940054495914e-06,
+      "loss": 1.5369,
+      "step": 11653
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.213896457765668e-06,
+      "loss": 1.614,
+      "step": 11654
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.211852861035423e-06,
+      "loss": 1.7471,
+      "step": 11655
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.209809264305177e-06,
+      "loss": 1.7187,
+      "step": 11656
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.207765667574932e-06,
+      "loss": 1.6339,
+      "step": 11657
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.205722070844687e-06,
+      "loss": 1.7346,
+      "step": 11658
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.2036784741144416e-06,
+      "loss": 1.6312,
+      "step": 11659
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.201634877384196e-06,
+      "loss": 1.8266,
+      "step": 11660
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.199591280653951e-06,
+      "loss": 1.7191,
+      "step": 11661
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.197547683923706e-06,
+      "loss": 1.5647,
+      "step": 11662
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.195504087193461e-06,
+      "loss": 1.7053,
+      "step": 11663
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1934604904632155e-06,
+      "loss": 1.7542,
+      "step": 11664
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.19141689373297e-06,
+      "loss": 1.7104,
+      "step": 11665
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.189373297002725e-06,
+      "loss": 1.5227,
+      "step": 11666
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.18732970027248e-06,
+      "loss": 1.6847,
+      "step": 11667
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.185286103542235e-06,
+      "loss": 1.7669,
+      "step": 11668
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1832425068119894e-06,
+      "loss": 1.5791,
+      "step": 11669
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.181198910081744e-06,
+      "loss": 1.6119,
+      "step": 11670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.179155313351499e-06,
+      "loss": 1.5961,
+      "step": 11671
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.177111716621254e-06,
+      "loss": 1.7517,
+      "step": 11672
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.175068119891008e-06,
+      "loss": 1.6271,
+      "step": 11673
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.173024523160763e-06,
+      "loss": 1.5716,
+      "step": 11674
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.170980926430517e-06,
+      "loss": 1.7405,
+      "step": 11675
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.168937329700273e-06,
+      "loss": 1.5425,
+      "step": 11676
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.166893732970027e-06,
+      "loss": 1.4949,
+      "step": 11677
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1648501362397825e-06,
+      "loss": 1.758,
+      "step": 11678
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1628065395095364e-06,
+      "loss": 1.6792,
+      "step": 11679
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.160762942779292e-06,
+      "loss": 1.7503,
+      "step": 11680
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.158719346049047e-06,
+      "loss": 1.7088,
+      "step": 11681
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.156675749318801e-06,
+      "loss": 1.5748,
+      "step": 11682
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1546321525885564e-06,
+      "loss": 1.7114,
+      "step": 11683
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.15258855585831e-06,
+      "loss": 1.6186,
+      "step": 11684
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.150544959128066e-06,
+      "loss": 1.6714,
+      "step": 11685
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.14850136239782e-06,
+      "loss": 1.7295,
+      "step": 11686
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1464577656675756e-06,
+      "loss": 1.7532,
+      "step": 11687
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1444141689373295e-06,
+      "loss": 1.668,
+      "step": 11688
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.142370572207085e-06,
+      "loss": 1.6151,
+      "step": 11689
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.140326975476839e-06,
+      "loss": 1.7784,
+      "step": 11690
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.138283378746594e-06,
+      "loss": 1.6301,
+      "step": 11691
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.136239782016349e-06,
+      "loss": 1.5543,
+      "step": 11692
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1341961852861034e-06,
+      "loss": 1.5025,
+      "step": 11693
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.132152588555859e-06,
+      "loss": 1.6759,
+      "step": 11694
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.130108991825613e-06,
+      "loss": 1.6762,
+      "step": 11695
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.128065395095369e-06,
+      "loss": 1.6419,
+      "step": 11696
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1260217983651226e-06,
+      "loss": 1.5441,
+      "step": 11697
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.123978201634878e-06,
+      "loss": 1.5782,
+      "step": 11698
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.121934604904632e-06,
+      "loss": 1.6069,
+      "step": 11699
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.119891008174387e-06,
+      "loss": 1.7667,
+      "step": 11700
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.117847411444142e-06,
+      "loss": 1.8118,
+      "step": 11701
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1158038147138965e-06,
+      "loss": 1.7439,
+      "step": 11702
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.113760217983651e-06,
+      "loss": 1.5975,
+      "step": 11703
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.111716621253406e-06,
+      "loss": 1.6993,
+      "step": 11704
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.109673024523161e-06,
+      "loss": 1.7105,
+      "step": 11705
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.107629427792916e-06,
+      "loss": 1.6994,
+      "step": 11706
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.1055858310626704e-06,
+      "loss": 1.6468,
+      "step": 11707
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.103542234332425e-06,
+      "loss": 1.6229,
+      "step": 11708
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.10149863760218e-06,
+      "loss": 1.6227,
+      "step": 11709
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.099455040871935e-06,
+      "loss": 1.6218,
+      "step": 11710
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.0974114441416896e-06,
+      "loss": 1.7351,
+      "step": 11711
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.095367847411444e-06,
+      "loss": 1.7075,
+      "step": 11712
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.093324250681199e-06,
+      "loss": 1.5643,
+      "step": 11713
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.091280653950954e-06,
+      "loss": 1.6332,
+      "step": 11714
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.089237057220709e-06,
+      "loss": 1.6229,
+      "step": 11715
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0871934604904635e-06,
+      "loss": 1.6144,
+      "step": 11716
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.085149863760218e-06,
+      "loss": 1.7594,
+      "step": 11717
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.083106267029972e-06,
+      "loss": 1.7272,
+      "step": 11718
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.081062670299728e-06,
+      "loss": 1.6954,
+      "step": 11719
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.079019073569482e-06,
+      "loss": 1.8025,
+      "step": 11720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0769754768392374e-06,
+      "loss": 1.5778,
+      "step": 11721
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.074931880108992e-06,
+      "loss": 1.6788,
+      "step": 11722
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.072888283378747e-06,
+      "loss": 1.6701,
+      "step": 11723
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.070844686648502e-06,
+      "loss": 1.6551,
+      "step": 11724
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0688010899182566e-06,
+      "loss": 1.7802,
+      "step": 11725
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.066757493188011e-06,
+      "loss": 1.7538,
+      "step": 11726
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.064713896457765e-06,
+      "loss": 1.6697,
+      "step": 11727
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.062670299727521e-06,
+      "loss": 1.6998,
+      "step": 11728
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.060626702997275e-06,
+      "loss": 1.7746,
+      "step": 11729
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0585831062670305e-06,
+      "loss": 1.6932,
+      "step": 11730
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0565395095367844e-06,
+      "loss": 1.6403,
+      "step": 11731
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.05449591280654e-06,
+      "loss": 1.5056,
+      "step": 11732
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.052452316076294e-06,
+      "loss": 1.7627,
+      "step": 11733
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.05040871934605e-06,
+      "loss": 1.675,
+      "step": 11734
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0483651226158044e-06,
+      "loss": 1.6242,
+      "step": 11735
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.046321525885558e-06,
+      "loss": 1.7459,
+      "step": 11736
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.044277929155314e-06,
+      "loss": 1.765,
+      "step": 11737
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.042234332425068e-06,
+      "loss": 1.6686,
+      "step": 11738
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0401907356948236e-06,
+      "loss": 1.6986,
+      "step": 11739
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0381471389645775e-06,
+      "loss": 1.8124,
+      "step": 11740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.036103542234333e-06,
+      "loss": 1.7076,
+      "step": 11741
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.034059945504087e-06,
+      "loss": 1.6495,
+      "step": 11742
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.032016348773843e-06,
+      "loss": 1.5348,
+      "step": 11743
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.029972752043597e-06,
+      "loss": 1.533,
+      "step": 11744
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0279291553133514e-06,
+      "loss": 1.5787,
+      "step": 11745
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.025885558583106e-06,
+      "loss": 1.6446,
+      "step": 11746
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.023841961852861e-06,
+      "loss": 1.709,
+      "step": 11747
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.021798365122616e-06,
+      "loss": 1.5457,
+      "step": 11748
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.019754768392371e-06,
+      "loss": 1.7605,
+      "step": 11749
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.017711171662126e-06,
+      "loss": 1.9197,
+      "step": 11750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.01566757493188e-06,
+      "loss": 1.6177,
+      "step": 11751
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.013623978201636e-06,
+      "loss": 1.6623,
+      "step": 11752
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.01158038147139e-06,
+      "loss": 1.6353,
+      "step": 11753
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.0095367847411445e-06,
+      "loss": 1.6811,
+      "step": 11754
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.007493188010899e-06,
+      "loss": 1.7144,
+      "step": 11755
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.005449591280654e-06,
+      "loss": 1.6152,
+      "step": 11756
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.003405994550409e-06,
+      "loss": 1.5477,
+      "step": 11757
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.001362397820164e-06,
+      "loss": 1.6596,
+      "step": 11758
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.9993188010899184e-06,
+      "loss": 1.6569,
+      "step": 11759
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.997275204359673e-06,
+      "loss": 1.5943,
+      "step": 11760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.995231607629428e-06,
+      "loss": 1.7569,
+      "step": 11761
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.993188010899183e-06,
+      "loss": 1.6972,
+      "step": 11762
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.991144414168938e-06,
+      "loss": 1.5507,
+      "step": 11763
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.989100817438692e-06,
+      "loss": 1.8175,
+      "step": 11764
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.987057220708447e-06,
+      "loss": 1.5649,
+      "step": 11765
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.985013623978202e-06,
+      "loss": 1.6661,
+      "step": 11766
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.982970027247957e-06,
+      "loss": 1.6777,
+      "step": 11767
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.9809264305177115e-06,
+      "loss": 1.7949,
+      "step": 11768
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.978882833787466e-06,
+      "loss": 1.6454,
+      "step": 11769
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.97683923705722e-06,
+      "loss": 1.6459,
+      "step": 11770
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.974795640326976e-06,
+      "loss": 1.6298,
+      "step": 11771
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.97275204359673e-06,
+      "loss": 1.711,
+      "step": 11772
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.9707084468664854e-06,
+      "loss": 1.7004,
+      "step": 11773
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.968664850136239e-06,
+      "loss": 1.6745,
+      "step": 11774
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.966621253405995e-06,
+      "loss": 1.7255,
+      "step": 11775
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.964577656675749e-06,
+      "loss": 1.6886,
+      "step": 11776
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.962534059945505e-06,
+      "loss": 1.6926,
+      "step": 11777
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.960490463215259e-06,
+      "loss": 1.6911,
+      "step": 11778
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.958446866485013e-06,
+      "loss": 1.6053,
+      "step": 11779
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.956403269754769e-06,
+      "loss": 1.6474,
+      "step": 11780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.954359673024523e-06,
+      "loss": 1.6788,
+      "step": 11781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.9523160762942785e-06,
+      "loss": 1.6128,
+      "step": 11782
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.9502724795640324e-06,
+      "loss": 1.7223,
+      "step": 11783
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.948228882833788e-06,
+      "loss": 1.6275,
+      "step": 11784
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.946185286103542e-06,
+      "loss": 1.6164,
+      "step": 11785
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.944141689373298e-06,
+      "loss": 1.7771,
+      "step": 11786
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.942098092643052e-06,
+      "loss": 1.5736,
+      "step": 11787
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.940054495912806e-06,
+      "loss": 1.6293,
+      "step": 11788
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.938010899182561e-06,
+      "loss": 1.6686,
+      "step": 11789
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.935967302452316e-06,
+      "loss": 1.5995,
+      "step": 11790
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.933923705722072e-06,
+      "loss": 1.6454,
+      "step": 11791
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.9318801089918255e-06,
+      "loss": 1.7153,
+      "step": 11792
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.929836512261581e-06,
+      "loss": 1.5908,
+      "step": 11793
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.927792915531335e-06,
+      "loss": 1.5863,
+      "step": 11794
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.925749318801091e-06,
+      "loss": 1.765,
+      "step": 11795
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923705722070845e-06,
+      "loss": 1.5323,
+      "step": 11796
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.9216621253405994e-06,
+      "loss": 1.6609,
+      "step": 11797
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.919618528610354e-06,
+      "loss": 1.7223,
+      "step": 11798
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.917574931880109e-06,
+      "loss": 1.8312,
+      "step": 11799
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.915531335149864e-06,
+      "loss": 1.6798,
+      "step": 11800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.913487738419619e-06,
+      "loss": 1.6103,
+      "step": 11801
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.911444141689373e-06,
+      "loss": 1.6665,
+      "step": 11802
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.909400544959128e-06,
+      "loss": 1.6389,
+      "step": 11803
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.907356948228884e-06,
+      "loss": 1.7016,
+      "step": 11804
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.905313351498638e-06,
+      "loss": 1.6506,
+      "step": 11805
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.9032697547683925e-06,
+      "loss": 1.5956,
+      "step": 11806
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.901226158038147e-06,
+      "loss": 1.6528,
+      "step": 11807
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.899182561307902e-06,
+      "loss": 1.641,
+      "step": 11808
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.897138964577657e-06,
+      "loss": 1.6626,
+      "step": 11809
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.895095367847412e-06,
+      "loss": 1.6323,
+      "step": 11810
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.8930517711171664e-06,
+      "loss": 1.7,
+      "step": 11811
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.891008174386921e-06,
+      "loss": 1.7613,
+      "step": 11812
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.888964577656676e-06,
+      "loss": 1.6391,
+      "step": 11813
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.886920980926431e-06,
+      "loss": 1.6573,
+      "step": 11814
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.884877384196185e-06,
+      "loss": 1.6056,
+      "step": 11815
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.88283378746594e-06,
+      "loss": 1.7547,
+      "step": 11816
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.880790190735694e-06,
+      "loss": 1.6912,
+      "step": 11817
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.87874659400545e-06,
+      "loss": 1.6753,
+      "step": 11818
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.876702997275205e-06,
+      "loss": 1.7558,
+      "step": 11819
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.8746594005449595e-06,
+      "loss": 1.649,
+      "step": 11820
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.872615803814714e-06,
+      "loss": 1.8515,
+      "step": 11821
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.870572207084469e-06,
+      "loss": 1.6557,
+      "step": 11822
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.868528610354224e-06,
+      "loss": 1.6071,
+      "step": 11823
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.866485013623978e-06,
+      "loss": 1.7422,
+      "step": 11824
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.8644414168937334e-06,
+      "loss": 1.5795,
+      "step": 11825
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.862397820163487e-06,
+      "loss": 1.6792,
+      "step": 11826
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.860354223433243e-06,
+      "loss": 1.7465,
+      "step": 11827
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.858310626702997e-06,
+      "loss": 1.7411,
+      "step": 11828
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.856267029972753e-06,
+      "loss": 1.6,
+      "step": 11829
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.8542234332425065e-06,
+      "loss": 1.6146,
+      "step": 11830
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.852179836512262e-06,
+      "loss": 1.7735,
+      "step": 11831
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.850136239782017e-06,
+      "loss": 1.8479,
+      "step": 11832
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.848092643051771e-06,
+      "loss": 1.7266,
+      "step": 11833
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.8460490463215265e-06,
+      "loss": 1.6082,
+      "step": 11834
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.8440054495912805e-06,
+      "loss": 1.6542,
+      "step": 11835
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.841961852861036e-06,
+      "loss": 1.6995,
+      "step": 11836
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.83991825613079e-06,
+      "loss": 1.5754,
+      "step": 11837
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.837874659400546e-06,
+      "loss": 1.8383,
+      "step": 11838
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.8358310626703e-06,
+      "loss": 1.6195,
+      "step": 11839
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.833787465940055e-06,
+      "loss": 1.7288,
+      "step": 11840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.831743869209809e-06,
+      "loss": 1.7681,
+      "step": 11841
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.829700272479564e-06,
+      "loss": 1.6716,
+      "step": 11842
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.827656675749319e-06,
+      "loss": 1.8121,
+      "step": 11843
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.8256130790190735e-06,
+      "loss": 1.7277,
+      "step": 11844
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.823569482288828e-06,
+      "loss": 1.8025,
+      "step": 11845
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.821525885558583e-06,
+      "loss": 1.7754,
+      "step": 11846
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.819482288828339e-06,
+      "loss": 1.7203,
+      "step": 11847
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.817438692098093e-06,
+      "loss": 1.6427,
+      "step": 11848
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.815395095367848e-06,
+      "loss": 1.6817,
+      "step": 11849
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.813351498637602e-06,
+      "loss": 1.645,
+      "step": 11850
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.811307901907357e-06,
+      "loss": 1.6768,
+      "step": 11851
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.809264305177112e-06,
+      "loss": 1.9144,
+      "step": 11852
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.807220708446867e-06,
+      "loss": 1.7245,
+      "step": 11853
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.805177111716621e-06,
+      "loss": 1.6671,
+      "step": 11854
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.803133514986376e-06,
+      "loss": 1.7817,
+      "step": 11855
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.801089918256131e-06,
+      "loss": 1.6455,
+      "step": 11856
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.799046321525886e-06,
+      "loss": 1.6851,
+      "step": 11857
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.7970027247956405e-06,
+      "loss": 1.7152,
+      "step": 11858
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.794959128065395e-06,
+      "loss": 1.663,
+      "step": 11859
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.79291553133515e-06,
+      "loss": 1.7252,
+      "step": 11860
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.790871934604905e-06,
+      "loss": 1.7537,
+      "step": 11861
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.78882833787466e-06,
+      "loss": 1.7341,
+      "step": 11862
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.7867847411444145e-06,
+      "loss": 1.6375,
+      "step": 11863
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.784741144414169e-06,
+      "loss": 1.6658,
+      "step": 11864
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.782697547683924e-06,
+      "loss": 1.6021,
+      "step": 11865
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.780653950953679e-06,
+      "loss": 1.734,
+      "step": 11866
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.778610354223433e-06,
+      "loss": 1.7404,
+      "step": 11867
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.776566757493188e-06,
+      "loss": 1.6909,
+      "step": 11868
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.774523160762942e-06,
+      "loss": 1.6312,
+      "step": 11869
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.772479564032698e-06,
+      "loss": 1.6695,
+      "step": 11870
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.770435967302452e-06,
+      "loss": 1.6601,
+      "step": 11871
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.7683923705722075e-06,
+      "loss": 1.6383,
+      "step": 11872
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.766348773841962e-06,
+      "loss": 1.6111,
+      "step": 11873
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.764305177111717e-06,
+      "loss": 1.5067,
+      "step": 11874
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.762261580381472e-06,
+      "loss": 1.7482,
+      "step": 11875
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.760217983651226e-06,
+      "loss": 1.7268,
+      "step": 11876
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.7581743869209815e-06,
+      "loss": 1.65,
+      "step": 11877
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.756130790190735e-06,
+      "loss": 1.6976,
+      "step": 11878
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.754087193460491e-06,
+      "loss": 1.7086,
+      "step": 11879
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.752043596730245e-06,
+      "loss": 1.6181,
+      "step": 11880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.6798,
+      "step": 11881
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.7479564032697545e-06,
+      "loss": 1.7361,
+      "step": 11882
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.74591280653951e-06,
+      "loss": 1.6868,
+      "step": 11883
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.743869209809264e-06,
+      "loss": 1.7237,
+      "step": 11884
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.741825613079019e-06,
+      "loss": 1.6931,
+      "step": 11885
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.739782016348774e-06,
+      "loss": 1.652,
+      "step": 11886
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.7377384196185285e-06,
+      "loss": 1.6874,
+      "step": 11887
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.735694822888284e-06,
+      "loss": 1.7025,
+      "step": 11888
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.733651226158038e-06,
+      "loss": 1.625,
+      "step": 11889
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.731607629427794e-06,
+      "loss": 1.6536,
+      "step": 11890
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.729564032697548e-06,
+      "loss": 1.7112,
+      "step": 11891
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.727520435967303e-06,
+      "loss": 1.676,
+      "step": 11892
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.725476839237057e-06,
+      "loss": 1.621,
+      "step": 11893
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.723433242506812e-06,
+      "loss": 2.0529,
+      "step": 11894
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.721389645776567e-06,
+      "loss": 1.6857,
+      "step": 11895
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.7193460490463215e-06,
+      "loss": 1.7337,
+      "step": 11896
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.717302452316076e-06,
+      "loss": 1.7149,
+      "step": 11897
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.715258855585831e-06,
+      "loss": 1.6211,
+      "step": 11898
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.713215258855586e-06,
+      "loss": 1.6886,
+      "step": 11899
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.711171662125341e-06,
+      "loss": 1.7964,
+      "step": 11900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.709128065395096e-06,
+      "loss": 1.5865,
+      "step": 11901
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.70708446866485e-06,
+      "loss": 1.6386,
+      "step": 11902
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.705040871934605e-06,
+      "loss": 1.6791,
+      "step": 11903
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.70299727520436e-06,
+      "loss": 1.6195,
+      "step": 11904
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.700953678474115e-06,
+      "loss": 1.6824,
+      "step": 11905
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.698910081743869e-06,
+      "loss": 1.7855,
+      "step": 11906
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.696866485013624e-06,
+      "loss": 1.7979,
+      "step": 11907
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.694822888283379e-06,
+      "loss": 1.6415,
+      "step": 11908
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.692779291553134e-06,
+      "loss": 1.5067,
+      "step": 11909
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.6907356948228885e-06,
+      "loss": 1.6923,
+      "step": 11910
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.688692098092643e-06,
+      "loss": 1.5653,
+      "step": 11911
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.686648501362397e-06,
+      "loss": 1.7177,
+      "step": 11912
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.684604904632153e-06,
+      "loss": 1.708,
+      "step": 11913
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.682561307901907e-06,
+      "loss": 1.721,
+      "step": 11914
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.6805177111716625e-06,
+      "loss": 1.5731,
+      "step": 11915
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.678474114441417e-06,
+      "loss": 1.5664,
+      "step": 11916
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.676430517711172e-06,
+      "loss": 1.6788,
+      "step": 11917
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.674386920980927e-06,
+      "loss": 1.6831,
+      "step": 11918
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.672343324250682e-06,
+      "loss": 1.6065,
+      "step": 11919
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.670299727520436e-06,
+      "loss": 1.6519,
+      "step": 11920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.66825613079019e-06,
+      "loss": 1.6257,
+      "step": 11921
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.666212534059946e-06,
+      "loss": 1.6806,
+      "step": 11922
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6641689373297e-06,
+      "loss": 1.6583,
+      "step": 11923
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6621253405994555e-06,
+      "loss": 1.6212,
+      "step": 11924
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6600817438692095e-06,
+      "loss": 1.7069,
+      "step": 11925
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.658038147138965e-06,
+      "loss": 1.6493,
+      "step": 11926
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.655994550408719e-06,
+      "loss": 1.7236,
+      "step": 11927
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.653950953678475e-06,
+      "loss": 1.6367,
+      "step": 11928
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6519073569482295e-06,
+      "loss": 1.5052,
+      "step": 11929
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.649863760217983e-06,
+      "loss": 1.7575,
+      "step": 11930
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.647820163487739e-06,
+      "loss": 1.8488,
+      "step": 11931
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.645776566757493e-06,
+      "loss": 1.636,
+      "step": 11932
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.643732970027249e-06,
+      "loss": 1.6472,
+      "step": 11933
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6416893732970025e-06,
+      "loss": 1.6811,
+      "step": 11934
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.639645776566758e-06,
+      "loss": 1.5078,
+      "step": 11935
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.637602179836512e-06,
+      "loss": 1.6721,
+      "step": 11936
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.635558583106268e-06,
+      "loss": 1.6406,
+      "step": 11937
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.633514986376022e-06,
+      "loss": 1.6555,
+      "step": 11938
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6314713896457765e-06,
+      "loss": 1.6032,
+      "step": 11939
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.629427792915531e-06,
+      "loss": 1.6091,
+      "step": 11940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.627384196185286e-06,
+      "loss": 1.7749,
+      "step": 11941
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.625340599455042e-06,
+      "loss": 1.6531,
+      "step": 11942
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.623297002724796e-06,
+      "loss": 1.6296,
+      "step": 11943
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.621253405994551e-06,
+      "loss": 1.6185,
+      "step": 11944
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.619209809264305e-06,
+      "loss": 1.7038,
+      "step": 11945
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.617166212534061e-06,
+      "loss": 1.7301,
+      "step": 11946
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.615122615803815e-06,
+      "loss": 1.6589,
+      "step": 11947
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6130790190735695e-06,
+      "loss": 1.6055,
+      "step": 11948
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.611035422343324e-06,
+      "loss": 1.8704,
+      "step": 11949
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.608991825613079e-06,
+      "loss": 1.864,
+      "step": 11950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.606948228882834e-06,
+      "loss": 1.6738,
+      "step": 11951
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.604904632152589e-06,
+      "loss": 1.7331,
+      "step": 11952
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.6028610354223435e-06,
+      "loss": 1.7221,
+      "step": 11953
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.600817438692098e-06,
+      "loss": 1.7338,
+      "step": 11954
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.598773841961853e-06,
+      "loss": 1.6928,
+      "step": 11955
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.596730245231608e-06,
+      "loss": 1.678,
+      "step": 11956
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.594686648501363e-06,
+      "loss": 1.6797,
+      "step": 11957
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.592643051771117e-06,
+      "loss": 1.6239,
+      "step": 11958
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.590599455040872e-06,
+      "loss": 1.7,
+      "step": 11959
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.588555858310627e-06,
+      "loss": 1.7291,
+      "step": 11960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.586512261580382e-06,
+      "loss": 1.5889,
+      "step": 11961
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.5844686648501365e-06,
+      "loss": 1.6521,
+      "step": 11962
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.582425068119891e-06,
+      "loss": 1.6186,
+      "step": 11963
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.580381471389646e-06,
+      "loss": 1.6029,
+      "step": 11964
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.578337874659401e-06,
+      "loss": 1.6435,
+      "step": 11965
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.576294277929155e-06,
+      "loss": 1.6121,
+      "step": 11966
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.5742506811989105e-06,
+      "loss": 1.783,
+      "step": 11967
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.572207084468664e-06,
+      "loss": 1.6446,
+      "step": 11968
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.57016348773842e-06,
+      "loss": 1.6004,
+      "step": 11969
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.568119891008175e-06,
+      "loss": 1.498,
+      "step": 11970
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.56607629427793e-06,
+      "loss": 1.519,
+      "step": 11971
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.564032697547684e-06,
+      "loss": 1.5427,
+      "step": 11972
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.561989100817438e-06,
+      "loss": 1.6123,
+      "step": 11973
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.559945504087194e-06,
+      "loss": 1.6454,
+      "step": 11974
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.557901907356948e-06,
+      "loss": 1.571,
+      "step": 11975
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.5558583106267035e-06,
+      "loss": 1.6125,
+      "step": 11976
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.5538147138964575e-06,
+      "loss": 1.6147,
+      "step": 11977
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.551771117166213e-06,
+      "loss": 1.6696,
+      "step": 11978
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.549727520435967e-06,
+      "loss": 1.6732,
+      "step": 11979
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.547683923705723e-06,
+      "loss": 1.8333,
+      "step": 11980
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.545640326975477e-06,
+      "loss": 1.6504,
+      "step": 11981
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.543596730245231e-06,
+      "loss": 1.5672,
+      "step": 11982
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.541553133514986e-06,
+      "loss": 1.5366,
+      "step": 11983
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.539509536784741e-06,
+      "loss": 1.5343,
+      "step": 11984
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.537465940054497e-06,
+      "loss": 1.6841,
+      "step": 11985
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.5354223433242505e-06,
+      "loss": 1.6266,
+      "step": 11986
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.533378746594006e-06,
+      "loss": 1.6132,
+      "step": 11987
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.53133514986376e-06,
+      "loss": 1.599,
+      "step": 11988
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.529291553133516e-06,
+      "loss": 1.612,
+      "step": 11989
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.52724795640327e-06,
+      "loss": 1.6803,
+      "step": 11990
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.5252043596730245e-06,
+      "loss": 1.6146,
+      "step": 11991
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.523160762942779e-06,
+      "loss": 1.7871,
+      "step": 11992
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.521117166212534e-06,
+      "loss": 1.6006,
+      "step": 11993
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.519073569482289e-06,
+      "loss": 1.6746,
+      "step": 11994
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.517029972752044e-06,
+      "loss": 1.6431,
+      "step": 11995
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.514986376021798e-06,
+      "loss": 1.7247,
+      "step": 11996
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.512942779291553e-06,
+      "loss": 1.6265,
+      "step": 11997
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.510899182561309e-06,
+      "loss": 1.618,
+      "step": 11998
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.508855585831063e-06,
+      "loss": 1.6945,
+      "step": 11999
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.5068119891008175e-06,
+      "loss": 1.6707,
+      "step": 12000
+    },
+    {
+      "epoch": 2.37,
+      "eval_loss": 1.5096793174743652,
+      "eval_runtime": 720.7111,
+      "eval_samples_per_second": 5.758,
+      "eval_steps_per_second": 1.44,
+      "step": 12000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.504768392370572e-06,
+      "loss": 1.5796,
+      "step": 12001
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.502724795640327e-06,
+      "loss": 1.7044,
+      "step": 12002
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.500681198910082e-06,
+      "loss": 1.6998,
+      "step": 12003
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.498637602179837e-06,
+      "loss": 1.7181,
+      "step": 12004
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.4965940054495915e-06,
+      "loss": 1.7768,
+      "step": 12005
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.494550408719346e-06,
+      "loss": 1.7021,
+      "step": 12006
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.492506811989101e-06,
+      "loss": 1.6219,
+      "step": 12007
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.490463215258856e-06,
+      "loss": 1.6503,
+      "step": 12008
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.48841961852861e-06,
+      "loss": 1.6392,
+      "step": 12009
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.486376021798365e-06,
+      "loss": 1.6175,
+      "step": 12010
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.48433242506812e-06,
+      "loss": 1.5987,
+      "step": 12011
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.482288828337875e-06,
+      "loss": 1.7597,
+      "step": 12012
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.48024523160763e-06,
+      "loss": 1.614,
+      "step": 12013
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.4782016348773845e-06,
+      "loss": 1.6278,
+      "step": 12014
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.476158038147139e-06,
+      "loss": 1.6044,
+      "step": 12015
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.474114441416894e-06,
+      "loss": 1.6901,
+      "step": 12016
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.472070844686649e-06,
+      "loss": 1.7019,
+      "step": 12017
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.470027247956403e-06,
+      "loss": 1.7358,
+      "step": 12018
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.4679836512261585e-06,
+      "loss": 1.6644,
+      "step": 12019
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.465940054495912e-06,
+      "loss": 1.6561,
+      "step": 12020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.463896457765668e-06,
+      "loss": 1.6378,
+      "step": 12021
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.461852861035422e-06,
+      "loss": 1.7825,
+      "step": 12022
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.459809264305178e-06,
+      "loss": 1.6251,
+      "step": 12023
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.4577656675749316e-06,
+      "loss": 1.6366,
+      "step": 12024
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.455722070844687e-06,
+      "loss": 1.7456,
+      "step": 12025
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.453678474114442e-06,
+      "loss": 1.7203,
+      "step": 12026
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.451634877384196e-06,
+      "loss": 1.7211,
+      "step": 12027
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.4495912806539515e-06,
+      "loss": 1.7373,
+      "step": 12028
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.4475476839237055e-06,
+      "loss": 1.5951,
+      "step": 12029
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.445504087193461e-06,
+      "loss": 1.7935,
+      "step": 12030
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.443460490463215e-06,
+      "loss": 1.6197,
+      "step": 12031
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.441416893732971e-06,
+      "loss": 1.5362,
+      "step": 12032
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.439373297002725e-06,
+      "loss": 1.5492,
+      "step": 12033
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.43732970027248e-06,
+      "loss": 1.7593,
+      "step": 12034
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.435286103542234e-06,
+      "loss": 1.7527,
+      "step": 12035
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.433242506811989e-06,
+      "loss": 1.7481,
+      "step": 12036
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.431198910081744e-06,
+      "loss": 1.6174,
+      "step": 12037
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.4291553133514986e-06,
+      "loss": 1.6109,
+      "step": 12038
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.427111716621254e-06,
+      "loss": 1.699,
+      "step": 12039
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.425068119891008e-06,
+      "loss": 1.5697,
+      "step": 12040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.423024523160764e-06,
+      "loss": 1.894,
+      "step": 12041
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.420980926430518e-06,
+      "loss": 1.5937,
+      "step": 12042
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.418937329700273e-06,
+      "loss": 1.6187,
+      "step": 12043
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.416893732970027e-06,
+      "loss": 1.6456,
+      "step": 12044
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.414850136239782e-06,
+      "loss": 1.7091,
+      "step": 12045
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.412806539509537e-06,
+      "loss": 1.6249,
+      "step": 12046
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.410762942779292e-06,
+      "loss": 1.6794,
+      "step": 12047
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.408719346049046e-06,
+      "loss": 1.6483,
+      "step": 12048
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.406675749318801e-06,
+      "loss": 1.6454,
+      "step": 12049
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.404632152588556e-06,
+      "loss": 1.6136,
+      "step": 12050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.402588555858311e-06,
+      "loss": 1.8129,
+      "step": 12051
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.4005449591280656e-06,
+      "loss": 1.7528,
+      "step": 12052
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.39850136239782e-06,
+      "loss": 1.7259,
+      "step": 12053
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.396457765667575e-06,
+      "loss": 1.6824,
+      "step": 12054
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.39441416893733e-06,
+      "loss": 1.6178,
+      "step": 12055
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.392370572207085e-06,
+      "loss": 1.6256,
+      "step": 12056
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.3903269754768395e-06,
+      "loss": 1.7748,
+      "step": 12057
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.388283378746594e-06,
+      "loss": 1.6166,
+      "step": 12058
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.386239782016349e-06,
+      "loss": 1.7168,
+      "step": 12059
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.384196185286104e-06,
+      "loss": 1.5898,
+      "step": 12060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.382152588555859e-06,
+      "loss": 1.6728,
+      "step": 12061
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.380108991825613e-06,
+      "loss": 1.6045,
+      "step": 12062
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.378065395095367e-06,
+      "loss": 1.7623,
+      "step": 12063
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.376021798365123e-06,
+      "loss": 1.6341,
+      "step": 12064
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.373978201634877e-06,
+      "loss": 1.6358,
+      "step": 12065
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.3719346049046326e-06,
+      "loss": 1.6752,
+      "step": 12066
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.369891008174387e-06,
+      "loss": 1.6951,
+      "step": 12067
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.367847411444142e-06,
+      "loss": 1.6498,
+      "step": 12068
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.365803814713897e-06,
+      "loss": 1.6331,
+      "step": 12069
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.363760217983651e-06,
+      "loss": 1.6338,
+      "step": 12070
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.3617166212534065e-06,
+      "loss": 1.639,
+      "step": 12071
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.35967302452316e-06,
+      "loss": 1.6414,
+      "step": 12072
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.357629427792916e-06,
+      "loss": 1.5326,
+      "step": 12073
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.35558583106267e-06,
+      "loss": 1.7791,
+      "step": 12074
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.353542234332426e-06,
+      "loss": 1.6546,
+      "step": 12075
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.3514986376021796e-06,
+      "loss": 1.5534,
+      "step": 12076
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.349455040871935e-06,
+      "loss": 1.5954,
+      "step": 12077
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.347411444141689e-06,
+      "loss": 1.6311,
+      "step": 12078
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.345367847411444e-06,
+      "loss": 1.571,
+      "step": 12079
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.3433242506811996e-06,
+      "loss": 1.6278,
+      "step": 12080
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.3412806539509535e-06,
+      "loss": 1.5954,
+      "step": 12081
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.339237057220709e-06,
+      "loss": 1.7381,
+      "step": 12082
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.337193460490463e-06,
+      "loss": 1.6478,
+      "step": 12083
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.335149863760219e-06,
+      "loss": 1.6125,
+      "step": 12084
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.333106267029973e-06,
+      "loss": 1.6007,
+      "step": 12085
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.331062670299728e-06,
+      "loss": 1.6599,
+      "step": 12086
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.329019073569482e-06,
+      "loss": 1.7122,
+      "step": 12087
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.326975476839237e-06,
+      "loss": 1.5876,
+      "step": 12088
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.324931880108992e-06,
+      "loss": 1.7359,
+      "step": 12089
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.3228882833787466e-06,
+      "loss": 1.7427,
+      "step": 12090
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.320844686648501e-06,
+      "loss": 1.578,
+      "step": 12091
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.318801089918256e-06,
+      "loss": 1.8385,
+      "step": 12092
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.316757493188011e-06,
+      "loss": 1.6161,
+      "step": 12093
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.314713896457766e-06,
+      "loss": 1.6169,
+      "step": 12094
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.312670299727521e-06,
+      "loss": 1.7236,
+      "step": 12095
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.310626702997275e-06,
+      "loss": 1.6151,
+      "step": 12096
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.30858310626703e-06,
+      "loss": 1.5241,
+      "step": 12097
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.306539509536785e-06,
+      "loss": 1.6782,
+      "step": 12098
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.30449591280654e-06,
+      "loss": 1.6218,
+      "step": 12099
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.302452316076294e-06,
+      "loss": 1.6767,
+      "step": 12100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.300408719346049e-06,
+      "loss": 1.6589,
+      "step": 12101
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.298365122615804e-06,
+      "loss": 1.7172,
+      "step": 12102
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.296321525885559e-06,
+      "loss": 1.6807,
+      "step": 12103
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.2942779291553136e-06,
+      "loss": 1.7205,
+      "step": 12104
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.292234332425068e-06,
+      "loss": 1.5733,
+      "step": 12105
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.290190735694822e-06,
+      "loss": 1.6949,
+      "step": 12106
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.288147138964578e-06,
+      "loss": 1.6911,
+      "step": 12107
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.286103542234333e-06,
+      "loss": 1.626,
+      "step": 12108
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.2840599455040875e-06,
+      "loss": 1.5971,
+      "step": 12109
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.282016348773842e-06,
+      "loss": 1.6572,
+      "step": 12110
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.279972752043597e-06,
+      "loss": 1.6833,
+      "step": 12111
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.277929155313352e-06,
+      "loss": 1.65,
+      "step": 12112
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.275885558583107e-06,
+      "loss": 1.5038,
+      "step": 12113
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.273841961852861e-06,
+      "loss": 1.6407,
+      "step": 12114
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.271798365122615e-06,
+      "loss": 1.6734,
+      "step": 12115
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.269754768392371e-06,
+      "loss": 1.6269,
+      "step": 12116
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.267711171662125e-06,
+      "loss": 1.7607,
+      "step": 12117
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.2656675749318806e-06,
+      "loss": 1.8463,
+      "step": 12118
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2636239782016345e-06,
+      "loss": 1.7927,
+      "step": 12119
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.26158038147139e-06,
+      "loss": 1.7908,
+      "step": 12120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.259536784741144e-06,
+      "loss": 1.7737,
+      "step": 12121
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2574931880109e-06,
+      "loss": 1.6853,
+      "step": 12122
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2554495912806545e-06,
+      "loss": 1.8106,
+      "step": 12123
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2534059945504084e-06,
+      "loss": 1.4542,
+      "step": 12124
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.251362397820164e-06,
+      "loss": 1.7208,
+      "step": 12125
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.249318801089918e-06,
+      "loss": 1.6276,
+      "step": 12126
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.247275204359674e-06,
+      "loss": 1.6475,
+      "step": 12127
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2452316076294276e-06,
+      "loss": 1.5834,
+      "step": 12128
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.243188010899183e-06,
+      "loss": 1.7084,
+      "step": 12129
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.241144414168937e-06,
+      "loss": 1.6647,
+      "step": 12130
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.239100817438693e-06,
+      "loss": 1.6082,
+      "step": 12131
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.237057220708447e-06,
+      "loss": 1.6482,
+      "step": 12132
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2350136239782015e-06,
+      "loss": 1.7174,
+      "step": 12133
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.232970027247956e-06,
+      "loss": 1.6348,
+      "step": 12134
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.230926430517711e-06,
+      "loss": 1.661,
+      "step": 12135
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.228882833787467e-06,
+      "loss": 1.6246,
+      "step": 12136
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.226839237057221e-06,
+      "loss": 1.6176,
+      "step": 12137
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.224795640326976e-06,
+      "loss": 1.6352,
+      "step": 12138
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.22275204359673e-06,
+      "loss": 1.732,
+      "step": 12139
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.220708446866486e-06,
+      "loss": 1.5552,
+      "step": 12140
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.21866485013624e-06,
+      "loss": 1.5351,
+      "step": 12141
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2166212534059946e-06,
+      "loss": 1.7322,
+      "step": 12142
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.214577656675749e-06,
+      "loss": 1.7772,
+      "step": 12143
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.212534059945504e-06,
+      "loss": 1.6138,
+      "step": 12144
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.210490463215259e-06,
+      "loss": 1.6693,
+      "step": 12145
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.208446866485014e-06,
+      "loss": 1.5389,
+      "step": 12146
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.2064032697547685e-06,
+      "loss": 1.6339,
+      "step": 12147
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.204359673024523e-06,
+      "loss": 1.5859,
+      "step": 12148
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.202316076294279e-06,
+      "loss": 1.7001,
+      "step": 12149
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.200272479564033e-06,
+      "loss": 1.5679,
+      "step": 12150
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.198228882833788e-06,
+      "loss": 1.7712,
+      "step": 12151
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.1961852861035424e-06,
+      "loss": 1.6274,
+      "step": 12152
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.194141689373297e-06,
+      "loss": 1.642,
+      "step": 12153
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.192098092643052e-06,
+      "loss": 1.588,
+      "step": 12154
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.190054495912807e-06,
+      "loss": 1.5728,
+      "step": 12155
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.1880108991825616e-06,
+      "loss": 1.5961,
+      "step": 12156
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.185967302452316e-06,
+      "loss": 1.5884,
+      "step": 12157
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.183923705722071e-06,
+      "loss": 1.6269,
+      "step": 12158
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.181880108991826e-06,
+      "loss": 1.6328,
+      "step": 12159
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.17983651226158e-06,
+      "loss": 1.6373,
+      "step": 12160
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.1777929155313355e-06,
+      "loss": 1.7572,
+      "step": 12161
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.1757493188010894e-06,
+      "loss": 1.647,
+      "step": 12162
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.173705722070845e-06,
+      "loss": 1.7398,
+      "step": 12163
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.1716621253406e-06,
+      "loss": 1.6907,
+      "step": 12164
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.169618528610355e-06,
+      "loss": 1.7866,
+      "step": 12165
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.167574931880109e-06,
+      "loss": 1.7588,
+      "step": 12166
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.165531335149863e-06,
+      "loss": 1.6433,
+      "step": 12167
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.163487738419619e-06,
+      "loss": 1.606,
+      "step": 12168
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.161444141689373e-06,
+      "loss": 1.6022,
+      "step": 12169
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1594005449591286e-06,
+      "loss": 1.6883,
+      "step": 12170
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1573569482288825e-06,
+      "loss": 1.5411,
+      "step": 12171
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.155313351498638e-06,
+      "loss": 1.714,
+      "step": 12172
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.153269754768392e-06,
+      "loss": 1.6106,
+      "step": 12173
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.151226158038148e-06,
+      "loss": 1.6088,
+      "step": 12174
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.149182561307902e-06,
+      "loss": 1.7345,
+      "step": 12175
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1471389645776564e-06,
+      "loss": 1.6358,
+      "step": 12176
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.145095367847412e-06,
+      "loss": 1.798,
+      "step": 12177
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.143051771117166e-06,
+      "loss": 1.5542,
+      "step": 12178
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.141008174386922e-06,
+      "loss": 1.7035,
+      "step": 12179
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1389645776566756e-06,
+      "loss": 1.723,
+      "step": 12180
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.136920980926431e-06,
+      "loss": 1.7127,
+      "step": 12181
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.134877384196185e-06,
+      "loss": 1.5848,
+      "step": 12182
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.132833787465941e-06,
+      "loss": 1.6466,
+      "step": 12183
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.130790190735695e-06,
+      "loss": 1.6484,
+      "step": 12184
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1287465940054495e-06,
+      "loss": 1.7167,
+      "step": 12185
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.126702997275204e-06,
+      "loss": 1.6174,
+      "step": 12186
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.124659400544959e-06,
+      "loss": 1.5933,
+      "step": 12187
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.122615803814714e-06,
+      "loss": 1.8143,
+      "step": 12188
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.120572207084469e-06,
+      "loss": 1.6415,
+      "step": 12189
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1185286103542234e-06,
+      "loss": 1.6461,
+      "step": 12190
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.116485013623978e-06,
+      "loss": 1.6127,
+      "step": 12191
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.114441416893734e-06,
+      "loss": 1.6641,
+      "step": 12192
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.112397820163488e-06,
+      "loss": 1.7202,
+      "step": 12193
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1103542234332426e-06,
+      "loss": 1.8548,
+      "step": 12194
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.108310626702997e-06,
+      "loss": 1.831,
+      "step": 12195
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.106267029972752e-06,
+      "loss": 1.6792,
+      "step": 12196
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.104223433242507e-06,
+      "loss": 1.4993,
+      "step": 12197
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.104223433242507e-06,
+      "loss": 1.6275,
+      "step": 12198
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.102179836512262e-06,
+      "loss": 1.7012,
+      "step": 12199
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.1001362397820165e-06,
+      "loss": 1.6904,
+      "step": 12200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.098092643051771e-06,
+      "loss": 1.6951,
+      "step": 12201
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.096049046321526e-06,
+      "loss": 1.5778,
+      "step": 12202
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.094005449591281e-06,
+      "loss": 1.7267,
+      "step": 12203
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.091961852861035e-06,
+      "loss": 1.6453,
+      "step": 12204
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.0899182561307904e-06,
+      "loss": 1.5923,
+      "step": 12205
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.087874659400545e-06,
+      "loss": 1.5811,
+      "step": 12206
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.0858310626703e-06,
+      "loss": 1.7373,
+      "step": 12207
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.083787465940055e-06,
+      "loss": 1.785,
+      "step": 12208
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.0817438692098096e-06,
+      "loss": 1.6526,
+      "step": 12209
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.079700272479564e-06,
+      "loss": 1.5862,
+      "step": 12210
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.077656675749319e-06,
+      "loss": 1.7092,
+      "step": 12211
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.075613079019074e-06,
+      "loss": 1.6687,
+      "step": 12212
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.073569482288828e-06,
+      "loss": 1.6303,
+      "step": 12213
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.0715258855585835e-06,
+      "loss": 1.7387,
+      "step": 12214
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.0694822888283374e-06,
+      "loss": 1.7519,
+      "step": 12215
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.067438692098093e-06,
+      "loss": 1.8413,
+      "step": 12216
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.065395095367847e-06,
+      "loss": 1.6005,
+      "step": 12217
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.063351498637603e-06,
+      "loss": 1.5671,
+      "step": 12218
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.0613079019073574e-06,
+      "loss": 1.5586,
+      "step": 12219
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.059264305177112e-06,
+      "loss": 1.6812,
+      "step": 12220
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.057220708446867e-06,
+      "loss": 1.6549,
+      "step": 12221
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.055177111716621e-06,
+      "loss": 1.5632,
+      "step": 12222
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.0531335149863766e-06,
+      "loss": 1.53,
+      "step": 12223
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.0510899182561305e-06,
+      "loss": 1.6239,
+      "step": 12224
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.049046321525886e-06,
+      "loss": 1.652,
+      "step": 12225
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.04700272479564e-06,
+      "loss": 1.6373,
+      "step": 12226
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.044959128065396e-06,
+      "loss": 1.6633,
+      "step": 12227
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.04291553133515e-06,
+      "loss": 1.6963,
+      "step": 12228
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.040871934604905e-06,
+      "loss": 1.6668,
+      "step": 12229
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.038828337874659e-06,
+      "loss": 1.6769,
+      "step": 12230
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.036784741144414e-06,
+      "loss": 1.5858,
+      "step": 12231
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.034741144414169e-06,
+      "loss": 1.8557,
+      "step": 12232
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.032697547683924e-06,
+      "loss": 1.5476,
+      "step": 12233
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.030653950953679e-06,
+      "loss": 1.7797,
+      "step": 12234
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.028610354223433e-06,
+      "loss": 1.6074,
+      "step": 12235
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.026566757493189e-06,
+      "loss": 1.7421,
+      "step": 12236
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.024523160762943e-06,
+      "loss": 1.5979,
+      "step": 12237
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.022479564032698e-06,
+      "loss": 1.7386,
+      "step": 12238
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.020435967302452e-06,
+      "loss": 1.7173,
+      "step": 12239
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.018392370572207e-06,
+      "loss": 1.6262,
+      "step": 12240
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.016348773841962e-06,
+      "loss": 1.712,
+      "step": 12241
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.014305177111717e-06,
+      "loss": 1.4665,
+      "step": 12242
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.0122615803814714e-06,
+      "loss": 1.6823,
+      "step": 12243
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.010217983651226e-06,
+      "loss": 1.759,
+      "step": 12244
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.008174386920981e-06,
+      "loss": 1.55,
+      "step": 12245
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.006130790190736e-06,
+      "loss": 1.6429,
+      "step": 12246
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.0040871934604914e-06,
+      "loss": 1.4785,
+      "step": 12247
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.002043596730245e-06,
+      "loss": 1.676,
+      "step": 12248
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6e-06,
+      "loss": 1.6601,
+      "step": 12249
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.997956403269755e-06,
+      "loss": 1.6048,
+      "step": 12250
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.99591280653951e-06,
+      "loss": 1.6291,
+      "step": 12251
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.9938692098092645e-06,
+      "loss": 1.7096,
+      "step": 12252
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.991825613079019e-06,
+      "loss": 1.6268,
+      "step": 12253
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.989782016348774e-06,
+      "loss": 1.6155,
+      "step": 12254
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.987738419618529e-06,
+      "loss": 1.6661,
+      "step": 12255
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.985694822888284e-06,
+      "loss": 1.6538,
+      "step": 12256
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.9836512261580384e-06,
+      "loss": 1.614,
+      "step": 12257
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.981607629427792e-06,
+      "loss": 1.7976,
+      "step": 12258
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.979564032697548e-06,
+      "loss": 1.6837,
+      "step": 12259
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.977520435967302e-06,
+      "loss": 1.5736,
+      "step": 12260
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.975476839237058e-06,
+      "loss": 1.6254,
+      "step": 12261
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.973433242506812e-06,
+      "loss": 1.8434,
+      "step": 12262
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.971389645776567e-06,
+      "loss": 1.6435,
+      "step": 12263
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.969346049046322e-06,
+      "loss": 1.7133,
+      "step": 12264
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.967302452316076e-06,
+      "loss": 1.6851,
+      "step": 12265
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.9652588555858315e-06,
+      "loss": 1.6872,
+      "step": 12266
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.9632152588555854e-06,
+      "loss": 1.5433,
+      "step": 12267
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.961171662125341e-06,
+      "loss": 1.6784,
+      "step": 12268
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.959128065395095e-06,
+      "loss": 1.6155,
+      "step": 12269
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.957084468664851e-06,
+      "loss": 1.583,
+      "step": 12270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.955040871934605e-06,
+      "loss": 1.481,
+      "step": 12271
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.95299727520436e-06,
+      "loss": 1.7724,
+      "step": 12272
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.950953678474114e-06,
+      "loss": 1.6189,
+      "step": 12273
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.948910081743869e-06,
+      "loss": 1.6092,
+      "step": 12274
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.946866485013625e-06,
+      "loss": 1.5656,
+      "step": 12275
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.9448228882833785e-06,
+      "loss": 1.7578,
+      "step": 12276
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.942779291553134e-06,
+      "loss": 1.7354,
+      "step": 12277
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.940735694822888e-06,
+      "loss": 1.6252,
+      "step": 12278
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.938692098092644e-06,
+      "loss": 1.7296,
+      "step": 12279
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.936648501362398e-06,
+      "loss": 1.8046,
+      "step": 12280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.934604904632153e-06,
+      "loss": 1.7057,
+      "step": 12281
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.932561307901907e-06,
+      "loss": 1.6749,
+      "step": 12282
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.930517711171662e-06,
+      "loss": 1.6486,
+      "step": 12283
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.928474114441417e-06,
+      "loss": 1.8989,
+      "step": 12284
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.926430517711172e-06,
+      "loss": 1.7343,
+      "step": 12285
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.924386920980926e-06,
+      "loss": 1.5878,
+      "step": 12286
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.922343324250681e-06,
+      "loss": 1.6807,
+      "step": 12287
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.920299727520437e-06,
+      "loss": 1.5873,
+      "step": 12288
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.918256130790191e-06,
+      "loss": 1.6172,
+      "step": 12289
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.916212534059946e-06,
+      "loss": 1.5382,
+      "step": 12290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.9141689373297e-06,
+      "loss": 1.6692,
+      "step": 12291
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.912125340599455e-06,
+      "loss": 1.6857,
+      "step": 12292
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.91008174386921e-06,
+      "loss": 1.7562,
+      "step": 12293
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.908038147138965e-06,
+      "loss": 1.6354,
+      "step": 12294
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.9059945504087194e-06,
+      "loss": 1.5826,
+      "step": 12295
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.903950953678474e-06,
+      "loss": 1.6704,
+      "step": 12296
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.901907356948229e-06,
+      "loss": 1.6837,
+      "step": 12297
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.899863760217984e-06,
+      "loss": 1.6681,
+      "step": 12298
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.897820163487739e-06,
+      "loss": 1.7381,
+      "step": 12299
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.895776566757493e-06,
+      "loss": 1.7118,
+      "step": 12300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.893732970027247e-06,
+      "loss": 1.6243,
+      "step": 12301
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.891689373297003e-06,
+      "loss": 1.6915,
+      "step": 12302
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.889645776566758e-06,
+      "loss": 1.7463,
+      "step": 12303
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.8876021798365125e-06,
+      "loss": 1.652,
+      "step": 12304
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.885558583106267e-06,
+      "loss": 1.6871,
+      "step": 12305
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.883514986376022e-06,
+      "loss": 1.6862,
+      "step": 12306
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.881471389645777e-06,
+      "loss": 1.6283,
+      "step": 12307
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.879427792915532e-06,
+      "loss": 1.6408,
+      "step": 12308
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.8773841961852864e-06,
+      "loss": 1.6703,
+      "step": 12309
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.87534059945504e-06,
+      "loss": 1.6157,
+      "step": 12310
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.873297002724796e-06,
+      "loss": 1.8485,
+      "step": 12311
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.87125340599455e-06,
+      "loss": 1.7727,
+      "step": 12312
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.869209809264306e-06,
+      "loss": 1.6816,
+      "step": 12313
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.8671662125340595e-06,
+      "loss": 1.5589,
+      "step": 12314
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.865122615803815e-06,
+      "loss": 1.7091,
+      "step": 12315
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.86307901907357e-06,
+      "loss": 1.7266,
+      "step": 12316
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.861035422343325e-06,
+      "loss": 1.7701,
+      "step": 12317
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.8589918256130795e-06,
+      "loss": 1.5448,
+      "step": 12318
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.8569482288828335e-06,
+      "loss": 1.6412,
+      "step": 12319
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.854904632152589e-06,
+      "loss": 1.6795,
+      "step": 12320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.852861035422343e-06,
+      "loss": 1.7349,
+      "step": 12321
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.850817438692099e-06,
+      "loss": 1.5767,
+      "step": 12322
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.848773841961853e-06,
+      "loss": 1.5907,
+      "step": 12323
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.846730245231608e-06,
+      "loss": 1.5594,
+      "step": 12324
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.844686648501362e-06,
+      "loss": 1.7226,
+      "step": 12325
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.842643051771118e-06,
+      "loss": 1.7251,
+      "step": 12326
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.840599455040872e-06,
+      "loss": 1.5819,
+      "step": 12327
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.8385558583106265e-06,
+      "loss": 1.7649,
+      "step": 12328
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.836512261580381e-06,
+      "loss": 1.5361,
+      "step": 12329
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.834468664850136e-06,
+      "loss": 1.5803,
+      "step": 12330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.832425068119892e-06,
+      "loss": 1.6988,
+      "step": 12331
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.830381471389646e-06,
+      "loss": 1.7584,
+      "step": 12332
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.828337874659401e-06,
+      "loss": 1.5917,
+      "step": 12333
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.826294277929155e-06,
+      "loss": 1.6249,
+      "step": 12334
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.824250681198911e-06,
+      "loss": 1.7205,
+      "step": 12335
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.822207084468665e-06,
+      "loss": 1.7943,
+      "step": 12336
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.82016348773842e-06,
+      "loss": 1.7852,
+      "step": 12337
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.818119891008174e-06,
+      "loss": 1.6756,
+      "step": 12338
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.816076294277929e-06,
+      "loss": 1.6511,
+      "step": 12339
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.814032697547684e-06,
+      "loss": 1.6703,
+      "step": 12340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.811989100817439e-06,
+      "loss": 1.649,
+      "step": 12341
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.8099455040871935e-06,
+      "loss": 1.5579,
+      "step": 12342
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.807901907356948e-06,
+      "loss": 1.7366,
+      "step": 12343
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.805858310626704e-06,
+      "loss": 1.6581,
+      "step": 12344
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.803814713896458e-06,
+      "loss": 1.749,
+      "step": 12345
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.801771117166213e-06,
+      "loss": 1.5769,
+      "step": 12346
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7997275204359675e-06,
+      "loss": 1.6511,
+      "step": 12347
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.797683923705722e-06,
+      "loss": 1.7291,
+      "step": 12348
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.795640326975477e-06,
+      "loss": 1.5655,
+      "step": 12349
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.793596730245232e-06,
+      "loss": 1.5884,
+      "step": 12350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.791553133514987e-06,
+      "loss": 1.6636,
+      "step": 12351
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.789509536784741e-06,
+      "loss": 1.6985,
+      "step": 12352
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.787465940054496e-06,
+      "loss": 1.593,
+      "step": 12353
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.785422343324251e-06,
+      "loss": 1.7234,
+      "step": 12354
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.783378746594005e-06,
+      "loss": 1.7196,
+      "step": 12355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7813351498637605e-06,
+      "loss": 1.6225,
+      "step": 12356
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.779291553133515e-06,
+      "loss": 1.728,
+      "step": 12357
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.77724795640327e-06,
+      "loss": 1.6899,
+      "step": 12358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.775204359673025e-06,
+      "loss": 1.7157,
+      "step": 12359
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.77316076294278e-06,
+      "loss": 1.7636,
+      "step": 12360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7711171662125345e-06,
+      "loss": 1.6942,
+      "step": 12361
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.769073569482288e-06,
+      "loss": 1.7104,
+      "step": 12362
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.767029972752044e-06,
+      "loss": 1.6363,
+      "step": 12363
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.764986376021798e-06,
+      "loss": 1.6486,
+      "step": 12364
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.762942779291554e-06,
+      "loss": 1.6676,
+      "step": 12365
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7608991825613075e-06,
+      "loss": 1.7925,
+      "step": 12366
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.758855585831063e-06,
+      "loss": 1.6179,
+      "step": 12367
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.756811989100817e-06,
+      "loss": 1.6145,
+      "step": 12368
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.754768392370573e-06,
+      "loss": 1.703,
+      "step": 12369
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.752724795640327e-06,
+      "loss": 1.6854,
+      "step": 12370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7506811989100815e-06,
+      "loss": 1.5768,
+      "step": 12371
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.748637602179837e-06,
+      "loss": 1.6774,
+      "step": 12372
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.746594005449591e-06,
+      "loss": 1.6997,
+      "step": 12373
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.744550408719347e-06,
+      "loss": 1.6548,
+      "step": 12374
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.742506811989101e-06,
+      "loss": 1.5336,
+      "step": 12375
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.740463215258856e-06,
+      "loss": 1.7958,
+      "step": 12376
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.73841961852861e-06,
+      "loss": 1.6268,
+      "step": 12377
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.736376021798366e-06,
+      "loss": 1.5807,
+      "step": 12378
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.73433242506812e-06,
+      "loss": 1.6095,
+      "step": 12379
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7322888283378745e-06,
+      "loss": 1.6929,
+      "step": 12380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.730245231607629e-06,
+      "loss": 1.5511,
+      "step": 12381
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.728201634877384e-06,
+      "loss": 1.6809,
+      "step": 12382
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.726158038147139e-06,
+      "loss": 1.7199,
+      "step": 12383
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.724114441416894e-06,
+      "loss": 1.6279,
+      "step": 12384
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.722070844686649e-06,
+      "loss": 1.7484,
+      "step": 12385
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.720027247956403e-06,
+      "loss": 1.6026,
+      "step": 12386
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.717983651226159e-06,
+      "loss": 1.6087,
+      "step": 12387
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.715940054495913e-06,
+      "loss": 1.7573,
+      "step": 12388
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.713896457765668e-06,
+      "loss": 1.6038,
+      "step": 12389
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.711852861035422e-06,
+      "loss": 1.6057,
+      "step": 12390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.709809264305177e-06,
+      "loss": 1.6236,
+      "step": 12391
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.707765667574932e-06,
+      "loss": 1.6718,
+      "step": 12392
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.705722070844687e-06,
+      "loss": 1.6549,
+      "step": 12393
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7036784741144415e-06,
+      "loss": 1.7285,
+      "step": 12394
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.701634877384196e-06,
+      "loss": 1.6565,
+      "step": 12395
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.699591280653951e-06,
+      "loss": 1.7463,
+      "step": 12396
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.697547683923706e-06,
+      "loss": 1.7943,
+      "step": 12397
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.69550408719346e-06,
+      "loss": 1.6582,
+      "step": 12398
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.6934604904632155e-06,
+      "loss": 1.6728,
+      "step": 12399
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.69141689373297e-06,
+      "loss": 1.7861,
+      "step": 12400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.689373297002725e-06,
+      "loss": 1.6914,
+      "step": 12401
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.68732970027248e-06,
+      "loss": 1.7,
+      "step": 12402
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.685286103542235e-06,
+      "loss": 1.5896,
+      "step": 12403
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.683242506811989e-06,
+      "loss": 1.6581,
+      "step": 12404
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.681198910081744e-06,
+      "loss": 1.6911,
+      "step": 12405
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.679155313351499e-06,
+      "loss": 1.6905,
+      "step": 12406
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.677111716621253e-06,
+      "loss": 1.5865,
+      "step": 12407
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.6750681198910085e-06,
+      "loss": 1.7141,
+      "step": 12408
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.6730245231607625e-06,
+      "loss": 1.5947,
+      "step": 12409
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.670980926430518e-06,
+      "loss": 1.7544,
+      "step": 12410
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.668937329700272e-06,
+      "loss": 1.7078,
+      "step": 12411
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.666893732970028e-06,
+      "loss": 1.7008,
+      "step": 12412
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.6648501362397825e-06,
+      "loss": 1.6405,
+      "step": 12413
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.662806539509537e-06,
+      "loss": 1.7344,
+      "step": 12414
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.660762942779292e-06,
+      "loss": 1.8034,
+      "step": 12415
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.658719346049046e-06,
+      "loss": 1.6617,
+      "step": 12416
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.656675749318802e-06,
+      "loss": 1.8489,
+      "step": 12417
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.6546321525885555e-06,
+      "loss": 1.6187,
+      "step": 12418
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.652588555858311e-06,
+      "loss": 1.6031,
+      "step": 12419
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.650544959128065e-06,
+      "loss": 1.705,
+      "step": 12420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.648501362397821e-06,
+      "loss": 1.6899,
+      "step": 12421
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.646457765667575e-06,
+      "loss": 1.7148,
+      "step": 12422
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.64441416893733e-06,
+      "loss": 1.6401,
+      "step": 12423
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.642370572207084e-06,
+      "loss": 1.5761,
+      "step": 12424
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.640326975476839e-06,
+      "loss": 1.7084,
+      "step": 12425
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.638283378746595e-06,
+      "loss": 1.6644,
+      "step": 12426
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.636239782016349e-06,
+      "loss": 1.681,
+      "step": 12427
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.634196185286104e-06,
+      "loss": 1.6482,
+      "step": 12428
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.632152588555858e-06,
+      "loss": 1.6989,
+      "step": 12429
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.630108991825614e-06,
+      "loss": 1.6981,
+      "step": 12430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.628065395095368e-06,
+      "loss": 1.6805,
+      "step": 12431
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.626021798365123e-06,
+      "loss": 1.6568,
+      "step": 12432
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.623978201634877e-06,
+      "loss": 1.6542,
+      "step": 12433
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.621934604904632e-06,
+      "loss": 1.6667,
+      "step": 12434
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.619891008174387e-06,
+      "loss": 1.6292,
+      "step": 12435
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.617847411444142e-06,
+      "loss": 1.7097,
+      "step": 12436
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.6158038147138965e-06,
+      "loss": 1.8877,
+      "step": 12437
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.613760217983651e-06,
+      "loss": 1.6539,
+      "step": 12438
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.611716621253406e-06,
+      "loss": 1.6584,
+      "step": 12439
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.609673024523161e-06,
+      "loss": 1.5736,
+      "step": 12440
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.6076294277929165e-06,
+      "loss": 1.7222,
+      "step": 12441
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.60558583106267e-06,
+      "loss": 1.7613,
+      "step": 12442
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.603542234332425e-06,
+      "loss": 1.6324,
+      "step": 12443
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.60149863760218e-06,
+      "loss": 1.6163,
+      "step": 12444
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.599455040871935e-06,
+      "loss": 1.632,
+      "step": 12445
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.5974114441416895e-06,
+      "loss": 1.5593,
+      "step": 12446
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.595367847411444e-06,
+      "loss": 1.5977,
+      "step": 12447
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.593324250681199e-06,
+      "loss": 1.5784,
+      "step": 12448
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.591280653950954e-06,
+      "loss": 1.727,
+      "step": 12449
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.589237057220709e-06,
+      "loss": 1.6287,
+      "step": 12450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.5871934604904635e-06,
+      "loss": 1.752,
+      "step": 12451
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.585149863760217e-06,
+      "loss": 1.6373,
+      "step": 12452
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.583106267029973e-06,
+      "loss": 1.7346,
+      "step": 12453
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.581062670299728e-06,
+      "loss": 1.6438,
+      "step": 12454
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.579019073569483e-06,
+      "loss": 1.5905,
+      "step": 12455
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.576975476839237e-06,
+      "loss": 1.782,
+      "step": 12456
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.574931880108992e-06,
+      "loss": 1.6958,
+      "step": 12457
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.572888283378747e-06,
+      "loss": 1.6413,
+      "step": 12458
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.570844686648502e-06,
+      "loss": 1.8564,
+      "step": 12459
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.5688010899182565e-06,
+      "loss": 1.7546,
+      "step": 12460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.5667574931880105e-06,
+      "loss": 1.7324,
+      "step": 12461
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.564713896457766e-06,
+      "loss": 1.8726,
+      "step": 12462
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.56267029972752e-06,
+      "loss": 1.7109,
+      "step": 12463
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.560626702997276e-06,
+      "loss": 1.6914,
+      "step": 12464
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.55858310626703e-06,
+      "loss": 1.6542,
+      "step": 12465
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.556539509536785e-06,
+      "loss": 1.5932,
+      "step": 12466
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.55449591280654e-06,
+      "loss": 1.6538,
+      "step": 12467
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.552452316076294e-06,
+      "loss": 1.6196,
+      "step": 12468
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.55040871934605e-06,
+      "loss": 1.7945,
+      "step": 12469
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.5483651226158035e-06,
+      "loss": 1.7144,
+      "step": 12470
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.546321525885559e-06,
+      "loss": 1.6681,
+      "step": 12471
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.544277929155313e-06,
+      "loss": 1.6831,
+      "step": 12472
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.542234332425069e-06,
+      "loss": 1.6014,
+      "step": 12473
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.540190735694823e-06,
+      "loss": 1.6286,
+      "step": 12474
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.538147138964578e-06,
+      "loss": 1.5714,
+      "step": 12475
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.536103542234332e-06,
+      "loss": 1.6294,
+      "step": 12476
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.534059945504087e-06,
+      "loss": 1.6211,
+      "step": 12477
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.532016348773842e-06,
+      "loss": 1.6326,
+      "step": 12478
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.529972752043597e-06,
+      "loss": 1.7382,
+      "step": 12479
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.527929155313351e-06,
+      "loss": 1.5373,
+      "step": 12480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.525885558583106e-06,
+      "loss": 1.541,
+      "step": 12481
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.523841961852862e-06,
+      "loss": 1.6583,
+      "step": 12482
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.521798365122616e-06,
+      "loss": 1.7364,
+      "step": 12483
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.519754768392371e-06,
+      "loss": 1.5746,
+      "step": 12484
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.517711171662125e-06,
+      "loss": 1.703,
+      "step": 12485
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.51566757493188e-06,
+      "loss": 1.7109,
+      "step": 12486
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.513623978201635e-06,
+      "loss": 1.5907,
+      "step": 12487
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.51158038147139e-06,
+      "loss": 1.7636,
+      "step": 12488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.5095367847411445e-06,
+      "loss": 1.6755,
+      "step": 12489
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.507493188010899e-06,
+      "loss": 1.6652,
+      "step": 12490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.505449591280654e-06,
+      "loss": 1.504,
+      "step": 12491
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.503405994550409e-06,
+      "loss": 1.604,
+      "step": 12492
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.501362397820164e-06,
+      "loss": 1.6936,
+      "step": 12493
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.499318801089918e-06,
+      "loss": 1.5634,
+      "step": 12494
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.497275204359673e-06,
+      "loss": 1.6882,
+      "step": 12495
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.495231607629428e-06,
+      "loss": 1.7148,
+      "step": 12496
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.493188010899183e-06,
+      "loss": 1.6044,
+      "step": 12497
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4911444141689375e-06,
+      "loss": 1.7961,
+      "step": 12498
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.489100817438692e-06,
+      "loss": 1.6551,
+      "step": 12499
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.487057220708447e-06,
+      "loss": 1.6835,
+      "step": 12500
+    },
+    {
+      "epoch": 2.47,
+      "eval_loss": 1.5040361881256104,
+      "eval_runtime": 720.7763,
+      "eval_samples_per_second": 5.758,
+      "eval_steps_per_second": 1.44,
+      "step": 12500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.485013623978202e-06,
+      "loss": 1.6302,
+      "step": 12501
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.482970027247957e-06,
+      "loss": 1.6393,
+      "step": 12502
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4809264305177115e-06,
+      "loss": 1.5701,
+      "step": 12503
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.478882833787465e-06,
+      "loss": 1.6861,
+      "step": 12504
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.476839237057221e-06,
+      "loss": 1.6644,
+      "step": 12505
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.474795640326975e-06,
+      "loss": 1.819,
+      "step": 12506
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.472752043596731e-06,
+      "loss": 1.5991,
+      "step": 12507
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4707084468664846e-06,
+      "loss": 1.6569,
+      "step": 12508
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.46866485013624e-06,
+      "loss": 1.6023,
+      "step": 12509
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.466621253405995e-06,
+      "loss": 1.5676,
+      "step": 12510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.46457765667575e-06,
+      "loss": 1.7585,
+      "step": 12511
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4625340599455045e-06,
+      "loss": 1.6458,
+      "step": 12512
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4604904632152585e-06,
+      "loss": 1.741,
+      "step": 12513
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.458446866485014e-06,
+      "loss": 1.7024,
+      "step": 12514
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.456403269754768e-06,
+      "loss": 1.5878,
+      "step": 12515
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.454359673024524e-06,
+      "loss": 1.6432,
+      "step": 12516
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.452316076294278e-06,
+      "loss": 1.716,
+      "step": 12517
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.450272479564033e-06,
+      "loss": 1.5862,
+      "step": 12518
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.448228882833787e-06,
+      "loss": 1.5943,
+      "step": 12519
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.446185286103543e-06,
+      "loss": 1.727,
+      "step": 12520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.444141689373297e-06,
+      "loss": 1.735,
+      "step": 12521
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.4420980926430516e-06,
+      "loss": 1.6741,
+      "step": 12522
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.440054495912807e-06,
+      "loss": 1.5535,
+      "step": 12523
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.438010899182561e-06,
+      "loss": 1.7283,
+      "step": 12524
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.435967302452317e-06,
+      "loss": 1.7411,
+      "step": 12525
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.433923705722071e-06,
+      "loss": 1.6686,
+      "step": 12526
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.431880108991826e-06,
+      "loss": 1.6276,
+      "step": 12527
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.42983651226158e-06,
+      "loss": 1.6942,
+      "step": 12528
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.427792915531336e-06,
+      "loss": 1.765,
+      "step": 12529
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.42574931880109e-06,
+      "loss": 1.7728,
+      "step": 12530
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.423705722070845e-06,
+      "loss": 1.6664,
+      "step": 12531
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.421662125340599e-06,
+      "loss": 1.7669,
+      "step": 12532
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.419618528610354e-06,
+      "loss": 1.6888,
+      "step": 12533
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.417574931880109e-06,
+      "loss": 1.6635,
+      "step": 12534
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.415531335149864e-06,
+      "loss": 1.5602,
+      "step": 12535
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.413487738419619e-06,
+      "loss": 1.7156,
+      "step": 12536
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.411444141689373e-06,
+      "loss": 1.6604,
+      "step": 12537
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.409400544959129e-06,
+      "loss": 1.6993,
+      "step": 12538
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.407356948228883e-06,
+      "loss": 1.6649,
+      "step": 12539
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.405313351498638e-06,
+      "loss": 1.6766,
+      "step": 12540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.4032697547683925e-06,
+      "loss": 1.6437,
+      "step": 12541
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.401226158038147e-06,
+      "loss": 1.7218,
+      "step": 12542
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.399182561307902e-06,
+      "loss": 1.6179,
+      "step": 12543
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.397138964577657e-06,
+      "loss": 1.6823,
+      "step": 12544
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.395095367847412e-06,
+      "loss": 1.6339,
+      "step": 12545
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.393051771117166e-06,
+      "loss": 1.5675,
+      "step": 12546
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.391008174386921e-06,
+      "loss": 1.6174,
+      "step": 12547
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.388964577656676e-06,
+      "loss": 1.7311,
+      "step": 12548
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.38692098092643e-06,
+      "loss": 1.6609,
+      "step": 12549
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3848773841961856e-06,
+      "loss": 1.6784,
+      "step": 12550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.38283378746594e-06,
+      "loss": 1.6978,
+      "step": 12551
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.380790190735695e-06,
+      "loss": 1.822,
+      "step": 12552
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.37874659400545e-06,
+      "loss": 1.7888,
+      "step": 12553
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.376702997275205e-06,
+      "loss": 1.774,
+      "step": 12554
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3746594005449595e-06,
+      "loss": 1.5617,
+      "step": 12555
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.372615803814714e-06,
+      "loss": 1.701,
+      "step": 12556
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.370572207084469e-06,
+      "loss": 1.6611,
+      "step": 12557
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.368528610354223e-06,
+      "loss": 1.6771,
+      "step": 12558
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.366485013623979e-06,
+      "loss": 1.5618,
+      "step": 12559
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3644414168937326e-06,
+      "loss": 1.6985,
+      "step": 12560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.362397820163488e-06,
+      "loss": 1.577,
+      "step": 12561
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.360354223433242e-06,
+      "loss": 1.7438,
+      "step": 12562
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.358310626702998e-06,
+      "loss": 1.6288,
+      "step": 12563
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3562670299727525e-06,
+      "loss": 1.7014,
+      "step": 12564
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3542234332425065e-06,
+      "loss": 1.5978,
+      "step": 12565
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.352179836512262e-06,
+      "loss": 1.6803,
+      "step": 12566
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.350136239782016e-06,
+      "loss": 1.6555,
+      "step": 12567
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.348092643051772e-06,
+      "loss": 1.6458,
+      "step": 12568
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.346049046321526e-06,
+      "loss": 1.6618,
+      "step": 12569
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.344005449591281e-06,
+      "loss": 1.7891,
+      "step": 12570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.341961852861035e-06,
+      "loss": 1.6959,
+      "step": 12571
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.339918256130791e-06,
+      "loss": 1.7318,
+      "step": 12572
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.337874659400545e-06,
+      "loss": 1.7419,
+      "step": 12573
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3358310626702996e-06,
+      "loss": 1.7448,
+      "step": 12574
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.333787465940054e-06,
+      "loss": 1.5997,
+      "step": 12575
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.331743869209809e-06,
+      "loss": 1.6605,
+      "step": 12576
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.329700272479564e-06,
+      "loss": 1.6403,
+      "step": 12577
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.327656675749319e-06,
+      "loss": 1.7152,
+      "step": 12578
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.325613079019074e-06,
+      "loss": 1.7023,
+      "step": 12579
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.323569482288828e-06,
+      "loss": 1.6074,
+      "step": 12580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.321525885558584e-06,
+      "loss": 1.6497,
+      "step": 12581
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.319482288828338e-06,
+      "loss": 1.7687,
+      "step": 12582
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.317438692098093e-06,
+      "loss": 1.703,
+      "step": 12583
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.315395095367847e-06,
+      "loss": 1.7888,
+      "step": 12584
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.313351498637602e-06,
+      "loss": 1.6567,
+      "step": 12585
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.311307901907357e-06,
+      "loss": 1.5843,
+      "step": 12586
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.309264305177112e-06,
+      "loss": 1.8197,
+      "step": 12587
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.3072207084468666e-06,
+      "loss": 1.5985,
+      "step": 12588
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.305177111716621e-06,
+      "loss": 1.7294,
+      "step": 12589
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.303133514986376e-06,
+      "loss": 1.6631,
+      "step": 12590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.301089918256131e-06,
+      "loss": 1.469,
+      "step": 12591
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.299046321525886e-06,
+      "loss": 1.7118,
+      "step": 12592
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2970027247956405e-06,
+      "loss": 1.53,
+      "step": 12593
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.294959128065395e-06,
+      "loss": 1.6658,
+      "step": 12594
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.29291553133515e-06,
+      "loss": 1.662,
+      "step": 12595
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.290871934604905e-06,
+      "loss": 1.7005,
+      "step": 12596
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.28882833787466e-06,
+      "loss": 1.6239,
+      "step": 12597
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.286784741144414e-06,
+      "loss": 1.6248,
+      "step": 12598
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.284741144414169e-06,
+      "loss": 1.6261,
+      "step": 12599
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.282697547683924e-06,
+      "loss": 1.8775,
+      "step": 12600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.280653950953678e-06,
+      "loss": 1.8255,
+      "step": 12601
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2786103542234336e-06,
+      "loss": 1.6981,
+      "step": 12602
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2765667574931875e-06,
+      "loss": 1.6279,
+      "step": 12603
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.274523160762943e-06,
+      "loss": 1.6527,
+      "step": 12604
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.272479564032698e-06,
+      "loss": 1.66,
+      "step": 12605
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.270435967302453e-06,
+      "loss": 1.6728,
+      "step": 12606
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2683923705722075e-06,
+      "loss": 1.6715,
+      "step": 12607
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.266348773841962e-06,
+      "loss": 1.5725,
+      "step": 12608
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.264305177111717e-06,
+      "loss": 1.6953,
+      "step": 12609
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.262261580381471e-06,
+      "loss": 1.6035,
+      "step": 12610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.260217983651227e-06,
+      "loss": 1.7787,
+      "step": 12611
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2581743869209806e-06,
+      "loss": 1.6968,
+      "step": 12612
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.256130790190736e-06,
+      "loss": 1.9027,
+      "step": 12613
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.25408719346049e-06,
+      "loss": 1.5511,
+      "step": 12614
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.252043596730246e-06,
+      "loss": 1.6944,
+      "step": 12615
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.25e-06,
+      "loss": 1.6875,
+      "step": 12616
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.247956403269755e-06,
+      "loss": 1.5192,
+      "step": 12617
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.245912806539509e-06,
+      "loss": 1.7347,
+      "step": 12618
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.243869209809264e-06,
+      "loss": 1.6408,
+      "step": 12619
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.24182561307902e-06,
+      "loss": 1.7424,
+      "step": 12620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.239782016348774e-06,
+      "loss": 1.6184,
+      "step": 12621
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.237738419618529e-06,
+      "loss": 1.5898,
+      "step": 12622
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.235694822888283e-06,
+      "loss": 1.7465,
+      "step": 12623
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.233651226158039e-06,
+      "loss": 1.7024,
+      "step": 12624
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.231607629427793e-06,
+      "loss": 1.6704,
+      "step": 12625
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.229564032697548e-06,
+      "loss": 1.6065,
+      "step": 12626
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.227520435967302e-06,
+      "loss": 1.716,
+      "step": 12627
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.225476839237057e-06,
+      "loss": 1.6024,
+      "step": 12628
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.223433242506812e-06,
+      "loss": 1.6757,
+      "step": 12629
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.221389645776567e-06,
+      "loss": 1.6566,
+      "step": 12630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.2193460490463215e-06,
+      "loss": 1.6623,
+      "step": 12631
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.217302452316076e-06,
+      "loss": 1.7872,
+      "step": 12632
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.215258855585832e-06,
+      "loss": 1.6649,
+      "step": 12633
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.213215258855586e-06,
+      "loss": 1.6121,
+      "step": 12634
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.2111716621253415e-06,
+      "loss": 1.7035,
+      "step": 12635
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.209128065395095e-06,
+      "loss": 1.6542,
+      "step": 12636
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.20708446866485e-06,
+      "loss": 1.7082,
+      "step": 12637
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.205040871934605e-06,
+      "loss": 1.8173,
+      "step": 12638
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.20299727520436e-06,
+      "loss": 1.5687,
+      "step": 12639
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.2009536784741146e-06,
+      "loss": 1.5387,
+      "step": 12640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.198910081743869e-06,
+      "loss": 1.6659,
+      "step": 12641
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.196866485013624e-06,
+      "loss": 1.6217,
+      "step": 12642
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.194822888283379e-06,
+      "loss": 1.8004,
+      "step": 12643
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.192779291553134e-06,
+      "loss": 1.6191,
+      "step": 12644
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.1907356948228885e-06,
+      "loss": 1.7456,
+      "step": 12645
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.1886920980926424e-06,
+      "loss": 1.6873,
+      "step": 12646
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.186648501362398e-06,
+      "loss": 1.7172,
+      "step": 12647
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.184604904632153e-06,
+      "loss": 1.7792,
+      "step": 12648
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.182561307901908e-06,
+      "loss": 1.8127,
+      "step": 12649
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.180517711171662e-06,
+      "loss": 1.7437,
+      "step": 12650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.178474114441417e-06,
+      "loss": 1.6472,
+      "step": 12651
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.176430517711172e-06,
+      "loss": 1.6263,
+      "step": 12652
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.174386920980927e-06,
+      "loss": 1.5336,
+      "step": 12653
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.1723433242506816e-06,
+      "loss": 1.573,
+      "step": 12654
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.1702997275204355e-06,
+      "loss": 1.5812,
+      "step": 12655
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.168256130790191e-06,
+      "loss": 1.7413,
+      "step": 12656
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.166212534059945e-06,
+      "loss": 1.5226,
+      "step": 12657
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.164168937329701e-06,
+      "loss": 1.7582,
+      "step": 12658
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.162125340599455e-06,
+      "loss": 1.5662,
+      "step": 12659
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.16008174386921e-06,
+      "loss": 1.7846,
+      "step": 12660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.158038147138965e-06,
+      "loss": 1.5686,
+      "step": 12661
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.155994550408719e-06,
+      "loss": 1.6621,
+      "step": 12662
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.153950953678475e-06,
+      "loss": 1.6391,
+      "step": 12663
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.1519073569482286e-06,
+      "loss": 1.7373,
+      "step": 12664
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.149863760217984e-06,
+      "loss": 1.658,
+      "step": 12665
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.147820163487738e-06,
+      "loss": 1.6873,
+      "step": 12666
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.145776566757494e-06,
+      "loss": 1.7154,
+      "step": 12667
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.143732970027248e-06,
+      "loss": 1.7055,
+      "step": 12668
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.141689373297003e-06,
+      "loss": 1.7407,
+      "step": 12669
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.139645776566757e-06,
+      "loss": 1.7973,
+      "step": 12670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.137602179836512e-06,
+      "loss": 1.6421,
+      "step": 12671
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.135558583106267e-06,
+      "loss": 1.5222,
+      "step": 12672
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.133514986376022e-06,
+      "loss": 1.5919,
+      "step": 12673
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.131471389645777e-06,
+      "loss": 1.696,
+      "step": 12674
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.129427792915531e-06,
+      "loss": 1.6558,
+      "step": 12675
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.127384196185287e-06,
+      "loss": 1.6518,
+      "step": 12676
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.125340599455041e-06,
+      "loss": 1.5545,
+      "step": 12677
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.123297002724796e-06,
+      "loss": 1.7398,
+      "step": 12678
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.12125340599455e-06,
+      "loss": 1.7851,
+      "step": 12679
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.119209809264305e-06,
+      "loss": 1.663,
+      "step": 12680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.11716621253406e-06,
+      "loss": 1.5497,
+      "step": 12681
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.115122615803815e-06,
+      "loss": 1.5339,
+      "step": 12682
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.1130790190735695e-06,
+      "loss": 1.6692,
+      "step": 12683
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.111035422343324e-06,
+      "loss": 1.6134,
+      "step": 12684
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.108991825613079e-06,
+      "loss": 1.6163,
+      "step": 12685
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.106948228882834e-06,
+      "loss": 1.6115,
+      "step": 12686
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.104904632152589e-06,
+      "loss": 1.671,
+      "step": 12687
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.1028610354223434e-06,
+      "loss": 1.6592,
+      "step": 12688
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.100817438692098e-06,
+      "loss": 1.6802,
+      "step": 12689
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.098773841961853e-06,
+      "loss": 1.6872,
+      "step": 12690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.096730245231608e-06,
+      "loss": 1.5938,
+      "step": 12691
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0946866485013626e-06,
+      "loss": 1.6675,
+      "step": 12692
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.092643051771117e-06,
+      "loss": 1.646,
+      "step": 12693
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.090599455040872e-06,
+      "loss": 1.5789,
+      "step": 12694
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.088555858310627e-06,
+      "loss": 1.7149,
+      "step": 12695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.086512261580382e-06,
+      "loss": 1.7334,
+      "step": 12696
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0844686648501365e-06,
+      "loss": 1.6756,
+      "step": 12697
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0824250681198904e-06,
+      "loss": 1.6232,
+      "step": 12698
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.080381471389646e-06,
+      "loss": 1.7206,
+      "step": 12699
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0783378746594e-06,
+      "loss": 1.7389,
+      "step": 12700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.076294277929156e-06,
+      "loss": 1.5833,
+      "step": 12701
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0742506811989104e-06,
+      "loss": 1.5544,
+      "step": 12702
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.072207084468665e-06,
+      "loss": 1.7019,
+      "step": 12703
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.07016348773842e-06,
+      "loss": 1.5953,
+      "step": 12704
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.068119891008175e-06,
+      "loss": 1.6746,
+      "step": 12705
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0660762942779296e-06,
+      "loss": 1.4937,
+      "step": 12706
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0640326975476835e-06,
+      "loss": 1.6469,
+      "step": 12707
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.061989100817439e-06,
+      "loss": 1.5502,
+      "step": 12708
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.059945504087193e-06,
+      "loss": 1.6281,
+      "step": 12709
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.057901907356949e-06,
+      "loss": 1.6128,
+      "step": 12710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.055858310626703e-06,
+      "loss": 1.7047,
+      "step": 12711
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.053814713896458e-06,
+      "loss": 1.6496,
+      "step": 12712
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.051771117166212e-06,
+      "loss": 1.6188,
+      "step": 12713
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.049727520435968e-06,
+      "loss": 1.609,
+      "step": 12714
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.047683923705722e-06,
+      "loss": 1.5664,
+      "step": 12715
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.045640326975477e-06,
+      "loss": 1.7126,
+      "step": 12716
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.043596730245232e-06,
+      "loss": 1.615,
+      "step": 12717
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.041553133514986e-06,
+      "loss": 1.6556,
+      "step": 12718
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.039509536784742e-06,
+      "loss": 1.6362,
+      "step": 12719
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.037465940054496e-06,
+      "loss": 1.7177,
+      "step": 12720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.035422343324251e-06,
+      "loss": 1.7098,
+      "step": 12721
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.033378746594005e-06,
+      "loss": 1.6395,
+      "step": 12722
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.031335149863761e-06,
+      "loss": 1.6046,
+      "step": 12723
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.029291553133515e-06,
+      "loss": 1.731,
+      "step": 12724
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.02724795640327e-06,
+      "loss": 1.729,
+      "step": 12725
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0252043596730244e-06,
+      "loss": 1.7057,
+      "step": 12726
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.023160762942779e-06,
+      "loss": 1.6782,
+      "step": 12727
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.021117166212534e-06,
+      "loss": 1.6834,
+      "step": 12728
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.019073569482289e-06,
+      "loss": 1.6036,
+      "step": 12729
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.0170299727520444e-06,
+      "loss": 1.7099,
+      "step": 12730
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.014986376021798e-06,
+      "loss": 1.6106,
+      "step": 12731
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.012942779291554e-06,
+      "loss": 1.6734,
+      "step": 12732
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.010899182561308e-06,
+      "loss": 1.6439,
+      "step": 12733
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.008855585831063e-06,
+      "loss": 1.6093,
+      "step": 12734
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.0068119891008175e-06,
+      "loss": 1.6007,
+      "step": 12735
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.004768392370572e-06,
+      "loss": 1.6935,
+      "step": 12736
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.002724795640327e-06,
+      "loss": 1.742,
+      "step": 12737
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.000681198910082e-06,
+      "loss": 1.7357,
+      "step": 12738
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.998637602179837e-06,
+      "loss": 1.7656,
+      "step": 12739
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.9965940054495914e-06,
+      "loss": 1.8188,
+      "step": 12740
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.994550408719346e-06,
+      "loss": 1.5591,
+      "step": 12741
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.992506811989101e-06,
+      "loss": 1.7153,
+      "step": 12742
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.990463215258856e-06,
+      "loss": 1.6655,
+      "step": 12743
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.988419618528611e-06,
+      "loss": 1.6463,
+      "step": 12744
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.986376021798365e-06,
+      "loss": 1.5848,
+      "step": 12745
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.98433242506812e-06,
+      "loss": 1.5954,
+      "step": 12746
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.982288828337875e-06,
+      "loss": 1.5866,
+      "step": 12747
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.98024523160763e-06,
+      "loss": 1.5968,
+      "step": 12748
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.9782016348773845e-06,
+      "loss": 1.6391,
+      "step": 12749
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.976158038147139e-06,
+      "loss": 1.632,
+      "step": 12750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.974114441416894e-06,
+      "loss": 1.6501,
+      "step": 12751
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.972070844686648e-06,
+      "loss": 1.6495,
+      "step": 12752
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.970027247956404e-06,
+      "loss": 1.5897,
+      "step": 12753
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.967983651226158e-06,
+      "loss": 1.7644,
+      "step": 12754
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.965940054495913e-06,
+      "loss": 1.7349,
+      "step": 12755
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.963896457765667e-06,
+      "loss": 1.7175,
+      "step": 12756
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.961852861035423e-06,
+      "loss": 1.8723,
+      "step": 12757
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.959809264305178e-06,
+      "loss": 1.6229,
+      "step": 12758
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.9577656675749315e-06,
+      "loss": 1.5515,
+      "step": 12759
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.955722070844687e-06,
+      "loss": 1.6487,
+      "step": 12760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.953678474114441e-06,
+      "loss": 1.7237,
+      "step": 12761
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.951634877384197e-06,
+      "loss": 1.6038,
+      "step": 12762
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.949591280653951e-06,
+      "loss": 1.5428,
+      "step": 12763
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.947547683923706e-06,
+      "loss": 1.6553,
+      "step": 12764
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.94550408719346e-06,
+      "loss": 1.6852,
+      "step": 12765
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.943460490463216e-06,
+      "loss": 1.5926,
+      "step": 12766
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.94141689373297e-06,
+      "loss": 1.7042,
+      "step": 12767
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.939373297002725e-06,
+      "loss": 1.6659,
+      "step": 12768
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.937329700272479e-06,
+      "loss": 1.657,
+      "step": 12769
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.935286103542234e-06,
+      "loss": 1.6757,
+      "step": 12770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.93324250681199e-06,
+      "loss": 1.6061,
+      "step": 12771
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.931198910081744e-06,
+      "loss": 1.6773,
+      "step": 12772
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.929155313351499e-06,
+      "loss": 1.5407,
+      "step": 12773
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.927111716621253e-06,
+      "loss": 1.6518,
+      "step": 12774
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.925068119891009e-06,
+      "loss": 1.6848,
+      "step": 12775
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.923024523160763e-06,
+      "loss": 1.6895,
+      "step": 12776
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.920980926430518e-06,
+      "loss": 1.6009,
+      "step": 12777
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.9189373297002724e-06,
+      "loss": 1.6636,
+      "step": 12778
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.916893732970027e-06,
+      "loss": 1.5446,
+      "step": 12779
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.914850136239782e-06,
+      "loss": 1.6143,
+      "step": 12780
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.912806539509537e-06,
+      "loss": 1.6119,
+      "step": 12781
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.910762942779292e-06,
+      "loss": 1.6568,
+      "step": 12782
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.908719346049046e-06,
+      "loss": 1.5336,
+      "step": 12783
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.906675749318801e-06,
+      "loss": 1.7667,
+      "step": 12784
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.904632152588556e-06,
+      "loss": 1.7332,
+      "step": 12785
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.902588555858311e-06,
+      "loss": 1.6326,
+      "step": 12786
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.9005449591280655e-06,
+      "loss": 1.7016,
+      "step": 12787
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.89850136239782e-06,
+      "loss": 1.65,
+      "step": 12788
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.896457765667575e-06,
+      "loss": 1.6364,
+      "step": 12789
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.89441416893733e-06,
+      "loss": 1.6183,
+      "step": 12790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.892370572207085e-06,
+      "loss": 1.6294,
+      "step": 12791
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8903269754768394e-06,
+      "loss": 1.6616,
+      "step": 12792
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.888283378746594e-06,
+      "loss": 1.6179,
+      "step": 12793
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.886239782016349e-06,
+      "loss": 1.6579,
+      "step": 12794
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.884196185286103e-06,
+      "loss": 1.66,
+      "step": 12795
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.882152588555859e-06,
+      "loss": 1.7486,
+      "step": 12796
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8801089918256125e-06,
+      "loss": 1.7872,
+      "step": 12797
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.878065395095368e-06,
+      "loss": 1.7122,
+      "step": 12798
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.876021798365123e-06,
+      "loss": 1.6852,
+      "step": 12799
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.873978201634878e-06,
+      "loss": 1.6633,
+      "step": 12800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8719346049046325e-06,
+      "loss": 1.7973,
+      "step": 12801
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.869891008174387e-06,
+      "loss": 1.6324,
+      "step": 12802
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.867847411444142e-06,
+      "loss": 1.6167,
+      "step": 12803
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.865803814713896e-06,
+      "loss": 1.7543,
+      "step": 12804
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.863760217983652e-06,
+      "loss": 1.6539,
+      "step": 12805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.861716621253406e-06,
+      "loss": 1.6009,
+      "step": 12806
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.859673024523161e-06,
+      "loss": 1.6947,
+      "step": 12807
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.857629427792915e-06,
+      "loss": 1.6826,
+      "step": 12808
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.855585831062671e-06,
+      "loss": 1.6456,
+      "step": 12809
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.853542234332425e-06,
+      "loss": 1.5942,
+      "step": 12810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.85149863760218e-06,
+      "loss": 1.5742,
+      "step": 12811
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.849455040871935e-06,
+      "loss": 1.7204,
+      "step": 12812
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.847411444141689e-06,
+      "loss": 1.6262,
+      "step": 12813
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.845367847411445e-06,
+      "loss": 1.6124,
+      "step": 12814
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.843324250681199e-06,
+      "loss": 1.7135,
+      "step": 12815
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.841280653950954e-06,
+      "loss": 1.5562,
+      "step": 12816
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.839237057220708e-06,
+      "loss": 1.6947,
+      "step": 12817
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.837193460490464e-06,
+      "loss": 1.6004,
+      "step": 12818
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.835149863760218e-06,
+      "loss": 1.679,
+      "step": 12819
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8331062670299734e-06,
+      "loss": 1.6703,
+      "step": 12820
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.831062670299727e-06,
+      "loss": 1.5925,
+      "step": 12821
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.829019073569482e-06,
+      "loss": 1.7124,
+      "step": 12822
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.826975476839237e-06,
+      "loss": 1.5721,
+      "step": 12823
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.824931880108992e-06,
+      "loss": 1.6409,
+      "step": 12824
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8228882833787465e-06,
+      "loss": 1.7069,
+      "step": 12825
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.820844686648501e-06,
+      "loss": 1.6198,
+      "step": 12826
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.818801089918257e-06,
+      "loss": 1.7731,
+      "step": 12827
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.816757493188011e-06,
+      "loss": 1.5945,
+      "step": 12828
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8147138964577665e-06,
+      "loss": 1.6861,
+      "step": 12829
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8126702997275205e-06,
+      "loss": 1.6333,
+      "step": 12830
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.810626702997275e-06,
+      "loss": 1.6088,
+      "step": 12831
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.80858310626703e-06,
+      "loss": 1.7924,
+      "step": 12832
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.806539509536785e-06,
+      "loss": 1.6578,
+      "step": 12833
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.80449591280654e-06,
+      "loss": 1.7399,
+      "step": 12834
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.802452316076294e-06,
+      "loss": 1.7409,
+      "step": 12835
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.800408719346049e-06,
+      "loss": 1.6473,
+      "step": 12836
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.798365122615804e-06,
+      "loss": 1.6218,
+      "step": 12837
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.796321525885559e-06,
+      "loss": 1.6856,
+      "step": 12838
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.7942779291553135e-06,
+      "loss": 1.6516,
+      "step": 12839
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.792234332425068e-06,
+      "loss": 1.6347,
+      "step": 12840
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.790190735694823e-06,
+      "loss": 1.6976,
+      "step": 12841
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.788147138964578e-06,
+      "loss": 1.6821,
+      "step": 12842
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.786103542234333e-06,
+      "loss": 1.6907,
+      "step": 12843
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.7840599455040874e-06,
+      "loss": 1.5614,
+      "step": 12844
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.782016348773842e-06,
+      "loss": 1.7012,
+      "step": 12845
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.779972752043597e-06,
+      "loss": 1.6186,
+      "step": 12846
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.777929155313352e-06,
+      "loss": 1.6954,
+      "step": 12847
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.775885558583107e-06,
+      "loss": 1.6256,
+      "step": 12848
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.7738419618528605e-06,
+      "loss": 1.6355,
+      "step": 12849
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.771798365122616e-06,
+      "loss": 1.6337,
+      "step": 12850
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.76975476839237e-06,
+      "loss": 1.6218,
+      "step": 12851
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.767711171662126e-06,
+      "loss": 1.5838,
+      "step": 12852
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.76566757493188e-06,
+      "loss": 1.6836,
+      "step": 12853
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.763623978201635e-06,
+      "loss": 1.7085,
+      "step": 12854
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.76158038147139e-06,
+      "loss": 1.7004,
+      "step": 12855
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.759536784741144e-06,
+      "loss": 1.6051,
+      "step": 12856
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.7574931880109e-06,
+      "loss": 1.6218,
+      "step": 12857
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.755449591280654e-06,
+      "loss": 1.7016,
+      "step": 12858
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.753405994550409e-06,
+      "loss": 1.5658,
+      "step": 12859
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.751362397820163e-06,
+      "loss": 1.6421,
+      "step": 12860
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.749318801089919e-06,
+      "loss": 1.7098,
+      "step": 12861
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.747275204359673e-06,
+      "loss": 1.7094,
+      "step": 12862
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.745231607629428e-06,
+      "loss": 1.6249,
+      "step": 12863
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.743188010899182e-06,
+      "loss": 1.6306,
+      "step": 12864
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.741144414168937e-06,
+      "loss": 1.7094,
+      "step": 12865
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.739100817438692e-06,
+      "loss": 1.6347,
+      "step": 12866
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.737057220708447e-06,
+      "loss": 1.577,
+      "step": 12867
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.735013623978202e-06,
+      "loss": 1.5985,
+      "step": 12868
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.732970027247956e-06,
+      "loss": 1.7584,
+      "step": 12869
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.730926430517712e-06,
+      "loss": 1.5779,
+      "step": 12870
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.728882833787466e-06,
+      "loss": 1.6328,
+      "step": 12871
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.7268392370572214e-06,
+      "loss": 1.6212,
+      "step": 12872
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.724795640326975e-06,
+      "loss": 1.6207,
+      "step": 12873
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.72275204359673e-06,
+      "loss": 1.6917,
+      "step": 12874
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.720708446866485e-06,
+      "loss": 1.7972,
+      "step": 12875
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.71866485013624e-06,
+      "loss": 1.7136,
+      "step": 12876
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.7166212534059945e-06,
+      "loss": 1.6395,
+      "step": 12877
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.714577656675749e-06,
+      "loss": 1.6105,
+      "step": 12878
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.712534059945504e-06,
+      "loss": 1.6269,
+      "step": 12879
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.710490463215259e-06,
+      "loss": 1.6995,
+      "step": 12880
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.7084468664850145e-06,
+      "loss": 1.7551,
+      "step": 12881
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.7064032697547685e-06,
+      "loss": 1.7962,
+      "step": 12882
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.704359673024523e-06,
+      "loss": 1.6018,
+      "step": 12883
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.702316076294278e-06,
+      "loss": 1.5903,
+      "step": 12884
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.700272479564033e-06,
+      "loss": 1.608,
+      "step": 12885
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.698228882833788e-06,
+      "loss": 1.6869,
+      "step": 12886
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.696185286103542e-06,
+      "loss": 1.5404,
+      "step": 12887
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.694141689373297e-06,
+      "loss": 1.517,
+      "step": 12888
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.692098092643052e-06,
+      "loss": 1.5449,
+      "step": 12889
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.690054495912807e-06,
+      "loss": 1.6563,
+      "step": 12890
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6880108991825615e-06,
+      "loss": 1.6476,
+      "step": 12891
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6859673024523155e-06,
+      "loss": 1.5578,
+      "step": 12892
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.683923705722071e-06,
+      "loss": 1.6608,
+      "step": 12893
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.681880108991825e-06,
+      "loss": 1.7252,
+      "step": 12894
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.679836512261581e-06,
+      "loss": 1.648,
+      "step": 12895
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6777929155313355e-06,
+      "loss": 1.7077,
+      "step": 12896
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.67574931880109e-06,
+      "loss": 1.6899,
+      "step": 12897
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.673705722070845e-06,
+      "loss": 1.7328,
+      "step": 12898
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6716621253406e-06,
+      "loss": 1.6653,
+      "step": 12899
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.669618528610355e-06,
+      "loss": 1.7134,
+      "step": 12900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6675749318801085e-06,
+      "loss": 1.7203,
+      "step": 12901
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.665531335149864e-06,
+      "loss": 1.7171,
+      "step": 12902
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.663487738419618e-06,
+      "loss": 1.5585,
+      "step": 12903
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.661444141689374e-06,
+      "loss": 1.6921,
+      "step": 12904
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.659400544959128e-06,
+      "loss": 1.6472,
+      "step": 12905
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.657356948228883e-06,
+      "loss": 1.6633,
+      "step": 12906
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.655313351498637e-06,
+      "loss": 1.5969,
+      "step": 12907
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.653269754768393e-06,
+      "loss": 1.7797,
+      "step": 12908
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.651226158038148e-06,
+      "loss": 1.6091,
+      "step": 12909
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.649182561307902e-06,
+      "loss": 1.6311,
+      "step": 12910
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.647138964577657e-06,
+      "loss": 1.728,
+      "step": 12911
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.645095367847411e-06,
+      "loss": 1.6955,
+      "step": 12912
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.643051771117167e-06,
+      "loss": 1.5708,
+      "step": 12913
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.641008174386921e-06,
+      "loss": 1.6308,
+      "step": 12914
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.638964577656676e-06,
+      "loss": 1.6982,
+      "step": 12915
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.63692098092643e-06,
+      "loss": 1.7958,
+      "step": 12916
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.634877384196186e-06,
+      "loss": 1.5567,
+      "step": 12917
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.63283378746594e-06,
+      "loss": 1.6686,
+      "step": 12918
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.630790190735695e-06,
+      "loss": 1.751,
+      "step": 12919
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6287465940054495e-06,
+      "loss": 1.6423,
+      "step": 12920
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.626702997275204e-06,
+      "loss": 1.6865,
+      "step": 12921
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.624659400544959e-06,
+      "loss": 1.6791,
+      "step": 12922
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.622615803814714e-06,
+      "loss": 1.6011,
+      "step": 12923
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6205722070844695e-06,
+      "loss": 1.6305,
+      "step": 12924
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.618528610354223e-06,
+      "loss": 1.7743,
+      "step": 12925
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.616485013623979e-06,
+      "loss": 1.7118,
+      "step": 12926
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.614441416893733e-06,
+      "loss": 1.5774,
+      "step": 12927
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.612397820163488e-06,
+      "loss": 1.748,
+      "step": 12928
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.6103542234332425e-06,
+      "loss": 1.7366,
+      "step": 12929
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.608310626702997e-06,
+      "loss": 1.5731,
+      "step": 12930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.606267029972752e-06,
+      "loss": 1.6404,
+      "step": 12931
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.604223433242507e-06,
+      "loss": 1.6919,
+      "step": 12932
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.602179836512262e-06,
+      "loss": 1.6103,
+      "step": 12933
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.6001362397820165e-06,
+      "loss": 1.6708,
+      "step": 12934
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.598092643051771e-06,
+      "loss": 1.6145,
+      "step": 12935
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.596049046321526e-06,
+      "loss": 1.7036,
+      "step": 12936
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.594005449591281e-06,
+      "loss": 1.6799,
+      "step": 12937
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.591961852861036e-06,
+      "loss": 1.6751,
+      "step": 12938
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.58991825613079e-06,
+      "loss": 1.7908,
+      "step": 12939
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.587874659400545e-06,
+      "loss": 1.7045,
+      "step": 12940
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5858310626703e-06,
+      "loss": 1.6401,
+      "step": 12941
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.583787465940055e-06,
+      "loss": 1.634,
+      "step": 12942
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5817438692098095e-06,
+      "loss": 1.6149,
+      "step": 12943
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.579700272479564e-06,
+      "loss": 1.6634,
+      "step": 12944
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.577656675749319e-06,
+      "loss": 1.6196,
+      "step": 12945
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.575613079019073e-06,
+      "loss": 1.7308,
+      "step": 12946
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573569482288829e-06,
+      "loss": 1.6943,
+      "step": 12947
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.571525885558583e-06,
+      "loss": 1.7196,
+      "step": 12948
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.569482288828338e-06,
+      "loss": 1.7192,
+      "step": 12949
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.567438692098093e-06,
+      "loss": 1.5948,
+      "step": 12950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.565395095367848e-06,
+      "loss": 1.6639,
+      "step": 12951
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.563351498637603e-06,
+      "loss": 1.5724,
+      "step": 12952
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.561307901907357e-06,
+      "loss": 1.5352,
+      "step": 12953
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.559264305177112e-06,
+      "loss": 1.6084,
+      "step": 12954
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.557220708446866e-06,
+      "loss": 1.6513,
+      "step": 12955
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.555177111716622e-06,
+      "loss": 1.7353,
+      "step": 12956
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.553133514986376e-06,
+      "loss": 1.5757,
+      "step": 12957
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.551089918256131e-06,
+      "loss": 1.787,
+      "step": 12958
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.549046321525885e-06,
+      "loss": 1.5889,
+      "step": 12959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.547002724795641e-06,
+      "loss": 1.6093,
+      "step": 12960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.544959128065395e-06,
+      "loss": 1.7016,
+      "step": 12961
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.54291553133515e-06,
+      "loss": 1.7797,
+      "step": 12962
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.540871934604904e-06,
+      "loss": 1.6052,
+      "step": 12963
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.538828337874659e-06,
+      "loss": 1.6523,
+      "step": 12964
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.536784741144415e-06,
+      "loss": 1.7815,
+      "step": 12965
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.534741144414169e-06,
+      "loss": 1.6816,
+      "step": 12966
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.532697547683924e-06,
+      "loss": 1.686,
+      "step": 12967
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.530653950953678e-06,
+      "loss": 1.5568,
+      "step": 12968
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.528610354223434e-06,
+      "loss": 1.8075,
+      "step": 12969
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.526566757493188e-06,
+      "loss": 1.647,
+      "step": 12970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.524523160762943e-06,
+      "loss": 1.7383,
+      "step": 12971
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5224795640326975e-06,
+      "loss": 1.6664,
+      "step": 12972
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.520435967302452e-06,
+      "loss": 1.7796,
+      "step": 12973
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.518392370572207e-06,
+      "loss": 1.7234,
+      "step": 12974
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.516348773841962e-06,
+      "loss": 1.7532,
+      "step": 12975
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.514305177111717e-06,
+      "loss": 1.728,
+      "step": 12976
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.512261580381471e-06,
+      "loss": 1.6,
+      "step": 12977
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.510217983651227e-06,
+      "loss": 1.7337,
+      "step": 12978
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.508174386920981e-06,
+      "loss": 1.7239,
+      "step": 12979
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.506130790190736e-06,
+      "loss": 1.6802,
+      "step": 12980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5040871934604905e-06,
+      "loss": 1.5516,
+      "step": 12981
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.502043596730245e-06,
+      "loss": 1.6886,
+      "step": 12982
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5e-06,
+      "loss": 1.6638,
+      "step": 12983
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.497956403269755e-06,
+      "loss": 1.6939,
+      "step": 12984
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.49591280653951e-06,
+      "loss": 1.6454,
+      "step": 12985
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.4938692098092645e-06,
+      "loss": 1.7407,
+      "step": 12986
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.491825613079019e-06,
+      "loss": 1.5747,
+      "step": 12987
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.489782016348774e-06,
+      "loss": 1.6858,
+      "step": 12988
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.487738419618528e-06,
+      "loss": 1.5273,
+      "step": 12989
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.485694822888284e-06,
+      "loss": 1.7699,
+      "step": 12990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.4836512261580376e-06,
+      "loss": 1.7088,
+      "step": 12991
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.481607629427793e-06,
+      "loss": 1.5864,
+      "step": 12992
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.479564032697548e-06,
+      "loss": 1.5294,
+      "step": 12993
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.477520435967303e-06,
+      "loss": 1.7508,
+      "step": 12994
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.4754768392370575e-06,
+      "loss": 1.5985,
+      "step": 12995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.473433242506812e-06,
+      "loss": 1.718,
+      "step": 12996
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.471389645776567e-06,
+      "loss": 1.5499,
+      "step": 12997
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.469346049046321e-06,
+      "loss": 1.7147,
+      "step": 12998
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.467302452316077e-06,
+      "loss": 1.6832,
+      "step": 12999
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.465258855585831e-06,
+      "loss": 1.7171,
+      "step": 13000
+    },
+    {
+      "epoch": 2.57,
+      "eval_loss": 1.5040637254714966,
+      "eval_runtime": 721.3993,
+      "eval_samples_per_second": 5.753,
+      "eval_steps_per_second": 1.439,
+      "step": 13000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.463215258855586e-06,
+      "loss": 1.7077,
+      "step": 13001
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.46117166212534e-06,
+      "loss": 1.5912,
+      "step": 13002
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.459128065395096e-06,
+      "loss": 1.5362,
+      "step": 13003
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.45708446866485e-06,
+      "loss": 1.5306,
+      "step": 13004
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.455040871934605e-06,
+      "loss": 1.6548,
+      "step": 13005
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.45299727520436e-06,
+      "loss": 1.7191,
+      "step": 13006
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.450953678474114e-06,
+      "loss": 1.5627,
+      "step": 13007
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.44891008174387e-06,
+      "loss": 1.7108,
+      "step": 13008
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.446866485013624e-06,
+      "loss": 1.8175,
+      "step": 13009
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.444822888283379e-06,
+      "loss": 1.6073,
+      "step": 13010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.442779291553133e-06,
+      "loss": 1.7917,
+      "step": 13011
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.440735694822889e-06,
+      "loss": 1.6478,
+      "step": 13012
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.438692098092643e-06,
+      "loss": 1.6402,
+      "step": 13013
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.4366485013623985e-06,
+      "loss": 1.6449,
+      "step": 13014
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.434604904632152e-06,
+      "loss": 1.7151,
+      "step": 13015
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.432561307901907e-06,
+      "loss": 1.6604,
+      "step": 13016
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.430517711171662e-06,
+      "loss": 1.5798,
+      "step": 13017
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.428474114441417e-06,
+      "loss": 1.5996,
+      "step": 13018
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.426430517711172e-06,
+      "loss": 1.6349,
+      "step": 13019
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.424386920980926e-06,
+      "loss": 1.5562,
+      "step": 13020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.422343324250682e-06,
+      "loss": 1.607,
+      "step": 13021
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.420299727520436e-06,
+      "loss": 1.6279,
+      "step": 13022
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.4182561307901915e-06,
+      "loss": 1.6419,
+      "step": 13023
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.4162125340599455e-06,
+      "loss": 1.6751,
+      "step": 13024
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.4141689373297e-06,
+      "loss": 1.763,
+      "step": 13025
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.412125340599455e-06,
+      "loss": 1.6491,
+      "step": 13026
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.41008174386921e-06,
+      "loss": 1.6799,
+      "step": 13027
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.408038147138965e-06,
+      "loss": 1.615,
+      "step": 13028
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.405994550408719e-06,
+      "loss": 1.738,
+      "step": 13029
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.403950953678474e-06,
+      "loss": 1.6061,
+      "step": 13030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.401907356948229e-06,
+      "loss": 1.7231,
+      "step": 13031
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.399863760217984e-06,
+      "loss": 1.6993,
+      "step": 13032
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3978201634877385e-06,
+      "loss": 1.7417,
+      "step": 13033
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.395776566757493e-06,
+      "loss": 1.5826,
+      "step": 13034
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.393732970027248e-06,
+      "loss": 1.6507,
+      "step": 13035
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.391689373297003e-06,
+      "loss": 1.5937,
+      "step": 13036
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.389645776566758e-06,
+      "loss": 1.7039,
+      "step": 13037
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3876021798365125e-06,
+      "loss": 1.7853,
+      "step": 13038
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.385558583106267e-06,
+      "loss": 1.6446,
+      "step": 13039
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.383514986376022e-06,
+      "loss": 1.673,
+      "step": 13040
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.381471389645777e-06,
+      "loss": 1.7778,
+      "step": 13041
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.379427792915532e-06,
+      "loss": 1.5218,
+      "step": 13042
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3773841961852856e-06,
+      "loss": 1.6522,
+      "step": 13043
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.375340599455041e-06,
+      "loss": 1.7662,
+      "step": 13044
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.373297002724795e-06,
+      "loss": 1.848,
+      "step": 13045
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.371253405994551e-06,
+      "loss": 1.5832,
+      "step": 13046
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3692098092643055e-06,
+      "loss": 1.6617,
+      "step": 13047
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.36716621253406e-06,
+      "loss": 1.692,
+      "step": 13048
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.365122615803815e-06,
+      "loss": 1.6122,
+      "step": 13049
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.36307901907357e-06,
+      "loss": 1.6659,
+      "step": 13050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.361035422343325e-06,
+      "loss": 1.7543,
+      "step": 13051
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.358991825613079e-06,
+      "loss": 1.5048,
+      "step": 13052
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.356948228882834e-06,
+      "loss": 1.7169,
+      "step": 13053
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.354904632152588e-06,
+      "loss": 1.6692,
+      "step": 13054
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.352861035422344e-06,
+      "loss": 1.7646,
+      "step": 13055
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.350817438692098e-06,
+      "loss": 1.7745,
+      "step": 13056
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.348773841961853e-06,
+      "loss": 1.6362,
+      "step": 13057
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.346730245231607e-06,
+      "loss": 1.5449,
+      "step": 13058
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.344686648501362e-06,
+      "loss": 1.6501,
+      "step": 13059
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.342643051771117e-06,
+      "loss": 1.6385,
+      "step": 13060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.340599455040872e-06,
+      "loss": 1.6367,
+      "step": 13061
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.338555858310627e-06,
+      "loss": 1.7547,
+      "step": 13062
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.336512261580381e-06,
+      "loss": 1.6876,
+      "step": 13063
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.334468664850137e-06,
+      "loss": 1.6252,
+      "step": 13064
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.332425068119891e-06,
+      "loss": 1.6782,
+      "step": 13065
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3303814713896465e-06,
+      "loss": 1.6788,
+      "step": 13066
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3283378746594e-06,
+      "loss": 1.5973,
+      "step": 13067
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.326294277929155e-06,
+      "loss": 1.5975,
+      "step": 13068
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.32425068119891e-06,
+      "loss": 1.6998,
+      "step": 13069
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.322207084468665e-06,
+      "loss": 1.6003,
+      "step": 13070
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3201634877384196e-06,
+      "loss": 1.7412,
+      "step": 13071
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.318119891008174e-06,
+      "loss": 1.6651,
+      "step": 13072
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.316076294277929e-06,
+      "loss": 1.7337,
+      "step": 13073
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.314032697547684e-06,
+      "loss": 1.7857,
+      "step": 13074
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3119891008174395e-06,
+      "loss": 1.6488,
+      "step": 13075
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.3099455040871935e-06,
+      "loss": 1.6685,
+      "step": 13076
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.307901907356948e-06,
+      "loss": 1.624,
+      "step": 13077
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.305858310626703e-06,
+      "loss": 1.532,
+      "step": 13078
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.303814713896458e-06,
+      "loss": 1.8211,
+      "step": 13079
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.301771117166213e-06,
+      "loss": 1.6756,
+      "step": 13080
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.299727520435967e-06,
+      "loss": 1.7214,
+      "step": 13081
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.297683923705722e-06,
+      "loss": 1.5891,
+      "step": 13082
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.295640326975477e-06,
+      "loss": 1.8107,
+      "step": 13083
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.293596730245232e-06,
+      "loss": 1.7227,
+      "step": 13084
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2915531335149866e-06,
+      "loss": 1.6116,
+      "step": 13085
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2895095367847405e-06,
+      "loss": 1.5412,
+      "step": 13086
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.287465940054496e-06,
+      "loss": 1.5626,
+      "step": 13087
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.285422343324251e-06,
+      "loss": 1.7089,
+      "step": 13088
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.283378746594006e-06,
+      "loss": 1.6486,
+      "step": 13089
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2813351498637605e-06,
+      "loss": 1.6705,
+      "step": 13090
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.279291553133515e-06,
+      "loss": 1.5872,
+      "step": 13091
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.27724795640327e-06,
+      "loss": 1.6618,
+      "step": 13092
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.275204359673025e-06,
+      "loss": 1.6136,
+      "step": 13093
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.27316076294278e-06,
+      "loss": 1.7255,
+      "step": 13094
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2711171662125336e-06,
+      "loss": 1.6482,
+      "step": 13095
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.269073569482289e-06,
+      "loss": 1.5839,
+      "step": 13096
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.267029972752043e-06,
+      "loss": 1.7211,
+      "step": 13097
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.264986376021799e-06,
+      "loss": 1.6286,
+      "step": 13098
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.262942779291553e-06,
+      "loss": 1.7421,
+      "step": 13099
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.260899182561308e-06,
+      "loss": 1.5986,
+      "step": 13100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.258855585831062e-06,
+      "loss": 1.7097,
+      "step": 13101
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.256811989100818e-06,
+      "loss": 1.6577,
+      "step": 13102
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.254768392370573e-06,
+      "loss": 1.7017,
+      "step": 13103
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.252724795640327e-06,
+      "loss": 1.6493,
+      "step": 13104
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.250681198910082e-06,
+      "loss": 1.783,
+      "step": 13105
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.248637602179836e-06,
+      "loss": 1.709,
+      "step": 13106
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.246594005449592e-06,
+      "loss": 1.6413,
+      "step": 13107
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.244550408719346e-06,
+      "loss": 1.5436,
+      "step": 13108
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.242506811989101e-06,
+      "loss": 1.709,
+      "step": 13109
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.240463215258855e-06,
+      "loss": 1.6702,
+      "step": 13110
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.238419618528611e-06,
+      "loss": 1.6542,
+      "step": 13111
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.236376021798365e-06,
+      "loss": 1.6309,
+      "step": 13112
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.23433242506812e-06,
+      "loss": 1.7818,
+      "step": 13113
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2322888283378745e-06,
+      "loss": 1.6768,
+      "step": 13114
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.230245231607629e-06,
+      "loss": 1.5859,
+      "step": 13115
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.228201634877385e-06,
+      "loss": 1.7383,
+      "step": 13116
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.226158038147139e-06,
+      "loss": 1.6583,
+      "step": 13117
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2241144414168945e-06,
+      "loss": 1.5188,
+      "step": 13118
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.222070844686648e-06,
+      "loss": 1.696,
+      "step": 13119
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.220027247956404e-06,
+      "loss": 1.5964,
+      "step": 13120
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.217983651226158e-06,
+      "loss": 1.6763,
+      "step": 13121
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.215940054495913e-06,
+      "loss": 1.706,
+      "step": 13122
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2138964577656676e-06,
+      "loss": 1.7286,
+      "step": 13123
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.211852861035422e-06,
+      "loss": 1.7127,
+      "step": 13124
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.209809264305177e-06,
+      "loss": 1.6933,
+      "step": 13125
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.207765667574932e-06,
+      "loss": 1.6182,
+      "step": 13126
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.205722070844687e-06,
+      "loss": 1.6274,
+      "step": 13127
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.2036784741144415e-06,
+      "loss": 1.7286,
+      "step": 13128
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.201634877384196e-06,
+      "loss": 1.5481,
+      "step": 13129
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.199591280653951e-06,
+      "loss": 1.6045,
+      "step": 13130
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.197547683923706e-06,
+      "loss": 1.6873,
+      "step": 13131
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.195504087193461e-06,
+      "loss": 1.632,
+      "step": 13132
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.193460490463215e-06,
+      "loss": 1.5788,
+      "step": 13133
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.19141689373297e-06,
+      "loss": 1.77,
+      "step": 13134
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.189373297002725e-06,
+      "loss": 1.7654,
+      "step": 13135
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.18732970027248e-06,
+      "loss": 1.5797,
+      "step": 13136
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1852861035422346e-06,
+      "loss": 1.6731,
+      "step": 13137
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.183242506811989e-06,
+      "loss": 1.5929,
+      "step": 13138
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.181198910081744e-06,
+      "loss": 1.7791,
+      "step": 13139
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.179155313351498e-06,
+      "loss": 1.6784,
+      "step": 13140
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.177111716621254e-06,
+      "loss": 1.6116,
+      "step": 13141
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.175068119891008e-06,
+      "loss": 1.6951,
+      "step": 13142
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.173024523160763e-06,
+      "loss": 1.8023,
+      "step": 13143
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.170980926430518e-06,
+      "loss": 1.5596,
+      "step": 13144
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.168937329700273e-06,
+      "loss": 1.6356,
+      "step": 13145
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.166893732970028e-06,
+      "loss": 1.7038,
+      "step": 13146
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.164850136239782e-06,
+      "loss": 1.629,
+      "step": 13147
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.162806539509537e-06,
+      "loss": 1.6455,
+      "step": 13148
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.160762942779291e-06,
+      "loss": 1.5967,
+      "step": 13149
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.158719346049047e-06,
+      "loss": 1.685,
+      "step": 13150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.156675749318801e-06,
+      "loss": 1.6381,
+      "step": 13151
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.154632152588556e-06,
+      "loss": 1.5836,
+      "step": 13152
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.15258855585831e-06,
+      "loss": 1.6342,
+      "step": 13153
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.150544959128066e-06,
+      "loss": 1.6152,
+      "step": 13154
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.14850136239782e-06,
+      "loss": 1.6138,
+      "step": 13155
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.146457765667575e-06,
+      "loss": 1.7465,
+      "step": 13156
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.14441416893733e-06,
+      "loss": 1.6379,
+      "step": 13157
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.142370572207084e-06,
+      "loss": 1.6514,
+      "step": 13158
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.14032697547684e-06,
+      "loss": 1.586,
+      "step": 13159
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.138283378746594e-06,
+      "loss": 1.8186,
+      "step": 13160
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.136239782016349e-06,
+      "loss": 1.5174,
+      "step": 13161
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.134196185286103e-06,
+      "loss": 1.6922,
+      "step": 13162
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.132152588555859e-06,
+      "loss": 1.7142,
+      "step": 13163
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.130108991825613e-06,
+      "loss": 1.5444,
+      "step": 13164
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.128065395095368e-06,
+      "loss": 1.7067,
+      "step": 13165
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1260217983651225e-06,
+      "loss": 1.792,
+      "step": 13166
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.123978201634877e-06,
+      "loss": 1.5851,
+      "step": 13167
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.121934604904632e-06,
+      "loss": 1.7117,
+      "step": 13168
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.119891008174387e-06,
+      "loss": 1.718,
+      "step": 13169
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.117847411444142e-06,
+      "loss": 1.7096,
+      "step": 13170
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1158038147138964e-06,
+      "loss": 1.5657,
+      "step": 13171
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.113760217983652e-06,
+      "loss": 1.674,
+      "step": 13172
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.111716621253406e-06,
+      "loss": 1.6589,
+      "step": 13173
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.109673024523161e-06,
+      "loss": 1.7165,
+      "step": 13174
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1076294277929156e-06,
+      "loss": 1.6407,
+      "step": 13175
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.10558583106267e-06,
+      "loss": 1.6281,
+      "step": 13176
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.103542234332425e-06,
+      "loss": 1.7431,
+      "step": 13177
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.10149863760218e-06,
+      "loss": 1.6065,
+      "step": 13178
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.099455040871935e-06,
+      "loss": 1.6204,
+      "step": 13179
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.0974114441416895e-06,
+      "loss": 1.5695,
+      "step": 13180
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.095367847411444e-06,
+      "loss": 1.659,
+      "step": 13181
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.093324250681199e-06,
+      "loss": 1.6144,
+      "step": 13182
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.091280653950953e-06,
+      "loss": 1.6735,
+      "step": 13183
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.089237057220709e-06,
+      "loss": 1.6217,
+      "step": 13184
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0871934604904634e-06,
+      "loss": 1.6424,
+      "step": 13185
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.085149863760218e-06,
+      "loss": 1.7121,
+      "step": 13186
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.083106267029973e-06,
+      "loss": 1.6851,
+      "step": 13187
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.081062670299728e-06,
+      "loss": 1.68,
+      "step": 13188
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0790190735694826e-06,
+      "loss": 1.697,
+      "step": 13189
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.076975476839237e-06,
+      "loss": 1.6923,
+      "step": 13190
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.074931880108992e-06,
+      "loss": 1.7065,
+      "step": 13191
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.072888283378746e-06,
+      "loss": 1.6865,
+      "step": 13192
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.070844686648502e-06,
+      "loss": 1.5737,
+      "step": 13193
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.068801089918256e-06,
+      "loss": 1.7356,
+      "step": 13194
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.066757493188011e-06,
+      "loss": 1.7493,
+      "step": 13195
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.064713896457765e-06,
+      "loss": 1.6914,
+      "step": 13196
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.062670299727521e-06,
+      "loss": 1.6199,
+      "step": 13197
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.060626702997275e-06,
+      "loss": 1.6458,
+      "step": 13198
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0585831062670304e-06,
+      "loss": 1.64,
+      "step": 13199
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.056539509536785e-06,
+      "loss": 1.7332,
+      "step": 13200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.054495912806539e-06,
+      "loss": 1.6154,
+      "step": 13201
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.052452316076295e-06,
+      "loss": 1.6252,
+      "step": 13202
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.050408719346049e-06,
+      "loss": 1.6021,
+      "step": 13203
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.048365122615804e-06,
+      "loss": 1.6055,
+      "step": 13204
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.046321525885558e-06,
+      "loss": 1.6008,
+      "step": 13205
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.044277929155314e-06,
+      "loss": 1.7593,
+      "step": 13206
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.042234332425068e-06,
+      "loss": 1.6392,
+      "step": 13207
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0401907356948235e-06,
+      "loss": 1.6361,
+      "step": 13208
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0381471389645774e-06,
+      "loss": 1.802,
+      "step": 13209
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.036103542234332e-06,
+      "loss": 1.889,
+      "step": 13210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.034059945504087e-06,
+      "loss": 1.6267,
+      "step": 13211
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.032016348773842e-06,
+      "loss": 1.6019,
+      "step": 13212
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0299727520435974e-06,
+      "loss": 1.8305,
+      "step": 13213
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.027929155313351e-06,
+      "loss": 1.6877,
+      "step": 13214
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.025885558583107e-06,
+      "loss": 1.7591,
+      "step": 13215
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.023841961852861e-06,
+      "loss": 1.6023,
+      "step": 13216
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0217983651226166e-06,
+      "loss": 1.6177,
+      "step": 13217
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0197547683923705e-06,
+      "loss": 1.6718,
+      "step": 13218
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.017711171662125e-06,
+      "loss": 1.8042,
+      "step": 13219
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.01566757493188e-06,
+      "loss": 1.7408,
+      "step": 13220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.013623978201635e-06,
+      "loss": 1.7413,
+      "step": 13221
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.01158038147139e-06,
+      "loss": 2.0248,
+      "step": 13222
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.0095367847411444e-06,
+      "loss": 1.7704,
+      "step": 13223
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.007493188010899e-06,
+      "loss": 1.7466,
+      "step": 13224
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.005449591280654e-06,
+      "loss": 1.7465,
+      "step": 13225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.00340599455041e-06,
+      "loss": 1.6184,
+      "step": 13226
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.001362397820164e-06,
+      "loss": 1.7271,
+      "step": 13227
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.999318801089918e-06,
+      "loss": 1.5873,
+      "step": 13228
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.997275204359673e-06,
+      "loss": 1.6556,
+      "step": 13229
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.995231607629428e-06,
+      "loss": 1.7974,
+      "step": 13230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.993188010899183e-06,
+      "loss": 1.7327,
+      "step": 13231
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.9911444141689375e-06,
+      "loss": 1.6227,
+      "step": 13232
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.989100817438692e-06,
+      "loss": 1.6562,
+      "step": 13233
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.987057220708447e-06,
+      "loss": 1.6948,
+      "step": 13234
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.985013623978202e-06,
+      "loss": 1.7303,
+      "step": 13235
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.982970027247957e-06,
+      "loss": 1.6244,
+      "step": 13236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.980926430517711e-06,
+      "loss": 1.6397,
+      "step": 13237
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.978882833787466e-06,
+      "loss": 1.7106,
+      "step": 13238
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.97683923705722e-06,
+      "loss": 1.6743,
+      "step": 13239
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.974795640326976e-06,
+      "loss": 1.6451,
+      "step": 13240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.972752043596731e-06,
+      "loss": 1.524,
+      "step": 13241
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.970708446866485e-06,
+      "loss": 1.6835,
+      "step": 13242
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.96866485013624e-06,
+      "loss": 1.6326,
+      "step": 13243
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.966621253405995e-06,
+      "loss": 1.6735,
+      "step": 13244
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.96457765667575e-06,
+      "loss": 1.6296,
+      "step": 13245
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.962534059945504e-06,
+      "loss": 1.6231,
+      "step": 13246
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.960490463215259e-06,
+      "loss": 1.7117,
+      "step": 13247
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.958446866485013e-06,
+      "loss": 1.6559,
+      "step": 13248
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.956403269754769e-06,
+      "loss": 1.6173,
+      "step": 13249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.954359673024523e-06,
+      "loss": 1.6936,
+      "step": 13250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.9523160762942784e-06,
+      "loss": 1.6752,
+      "step": 13251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.950272479564032e-06,
+      "loss": 1.7524,
+      "step": 13252
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.948228882833787e-06,
+      "loss": 1.6417,
+      "step": 13253
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.946185286103543e-06,
+      "loss": 1.7702,
+      "step": 13254
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.944141689373297e-06,
+      "loss": 1.7127,
+      "step": 13255
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.942098092643052e-06,
+      "loss": 1.6504,
+      "step": 13256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.940054495912806e-06,
+      "loss": 1.562,
+      "step": 13257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.938010899182562e-06,
+      "loss": 1.7112,
+      "step": 13258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.935967302452316e-06,
+      "loss": 1.5864,
+      "step": 13259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.9339237057220715e-06,
+      "loss": 1.7054,
+      "step": 13260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.9318801089918254e-06,
+      "loss": 1.5584,
+      "step": 13261
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.92983651226158e-06,
+      "loss": 1.8037,
+      "step": 13262
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.927792915531335e-06,
+      "loss": 1.6109,
+      "step": 13263
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.92574931880109e-06,
+      "loss": 1.6225,
+      "step": 13264
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.923705722070845e-06,
+      "loss": 1.6656,
+      "step": 13265
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.921662125340599e-06,
+      "loss": 1.6215,
+      "step": 13266
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.919618528610354e-06,
+      "loss": 1.689,
+      "step": 13267
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.917574931880109e-06,
+      "loss": 1.6529,
+      "step": 13268
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.9155313351498646e-06,
+      "loss": 1.5273,
+      "step": 13269
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.9134877384196185e-06,
+      "loss": 1.7421,
+      "step": 13270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.911444141689373e-06,
+      "loss": 1.6369,
+      "step": 13271
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.909400544959128e-06,
+      "loss": 1.6643,
+      "step": 13272
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.907356948228883e-06,
+      "loss": 1.7274,
+      "step": 13273
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.905313351498638e-06,
+      "loss": 1.6098,
+      "step": 13274
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.9032697547683924e-06,
+      "loss": 1.6704,
+      "step": 13275
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.901226158038147e-06,
+      "loss": 1.621,
+      "step": 13276
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.899182561307902e-06,
+      "loss": 1.7968,
+      "step": 13277
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.897138964577657e-06,
+      "loss": 1.6678,
+      "step": 13278
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.895095367847412e-06,
+      "loss": 1.7328,
+      "step": 13279
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.8930517711171655e-06,
+      "loss": 1.6339,
+      "step": 13280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.891008174386921e-06,
+      "loss": 1.6641,
+      "step": 13281
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.888964577656676e-06,
+      "loss": 1.6194,
+      "step": 13282
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.886920980926431e-06,
+      "loss": 1.651,
+      "step": 13283
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8848773841961855e-06,
+      "loss": 1.6051,
+      "step": 13284
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.88283378746594e-06,
+      "loss": 1.6799,
+      "step": 13285
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.880790190735695e-06,
+      "loss": 1.6956,
+      "step": 13286
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.87874659400545e-06,
+      "loss": 1.6363,
+      "step": 13287
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.876702997275205e-06,
+      "loss": 1.7405,
+      "step": 13288
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.874659400544959e-06,
+      "loss": 1.716,
+      "step": 13289
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.872615803814714e-06,
+      "loss": 1.6811,
+      "step": 13290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.870572207084468e-06,
+      "loss": 1.5363,
+      "step": 13291
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.868528610354224e-06,
+      "loss": 1.6631,
+      "step": 13292
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.866485013623978e-06,
+      "loss": 1.6813,
+      "step": 13293
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.864441416893733e-06,
+      "loss": 1.5825,
+      "step": 13294
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.862397820163488e-06,
+      "loss": 1.5889,
+      "step": 13295
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.860354223433243e-06,
+      "loss": 1.587,
+      "step": 13296
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.858310626702998e-06,
+      "loss": 1.6097,
+      "step": 13297
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.856267029972752e-06,
+      "loss": 1.6972,
+      "step": 13298
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.854223433242507e-06,
+      "loss": 1.6356,
+      "step": 13299
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.852179836512261e-06,
+      "loss": 1.7641,
+      "step": 13300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.850136239782017e-06,
+      "loss": 1.5586,
+      "step": 13301
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.848092643051771e-06,
+      "loss": 1.6056,
+      "step": 13302
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8460490463215264e-06,
+      "loss": 1.6842,
+      "step": 13303
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.84400544959128e-06,
+      "loss": 1.7043,
+      "step": 13304
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.841961852861036e-06,
+      "loss": 1.6317,
+      "step": 13305
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.83991825613079e-06,
+      "loss": 1.6581,
+      "step": 13306
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.837874659400545e-06,
+      "loss": 1.876,
+      "step": 13307
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8358310626702995e-06,
+      "loss": 1.6514,
+      "step": 13308
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.833787465940054e-06,
+      "loss": 1.5578,
+      "step": 13309
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.83174386920981e-06,
+      "loss": 1.6031,
+      "step": 13310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.829700272479564e-06,
+      "loss": 1.5937,
+      "step": 13311
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8276566757493195e-06,
+      "loss": 1.7795,
+      "step": 13312
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8256130790190734e-06,
+      "loss": 1.5882,
+      "step": 13313
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.823569482288829e-06,
+      "loss": 1.6262,
+      "step": 13314
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.821525885558583e-06,
+      "loss": 1.6281,
+      "step": 13315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.819482288828338e-06,
+      "loss": 1.7545,
+      "step": 13316
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.817438692098093e-06,
+      "loss": 1.5985,
+      "step": 13317
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.815395095367847e-06,
+      "loss": 1.6055,
+      "step": 13318
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.813351498637602e-06,
+      "loss": 1.6325,
+      "step": 13319
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.811307901907357e-06,
+      "loss": 1.7402,
+      "step": 13320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8092643051771113e-06,
+      "loss": 1.6848,
+      "step": 13321
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8072207084468665e-06,
+      "loss": 1.6679,
+      "step": 13322
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8051771117166217e-06,
+      "loss": 1.5523,
+      "step": 13323
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.803133514986376e-06,
+      "loss": 1.71,
+      "step": 13324
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.8010899182561313e-06,
+      "loss": 1.687,
+      "step": 13325
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7990463215258857e-06,
+      "loss": 1.6268,
+      "step": 13326
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.797002724795641e-06,
+      "loss": 1.634,
+      "step": 13327
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7949591280653952e-06,
+      "loss": 1.6197,
+      "step": 13328
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.79291553133515e-06,
+      "loss": 1.5178,
+      "step": 13329
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7908719346049044e-06,
+      "loss": 1.5645,
+      "step": 13330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7888283378746596e-06,
+      "loss": 1.7177,
+      "step": 13331
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.786784741144414e-06,
+      "loss": 1.7095,
+      "step": 13332
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.784741144414169e-06,
+      "loss": 1.6887,
+      "step": 13333
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7826975476839235e-06,
+      "loss": 1.7475,
+      "step": 13334
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7806539509536787e-06,
+      "loss": 1.6949,
+      "step": 13335
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.778610354223433e-06,
+      "loss": 1.7452,
+      "step": 13336
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.776566757493188e-06,
+      "loss": 1.4657,
+      "step": 13337
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.774523160762943e-06,
+      "loss": 1.7645,
+      "step": 13338
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7724795640326975e-06,
+      "loss": 1.7423,
+      "step": 13339
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7704359673024527e-06,
+      "loss": 1.7937,
+      "step": 13340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.768392370572207e-06,
+      "loss": 1.7809,
+      "step": 13341
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7663487738419622e-06,
+      "loss": 1.8381,
+      "step": 13342
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7643051771117166e-06,
+      "loss": 1.6408,
+      "step": 13343
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.762261580381472e-06,
+      "loss": 1.4874,
+      "step": 13344
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.760217983651226e-06,
+      "loss": 1.6505,
+      "step": 13345
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.758174386920981e-06,
+      "loss": 1.6057,
+      "step": 13346
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7561307901907353e-06,
+      "loss": 1.7663,
+      "step": 13347
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7540871934604905e-06,
+      "loss": 1.6111,
+      "step": 13348
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.752043596730245e-06,
+      "loss": 1.6761,
+      "step": 13349
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.75e-06,
+      "loss": 1.7498,
+      "step": 13350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.747956403269755e-06,
+      "loss": 1.7104,
+      "step": 13351
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7459128065395097e-06,
+      "loss": 1.6656,
+      "step": 13352
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7438692098092645e-06,
+      "loss": 1.6663,
+      "step": 13353
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7418256130790192e-06,
+      "loss": 1.6784,
+      "step": 13354
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7397820163487736e-06,
+      "loss": 1.6389,
+      "step": 13355
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7377384196185284e-06,
+      "loss": 1.7926,
+      "step": 13356
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7356948228882836e-06,
+      "loss": 1.6157,
+      "step": 13357
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7336512261580384e-06,
+      "loss": 1.7001,
+      "step": 13358
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.731607629427793e-06,
+      "loss": 1.894,
+      "step": 13359
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.729564032697548e-06,
+      "loss": 1.6078,
+      "step": 13360
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7275204359673027e-06,
+      "loss": 1.6118,
+      "step": 13361
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7254768392370575e-06,
+      "loss": 1.7488,
+      "step": 13362
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7234332425068123e-06,
+      "loss": 1.6338,
+      "step": 13363
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7213896457765667e-06,
+      "loss": 1.6586,
+      "step": 13364
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7193460490463215e-06,
+      "loss": 1.6778,
+      "step": 13365
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7173024523160762e-06,
+      "loss": 1.7016,
+      "step": 13366
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.715258855585831e-06,
+      "loss": 1.6402,
+      "step": 13367
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.713215258855586e-06,
+      "loss": 1.5714,
+      "step": 13368
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7111716621253406e-06,
+      "loss": 1.5864,
+      "step": 13369
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7091280653950954e-06,
+      "loss": 1.646,
+      "step": 13370
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7070844686648506e-06,
+      "loss": 1.6012,
+      "step": 13371
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.705040871934605e-06,
+      "loss": 1.5889,
+      "step": 13372
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7029972752043597e-06,
+      "loss": 1.8779,
+      "step": 13373
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7009536784741145e-06,
+      "loss": 1.8543,
+      "step": 13374
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6989100817438693e-06,
+      "loss": 1.6845,
+      "step": 13375
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.696866485013624e-06,
+      "loss": 1.5913,
+      "step": 13376
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.694822888283379e-06,
+      "loss": 1.7969,
+      "step": 13377
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6927792915531337e-06,
+      "loss": 1.6832,
+      "step": 13378
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6907356948228885e-06,
+      "loss": 1.5818,
+      "step": 13379
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6886920980926432e-06,
+      "loss": 1.6092,
+      "step": 13380
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6866485013623976e-06,
+      "loss": 1.5231,
+      "step": 13381
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6846049046321524e-06,
+      "loss": 1.6689,
+      "step": 13382
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682561307901907e-06,
+      "loss": 1.6548,
+      "step": 13383
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.680517711171662e-06,
+      "loss": 1.6489,
+      "step": 13384
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.678474114441417e-06,
+      "loss": 1.7156,
+      "step": 13385
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.676430517711172e-06,
+      "loss": 1.6803,
+      "step": 13386
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6743869209809267e-06,
+      "loss": 1.6505,
+      "step": 13387
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6723433242506815e-06,
+      "loss": 1.6039,
+      "step": 13388
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6702997275204363e-06,
+      "loss": 1.511,
+      "step": 13389
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6682561307901907e-06,
+      "loss": 1.7491,
+      "step": 13390
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6662125340599455e-06,
+      "loss": 1.6666,
+      "step": 13391
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6641689373297002e-06,
+      "loss": 1.6732,
+      "step": 13392
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.662125340599455e-06,
+      "loss": 1.6905,
+      "step": 13393
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.66008174386921e-06,
+      "loss": 1.624,
+      "step": 13394
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6580381471389646e-06,
+      "loss": 1.6858,
+      "step": 13395
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6559945504087194e-06,
+      "loss": 1.6763,
+      "step": 13396
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.653950953678474e-06,
+      "loss": 1.6359,
+      "step": 13397
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.651907356948229e-06,
+      "loss": 1.6486,
+      "step": 13398
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6498637602179837e-06,
+      "loss": 1.6994,
+      "step": 13399
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6478201634877385e-06,
+      "loss": 1.5498,
+      "step": 13400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6457765667574933e-06,
+      "loss": 1.6309,
+      "step": 13401
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.643732970027248e-06,
+      "loss": 1.7012,
+      "step": 13402
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.641689373297003e-06,
+      "loss": 1.6066,
+      "step": 13403
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6396457765667577e-06,
+      "loss": 1.5859,
+      "step": 13404
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6376021798365125e-06,
+      "loss": 1.6475,
+      "step": 13405
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6355585831062672e-06,
+      "loss": 1.5851,
+      "step": 13406
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.633514986376022e-06,
+      "loss": 1.5481,
+      "step": 13407
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6314713896457764e-06,
+      "loss": 1.5797,
+      "step": 13408
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.629427792915531e-06,
+      "loss": 1.6694,
+      "step": 13409
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.627384196185286e-06,
+      "loss": 1.5464,
+      "step": 13410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6253405994550407e-06,
+      "loss": 1.6135,
+      "step": 13411
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6232970027247955e-06,
+      "loss": 1.6928,
+      "step": 13412
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6212534059945507e-06,
+      "loss": 1.7289,
+      "step": 13413
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6192098092643055e-06,
+      "loss": 1.7756,
+      "step": 13414
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6171662125340603e-06,
+      "loss": 1.6768,
+      "step": 13415
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.615122615803815e-06,
+      "loss": 1.6557,
+      "step": 13416
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6130790190735695e-06,
+      "loss": 1.8226,
+      "step": 13417
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6110354223433242e-06,
+      "loss": 1.6463,
+      "step": 13418
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.608991825613079e-06,
+      "loss": 1.7172,
+      "step": 13419
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.606948228882834e-06,
+      "loss": 1.6613,
+      "step": 13420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6049046321525886e-06,
+      "loss": 1.6387,
+      "step": 13421
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6028610354223434e-06,
+      "loss": 1.7201,
+      "step": 13422
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.600817438692098e-06,
+      "loss": 1.621,
+      "step": 13423
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.598773841961853e-06,
+      "loss": 1.574,
+      "step": 13424
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5967302452316077e-06,
+      "loss": 1.6203,
+      "step": 13425
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5946866485013625e-06,
+      "loss": 1.6402,
+      "step": 13426
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5926430517711173e-06,
+      "loss": 1.671,
+      "step": 13427
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.590599455040872e-06,
+      "loss": 1.5548,
+      "step": 13428
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.588555858310627e-06,
+      "loss": 1.7056,
+      "step": 13429
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5865122615803817e-06,
+      "loss": 1.6357,
+      "step": 13430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5844686648501365e-06,
+      "loss": 1.7323,
+      "step": 13431
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5824250681198912e-06,
+      "loss": 1.7246,
+      "step": 13432
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.580381471389646e-06,
+      "loss": 1.8016,
+      "step": 13433
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5783378746594004e-06,
+      "loss": 1.6403,
+      "step": 13434
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.576294277929155e-06,
+      "loss": 1.6146,
+      "step": 13435
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.57425068119891e-06,
+      "loss": 1.762,
+      "step": 13436
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5722070844686648e-06,
+      "loss": 1.7662,
+      "step": 13437
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5701634877384195e-06,
+      "loss": 1.609,
+      "step": 13438
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5681198910081743e-06,
+      "loss": 1.6863,
+      "step": 13439
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5660762942779295e-06,
+      "loss": 1.626,
+      "step": 13440
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5640326975476843e-06,
+      "loss": 1.6659,
+      "step": 13441
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.561989100817439e-06,
+      "loss": 1.6754,
+      "step": 13442
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5599455040871935e-06,
+      "loss": 1.704,
+      "step": 13443
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5579019073569483e-06,
+      "loss": 1.7052,
+      "step": 13444
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.555858310626703e-06,
+      "loss": 1.6062,
+      "step": 13445
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.553814713896458e-06,
+      "loss": 1.5816,
+      "step": 13446
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5517711171662126e-06,
+      "loss": 1.6945,
+      "step": 13447
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5497275204359674e-06,
+      "loss": 1.5843,
+      "step": 13448
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.547683923705722e-06,
+      "loss": 1.6236,
+      "step": 13449
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.545640326975477e-06,
+      "loss": 1.6642,
+      "step": 13450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5435967302452318e-06,
+      "loss": 1.7093,
+      "step": 13451
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.541553133514986e-06,
+      "loss": 1.7153,
+      "step": 13452
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.539509536784741e-06,
+      "loss": 1.6159,
+      "step": 13453
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.537465940054496e-06,
+      "loss": 1.7576,
+      "step": 13454
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.535422343324251e-06,
+      "loss": 1.6084,
+      "step": 13455
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5333787465940057e-06,
+      "loss": 1.6481,
+      "step": 13456
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5313351498637605e-06,
+      "loss": 1.7231,
+      "step": 13457
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5292915531335153e-06,
+      "loss": 1.7065,
+      "step": 13458
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.52724795640327e-06,
+      "loss": 1.5896,
+      "step": 13459
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.525204359673025e-06,
+      "loss": 1.6593,
+      "step": 13460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.523160762942779e-06,
+      "loss": 1.7225,
+      "step": 13461
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.521117166212534e-06,
+      "loss": 1.6744,
+      "step": 13462
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5190735694822888e-06,
+      "loss": 1.6844,
+      "step": 13463
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5170299727520435e-06,
+      "loss": 1.6345,
+      "step": 13464
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5149863760217983e-06,
+      "loss": 1.5694,
+      "step": 13465
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.512942779291553e-06,
+      "loss": 1.5997,
+      "step": 13466
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.510899182561308e-06,
+      "loss": 1.6954,
+      "step": 13467
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.508855585831063e-06,
+      "loss": 1.6672,
+      "step": 13468
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.506811989100818e-06,
+      "loss": 1.7531,
+      "step": 13469
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.5047683923705723e-06,
+      "loss": 1.6149,
+      "step": 13470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.502724795640327e-06,
+      "loss": 1.69,
+      "step": 13471
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.500681198910082e-06,
+      "loss": 1.5954,
+      "step": 13472
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4986376021798366e-06,
+      "loss": 1.6791,
+      "step": 13473
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4965940054495914e-06,
+      "loss": 1.7392,
+      "step": 13474
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.494550408719346e-06,
+      "loss": 1.8608,
+      "step": 13475
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.492506811989101e-06,
+      "loss": 1.7014,
+      "step": 13476
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4904632152588558e-06,
+      "loss": 1.6853,
+      "step": 13477
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.48841961852861e-06,
+      "loss": 1.4695,
+      "step": 13478
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.486376021798365e-06,
+      "loss": 1.6975,
+      "step": 13479
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4843324250681197e-06,
+      "loss": 1.7022,
+      "step": 13480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4822888283378745e-06,
+      "loss": 1.6326,
+      "step": 13481
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4802452316076297e-06,
+      "loss": 1.6009,
+      "step": 13482
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4782016348773845e-06,
+      "loss": 1.6596,
+      "step": 13483
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4761580381471393e-06,
+      "loss": 1.7187,
+      "step": 13484
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.474114441416894e-06,
+      "loss": 1.6025,
+      "step": 13485
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.472070844686649e-06,
+      "loss": 1.7197,
+      "step": 13486
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.470027247956403e-06,
+      "loss": 1.6427,
+      "step": 13487
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.467983651226158e-06,
+      "loss": 1.5973,
+      "step": 13488
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4659400544959128e-06,
+      "loss": 1.6383,
+      "step": 13489
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4638964577656675e-06,
+      "loss": 1.6984,
+      "step": 13490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4618528610354223e-06,
+      "loss": 1.6341,
+      "step": 13491
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.459809264305177e-06,
+      "loss": 1.7714,
+      "step": 13492
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.457765667574932e-06,
+      "loss": 1.6273,
+      "step": 13493
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4557220708446867e-06,
+      "loss": 1.6711,
+      "step": 13494
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.453678474114442e-06,
+      "loss": 1.619,
+      "step": 13495
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4516348773841963e-06,
+      "loss": 1.5938,
+      "step": 13496
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.449591280653951e-06,
+      "loss": 1.6961,
+      "step": 13497
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.447547683923706e-06,
+      "loss": 1.6034,
+      "step": 13498
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4455040871934606e-06,
+      "loss": 1.7434,
+      "step": 13499
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4434604904632154e-06,
+      "loss": 1.7257,
+      "step": 13500
+    },
+    {
+      "epoch": 2.67,
+      "eval_loss": 1.4990413188934326,
+      "eval_runtime": 720.4662,
+      "eval_samples_per_second": 5.76,
+      "eval_steps_per_second": 1.441,
+      "step": 13500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.44141689373297e-06,
+      "loss": 1.5838,
+      "step": 13501
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.439373297002725e-06,
+      "loss": 1.6581,
+      "step": 13502
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4373297002724798e-06,
+      "loss": 1.6433,
+      "step": 13503
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4352861035422345e-06,
+      "loss": 1.6642,
+      "step": 13504
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.433242506811989e-06,
+      "loss": 1.8187,
+      "step": 13505
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4311989100817437e-06,
+      "loss": 1.6023,
+      "step": 13506
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4291553133514985e-06,
+      "loss": 1.6492,
+      "step": 13507
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4271117166212533e-06,
+      "loss": 1.6411,
+      "step": 13508
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4250681198910085e-06,
+      "loss": 1.684,
+      "step": 13509
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4230245231607633e-06,
+      "loss": 1.5276,
+      "step": 13510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.420980926430518e-06,
+      "loss": 1.6778,
+      "step": 13511
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.418937329700273e-06,
+      "loss": 1.7514,
+      "step": 13512
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4168937329700276e-06,
+      "loss": 1.6808,
+      "step": 13513
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.414850136239782e-06,
+      "loss": 1.772,
+      "step": 13514
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4128065395095368e-06,
+      "loss": 1.5737,
+      "step": 13515
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4107629427792915e-06,
+      "loss": 1.6667,
+      "step": 13516
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4087193460490463e-06,
+      "loss": 1.6538,
+      "step": 13517
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.406675749318801e-06,
+      "loss": 1.6888,
+      "step": 13518
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.404632152588556e-06,
+      "loss": 1.7035,
+      "step": 13519
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4025885558583107e-06,
+      "loss": 1.568,
+      "step": 13520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4005449591280655e-06,
+      "loss": 1.7534,
+      "step": 13521
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3985013623978203e-06,
+      "loss": 1.7427,
+      "step": 13522
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.396457765667575e-06,
+      "loss": 1.5806,
+      "step": 13523
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.39441416893733e-06,
+      "loss": 1.6072,
+      "step": 13524
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3923705722070846e-06,
+      "loss": 1.5749,
+      "step": 13525
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3903269754768394e-06,
+      "loss": 1.6375,
+      "step": 13526
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.388283378746594e-06,
+      "loss": 1.6742,
+      "step": 13527
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.386239782016349e-06,
+      "loss": 1.5674,
+      "step": 13528
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3841961852861038e-06,
+      "loss": 1.6043,
+      "step": 13529
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3821525885558585e-06,
+      "loss": 1.6019,
+      "step": 13530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.380108991825613e-06,
+      "loss": 1.6654,
+      "step": 13531
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3780653950953677e-06,
+      "loss": 1.6444,
+      "step": 13532
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3760217983651225e-06,
+      "loss": 1.6566,
+      "step": 13533
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3739782016348773e-06,
+      "loss": 1.5123,
+      "step": 13534
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.371934604904632e-06,
+      "loss": 1.7028,
+      "step": 13535
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.369891008174387e-06,
+      "loss": 1.5245,
+      "step": 13536
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.367847411444142e-06,
+      "loss": 1.7596,
+      "step": 13537
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.365803814713897e-06,
+      "loss": 1.6229,
+      "step": 13538
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3637602179836516e-06,
+      "loss": 1.6642,
+      "step": 13539
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.361716621253406e-06,
+      "loss": 1.7242,
+      "step": 13540
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3596730245231608e-06,
+      "loss": 1.5663,
+      "step": 13541
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3576294277929156e-06,
+      "loss": 1.6371,
+      "step": 13542
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3555858310626703e-06,
+      "loss": 1.6277,
+      "step": 13543
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.353542234332425e-06,
+      "loss": 1.6965,
+      "step": 13544
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.35149863760218e-06,
+      "loss": 1.6412,
+      "step": 13545
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3494550408719347e-06,
+      "loss": 1.706,
+      "step": 13546
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3474114441416895e-06,
+      "loss": 1.7925,
+      "step": 13547
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3453678474114443e-06,
+      "loss": 1.6615,
+      "step": 13548
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3433242506811986e-06,
+      "loss": 1.631,
+      "step": 13549
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3412806539509534e-06,
+      "loss": 1.8376,
+      "step": 13550
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3392370572207086e-06,
+      "loss": 1.575,
+      "step": 13551
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3371934604904634e-06,
+      "loss": 1.6039,
+      "step": 13552
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.335149863760218e-06,
+      "loss": 1.8231,
+      "step": 13553
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.333106267029973e-06,
+      "loss": 1.6019,
+      "step": 13554
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3310626702997278e-06,
+      "loss": 1.7396,
+      "step": 13555
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3290190735694826e-06,
+      "loss": 1.7134,
+      "step": 13556
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3269754768392373e-06,
+      "loss": 1.7032,
+      "step": 13557
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3249318801089917e-06,
+      "loss": 1.6564,
+      "step": 13558
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3228882833787465e-06,
+      "loss": 1.7573,
+      "step": 13559
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3208446866485013e-06,
+      "loss": 1.6151,
+      "step": 13560
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.318801089918256e-06,
+      "loss": 1.646,
+      "step": 13561
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.316757493188011e-06,
+      "loss": 1.7963,
+      "step": 13562
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3147138964577656e-06,
+      "loss": 1.6284,
+      "step": 13563
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.312670299727521e-06,
+      "loss": 1.5458,
+      "step": 13564
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3106267029972756e-06,
+      "loss": 1.6535,
+      "step": 13565
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3085831062670304e-06,
+      "loss": 1.6999,
+      "step": 13566
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3065395095367848e-06,
+      "loss": 1.6134,
+      "step": 13567
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3044959128065396e-06,
+      "loss": 1.5944,
+      "step": 13568
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3024523160762943e-06,
+      "loss": 1.6337,
+      "step": 13569
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.300408719346049e-06,
+      "loss": 1.6479,
+      "step": 13570
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.298365122615804e-06,
+      "loss": 1.7823,
+      "step": 13571
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2963215258855587e-06,
+      "loss": 1.6509,
+      "step": 13572
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2942779291553135e-06,
+      "loss": 1.6164,
+      "step": 13573
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2922343324250683e-06,
+      "loss": 1.5107,
+      "step": 13574
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.290190735694823e-06,
+      "loss": 1.597,
+      "step": 13575
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2881471389645774e-06,
+      "loss": 1.6295,
+      "step": 13576
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.286103542234332e-06,
+      "loss": 1.7158,
+      "step": 13577
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2840599455040874e-06,
+      "loss": 1.6343,
+      "step": 13578
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.282016348773842e-06,
+      "loss": 1.6993,
+      "step": 13579
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.279972752043597e-06,
+      "loss": 1.6262,
+      "step": 13580
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2779291553133518e-06,
+      "loss": 1.5793,
+      "step": 13581
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2758855585831066e-06,
+      "loss": 1.6677,
+      "step": 13582
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2738419618528613e-06,
+      "loss": 1.6925,
+      "step": 13583
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2717983651226157e-06,
+      "loss": 1.7202,
+      "step": 13584
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2697547683923705e-06,
+      "loss": 1.728,
+      "step": 13585
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2677111716621253e-06,
+      "loss": 1.656,
+      "step": 13586
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.26566757493188e-06,
+      "loss": 1.5735,
+      "step": 13587
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.263623978201635e-06,
+      "loss": 1.6621,
+      "step": 13588
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2615803814713896e-06,
+      "loss": 1.6243,
+      "step": 13589
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2595367847411444e-06,
+      "loss": 1.8415,
+      "step": 13590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.257493188010899e-06,
+      "loss": 1.5958,
+      "step": 13591
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2554495912806544e-06,
+      "loss": 1.6604,
+      "step": 13592
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2534059945504088e-06,
+      "loss": 1.7357,
+      "step": 13593
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2513623978201636e-06,
+      "loss": 1.6586,
+      "step": 13594
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2493188010899183e-06,
+      "loss": 1.6837,
+      "step": 13595
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.247275204359673e-06,
+      "loss": 1.6941,
+      "step": 13596
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.245231607629428e-06,
+      "loss": 1.6763,
+      "step": 13597
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2431880108991827e-06,
+      "loss": 1.6689,
+      "step": 13598
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2411444141689375e-06,
+      "loss": 1.5683,
+      "step": 13599
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2391008174386923e-06,
+      "loss": 1.6047,
+      "step": 13600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.237057220708447e-06,
+      "loss": 1.6904,
+      "step": 13601
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2350136239782014e-06,
+      "loss": 1.7976,
+      "step": 13602
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.232970027247956e-06,
+      "loss": 1.6966,
+      "step": 13603
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.230926430517711e-06,
+      "loss": 1.7064,
+      "step": 13604
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2288828337874658e-06,
+      "loss": 1.7581,
+      "step": 13605
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.226839237057221e-06,
+      "loss": 1.7277,
+      "step": 13606
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2247956403269758e-06,
+      "loss": 1.6886,
+      "step": 13607
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2227520435967306e-06,
+      "loss": 1.5651,
+      "step": 13608
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2207084468664853e-06,
+      "loss": 1.5579,
+      "step": 13609
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.21866485013624e-06,
+      "loss": 1.7387,
+      "step": 13610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2166212534059945e-06,
+      "loss": 1.6097,
+      "step": 13611
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2145776566757493e-06,
+      "loss": 1.5241,
+      "step": 13612
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.212534059945504e-06,
+      "loss": 1.6121,
+      "step": 13613
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.210490463215259e-06,
+      "loss": 1.711,
+      "step": 13614
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2084468664850136e-06,
+      "loss": 1.6868,
+      "step": 13615
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2064032697547684e-06,
+      "loss": 1.6157,
+      "step": 13616
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.204359673024523e-06,
+      "loss": 1.6604,
+      "step": 13617
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.202316076294278e-06,
+      "loss": 1.7386,
+      "step": 13618
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2002724795640328e-06,
+      "loss": 1.6343,
+      "step": 13619
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1982288828337876e-06,
+      "loss": 1.618,
+      "step": 13620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1961852861035423e-06,
+      "loss": 1.6523,
+      "step": 13621
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.194141689373297e-06,
+      "loss": 1.5891,
+      "step": 13622
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.192098092643052e-06,
+      "loss": 1.7053,
+      "step": 13623
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1900544959128067e-06,
+      "loss": 1.6665,
+      "step": 13624
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1880108991825615e-06,
+      "loss": 1.6282,
+      "step": 13625
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1859673024523163e-06,
+      "loss": 1.7584,
+      "step": 13626
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.183923705722071e-06,
+      "loss": 1.7165,
+      "step": 13627
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1818801089918254e-06,
+      "loss": 1.7326,
+      "step": 13628
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.17983651226158e-06,
+      "loss": 1.5738,
+      "step": 13629
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.177792915531335e-06,
+      "loss": 1.6559,
+      "step": 13630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1757493188010898e-06,
+      "loss": 1.6026,
+      "step": 13631
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1737057220708446e-06,
+      "loss": 1.7178,
+      "step": 13632
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1716621253405998e-06,
+      "loss": 1.6554,
+      "step": 13633
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1696185286103546e-06,
+      "loss": 1.6137,
+      "step": 13634
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1675749318801093e-06,
+      "loss": 1.6335,
+      "step": 13635
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.165531335149864e-06,
+      "loss": 1.5955,
+      "step": 13636
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1634877384196185e-06,
+      "loss": 1.6315,
+      "step": 13637
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1614441416893733e-06,
+      "loss": 1.6862,
+      "step": 13638
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.159400544959128e-06,
+      "loss": 1.541,
+      "step": 13639
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.157356948228883e-06,
+      "loss": 1.8179,
+      "step": 13640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1553133514986376e-06,
+      "loss": 1.7764,
+      "step": 13641
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1532697547683924e-06,
+      "loss": 1.6634,
+      "step": 13642
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.151226158038147e-06,
+      "loss": 1.7604,
+      "step": 13643
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.149182561307902e-06,
+      "loss": 1.5586,
+      "step": 13644
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1471389645776568e-06,
+      "loss": 1.7306,
+      "step": 13645
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.145095367847411e-06,
+      "loss": 1.7016,
+      "step": 13646
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1430517711171664e-06,
+      "loss": 1.69,
+      "step": 13647
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.141008174386921e-06,
+      "loss": 1.6622,
+      "step": 13648
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.138964577656676e-06,
+      "loss": 1.665,
+      "step": 13649
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1369209809264307e-06,
+      "loss": 1.6178,
+      "step": 13650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1348773841961855e-06,
+      "loss": 1.554,
+      "step": 13651
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1328337874659403e-06,
+      "loss": 1.7289,
+      "step": 13652
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.130790190735695e-06,
+      "loss": 1.6252,
+      "step": 13653
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.12874659400545e-06,
+      "loss": 1.664,
+      "step": 13654
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1267029972752042e-06,
+      "loss": 1.6817,
+      "step": 13655
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.124659400544959e-06,
+      "loss": 1.6115,
+      "step": 13656
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1226158038147138e-06,
+      "loss": 1.6211,
+      "step": 13657
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1205722070844686e-06,
+      "loss": 1.7206,
+      "step": 13658
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1185286103542234e-06,
+      "loss": 1.7065,
+      "step": 13659
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.116485013623978e-06,
+      "loss": 1.6931,
+      "step": 13660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1144414168937334e-06,
+      "loss": 1.7252,
+      "step": 13661
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.112397820163488e-06,
+      "loss": 1.6514,
+      "step": 13662
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.110354223433243e-06,
+      "loss": 1.7124,
+      "step": 13663
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1083106267029973e-06,
+      "loss": 1.6233,
+      "step": 13664
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.106267029972752e-06,
+      "loss": 1.671,
+      "step": 13665
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.104223433242507e-06,
+      "loss": 1.5965,
+      "step": 13666
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1021798365122616e-06,
+      "loss": 1.717,
+      "step": 13667
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.1001362397820164e-06,
+      "loss": 1.6343,
+      "step": 13668
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0980926430517712e-06,
+      "loss": 1.5263,
+      "step": 13669
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.096049046321526e-06,
+      "loss": 1.7208,
+      "step": 13670
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0940054495912808e-06,
+      "loss": 1.741,
+      "step": 13671
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0919618528610356e-06,
+      "loss": 1.6035,
+      "step": 13672
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.08991825613079e-06,
+      "loss": 1.6939,
+      "step": 13673
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0878746594005447e-06,
+      "loss": 1.5418,
+      "step": 13674
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0858310626703e-06,
+      "loss": 1.5254,
+      "step": 13675
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0837874659400547e-06,
+      "loss": 1.7458,
+      "step": 13676
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0817438692098095e-06,
+      "loss": 1.7756,
+      "step": 13677
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0797002724795643e-06,
+      "loss": 1.7283,
+      "step": 13678
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.077656675749319e-06,
+      "loss": 1.6267,
+      "step": 13679
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.075613079019074e-06,
+      "loss": 1.8001,
+      "step": 13680
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0735694822888282e-06,
+      "loss": 1.6888,
+      "step": 13681
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.071525885558583e-06,
+      "loss": 1.6731,
+      "step": 13682
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0694822888283378e-06,
+      "loss": 1.7154,
+      "step": 13683
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0674386920980926e-06,
+      "loss": 1.7118,
+      "step": 13684
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0653950953678474e-06,
+      "loss": 1.632,
+      "step": 13685
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.063351498637602e-06,
+      "loss": 1.6491,
+      "step": 13686
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.061307901907357e-06,
+      "loss": 1.6443,
+      "step": 13687
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0592643051771117e-06,
+      "loss": 1.6288,
+      "step": 13688
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.057220708446867e-06,
+      "loss": 1.6645,
+      "step": 13689
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0551771117166213e-06,
+      "loss": 1.6088,
+      "step": 13690
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.053133514986376e-06,
+      "loss": 1.5669,
+      "step": 13691
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.051089918256131e-06,
+      "loss": 1.6707,
+      "step": 13692
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0490463215258856e-06,
+      "loss": 1.6369,
+      "step": 13693
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0470027247956404e-06,
+      "loss": 1.6987,
+      "step": 13694
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0449591280653952e-06,
+      "loss": 1.6097,
+      "step": 13695
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.04291553133515e-06,
+      "loss": 1.6712,
+      "step": 13696
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0408719346049048e-06,
+      "loss": 1.7197,
+      "step": 13697
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0388283378746596e-06,
+      "loss": 1.6253,
+      "step": 13698
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.036784741144414e-06,
+      "loss": 1.7508,
+      "step": 13699
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0347411444141687e-06,
+      "loss": 1.7567,
+      "step": 13700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0326975476839235e-06,
+      "loss": 1.5954,
+      "step": 13701
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0306539509536787e-06,
+      "loss": 1.5737,
+      "step": 13702
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0286103542234335e-06,
+      "loss": 1.5981,
+      "step": 13703
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0265667574931883e-06,
+      "loss": 1.6711,
+      "step": 13704
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.024523160762943e-06,
+      "loss": 1.5341,
+      "step": 13705
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.022479564032698e-06,
+      "loss": 1.7654,
+      "step": 13706
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0204359673024526e-06,
+      "loss": 1.6663,
+      "step": 13707
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.018392370572207e-06,
+      "loss": 1.7277,
+      "step": 13708
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.016348773841962e-06,
+      "loss": 1.6009,
+      "step": 13709
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0143051771117166e-06,
+      "loss": 1.6336,
+      "step": 13710
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0122615803814714e-06,
+      "loss": 1.7206,
+      "step": 13711
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.010217983651226e-06,
+      "loss": 1.6288,
+      "step": 13712
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.008174386920981e-06,
+      "loss": 1.6537,
+      "step": 13713
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0061307901907357e-06,
+      "loss": 1.7077,
+      "step": 13714
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0040871934604905e-06,
+      "loss": 1.6384,
+      "step": 13715
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.0020435967302457e-06,
+      "loss": 1.6332,
+      "step": 13716
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3e-06,
+      "loss": 1.5852,
+      "step": 13717
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.997956403269755e-06,
+      "loss": 1.5937,
+      "step": 13718
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9959128065395096e-06,
+      "loss": 1.7121,
+      "step": 13719
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9938692098092644e-06,
+      "loss": 1.6999,
+      "step": 13720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9918256130790192e-06,
+      "loss": 1.5626,
+      "step": 13721
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.989782016348774e-06,
+      "loss": 1.5908,
+      "step": 13722
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.987738419618529e-06,
+      "loss": 1.936,
+      "step": 13723
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9856948228882836e-06,
+      "loss": 1.5983,
+      "step": 13724
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.983651226158038e-06,
+      "loss": 1.6944,
+      "step": 13725
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9816076294277927e-06,
+      "loss": 1.6428,
+      "step": 13726
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9795640326975475e-06,
+      "loss": 1.7519,
+      "step": 13727
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9775204359673023e-06,
+      "loss": 1.6948,
+      "step": 13728
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.975476839237057e-06,
+      "loss": 1.6109,
+      "step": 13729
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9734332425068123e-06,
+      "loss": 1.6397,
+      "step": 13730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.971389645776567e-06,
+      "loss": 1.7399,
+      "step": 13731
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.969346049046322e-06,
+      "loss": 1.6301,
+      "step": 13732
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9673024523160766e-06,
+      "loss": 1.7249,
+      "step": 13733
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.965258855585831e-06,
+      "loss": 1.6381,
+      "step": 13734
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.963215258855586e-06,
+      "loss": 1.6289,
+      "step": 13735
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9611716621253406e-06,
+      "loss": 1.5692,
+      "step": 13736
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9591280653950954e-06,
+      "loss": 1.6504,
+      "step": 13737
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.95708446866485e-06,
+      "loss": 1.6541,
+      "step": 13738
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.955040871934605e-06,
+      "loss": 1.5891,
+      "step": 13739
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9529972752043597e-06,
+      "loss": 1.6558,
+      "step": 13740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9509536784741145e-06,
+      "loss": 1.7497,
+      "step": 13741
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9489100817438693e-06,
+      "loss": 1.6152,
+      "step": 13742
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9468664850136237e-06,
+      "loss": 1.5777,
+      "step": 13743
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.944822888283379e-06,
+      "loss": 1.6523,
+      "step": 13744
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9427792915531337e-06,
+      "loss": 1.7118,
+      "step": 13745
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9407356948228884e-06,
+      "loss": 1.6238,
+      "step": 13746
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9386920980926432e-06,
+      "loss": 1.6953,
+      "step": 13747
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.936648501362398e-06,
+      "loss": 1.7346,
+      "step": 13748
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.934604904632153e-06,
+      "loss": 1.709,
+      "step": 13749
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9325613079019076e-06,
+      "loss": 1.6473,
+      "step": 13750
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9305177111716624e-06,
+      "loss": 1.5773,
+      "step": 13751
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9284741144414167e-06,
+      "loss": 1.6577,
+      "step": 13752
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9264305177111715e-06,
+      "loss": 1.771,
+      "step": 13753
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9243869209809263e-06,
+      "loss": 1.7668,
+      "step": 13754
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.922343324250681e-06,
+      "loss": 1.6406,
+      "step": 13755
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.920299727520436e-06,
+      "loss": 1.787,
+      "step": 13756
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9182561307901907e-06,
+      "loss": 1.6538,
+      "step": 13757
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.916212534059946e-06,
+      "loss": 1.6838,
+      "step": 13758
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9141689373297006e-06,
+      "loss": 1.7347,
+      "step": 13759
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9121253405994554e-06,
+      "loss": 1.615,
+      "step": 13760
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.91008174386921e-06,
+      "loss": 1.7816,
+      "step": 13761
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9080381471389646e-06,
+      "loss": 1.6806,
+      "step": 13762
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9059945504087194e-06,
+      "loss": 1.7177,
+      "step": 13763
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.903950953678474e-06,
+      "loss": 1.6229,
+      "step": 13764
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.901907356948229e-06,
+      "loss": 1.6527,
+      "step": 13765
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8998637602179837e-06,
+      "loss": 1.7087,
+      "step": 13766
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8978201634877385e-06,
+      "loss": 1.619,
+      "step": 13767
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8957765667574933e-06,
+      "loss": 1.5406,
+      "step": 13768
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.893732970027248e-06,
+      "loss": 1.615,
+      "step": 13769
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8916893732970024e-06,
+      "loss": 1.6487,
+      "step": 13770
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8896457765667577e-06,
+      "loss": 1.6862,
+      "step": 13771
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8876021798365124e-06,
+      "loss": 1.6621,
+      "step": 13772
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8855585831062672e-06,
+      "loss": 1.6127,
+      "step": 13773
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.883514986376022e-06,
+      "loss": 1.6508,
+      "step": 13774
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.881471389645777e-06,
+      "loss": 1.6253,
+      "step": 13775
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8794277929155316e-06,
+      "loss": 1.6639,
+      "step": 13776
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8773841961852864e-06,
+      "loss": 1.7301,
+      "step": 13777
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8753405994550407e-06,
+      "loss": 1.7129,
+      "step": 13778
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8732970027247955e-06,
+      "loss": 1.7828,
+      "step": 13779
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8712534059945503e-06,
+      "loss": 1.7233,
+      "step": 13780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.869209809264305e-06,
+      "loss": 1.7074,
+      "step": 13781
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.86716621253406e-06,
+      "loss": 1.6284,
+      "step": 13782
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8651226158038147e-06,
+      "loss": 1.571,
+      "step": 13783
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8630790190735694e-06,
+      "loss": 1.5907,
+      "step": 13784
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8610354223433247e-06,
+      "loss": 1.7231,
+      "step": 13785
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8589918256130794e-06,
+      "loss": 1.7171,
+      "step": 13786
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.856948228882834e-06,
+      "loss": 1.7095,
+      "step": 13787
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8549046321525886e-06,
+      "loss": 1.6374,
+      "step": 13788
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8528610354223434e-06,
+      "loss": 1.6907,
+      "step": 13789
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.850817438692098e-06,
+      "loss": 1.6957,
+      "step": 13790
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.848773841961853e-06,
+      "loss": 1.8038,
+      "step": 13791
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8467302452316077e-06,
+      "loss": 1.6588,
+      "step": 13792
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8446866485013625e-06,
+      "loss": 1.4843,
+      "step": 13793
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8426430517711173e-06,
+      "loss": 1.5723,
+      "step": 13794
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.840599455040872e-06,
+      "loss": 1.6508,
+      "step": 13795
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8385558583106264e-06,
+      "loss": 1.6873,
+      "step": 13796
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8365122615803812e-06,
+      "loss": 1.6429,
+      "step": 13797
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.834468664850136e-06,
+      "loss": 1.775,
+      "step": 13798
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8324250681198912e-06,
+      "loss": 1.838,
+      "step": 13799
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.830381471389646e-06,
+      "loss": 1.6332,
+      "step": 13800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.828337874659401e-06,
+      "loss": 1.5433,
+      "step": 13801
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8262942779291556e-06,
+      "loss": 1.5253,
+      "step": 13802
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8242506811989104e-06,
+      "loss": 1.8097,
+      "step": 13803
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.822207084468665e-06,
+      "loss": 1.6015,
+      "step": 13804
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8201634877384195e-06,
+      "loss": 1.5188,
+      "step": 13805
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8181198910081743e-06,
+      "loss": 1.6941,
+      "step": 13806
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.816076294277929e-06,
+      "loss": 1.5312,
+      "step": 13807
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.814032697547684e-06,
+      "loss": 1.6629,
+      "step": 13808
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8119891008174387e-06,
+      "loss": 1.6564,
+      "step": 13809
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8099455040871934e-06,
+      "loss": 1.666,
+      "step": 13810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8079019073569482e-06,
+      "loss": 1.5976,
+      "step": 13811
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.805858310626703e-06,
+      "loss": 1.5923,
+      "step": 13812
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8038147138964582e-06,
+      "loss": 1.6666,
+      "step": 13813
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8017711171662126e-06,
+      "loss": 1.744,
+      "step": 13814
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7997275204359674e-06,
+      "loss": 1.5794,
+      "step": 13815
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.797683923705722e-06,
+      "loss": 1.6011,
+      "step": 13816
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.795640326975477e-06,
+      "loss": 1.5995,
+      "step": 13817
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7935967302452317e-06,
+      "loss": 1.5403,
+      "step": 13818
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7915531335149865e-06,
+      "loss": 1.8509,
+      "step": 13819
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7895095367847413e-06,
+      "loss": 1.5415,
+      "step": 13820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.787465940054496e-06,
+      "loss": 1.6911,
+      "step": 13821
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.785422343324251e-06,
+      "loss": 1.6789,
+      "step": 13822
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7833787465940052e-06,
+      "loss": 1.6633,
+      "step": 13823
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.78133514986376e-06,
+      "loss": 1.6102,
+      "step": 13824
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.779291553133515e-06,
+      "loss": 1.6223,
+      "step": 13825
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.77724795640327e-06,
+      "loss": 1.6322,
+      "step": 13826
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.775204359673025e-06,
+      "loss": 1.6383,
+      "step": 13827
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7731607629427796e-06,
+      "loss": 1.6157,
+      "step": 13828
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7711171662125344e-06,
+      "loss": 1.6436,
+      "step": 13829
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.769073569482289e-06,
+      "loss": 1.5865,
+      "step": 13830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7670299727520435e-06,
+      "loss": 1.8797,
+      "step": 13831
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7649863760217983e-06,
+      "loss": 1.6895,
+      "step": 13832
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.762942779291553e-06,
+      "loss": 1.6745,
+      "step": 13833
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.760899182561308e-06,
+      "loss": 1.6091,
+      "step": 13834
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7588555858310627e-06,
+      "loss": 1.582,
+      "step": 13835
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7568119891008175e-06,
+      "loss": 1.6308,
+      "step": 13836
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7547683923705722e-06,
+      "loss": 1.6182,
+      "step": 13837
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.752724795640327e-06,
+      "loss": 1.4963,
+      "step": 13838
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.750681198910082e-06,
+      "loss": 1.6382,
+      "step": 13839
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7486376021798366e-06,
+      "loss": 1.7081,
+      "step": 13840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7465940054495914e-06,
+      "loss": 1.6608,
+      "step": 13841
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.744550408719346e-06,
+      "loss": 1.8365,
+      "step": 13842
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.742506811989101e-06,
+      "loss": 1.6366,
+      "step": 13843
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7404632152588557e-06,
+      "loss": 1.7446,
+      "step": 13844
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7384196185286105e-06,
+      "loss": 1.678,
+      "step": 13845
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7363760217983653e-06,
+      "loss": 1.6745,
+      "step": 13846
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.73433242506812e-06,
+      "loss": 1.7743,
+      "step": 13847
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.732288828337875e-06,
+      "loss": 1.7845,
+      "step": 13848
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7302452316076292e-06,
+      "loss": 1.6648,
+      "step": 13849
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.728201634877384e-06,
+      "loss": 1.5995,
+      "step": 13850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.726158038147139e-06,
+      "loss": 1.6547,
+      "step": 13851
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7241144414168936e-06,
+      "loss": 1.7112,
+      "step": 13852
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7220708446866484e-06,
+      "loss": 1.7251,
+      "step": 13853
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7200272479564036e-06,
+      "loss": 1.7426,
+      "step": 13854
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7179836512261584e-06,
+      "loss": 1.5862,
+      "step": 13855
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.715940054495913e-06,
+      "loss": 1.7749,
+      "step": 13856
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.713896457765668e-06,
+      "loss": 1.5792,
+      "step": 13857
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7118528610354223e-06,
+      "loss": 1.7631,
+      "step": 13858
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.709809264305177e-06,
+      "loss": 1.6383,
+      "step": 13859
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.707765667574932e-06,
+      "loss": 1.6529,
+      "step": 13860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7057220708446867e-06,
+      "loss": 1.7273,
+      "step": 13861
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7036784741144415e-06,
+      "loss": 1.6444,
+      "step": 13862
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7016348773841962e-06,
+      "loss": 1.7832,
+      "step": 13863
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.699591280653951e-06,
+      "loss": 1.6926,
+      "step": 13864
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.697547683923706e-06,
+      "loss": 1.5505,
+      "step": 13865
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6955040871934606e-06,
+      "loss": 1.6085,
+      "step": 13866
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.693460490463215e-06,
+      "loss": 1.7097,
+      "step": 13867
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.69141689373297e-06,
+      "loss": 1.7369,
+      "step": 13868
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.689373297002725e-06,
+      "loss": 1.6406,
+      "step": 13869
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6873297002724797e-06,
+      "loss": 1.6589,
+      "step": 13870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6852861035422345e-06,
+      "loss": 1.759,
+      "step": 13871
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6832425068119893e-06,
+      "loss": 1.7731,
+      "step": 13872
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.681198910081744e-06,
+      "loss": 1.6771,
+      "step": 13873
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.679155313351499e-06,
+      "loss": 1.711,
+      "step": 13874
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6771117166212532e-06,
+      "loss": 1.6234,
+      "step": 13875
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.675068119891008e-06,
+      "loss": 1.6337,
+      "step": 13876
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.673024523160763e-06,
+      "loss": 1.6271,
+      "step": 13877
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6709809264305176e-06,
+      "loss": 1.6033,
+      "step": 13878
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6689373297002724e-06,
+      "loss": 1.7708,
+      "step": 13879
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.666893732970027e-06,
+      "loss": 1.716,
+      "step": 13880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.664850136239782e-06,
+      "loss": 1.6828,
+      "step": 13881
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.662806539509537e-06,
+      "loss": 1.6143,
+      "step": 13882
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.660762942779292e-06,
+      "loss": 1.7054,
+      "step": 13883
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6587193460490463e-06,
+      "loss": 1.7338,
+      "step": 13884
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.656675749318801e-06,
+      "loss": 1.701,
+      "step": 13885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.654632152588556e-06,
+      "loss": 1.586,
+      "step": 13886
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6525885558583107e-06,
+      "loss": 1.586,
+      "step": 13887
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6505449591280655e-06,
+      "loss": 1.7388,
+      "step": 13888
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6485013623978202e-06,
+      "loss": 1.7131,
+      "step": 13889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.646457765667575e-06,
+      "loss": 1.6489,
+      "step": 13890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.64441416893733e-06,
+      "loss": 1.6085,
+      "step": 13891
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6423705722070846e-06,
+      "loss": 1.6826,
+      "step": 13892
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.640326975476839e-06,
+      "loss": 1.723,
+      "step": 13893
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6382833787465937e-06,
+      "loss": 1.7565,
+      "step": 13894
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.636239782016349e-06,
+      "loss": 1.7706,
+      "step": 13895
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6341961852861037e-06,
+      "loss": 1.6488,
+      "step": 13896
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6321525885558585e-06,
+      "loss": 1.6516,
+      "step": 13897
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6301089918256133e-06,
+      "loss": 1.8227,
+      "step": 13898
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.628065395095368e-06,
+      "loss": 1.5123,
+      "step": 13899
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.626021798365123e-06,
+      "loss": 1.6792,
+      "step": 13900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6239782016348777e-06,
+      "loss": 1.6179,
+      "step": 13901
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.621934604904632e-06,
+      "loss": 1.7628,
+      "step": 13902
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.619891008174387e-06,
+      "loss": 1.7025,
+      "step": 13903
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6178474114441416e-06,
+      "loss": 1.9369,
+      "step": 13904
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6158038147138964e-06,
+      "loss": 1.6902,
+      "step": 13905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.613760217983651e-06,
+      "loss": 1.5772,
+      "step": 13906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.611716621253406e-06,
+      "loss": 1.6248,
+      "step": 13907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6096730245231607e-06,
+      "loss": 1.7363,
+      "step": 13908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.607629427792916e-06,
+      "loss": 1.7082,
+      "step": 13909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.6055858310626707e-06,
+      "loss": 1.7045,
+      "step": 13910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.603542234332425e-06,
+      "loss": 1.6444,
+      "step": 13911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.60149863760218e-06,
+      "loss": 1.6287,
+      "step": 13912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5994550408719347e-06,
+      "loss": 1.6659,
+      "step": 13913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5974114441416895e-06,
+      "loss": 1.7692,
+      "step": 13914
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5953678474114442e-06,
+      "loss": 1.6635,
+      "step": 13915
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.593324250681199e-06,
+      "loss": 1.6481,
+      "step": 13916
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.591280653950954e-06,
+      "loss": 1.5242,
+      "step": 13917
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5892370572207086e-06,
+      "loss": 1.7339,
+      "step": 13918
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5871934604904634e-06,
+      "loss": 1.6841,
+      "step": 13919
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5851498637602178e-06,
+      "loss": 1.662,
+      "step": 13920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5831062670299725e-06,
+      "loss": 1.638,
+      "step": 13921
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5810626702997273e-06,
+      "loss": 1.6135,
+      "step": 13922
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5790190735694825e-06,
+      "loss": 1.8355,
+      "step": 13923
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5769754768392373e-06,
+      "loss": 1.5957,
+      "step": 13924
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.574931880108992e-06,
+      "loss": 1.4986,
+      "step": 13925
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.572888283378747e-06,
+      "loss": 1.6272,
+      "step": 13926
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5708446866485017e-06,
+      "loss": 1.7269,
+      "step": 13927
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.568801089918256e-06,
+      "loss": 1.7168,
+      "step": 13928
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.566757493188011e-06,
+      "loss": 1.7477,
+      "step": 13929
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5647138964577656e-06,
+      "loss": 1.6791,
+      "step": 13930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5626702997275204e-06,
+      "loss": 1.6555,
+      "step": 13931
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.560626702997275e-06,
+      "loss": 1.6272,
+      "step": 13932
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.55858310626703e-06,
+      "loss": 1.6997,
+      "step": 13933
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5565395095367848e-06,
+      "loss": 1.6133,
+      "step": 13934
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5544959128065395e-06,
+      "loss": 1.6851,
+      "step": 13935
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5524523160762943e-06,
+      "loss": 1.7878,
+      "step": 13936
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.550408719346049e-06,
+      "loss": 1.7224,
+      "step": 13937
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.548365122615804e-06,
+      "loss": 1.6928,
+      "step": 13938
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5463215258855587e-06,
+      "loss": 1.6859,
+      "step": 13939
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5442779291553135e-06,
+      "loss": 1.7474,
+      "step": 13940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5422343324250683e-06,
+      "loss": 1.7149,
+      "step": 13941
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.540190735694823e-06,
+      "loss": 1.6666,
+      "step": 13942
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.538147138964578e-06,
+      "loss": 1.7103,
+      "step": 13943
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5361035422343326e-06,
+      "loss": 1.8397,
+      "step": 13944
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5340599455040874e-06,
+      "loss": 1.7394,
+      "step": 13945
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5320163487738418e-06,
+      "loss": 1.7401,
+      "step": 13946
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5299727520435965e-06,
+      "loss": 1.6572,
+      "step": 13947
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5279291553133513e-06,
+      "loss": 1.5975,
+      "step": 13948
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.525885558583106e-06,
+      "loss": 1.6613,
+      "step": 13949
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.523841961852861e-06,
+      "loss": 1.6502,
+      "step": 13950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.521798365122616e-06,
+      "loss": 1.7234,
+      "step": 13951
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.519754768392371e-06,
+      "loss": 1.6245,
+      "step": 13952
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5177111716621257e-06,
+      "loss": 1.4482,
+      "step": 13953
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5156675749318805e-06,
+      "loss": 1.7108,
+      "step": 13954
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.513623978201635e-06,
+      "loss": 1.7071,
+      "step": 13955
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5115803814713896e-06,
+      "loss": 1.7235,
+      "step": 13956
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5095367847411444e-06,
+      "loss": 1.6425,
+      "step": 13957
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.507493188010899e-06,
+      "loss": 1.637,
+      "step": 13958
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.505449591280654e-06,
+      "loss": 1.7412,
+      "step": 13959
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5034059945504088e-06,
+      "loss": 1.6369,
+      "step": 13960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.5013623978201635e-06,
+      "loss": 1.6576,
+      "step": 13961
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4993188010899183e-06,
+      "loss": 1.5557,
+      "step": 13962
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.497275204359673e-06,
+      "loss": 1.6244,
+      "step": 13963
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.495231607629428e-06,
+      "loss": 1.589,
+      "step": 13964
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4931880108991827e-06,
+      "loss": 1.6151,
+      "step": 13965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4911444141689375e-06,
+      "loss": 1.6642,
+      "step": 13966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4891008174386923e-06,
+      "loss": 1.8763,
+      "step": 13967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.487057220708447e-06,
+      "loss": 1.8414,
+      "step": 13968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.485013623978202e-06,
+      "loss": 1.6051,
+      "step": 13969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4829700272479566e-06,
+      "loss": 1.6931,
+      "step": 13970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4809264305177114e-06,
+      "loss": 1.6214,
+      "step": 13971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4788828337874658e-06,
+      "loss": 1.7009,
+      "step": 13972
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4768392370572205e-06,
+      "loss": 1.765,
+      "step": 13973
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4747956403269753e-06,
+      "loss": 1.6659,
+      "step": 13974
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.47275204359673e-06,
+      "loss": 1.6244,
+      "step": 13975
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.470708446866485e-06,
+      "loss": 1.5783,
+      "step": 13976
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4686648501362397e-06,
+      "loss": 1.6381,
+      "step": 13977
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.466621253405995e-06,
+      "loss": 1.7249,
+      "step": 13978
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4645776566757497e-06,
+      "loss": 1.6209,
+      "step": 13979
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4625340599455045e-06,
+      "loss": 1.7296,
+      "step": 13980
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.460490463215259e-06,
+      "loss": 1.667,
+      "step": 13981
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4584468664850136e-06,
+      "loss": 1.707,
+      "step": 13982
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4564032697547684e-06,
+      "loss": 1.6792,
+      "step": 13983
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.454359673024523e-06,
+      "loss": 1.5918,
+      "step": 13984
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.452316076294278e-06,
+      "loss": 1.5639,
+      "step": 13985
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4502724795640328e-06,
+      "loss": 1.6127,
+      "step": 13986
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4482288828337875e-06,
+      "loss": 1.6458,
+      "step": 13987
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4461852861035423e-06,
+      "loss": 1.5965,
+      "step": 13988
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.444141689373297e-06,
+      "loss": 1.6483,
+      "step": 13989
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4420980926430515e-06,
+      "loss": 1.6671,
+      "step": 13990
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4400544959128063e-06,
+      "loss": 1.6721,
+      "step": 13991
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4380108991825615e-06,
+      "loss": 1.6575,
+      "step": 13992
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4359673024523163e-06,
+      "loss": 1.8207,
+      "step": 13993
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.433923705722071e-06,
+      "loss": 1.7376,
+      "step": 13994
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.431880108991826e-06,
+      "loss": 1.6657,
+      "step": 13995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4298365122615806e-06,
+      "loss": 1.5912,
+      "step": 13996
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4277929155313354e-06,
+      "loss": 1.6588,
+      "step": 13997
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.42574931880109e-06,
+      "loss": 1.6306,
+      "step": 13998
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4237057220708445e-06,
+      "loss": 1.6023,
+      "step": 13999
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4216621253405993e-06,
+      "loss": 1.6287,
+      "step": 14000
+    },
+    {
+      "epoch": 2.77,
+      "eval_loss": 1.5016776323318481,
+      "eval_runtime": 720.5301,
+      "eval_samples_per_second": 5.76,
+      "eval_steps_per_second": 1.441,
+      "step": 14000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.419618528610354e-06,
+      "loss": 1.6075,
+      "step": 14001
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.417574931880109e-06,
+      "loss": 1.5667,
+      "step": 14002
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4155313351498637e-06,
+      "loss": 1.7147,
+      "step": 14003
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4134877384196185e-06,
+      "loss": 1.7735,
+      "step": 14004
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4114441416893733e-06,
+      "loss": 1.7884,
+      "step": 14005
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4094005449591285e-06,
+      "loss": 1.6604,
+      "step": 14006
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4073569482288833e-06,
+      "loss": 1.7435,
+      "step": 14007
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4053133514986376e-06,
+      "loss": 1.7252,
+      "step": 14008
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4032697547683924e-06,
+      "loss": 1.7454,
+      "step": 14009
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.401226158038147e-06,
+      "loss": 1.6533,
+      "step": 14010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.399182561307902e-06,
+      "loss": 1.7356,
+      "step": 14011
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3971389645776568e-06,
+      "loss": 1.8424,
+      "step": 14012
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3950953678474115e-06,
+      "loss": 1.6631,
+      "step": 14013
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3930517711171663e-06,
+      "loss": 1.592,
+      "step": 14014
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.391008174386921e-06,
+      "loss": 1.6168,
+      "step": 14015
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.388964577656676e-06,
+      "loss": 1.569,
+      "step": 14016
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3869209809264303e-06,
+      "loss": 1.6385,
+      "step": 14017
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.384877384196185e-06,
+      "loss": 1.5903,
+      "step": 14018
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.38283378746594e-06,
+      "loss": 1.6041,
+      "step": 14019
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.380790190735695e-06,
+      "loss": 1.609,
+      "step": 14020
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.37874659400545e-06,
+      "loss": 1.6093,
+      "step": 14021
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3767029972752046e-06,
+      "loss": 1.6382,
+      "step": 14022
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3746594005449594e-06,
+      "loss": 1.5988,
+      "step": 14023
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.372615803814714e-06,
+      "loss": 1.5148,
+      "step": 14024
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3705722070844686e-06,
+      "loss": 1.6239,
+      "step": 14025
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3685286103542233e-06,
+      "loss": 1.7261,
+      "step": 14026
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.366485013623978e-06,
+      "loss": 1.6432,
+      "step": 14027
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.364441416893733e-06,
+      "loss": 1.648,
+      "step": 14028
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3623978201634877e-06,
+      "loss": 1.6558,
+      "step": 14029
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3603542234332425e-06,
+      "loss": 1.6056,
+      "step": 14030
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3583106267029973e-06,
+      "loss": 1.6295,
+      "step": 14031
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.356267029972752e-06,
+      "loss": 1.68,
+      "step": 14032
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3542234332425073e-06,
+      "loss": 1.5963,
+      "step": 14033
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3521798365122616e-06,
+      "loss": 1.7343,
+      "step": 14034
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3501362397820164e-06,
+      "loss": 1.7276,
+      "step": 14035
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.348092643051771e-06,
+      "loss": 1.6963,
+      "step": 14036
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.346049046321526e-06,
+      "loss": 1.6728,
+      "step": 14037
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3440054495912808e-06,
+      "loss": 1.645,
+      "step": 14038
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3419618528610355e-06,
+      "loss": 1.589,
+      "step": 14039
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3399182561307903e-06,
+      "loss": 1.6534,
+      "step": 14040
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.337874659400545e-06,
+      "loss": 1.6312,
+      "step": 14041
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3358310626703e-06,
+      "loss": 1.6773,
+      "step": 14042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3337874659400543e-06,
+      "loss": 1.6177,
+      "step": 14043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.331743869209809e-06,
+      "loss": 1.5891,
+      "step": 14044
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.329700272479564e-06,
+      "loss": 1.6345,
+      "step": 14045
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3276566757493186e-06,
+      "loss": 1.6041,
+      "step": 14046
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.325613079019074e-06,
+      "loss": 1.6003,
+      "step": 14047
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3235694822888286e-06,
+      "loss": 1.6547,
+      "step": 14048
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3215258855585834e-06,
+      "loss": 1.6326,
+      "step": 14049
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.319482288828338e-06,
+      "loss": 1.6659,
+      "step": 14050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.317438692098093e-06,
+      "loss": 1.8049,
+      "step": 14051
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3153950953678473e-06,
+      "loss": 1.7806,
+      "step": 14052
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.313351498637602e-06,
+      "loss": 1.673,
+      "step": 14053
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.311307901907357e-06,
+      "loss": 1.6514,
+      "step": 14054
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3092643051771117e-06,
+      "loss": 1.7139,
+      "step": 14055
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3072207084468665e-06,
+      "loss": 1.7775,
+      "step": 14056
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3051771117166213e-06,
+      "loss": 1.4644,
+      "step": 14057
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.303133514986376e-06,
+      "loss": 1.7413,
+      "step": 14058
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.301089918256131e-06,
+      "loss": 1.5655,
+      "step": 14059
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2990463215258856e-06,
+      "loss": 1.5966,
+      "step": 14060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2970027247956404e-06,
+      "loss": 1.6725,
+      "step": 14061
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.294959128065395e-06,
+      "loss": 1.5326,
+      "step": 14062
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.29291553133515e-06,
+      "loss": 1.7164,
+      "step": 14063
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2908719346049048e-06,
+      "loss": 1.763,
+      "step": 14064
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2888283378746596e-06,
+      "loss": 1.6222,
+      "step": 14065
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2867847411444143e-06,
+      "loss": 1.5726,
+      "step": 14066
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.284741144414169e-06,
+      "loss": 1.7255,
+      "step": 14067
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.282697547683924e-06,
+      "loss": 1.7907,
+      "step": 14068
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2806539509536787e-06,
+      "loss": 1.5543,
+      "step": 14069
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.278610354223433e-06,
+      "loss": 1.6205,
+      "step": 14070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.276566757493188e-06,
+      "loss": 1.5818,
+      "step": 14071
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2745231607629426e-06,
+      "loss": 1.6691,
+      "step": 14072
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2724795640326974e-06,
+      "loss": 1.7024,
+      "step": 14073
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.270435967302452e-06,
+      "loss": 1.6439,
+      "step": 14074
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2683923705722074e-06,
+      "loss": 1.7595,
+      "step": 14075
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.266348773841962e-06,
+      "loss": 1.5352,
+      "step": 14076
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.264305177111717e-06,
+      "loss": 1.5921,
+      "step": 14077
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2622615803814713e-06,
+      "loss": 1.6044,
+      "step": 14078
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.260217983651226e-06,
+      "loss": 1.649,
+      "step": 14079
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.258174386920981e-06,
+      "loss": 1.664,
+      "step": 14080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2561307901907357e-06,
+      "loss": 1.6277,
+      "step": 14081
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2540871934604905e-06,
+      "loss": 1.6545,
+      "step": 14082
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2520435967302453e-06,
+      "loss": 1.663,
+      "step": 14083
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.25e-06,
+      "loss": 1.7567,
+      "step": 14084
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.247956403269755e-06,
+      "loss": 1.6535,
+      "step": 14085
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2459128065395096e-06,
+      "loss": 1.7114,
+      "step": 14086
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.243869209809264e-06,
+      "loss": 1.6498,
+      "step": 14087
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2418256130790188e-06,
+      "loss": 1.6608,
+      "step": 14088
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.239782016348774e-06,
+      "loss": 1.6129,
+      "step": 14089
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2377384196185288e-06,
+      "loss": 1.5835,
+      "step": 14090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2356948228882836e-06,
+      "loss": 1.6472,
+      "step": 14091
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2336512261580383e-06,
+      "loss": 1.5173,
+      "step": 14092
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.231607629427793e-06,
+      "loss": 1.6628,
+      "step": 14093
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.229564032697548e-06,
+      "loss": 1.6837,
+      "step": 14094
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2275204359673027e-06,
+      "loss": 1.6113,
+      "step": 14095
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.225476839237057e-06,
+      "loss": 1.6242,
+      "step": 14096
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.223433242506812e-06,
+      "loss": 1.6507,
+      "step": 14097
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2213896457765666e-06,
+      "loss": 1.6997,
+      "step": 14098
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2193460490463214e-06,
+      "loss": 1.6042,
+      "step": 14099
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.217302452316076e-06,
+      "loss": 1.7538,
+      "step": 14100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.215258855585831e-06,
+      "loss": 1.7385,
+      "step": 14101
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.213215258855586e-06,
+      "loss": 1.5559,
+      "step": 14102
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.211171662125341e-06,
+      "loss": 1.6137,
+      "step": 14103
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2091280653950958e-06,
+      "loss": 1.6976,
+      "step": 14104
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.20708446866485e-06,
+      "loss": 1.6471,
+      "step": 14105
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.205040871934605e-06,
+      "loss": 1.7016,
+      "step": 14106
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2029972752043597e-06,
+      "loss": 1.7008,
+      "step": 14107
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2009536784741145e-06,
+      "loss": 1.7337,
+      "step": 14108
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1989100817438693e-06,
+      "loss": 1.6155,
+      "step": 14109
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.196866485013624e-06,
+      "loss": 1.7074,
+      "step": 14110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.194822888283379e-06,
+      "loss": 1.6369,
+      "step": 14111
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1927792915531336e-06,
+      "loss": 1.6625,
+      "step": 14112
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1907356948228884e-06,
+      "loss": 1.5628,
+      "step": 14113
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1886920980926428e-06,
+      "loss": 1.9264,
+      "step": 14114
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1866485013623976e-06,
+      "loss": 1.6271,
+      "step": 14115
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1846049046321528e-06,
+      "loss": 1.7261,
+      "step": 14116
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1825613079019076e-06,
+      "loss": 1.5874,
+      "step": 14117
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1805177111716623e-06,
+      "loss": 1.59,
+      "step": 14118
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.178474114441417e-06,
+      "loss": 1.6974,
+      "step": 14119
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.176430517711172e-06,
+      "loss": 1.6315,
+      "step": 14120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1743869209809267e-06,
+      "loss": 1.6506,
+      "step": 14121
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.172343324250681e-06,
+      "loss": 1.7082,
+      "step": 14122
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.170299727520436e-06,
+      "loss": 1.6946,
+      "step": 14123
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1682561307901906e-06,
+      "loss": 1.6245,
+      "step": 14124
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1662125340599454e-06,
+      "loss": 1.6779,
+      "step": 14125
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1641689373297e-06,
+      "loss": 1.6804,
+      "step": 14126
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.162125340599455e-06,
+      "loss": 1.5571,
+      "step": 14127
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1600817438692098e-06,
+      "loss": 1.5739,
+      "step": 14128
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1580381471389646e-06,
+      "loss": 1.7043,
+      "step": 14129
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1559945504087198e-06,
+      "loss": 1.6218,
+      "step": 14130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.153950953678474e-06,
+      "loss": 1.6506,
+      "step": 14131
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.151907356948229e-06,
+      "loss": 1.7592,
+      "step": 14132
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1498637602179837e-06,
+      "loss": 1.748,
+      "step": 14133
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1478201634877385e-06,
+      "loss": 1.5731,
+      "step": 14134
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1457765667574933e-06,
+      "loss": 1.6172,
+      "step": 14135
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.143732970027248e-06,
+      "loss": 1.7212,
+      "step": 14136
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.141689373297003e-06,
+      "loss": 1.681,
+      "step": 14137
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1396457765667576e-06,
+      "loss": 1.5771,
+      "step": 14138
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1376021798365124e-06,
+      "loss": 1.533,
+      "step": 14139
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1355585831062668e-06,
+      "loss": 1.8152,
+      "step": 14140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1335149863760216e-06,
+      "loss": 1.5871,
+      "step": 14141
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1314713896457764e-06,
+      "loss": 1.6987,
+      "step": 14142
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.129427792915531e-06,
+      "loss": 1.5675,
+      "step": 14143
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1273841961852863e-06,
+      "loss": 1.6349,
+      "step": 14144
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.125340599455041e-06,
+      "loss": 1.6811,
+      "step": 14145
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.123297002724796e-06,
+      "loss": 1.6529,
+      "step": 14146
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1212534059945507e-06,
+      "loss": 1.7491,
+      "step": 14147
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1192098092643055e-06,
+      "loss": 1.6193,
+      "step": 14148
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.11716621253406e-06,
+      "loss": 1.511,
+      "step": 14149
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1151226158038146e-06,
+      "loss": 1.5936,
+      "step": 14150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1130790190735694e-06,
+      "loss": 1.6985,
+      "step": 14151
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.111035422343324e-06,
+      "loss": 1.5839,
+      "step": 14152
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.108991825613079e-06,
+      "loss": 1.6446,
+      "step": 14153
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1069482288828338e-06,
+      "loss": 1.5941,
+      "step": 14154
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1049046321525886e-06,
+      "loss": 1.677,
+      "step": 14155
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1028610354223434e-06,
+      "loss": 1.8511,
+      "step": 14156
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.100817438692098e-06,
+      "loss": 1.6893,
+      "step": 14157
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.098773841961853e-06,
+      "loss": 1.5786,
+      "step": 14158
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0967302452316077e-06,
+      "loss": 1.5513,
+      "step": 14159
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0946866485013625e-06,
+      "loss": 1.6245,
+      "step": 14160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0926430517711173e-06,
+      "loss": 1.7124,
+      "step": 14161
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.090599455040872e-06,
+      "loss": 1.5508,
+      "step": 14162
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.088555858310627e-06,
+      "loss": 1.5943,
+      "step": 14163
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0865122615803816e-06,
+      "loss": 1.8425,
+      "step": 14164
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0844686648501364e-06,
+      "loss": 1.5246,
+      "step": 14165
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.082425068119891e-06,
+      "loss": 1.635,
+      "step": 14166
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0803814713896456e-06,
+      "loss": 1.7878,
+      "step": 14167
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0783378746594004e-06,
+      "loss": 1.6019,
+      "step": 14168
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.076294277929155e-06,
+      "loss": 1.6985,
+      "step": 14169
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.07425068119891e-06,
+      "loss": 1.5032,
+      "step": 14170
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.072207084468665e-06,
+      "loss": 1.6232,
+      "step": 14171
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.07016348773842e-06,
+      "loss": 1.6953,
+      "step": 14172
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0681198910081747e-06,
+      "loss": 1.6657,
+      "step": 14173
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0660762942779295e-06,
+      "loss": 1.6779,
+      "step": 14174
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.064032697547684e-06,
+      "loss": 1.7696,
+      "step": 14175
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0619891008174386e-06,
+      "loss": 1.6036,
+      "step": 14176
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0599455040871934e-06,
+      "loss": 1.5726,
+      "step": 14177
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0579019073569482e-06,
+      "loss": 1.6032,
+      "step": 14178
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.055858310626703e-06,
+      "loss": 1.6277,
+      "step": 14179
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0538147138964578e-06,
+      "loss": 1.6957,
+      "step": 14180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0517711171662126e-06,
+      "loss": 1.6702,
+      "step": 14181
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0497275204359674e-06,
+      "loss": 1.6236,
+      "step": 14182
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.047683923705722e-06,
+      "loss": 1.6863,
+      "step": 14183
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0456403269754765e-06,
+      "loss": 1.5595,
+      "step": 14184
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0435967302452317e-06,
+      "loss": 1.6043,
+      "step": 14185
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0415531335149865e-06,
+      "loss": 1.6506,
+      "step": 14186
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0395095367847413e-06,
+      "loss": 1.5847,
+      "step": 14187
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.037465940054496e-06,
+      "loss": 1.6878,
+      "step": 14188
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.035422343324251e-06,
+      "loss": 1.5525,
+      "step": 14189
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0333787465940056e-06,
+      "loss": 1.6622,
+      "step": 14190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0313351498637604e-06,
+      "loss": 1.6505,
+      "step": 14191
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0292915531335152e-06,
+      "loss": 1.6229,
+      "step": 14192
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0272479564032696e-06,
+      "loss": 1.6137,
+      "step": 14193
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0252043596730244e-06,
+      "loss": 1.7492,
+      "step": 14194
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.023160762942779e-06,
+      "loss": 1.5565,
+      "step": 14195
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.021117166212534e-06,
+      "loss": 1.5855,
+      "step": 14196
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0190735694822887e-06,
+      "loss": 1.6565,
+      "step": 14197
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0170299727520435e-06,
+      "loss": 1.6608,
+      "step": 14198
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0149863760217987e-06,
+      "loss": 1.5364,
+      "step": 14199
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0129427792915535e-06,
+      "loss": 1.6471,
+      "step": 14200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0108991825613083e-06,
+      "loss": 1.6688,
+      "step": 14201
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0088555858310626e-06,
+      "loss": 1.5476,
+      "step": 14202
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0068119891008174e-06,
+      "loss": 1.6324,
+      "step": 14203
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0047683923705722e-06,
+      "loss": 1.556,
+      "step": 14204
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.002724795640327e-06,
+      "loss": 1.7632,
+      "step": 14205
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.000681198910082e-06,
+      "loss": 1.7396,
+      "step": 14206
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9986376021798366e-06,
+      "loss": 1.7494,
+      "step": 14207
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9965940054495914e-06,
+      "loss": 1.6567,
+      "step": 14208
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.994550408719346e-06,
+      "loss": 1.5523,
+      "step": 14209
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.992506811989101e-06,
+      "loss": 1.7402,
+      "step": 14210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9904632152588553e-06,
+      "loss": 1.4539,
+      "step": 14211
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.98841961852861e-06,
+      "loss": 1.6344,
+      "step": 14212
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9863760217983653e-06,
+      "loss": 1.6888,
+      "step": 14213
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.98433242506812e-06,
+      "loss": 1.7607,
+      "step": 14214
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.982288828337875e-06,
+      "loss": 1.6642,
+      "step": 14215
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9802452316076296e-06,
+      "loss": 1.7233,
+      "step": 14216
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9782016348773844e-06,
+      "loss": 1.7075,
+      "step": 14217
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9761580381471392e-06,
+      "loss": 1.6565,
+      "step": 14218
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9741144414168936e-06,
+      "loss": 1.7932,
+      "step": 14219
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9720708446866484e-06,
+      "loss": 1.7005,
+      "step": 14220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.970027247956403e-06,
+      "loss": 1.6079,
+      "step": 14221
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.967983651226158e-06,
+      "loss": 1.5621,
+      "step": 14222
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9659400544959127e-06,
+      "loss": 1.7118,
+      "step": 14223
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9638964577656675e-06,
+      "loss": 1.713,
+      "step": 14224
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9618528610354223e-06,
+      "loss": 1.7009,
+      "step": 14225
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.959809264305177e-06,
+      "loss": 1.7529,
+      "step": 14226
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9577656675749323e-06,
+      "loss": 1.6836,
+      "step": 14227
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9557220708446866e-06,
+      "loss": 1.5911,
+      "step": 14228
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9536784741144414e-06,
+      "loss": 1.5518,
+      "step": 14229
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9516348773841962e-06,
+      "loss": 1.6907,
+      "step": 14230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.949591280653951e-06,
+      "loss": 1.7196,
+      "step": 14231
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.947547683923706e-06,
+      "loss": 1.6431,
+      "step": 14232
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9455040871934606e-06,
+      "loss": 1.4945,
+      "step": 14233
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9434604904632154e-06,
+      "loss": 1.5742,
+      "step": 14234
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.94141689373297e-06,
+      "loss": 1.6037,
+      "step": 14235
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.939373297002725e-06,
+      "loss": 1.6232,
+      "step": 14236
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9373297002724793e-06,
+      "loss": 1.63,
+      "step": 14237
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.935286103542234e-06,
+      "loss": 1.6886,
+      "step": 14238
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.933242506811989e-06,
+      "loss": 1.6878,
+      "step": 14239
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.931198910081744e-06,
+      "loss": 1.6116,
+      "step": 14240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.929155313351499e-06,
+      "loss": 1.7273,
+      "step": 14241
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9271117166212536e-06,
+      "loss": 1.5975,
+      "step": 14242
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9250681198910084e-06,
+      "loss": 1.721,
+      "step": 14243
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9230245231607632e-06,
+      "loss": 1.7098,
+      "step": 14244
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.920980926430518e-06,
+      "loss": 1.6004,
+      "step": 14245
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9189373297002724e-06,
+      "loss": 1.6999,
+      "step": 14246
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.916893732970027e-06,
+      "loss": 1.7376,
+      "step": 14247
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.914850136239782e-06,
+      "loss": 1.6436,
+      "step": 14248
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9128065395095367e-06,
+      "loss": 1.6629,
+      "step": 14249
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9107629427792915e-06,
+      "loss": 1.6839,
+      "step": 14250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9087193460490463e-06,
+      "loss": 1.7213,
+      "step": 14251
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906675749318801e-06,
+      "loss": 1.6643,
+      "step": 14252
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9046321525885557e-06,
+      "loss": 1.6354,
+      "step": 14253
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9025885558583109e-06,
+      "loss": 1.6225,
+      "step": 14254
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9005449591280657e-06,
+      "loss": 1.6075,
+      "step": 14255
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8985013623978204e-06,
+      "loss": 1.6783,
+      "step": 14256
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.896457765667575e-06,
+      "loss": 1.7218,
+      "step": 14257
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8944141689373298e-06,
+      "loss": 1.6594,
+      "step": 14258
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8923705722070846e-06,
+      "loss": 1.6978,
+      "step": 14259
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8903269754768394e-06,
+      "loss": 1.6491,
+      "step": 14260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.888283378746594e-06,
+      "loss": 1.631,
+      "step": 14261
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8862397820163487e-06,
+      "loss": 1.6341,
+      "step": 14262
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8841961852861035e-06,
+      "loss": 1.7318,
+      "step": 14263
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8821525885558583e-06,
+      "loss": 1.6071,
+      "step": 14264
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.880108991825613e-06,
+      "loss": 1.7323,
+      "step": 14265
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8780653950953677e-06,
+      "loss": 1.7756,
+      "step": 14266
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8760217983651224e-06,
+      "loss": 1.689,
+      "step": 14267
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8739782016348774e-06,
+      "loss": 1.6615,
+      "step": 14268
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8719346049046322e-06,
+      "loss": 1.6236,
+      "step": 14269
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8698910081743868e-06,
+      "loss": 1.6912,
+      "step": 14270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8678474114441418e-06,
+      "loss": 1.8797,
+      "step": 14271
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8658038147138966e-06,
+      "loss": 1.7879,
+      "step": 14272
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8637602179836514e-06,
+      "loss": 1.795,
+      "step": 14273
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8617166212534062e-06,
+      "loss": 1.536,
+      "step": 14274
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8596730245231607e-06,
+      "loss": 1.7772,
+      "step": 14275
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8576294277929155e-06,
+      "loss": 1.5862,
+      "step": 14276
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8555858310626703e-06,
+      "loss": 1.6427,
+      "step": 14277
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8535422343324253e-06,
+      "loss": 1.572,
+      "step": 14278
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8514986376021799e-06,
+      "loss": 1.6507,
+      "step": 14279
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8494550408719347e-06,
+      "loss": 1.6802,
+      "step": 14280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8474114441416894e-06,
+      "loss": 1.6147,
+      "step": 14281
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8453678474114442e-06,
+      "loss": 1.6635,
+      "step": 14282
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8433242506811988e-06,
+      "loss": 1.5437,
+      "step": 14283
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8412806539509536e-06,
+      "loss": 1.6883,
+      "step": 14284
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8392370572207086e-06,
+      "loss": 1.7047,
+      "step": 14285
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8371934604904634e-06,
+      "loss": 1.6531,
+      "step": 14286
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8351498637602182e-06,
+      "loss": 1.6084,
+      "step": 14287
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8331062670299727e-06,
+      "loss": 1.5649,
+      "step": 14288
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8310626702997275e-06,
+      "loss": 1.7604,
+      "step": 14289
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8290190735694823e-06,
+      "loss": 1.6788,
+      "step": 14290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.826975476839237e-06,
+      "loss": 1.5905,
+      "step": 14291
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8249318801089919e-06,
+      "loss": 1.591,
+      "step": 14292
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8228882833787467e-06,
+      "loss": 1.6349,
+      "step": 14293
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8208446866485014e-06,
+      "loss": 1.6172,
+      "step": 14294
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8188010899182562e-06,
+      "loss": 1.7958,
+      "step": 14295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.816757493188011e-06,
+      "loss": 1.6824,
+      "step": 14296
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8147138964577656e-06,
+      "loss": 1.737,
+      "step": 14297
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8126702997275204e-06,
+      "loss": 1.6727,
+      "step": 14298
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8106267029972754e-06,
+      "loss": 1.6073,
+      "step": 14299
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8085831062670302e-06,
+      "loss": 1.6372,
+      "step": 14300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8065395095367847e-06,
+      "loss": 1.773,
+      "step": 14301
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8044959128065395e-06,
+      "loss": 1.59,
+      "step": 14302
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8024523160762943e-06,
+      "loss": 1.652,
+      "step": 14303
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.800408719346049e-06,
+      "loss": 1.5813,
+      "step": 14304
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7983651226158039e-06,
+      "loss": 1.673,
+      "step": 14305
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7963215258855587e-06,
+      "loss": 1.723,
+      "step": 14306
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7942779291553134e-06,
+      "loss": 1.739,
+      "step": 14307
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7922343324250682e-06,
+      "loss": 1.599,
+      "step": 14308
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.790190735694823e-06,
+      "loss": 1.6051,
+      "step": 14309
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7881471389645776e-06,
+      "loss": 1.7011,
+      "step": 14310
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7861035422343324e-06,
+      "loss": 1.6512,
+      "step": 14311
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7840599455040872e-06,
+      "loss": 1.8286,
+      "step": 14312
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7820163487738422e-06,
+      "loss": 1.648,
+      "step": 14313
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7799727520435967e-06,
+      "loss": 1.7291,
+      "step": 14314
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7779291553133515e-06,
+      "loss": 1.6749,
+      "step": 14315
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7758855585831063e-06,
+      "loss": 1.6515,
+      "step": 14316
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.773841961852861e-06,
+      "loss": 1.6687,
+      "step": 14317
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7717983651226159e-06,
+      "loss": 1.6818,
+      "step": 14318
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7697547683923704e-06,
+      "loss": 1.7006,
+      "step": 14319
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7677111716621254e-06,
+      "loss": 1.7099,
+      "step": 14320
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7656675749318802e-06,
+      "loss": 1.6073,
+      "step": 14321
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.763623978201635e-06,
+      "loss": 1.6231,
+      "step": 14322
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7615803814713896e-06,
+      "loss": 1.7479,
+      "step": 14323
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7595367847411444e-06,
+      "loss": 1.9163,
+      "step": 14324
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7574931880108992e-06,
+      "loss": 1.6291,
+      "step": 14325
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.755449591280654e-06,
+      "loss": 1.6262,
+      "step": 14326
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.753405994550409e-06,
+      "loss": 1.9042,
+      "step": 14327
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7513623978201635e-06,
+      "loss": 1.6517,
+      "step": 14328
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7493188010899183e-06,
+      "loss": 1.5976,
+      "step": 14329
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.747275204359673e-06,
+      "loss": 1.6798,
+      "step": 14330
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7452316076294279e-06,
+      "loss": 1.5747,
+      "step": 14331
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7431880108991825e-06,
+      "loss": 1.6134,
+      "step": 14332
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7411444141689372e-06,
+      "loss": 1.6887,
+      "step": 14333
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7391008174386922e-06,
+      "loss": 1.5772,
+      "step": 14334
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.737057220708447e-06,
+      "loss": 1.4835,
+      "step": 14335
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7350136239782016e-06,
+      "loss": 1.6909,
+      "step": 14336
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7329700272479564e-06,
+      "loss": 1.5737,
+      "step": 14337
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7309264305177112e-06,
+      "loss": 1.7518,
+      "step": 14338
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.728882833787466e-06,
+      "loss": 1.6345,
+      "step": 14339
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.726839237057221e-06,
+      "loss": 1.61,
+      "step": 14340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7247956403269755e-06,
+      "loss": 1.7372,
+      "step": 14341
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7227520435967303e-06,
+      "loss": 1.669,
+      "step": 14342
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.720708446866485e-06,
+      "loss": 1.5855,
+      "step": 14343
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7186648501362399e-06,
+      "loss": 1.5605,
+      "step": 14344
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7166212534059945e-06,
+      "loss": 1.6855,
+      "step": 14345
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7145776566757492e-06,
+      "loss": 1.6809,
+      "step": 14346
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7125340599455042e-06,
+      "loss": 1.5732,
+      "step": 14347
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.710490463215259e-06,
+      "loss": 1.6184,
+      "step": 14348
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7084468664850138e-06,
+      "loss": 1.588,
+      "step": 14349
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7064032697547684e-06,
+      "loss": 1.5695,
+      "step": 14350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7043596730245232e-06,
+      "loss": 1.6826,
+      "step": 14351
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.702316076294278e-06,
+      "loss": 1.7258,
+      "step": 14352
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7002724795640327e-06,
+      "loss": 1.6191,
+      "step": 14353
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6982288828337875e-06,
+      "loss": 1.6941,
+      "step": 14354
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6961852861035423e-06,
+      "loss": 1.6174,
+      "step": 14355
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.694141689373297e-06,
+      "loss": 1.623,
+      "step": 14356
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6920980926430519e-06,
+      "loss": 1.6275,
+      "step": 14357
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6900544959128065e-06,
+      "loss": 1.7185,
+      "step": 14358
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6880108991825612e-06,
+      "loss": 1.5444,
+      "step": 14359
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.685967302452316e-06,
+      "loss": 1.5964,
+      "step": 14360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.683923705722071e-06,
+      "loss": 1.6766,
+      "step": 14361
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6818801089918258e-06,
+      "loss": 1.6539,
+      "step": 14362
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6798365122615804e-06,
+      "loss": 1.7148,
+      "step": 14363
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6777929155313352e-06,
+      "loss": 1.706,
+      "step": 14364
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.67574931880109e-06,
+      "loss": 1.6918,
+      "step": 14365
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6737057220708447e-06,
+      "loss": 1.6826,
+      "step": 14366
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6716621253405993e-06,
+      "loss": 1.6621,
+      "step": 14367
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6696185286103543e-06,
+      "loss": 1.6091,
+      "step": 14368
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.667574931880109e-06,
+      "loss": 1.7352,
+      "step": 14369
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6655313351498639e-06,
+      "loss": 1.7492,
+      "step": 14370
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6634877384196187e-06,
+      "loss": 1.602,
+      "step": 14371
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6614441416893732e-06,
+      "loss": 1.643,
+      "step": 14372
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.659400544959128e-06,
+      "loss": 1.6236,
+      "step": 14373
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6573569482288828e-06,
+      "loss": 1.6692,
+      "step": 14374
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6553133514986378e-06,
+      "loss": 1.5089,
+      "step": 14375
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6532697547683924e-06,
+      "loss": 1.5403,
+      "step": 14376
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6512261580381472e-06,
+      "loss": 1.6895,
+      "step": 14377
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.649182561307902e-06,
+      "loss": 1.7156,
+      "step": 14378
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6471389645776567e-06,
+      "loss": 1.6071,
+      "step": 14379
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6450953678474115e-06,
+      "loss": 1.7306,
+      "step": 14380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.643051771117166e-06,
+      "loss": 1.6446,
+      "step": 14381
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.641008174386921e-06,
+      "loss": 1.6148,
+      "step": 14382
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6389645776566759e-06,
+      "loss": 1.9295,
+      "step": 14383
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6369209809264307e-06,
+      "loss": 1.7384,
+      "step": 14384
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6348773841961852e-06,
+      "loss": 1.6218,
+      "step": 14385
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.63283378746594e-06,
+      "loss": 1.6924,
+      "step": 14386
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6307901907356948e-06,
+      "loss": 1.5768,
+      "step": 14387
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6287465940054496e-06,
+      "loss": 1.7606,
+      "step": 14388
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6267029972752044e-06,
+      "loss": 1.7098,
+      "step": 14389
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6246594005449592e-06,
+      "loss": 1.6127,
+      "step": 14390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.622615803814714e-06,
+      "loss": 1.6117,
+      "step": 14391
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6205722070844687e-06,
+      "loss": 1.637,
+      "step": 14392
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6185286103542235e-06,
+      "loss": 1.7086,
+      "step": 14393
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.616485013623978e-06,
+      "loss": 1.8085,
+      "step": 14394
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6144414168937329e-06,
+      "loss": 1.6991,
+      "step": 14395
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6123978201634879e-06,
+      "loss": 1.7791,
+      "step": 14396
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6103542234332427e-06,
+      "loss": 1.7078,
+      "step": 14397
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6083106267029972e-06,
+      "loss": 1.7094,
+      "step": 14398
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.606267029972752e-06,
+      "loss": 1.7087,
+      "step": 14399
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6042234332425068e-06,
+      "loss": 1.6953,
+      "step": 14400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6021798365122616e-06,
+      "loss": 1.7264,
+      "step": 14401
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6001362397820164e-06,
+      "loss": 1.5278,
+      "step": 14402
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5980926430517712e-06,
+      "loss": 1.6731,
+      "step": 14403
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.596049046321526e-06,
+      "loss": 1.6957,
+      "step": 14404
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5940054495912807e-06,
+      "loss": 1.7435,
+      "step": 14405
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5919618528610355e-06,
+      "loss": 1.8653,
+      "step": 14406
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.58991825613079e-06,
+      "loss": 1.6567,
+      "step": 14407
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5878746594005449e-06,
+      "loss": 1.6317,
+      "step": 14408
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5858310626702999e-06,
+      "loss": 1.6737,
+      "step": 14409
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5837874659400547e-06,
+      "loss": 1.5081,
+      "step": 14410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5817438692098092e-06,
+      "loss": 1.7454,
+      "step": 14411
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.579700272479564e-06,
+      "loss": 1.7332,
+      "step": 14412
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5776566757493188e-06,
+      "loss": 1.5631,
+      "step": 14413
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5756130790190736e-06,
+      "loss": 1.6642,
+      "step": 14414
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5735694822888284e-06,
+      "loss": 1.7221,
+      "step": 14415
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5715258855585832e-06,
+      "loss": 1.6643,
+      "step": 14416
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.569482288828338e-06,
+      "loss": 1.6907,
+      "step": 14417
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5674386920980927e-06,
+      "loss": 1.5508,
+      "step": 14418
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5653950953678475e-06,
+      "loss": 1.6041,
+      "step": 14419
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5633514986376021e-06,
+      "loss": 1.6644,
+      "step": 14420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5613079019073569e-06,
+      "loss": 1.5955,
+      "step": 14421
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5592643051771117e-06,
+      "loss": 1.7846,
+      "step": 14422
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5572207084468667e-06,
+      "loss": 1.6822,
+      "step": 14423
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5551771117166215e-06,
+      "loss": 1.6688,
+      "step": 14424
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.553133514986376e-06,
+      "loss": 1.5333,
+      "step": 14425
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5510899182561308e-06,
+      "loss": 1.669,
+      "step": 14426
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5490463215258856e-06,
+      "loss": 1.7211,
+      "step": 14427
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5470027247956404e-06,
+      "loss": 1.7773,
+      "step": 14428
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.544959128065395e-06,
+      "loss": 1.5119,
+      "step": 14429
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.54291553133515e-06,
+      "loss": 1.6734,
+      "step": 14430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5408719346049047e-06,
+      "loss": 1.5071,
+      "step": 14431
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5388283378746595e-06,
+      "loss": 1.744,
+      "step": 14432
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5367847411444141e-06,
+      "loss": 1.5715,
+      "step": 14433
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5347411444141689e-06,
+      "loss": 1.6255,
+      "step": 14434
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5326975476839237e-06,
+      "loss": 1.8179,
+      "step": 14435
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5306539509536785e-06,
+      "loss": 1.6584,
+      "step": 14436
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5286103542234335e-06,
+      "loss": 1.6358,
+      "step": 14437
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.526566757493188e-06,
+      "loss": 1.7159,
+      "step": 14438
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5245231607629428e-06,
+      "loss": 1.567,
+      "step": 14439
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5224795640326976e-06,
+      "loss": 1.638,
+      "step": 14440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5204359673024524e-06,
+      "loss": 1.5963,
+      "step": 14441
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.518392370572207e-06,
+      "loss": 1.6332,
+      "step": 14442
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5163487738419618e-06,
+      "loss": 1.7132,
+      "step": 14443
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5143051771117168e-06,
+      "loss": 1.6383,
+      "step": 14444
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5122615803814715e-06,
+      "loss": 1.555,
+      "step": 14445
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5102179836512263e-06,
+      "loss": 1.6905,
+      "step": 14446
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.508174386920981e-06,
+      "loss": 1.6659,
+      "step": 14447
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5061307901907357e-06,
+      "loss": 1.6593,
+      "step": 14448
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5040871934604905e-06,
+      "loss": 1.7128,
+      "step": 14449
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5020435967302453e-06,
+      "loss": 1.6698,
+      "step": 14450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5e-06,
+      "loss": 1.6523,
+      "step": 14451
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4979564032697548e-06,
+      "loss": 1.7303,
+      "step": 14452
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4959128065395096e-06,
+      "loss": 1.5715,
+      "step": 14453
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4938692098092644e-06,
+      "loss": 1.7319,
+      "step": 14454
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.491825613079019e-06,
+      "loss": 1.5567,
+      "step": 14455
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4897820163487738e-06,
+      "loss": 1.709,
+      "step": 14456
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4877384196185285e-06,
+      "loss": 1.6627,
+      "step": 14457
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4856948228882835e-06,
+      "loss": 1.6577,
+      "step": 14458
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4836512261580383e-06,
+      "loss": 1.7413,
+      "step": 14459
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.481607629427793e-06,
+      "loss": 1.6221,
+      "step": 14460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4795640326975477e-06,
+      "loss": 1.6056,
+      "step": 14461
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4775204359673025e-06,
+      "loss": 1.5851,
+      "step": 14462
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4754768392370573e-06,
+      "loss": 1.6605,
+      "step": 14463
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4734332425068118e-06,
+      "loss": 1.7025,
+      "step": 14464
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4713896457765668e-06,
+      "loss": 1.6871,
+      "step": 14465
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4693460490463216e-06,
+      "loss": 1.6387,
+      "step": 14466
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4673024523160764e-06,
+      "loss": 1.5624,
+      "step": 14467
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4652588555858312e-06,
+      "loss": 1.7454,
+      "step": 14468
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4632152588555858e-06,
+      "loss": 1.5643,
+      "step": 14469
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4611716621253405e-06,
+      "loss": 1.5069,
+      "step": 14470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4591280653950953e-06,
+      "loss": 1.6782,
+      "step": 14471
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4570844686648503e-06,
+      "loss": 1.7204,
+      "step": 14472
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.455040871934605e-06,
+      "loss": 1.5907,
+      "step": 14473
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4529972752043597e-06,
+      "loss": 1.7227,
+      "step": 14474
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4509536784741145e-06,
+      "loss": 1.7076,
+      "step": 14475
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4489100817438693e-06,
+      "loss": 1.6998,
+      "step": 14476
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.446866485013624e-06,
+      "loss": 1.6044,
+      "step": 14477
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4448228882833788e-06,
+      "loss": 1.6408,
+      "step": 14478
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4427792915531336e-06,
+      "loss": 1.6809,
+      "step": 14479
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4407356948228884e-06,
+      "loss": 1.7366,
+      "step": 14480
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4386920980926432e-06,
+      "loss": 1.6819,
+      "step": 14481
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4366485013623978e-06,
+      "loss": 1.5212,
+      "step": 14482
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4346049046321525e-06,
+      "loss": 1.7076,
+      "step": 14483
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4325613079019073e-06,
+      "loss": 1.7372,
+      "step": 14484
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4305177111716623e-06,
+      "loss": 1.5629,
+      "step": 14485
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.428474114441417e-06,
+      "loss": 1.6233,
+      "step": 14486
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4264305177111717e-06,
+      "loss": 1.614,
+      "step": 14487
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4243869209809265e-06,
+      "loss": 1.7721,
+      "step": 14488
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4223433242506813e-06,
+      "loss": 1.6832,
+      "step": 14489
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.420299727520436e-06,
+      "loss": 1.7083,
+      "step": 14490
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4182561307901906e-06,
+      "loss": 1.6619,
+      "step": 14491
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4162125340599456e-06,
+      "loss": 1.6359,
+      "step": 14492
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4141689373297004e-06,
+      "loss": 1.805,
+      "step": 14493
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4121253405994552e-06,
+      "loss": 1.7048,
+      "step": 14494
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4100817438692098e-06,
+      "loss": 1.7538,
+      "step": 14495
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4080381471389645e-06,
+      "loss": 1.8905,
+      "step": 14496
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4059945504087193e-06,
+      "loss": 1.8166,
+      "step": 14497
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4039509536784741e-06,
+      "loss": 1.5324,
+      "step": 14498
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4019073569482291e-06,
+      "loss": 1.5972,
+      "step": 14499
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3998637602179837e-06,
+      "loss": 1.7737,
+      "step": 14500
+    },
+    {
+      "epoch": 2.87,
+      "eval_loss": 1.4983410835266113,
+      "eval_runtime": 720.2447,
+      "eval_samples_per_second": 5.762,
+      "eval_steps_per_second": 1.441,
+      "step": 14500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3978201634877385e-06,
+      "loss": 1.5401,
+      "step": 14501
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3957765667574933e-06,
+      "loss": 1.8833,
+      "step": 14502
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.393732970027248e-06,
+      "loss": 1.6414,
+      "step": 14503
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3916893732970026e-06,
+      "loss": 1.6351,
+      "step": 14504
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3896457765667574e-06,
+      "loss": 1.6769,
+      "step": 14505
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3876021798365124e-06,
+      "loss": 1.7317,
+      "step": 14506
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3855585831062672e-06,
+      "loss": 1.7221,
+      "step": 14507
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3835149863760218e-06,
+      "loss": 1.5725,
+      "step": 14508
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3814713896457765e-06,
+      "loss": 1.6493,
+      "step": 14509
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3794277929155313e-06,
+      "loss": 1.6701,
+      "step": 14510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3773841961852861e-06,
+      "loss": 1.5768,
+      "step": 14511
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.375340599455041e-06,
+      "loss": 1.6036,
+      "step": 14512
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3732970027247957e-06,
+      "loss": 1.8309,
+      "step": 14513
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3712534059945505e-06,
+      "loss": 1.8547,
+      "step": 14514
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3692098092643053e-06,
+      "loss": 1.8548,
+      "step": 14515
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.36716621253406e-06,
+      "loss": 1.6523,
+      "step": 14516
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3651226158038146e-06,
+      "loss": 1.6463,
+      "step": 14517
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3651226158038146e-06,
+      "loss": 1.6121,
+      "step": 14518
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3630790190735694e-06,
+      "loss": 1.6055,
+      "step": 14519
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3610354223433242e-06,
+      "loss": 1.7841,
+      "step": 14520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3589918256130792e-06,
+      "loss": 1.6225,
+      "step": 14521
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.356948228882834e-06,
+      "loss": 1.5743,
+      "step": 14522
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3549046321525885e-06,
+      "loss": 1.6623,
+      "step": 14523
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3528610354223433e-06,
+      "loss": 1.73,
+      "step": 14524
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3508174386920981e-06,
+      "loss": 1.5498,
+      "step": 14525
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.348773841961853e-06,
+      "loss": 1.6336,
+      "step": 14526
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3467302452316075e-06,
+      "loss": 1.6915,
+      "step": 14527
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3446866485013625e-06,
+      "loss": 1.7261,
+      "step": 14528
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3426430517711173e-06,
+      "loss": 1.7697,
+      "step": 14529
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.340599455040872e-06,
+      "loss": 1.7871,
+      "step": 14530
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3385558583106266e-06,
+      "loss": 1.6041,
+      "step": 14531
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3365122615803814e-06,
+      "loss": 1.6295,
+      "step": 14532
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3344686648501362e-06,
+      "loss": 1.6097,
+      "step": 14533
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.332425068119891e-06,
+      "loss": 1.8647,
+      "step": 14534
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.330381471389646e-06,
+      "loss": 1.6312,
+      "step": 14535
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3283378746594006e-06,
+      "loss": 1.6865,
+      "step": 14536
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3262942779291553e-06,
+      "loss": 1.8272,
+      "step": 14537
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3242506811989101e-06,
+      "loss": 1.7067,
+      "step": 14538
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.322207084468665e-06,
+      "loss": 1.6987,
+      "step": 14539
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3201634877384195e-06,
+      "loss": 1.8044,
+      "step": 14540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3181198910081745e-06,
+      "loss": 1.7014,
+      "step": 14541
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3160762942779293e-06,
+      "loss": 1.6234,
+      "step": 14542
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.314032697547684e-06,
+      "loss": 1.6566,
+      "step": 14543
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3119891008174388e-06,
+      "loss": 1.6435,
+      "step": 14544
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3099455040871934e-06,
+      "loss": 1.7307,
+      "step": 14545
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3079019073569482e-06,
+      "loss": 1.5407,
+      "step": 14546
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.305858310626703e-06,
+      "loss": 1.6531,
+      "step": 14547
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.303814713896458e-06,
+      "loss": 1.9044,
+      "step": 14548
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3017711171662126e-06,
+      "loss": 1.6408,
+      "step": 14549
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2997275204359673e-06,
+      "loss": 1.7242,
+      "step": 14550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2976839237057221e-06,
+      "loss": 1.8337,
+      "step": 14551
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.295640326975477e-06,
+      "loss": 1.6644,
+      "step": 14552
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2935967302452317e-06,
+      "loss": 1.6842,
+      "step": 14553
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2915531335149863e-06,
+      "loss": 1.6783,
+      "step": 14554
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2895095367847413e-06,
+      "loss": 1.6726,
+      "step": 14555
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.287465940054496e-06,
+      "loss": 1.803,
+      "step": 14556
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2854223433242508e-06,
+      "loss": 1.6358,
+      "step": 14557
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2833787465940054e-06,
+      "loss": 1.6337,
+      "step": 14558
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2813351498637602e-06,
+      "loss": 1.6771,
+      "step": 14559
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.279291553133515e-06,
+      "loss": 1.6855,
+      "step": 14560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2772479564032698e-06,
+      "loss": 1.7124,
+      "step": 14561
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2752043596730246e-06,
+      "loss": 1.6519,
+      "step": 14562
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2731607629427793e-06,
+      "loss": 1.6678,
+      "step": 14563
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2711171662125341e-06,
+      "loss": 1.543,
+      "step": 14564
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.269073569482289e-06,
+      "loss": 1.6322,
+      "step": 14565
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2670299727520437e-06,
+      "loss": 1.6253,
+      "step": 14566
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2649863760217983e-06,
+      "loss": 1.621,
+      "step": 14567
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.262942779291553e-06,
+      "loss": 1.6513,
+      "step": 14568
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.260899182561308e-06,
+      "loss": 1.6987,
+      "step": 14569
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2588555858310628e-06,
+      "loss": 1.5359,
+      "step": 14570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2568119891008174e-06,
+      "loss": 1.5827,
+      "step": 14571
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2547683923705722e-06,
+      "loss": 1.6468,
+      "step": 14572
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.252724795640327e-06,
+      "loss": 1.5566,
+      "step": 14573
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2506811989100818e-06,
+      "loss": 1.7258,
+      "step": 14574
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2486376021798366e-06,
+      "loss": 1.5413,
+      "step": 14575
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2465940054495913e-06,
+      "loss": 1.6144,
+      "step": 14576
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2445504087193461e-06,
+      "loss": 1.758,
+      "step": 14577
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.242506811989101e-06,
+      "loss": 1.6085,
+      "step": 14578
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2404632152588557e-06,
+      "loss": 1.6882,
+      "step": 14579
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2384196185286103e-06,
+      "loss": 1.6966,
+      "step": 14580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.236376021798365e-06,
+      "loss": 1.7131,
+      "step": 14581
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2343324250681198e-06,
+      "loss": 1.5321,
+      "step": 14582
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2322888283378748e-06,
+      "loss": 1.658,
+      "step": 14583
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2302452316076294e-06,
+      "loss": 1.7317,
+      "step": 14584
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2282016348773842e-06,
+      "loss": 1.5293,
+      "step": 14585
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.226158038147139e-06,
+      "loss": 1.6586,
+      "step": 14586
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2241144414168938e-06,
+      "loss": 1.6392,
+      "step": 14587
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2220708446866486e-06,
+      "loss": 1.6692,
+      "step": 14588
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2200272479564031e-06,
+      "loss": 1.5377,
+      "step": 14589
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2179836512261581e-06,
+      "loss": 1.6897,
+      "step": 14590
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.215940054495913e-06,
+      "loss": 1.6463,
+      "step": 14591
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2138964577656677e-06,
+      "loss": 1.6735,
+      "step": 14592
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2118528610354223e-06,
+      "loss": 1.6588,
+      "step": 14593
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.209809264305177e-06,
+      "loss": 1.623,
+      "step": 14594
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2077656675749318e-06,
+      "loss": 1.6246,
+      "step": 14595
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2057220708446866e-06,
+      "loss": 1.6394,
+      "step": 14596
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2036784741144416e-06,
+      "loss": 1.6688,
+      "step": 14597
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2016348773841962e-06,
+      "loss": 1.694,
+      "step": 14598
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.199591280653951e-06,
+      "loss": 1.5719,
+      "step": 14599
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1975476839237058e-06,
+      "loss": 1.8103,
+      "step": 14600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1955040871934606e-06,
+      "loss": 1.6881,
+      "step": 14601
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1934604904632151e-06,
+      "loss": 1.6568,
+      "step": 14602
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.19141689373297e-06,
+      "loss": 1.6638,
+      "step": 14603
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.189373297002725e-06,
+      "loss": 1.6556,
+      "step": 14604
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1873297002724797e-06,
+      "loss": 1.7254,
+      "step": 14605
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1852861035422343e-06,
+      "loss": 1.7834,
+      "step": 14606
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.183242506811989e-06,
+      "loss": 1.6867,
+      "step": 14607
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1811989100817438e-06,
+      "loss": 1.6188,
+      "step": 14608
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1791553133514986e-06,
+      "loss": 1.7747,
+      "step": 14609
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1771117166212536e-06,
+      "loss": 1.5265,
+      "step": 14610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1750681198910082e-06,
+      "loss": 1.61,
+      "step": 14611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.173024523160763e-06,
+      "loss": 1.6226,
+      "step": 14612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1709809264305178e-06,
+      "loss": 1.6294,
+      "step": 14613
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1689373297002726e-06,
+      "loss": 1.6662,
+      "step": 14614
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1668937329700271e-06,
+      "loss": 1.6388,
+      "step": 14615
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.164850136239782e-06,
+      "loss": 1.5782,
+      "step": 14616
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.162806539509537e-06,
+      "loss": 1.5759,
+      "step": 14617
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1607629427792917e-06,
+      "loss": 1.8485,
+      "step": 14618
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1587193460490465e-06,
+      "loss": 1.6753,
+      "step": 14619
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.156675749318801e-06,
+      "loss": 1.7405,
+      "step": 14620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1546321525885558e-06,
+      "loss": 1.5243,
+      "step": 14621
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1525885558583106e-06,
+      "loss": 1.5529,
+      "step": 14622
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1505449591280654e-06,
+      "loss": 1.7718,
+      "step": 14623
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1485013623978202e-06,
+      "loss": 1.5963,
+      "step": 14624
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.146457765667575e-06,
+      "loss": 1.6498,
+      "step": 14625
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1444141689373298e-06,
+      "loss": 1.7503,
+      "step": 14626
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1423705722070846e-06,
+      "loss": 1.7274,
+      "step": 14627
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1403269754768393e-06,
+      "loss": 1.5573,
+      "step": 14628
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.138283378746594e-06,
+      "loss": 1.6172,
+      "step": 14629
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1362397820163487e-06,
+      "loss": 1.6574,
+      "step": 14630
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1341961852861037e-06,
+      "loss": 1.595,
+      "step": 14631
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1321525885558585e-06,
+      "loss": 1.9045,
+      "step": 14632
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.130108991825613e-06,
+      "loss": 1.7017,
+      "step": 14633
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1280653950953679e-06,
+      "loss": 1.7086,
+      "step": 14634
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1260217983651226e-06,
+      "loss": 1.9049,
+      "step": 14635
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1239782016348774e-06,
+      "loss": 1.6521,
+      "step": 14636
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.121934604904632e-06,
+      "loss": 1.6341,
+      "step": 14637
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.119891008174387e-06,
+      "loss": 1.6512,
+      "step": 14638
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1178474114441418e-06,
+      "loss": 1.6601,
+      "step": 14639
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1158038147138966e-06,
+      "loss": 1.7486,
+      "step": 14640
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1137602179836513e-06,
+      "loss": 1.6904,
+      "step": 14641
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.111716621253406e-06,
+      "loss": 1.628,
+      "step": 14642
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1096730245231607e-06,
+      "loss": 1.6523,
+      "step": 14643
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1076294277929155e-06,
+      "loss": 1.6214,
+      "step": 14644
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1055858310626705e-06,
+      "loss": 1.525,
+      "step": 14645
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.103542234332425e-06,
+      "loss": 1.6086,
+      "step": 14646
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1014986376021799e-06,
+      "loss": 1.657,
+      "step": 14647
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0994550408719346e-06,
+      "loss": 1.6272,
+      "step": 14648
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0974114441416894e-06,
+      "loss": 1.6935,
+      "step": 14649
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0953678474114442e-06,
+      "loss": 1.7959,
+      "step": 14650
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0933242506811988e-06,
+      "loss": 1.5292,
+      "step": 14651
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0912806539509538e-06,
+      "loss": 1.6166,
+      "step": 14652
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0892370572207086e-06,
+      "loss": 1.6599,
+      "step": 14653
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0871934604904634e-06,
+      "loss": 1.5739,
+      "step": 14654
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.085149863760218e-06,
+      "loss": 1.6005,
+      "step": 14655
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0831062670299727e-06,
+      "loss": 1.6421,
+      "step": 14656
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0810626702997275e-06,
+      "loss": 1.5165,
+      "step": 14657
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0790190735694823e-06,
+      "loss": 1.6891,
+      "step": 14658
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.076975476839237e-06,
+      "loss": 1.6247,
+      "step": 14659
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0749318801089919e-06,
+      "loss": 1.5988,
+      "step": 14660
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0728882833787466e-06,
+      "loss": 1.801,
+      "step": 14661
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0708446866485014e-06,
+      "loss": 1.6221,
+      "step": 14662
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0688010899182562e-06,
+      "loss": 1.4866,
+      "step": 14663
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0667574931880108e-06,
+      "loss": 1.6201,
+      "step": 14664
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0647138964577656e-06,
+      "loss": 1.5974,
+      "step": 14665
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0626702997275206e-06,
+      "loss": 1.69,
+      "step": 14666
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0606267029972754e-06,
+      "loss": 1.7216,
+      "step": 14667
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.05858310626703e-06,
+      "loss": 1.6576,
+      "step": 14668
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0565395095367847e-06,
+      "loss": 1.7634,
+      "step": 14669
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0544959128065395e-06,
+      "loss": 1.7179,
+      "step": 14670
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0524523160762943e-06,
+      "loss": 1.7337,
+      "step": 14671
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.050408719346049e-06,
+      "loss": 1.669,
+      "step": 14672
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0483651226158039e-06,
+      "loss": 1.6822,
+      "step": 14673
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0463215258855586e-06,
+      "loss": 1.7286,
+      "step": 14674
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0442779291553134e-06,
+      "loss": 1.7011,
+      "step": 14675
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0422343324250682e-06,
+      "loss": 1.8269,
+      "step": 14676
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0401907356948228e-06,
+      "loss": 1.7255,
+      "step": 14677
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0381471389645776e-06,
+      "loss": 1.6353,
+      "step": 14678
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0361035422343326e-06,
+      "loss": 1.6891,
+      "step": 14679
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0340599455040874e-06,
+      "loss": 1.6176,
+      "step": 14680
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.032016348773842e-06,
+      "loss": 1.7212,
+      "step": 14681
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0299727520435967e-06,
+      "loss": 1.727,
+      "step": 14682
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0279291553133515e-06,
+      "loss": 1.7157,
+      "step": 14683
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0258855585831063e-06,
+      "loss": 1.5576,
+      "step": 14684
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.023841961852861e-06,
+      "loss": 1.6283,
+      "step": 14685
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0217983651226159e-06,
+      "loss": 1.7228,
+      "step": 14686
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0197547683923706e-06,
+      "loss": 1.6058,
+      "step": 14687
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0177111716621254e-06,
+      "loss": 1.5523,
+      "step": 14688
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0156675749318802e-06,
+      "loss": 1.6277,
+      "step": 14689
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0136239782016348e-06,
+      "loss": 1.6805,
+      "step": 14690
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0115803814713896e-06,
+      "loss": 1.6496,
+      "step": 14691
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0095367847411444e-06,
+      "loss": 1.6057,
+      "step": 14692
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0074931880108994e-06,
+      "loss": 1.7983,
+      "step": 14693
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0054495912806541e-06,
+      "loss": 1.7799,
+      "step": 14694
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0034059945504087e-06,
+      "loss": 1.6506,
+      "step": 14695
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0013623978201635e-06,
+      "loss": 1.631,
+      "step": 14696
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.993188010899183e-07,
+      "loss": 1.7547,
+      "step": 14697
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.97275204359673e-07,
+      "loss": 1.7328,
+      "step": 14698
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.952316076294276e-07,
+      "loss": 1.6864,
+      "step": 14699
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.931880108991826e-07,
+      "loss": 1.6801,
+      "step": 14700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.911444141689374e-07,
+      "loss": 1.5323,
+      "step": 14701
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.891008174386922e-07,
+      "loss": 1.5523,
+      "step": 14702
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.870572207084468e-07,
+      "loss": 1.742,
+      "step": 14703
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.850136239782016e-07,
+      "loss": 1.6942,
+      "step": 14704
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.829700272479564e-07,
+      "loss": 1.6889,
+      "step": 14705
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.809264305177111e-07,
+      "loss": 1.707,
+      "step": 14706
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.788828337874661e-07,
+      "loss": 1.6891,
+      "step": 14707
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.768392370572207e-07,
+      "loss": 1.7119,
+      "step": 14708
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.747956403269755e-07,
+      "loss": 1.7912,
+      "step": 14709
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.727520435967303e-07,
+      "loss": 1.62,
+      "step": 14710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.70708446866485e-07,
+      "loss": 1.6707,
+      "step": 14711
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.686648501362396e-07,
+      "loss": 1.6004,
+      "step": 14712
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.666212534059944e-07,
+      "loss": 1.7489,
+      "step": 14713
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.645776566757494e-07,
+      "loss": 1.75,
+      "step": 14714
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.625340599455042e-07,
+      "loss": 1.7708,
+      "step": 14715
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.60490463215259e-07,
+      "loss": 1.5882,
+      "step": 14716
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.584468664850136e-07,
+      "loss": 1.6174,
+      "step": 14717
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.564032697547684e-07,
+      "loss": 1.6597,
+      "step": 14718
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.543596730245231e-07,
+      "loss": 1.7225,
+      "step": 14719
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.523160762942778e-07,
+      "loss": 1.6766,
+      "step": 14720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.502724795640328e-07,
+      "loss": 1.7049,
+      "step": 14721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.482288828337875e-07,
+      "loss": 1.7016,
+      "step": 14722
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.461852861035423e-07,
+      "loss": 1.6649,
+      "step": 14723
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.44141689373297e-07,
+      "loss": 1.5699,
+      "step": 14724
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.420980926430518e-07,
+      "loss": 1.6278,
+      "step": 14725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.400544959128065e-07,
+      "loss": 1.5548,
+      "step": 14726
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.380108991825612e-07,
+      "loss": 1.6488,
+      "step": 14727
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.359673024523161e-07,
+      "loss": 1.6168,
+      "step": 14728
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.339237057220709e-07,
+      "loss": 1.8161,
+      "step": 14729
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.318801089918257e-07,
+      "loss": 1.7074,
+      "step": 14730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.298365122615804e-07,
+      "loss": 1.75,
+      "step": 14731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.277929155313351e-07,
+      "loss": 1.5698,
+      "step": 14732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.257493188010899e-07,
+      "loss": 1.6288,
+      "step": 14733
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.237057220708447e-07,
+      "loss": 1.622,
+      "step": 14734
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.216621253405994e-07,
+      "loss": 1.7799,
+      "step": 14735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.196185286103543e-07,
+      "loss": 1.85,
+      "step": 14736
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.175749318801091e-07,
+      "loss": 1.6416,
+      "step": 14737
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.155313351498638e-07,
+      "loss": 1.7758,
+      "step": 14738
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.134877384196185e-07,
+      "loss": 1.5448,
+      "step": 14739
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.114441416893733e-07,
+      "loss": 1.6826,
+      "step": 14740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.094005449591281e-07,
+      "loss": 1.6212,
+      "step": 14741
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.073569482288828e-07,
+      "loss": 1.7563,
+      "step": 14742
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.053133514986377e-07,
+      "loss": 1.5717,
+      "step": 14743
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.032697547683924e-07,
+      "loss": 1.6583,
+      "step": 14744
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.012261580381472e-07,
+      "loss": 1.5891,
+      "step": 14745
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.991825613079019e-07,
+      "loss": 1.6385,
+      "step": 14746
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.971389645776567e-07,
+      "loss": 1.6908,
+      "step": 14747
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.950953678474115e-07,
+      "loss": 1.5954,
+      "step": 14748
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.930517711171662e-07,
+      "loss": 1.7114,
+      "step": 14749
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.910081743869211e-07,
+      "loss": 1.7153,
+      "step": 14750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.889645776566758e-07,
+      "loss": 1.6143,
+      "step": 14751
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.869209809264305e-07,
+      "loss": 1.7035,
+      "step": 14752
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.848773841961852e-07,
+      "loss": 1.6983,
+      "step": 14753
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.828337874659401e-07,
+      "loss": 1.7009,
+      "step": 14754
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.807901907356948e-07,
+      "loss": 1.672,
+      "step": 14755
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.787465940054496e-07,
+      "loss": 1.6316,
+      "step": 14756
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.767029972752045e-07,
+      "loss": 1.7276,
+      "step": 14757
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.746594005449592e-07,
+      "loss": 1.5834,
+      "step": 14758
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.726158038147139e-07,
+      "loss": 1.7265,
+      "step": 14759
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.705722070844686e-07,
+      "loss": 1.6252,
+      "step": 14760
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.685286103542235e-07,
+      "loss": 1.6996,
+      "step": 14761
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.664850136239782e-07,
+      "loss": 1.6809,
+      "step": 14762
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.64441416893733e-07,
+      "loss": 1.8441,
+      "step": 14763
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.623978201634878e-07,
+      "loss": 1.7331,
+      "step": 14764
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.603542234332425e-07,
+      "loss": 1.5843,
+      "step": 14765
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.583106267029972e-07,
+      "loss": 1.6802,
+      "step": 14766
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.562670299727521e-07,
+      "loss": 1.7106,
+      "step": 14767
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.542234332425069e-07,
+      "loss": 1.6395,
+      "step": 14768
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.521798365122616e-07,
+      "loss": 1.6289,
+      "step": 14769
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.501362397820164e-07,
+      "loss": 1.6424,
+      "step": 14770
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.480926430517712e-07,
+      "loss": 1.6768,
+      "step": 14771
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.460490463215259e-07,
+      "loss": 1.6542,
+      "step": 14772
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.440054495912806e-07,
+      "loss": 1.6522,
+      "step": 14773
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.419618528610355e-07,
+      "loss": 1.7315,
+      "step": 14774
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.399182561307902e-07,
+      "loss": 1.6675,
+      "step": 14775
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.37874659400545e-07,
+      "loss": 1.5732,
+      "step": 14776
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.358310626702997e-07,
+      "loss": 1.578,
+      "step": 14777
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.337874659400545e-07,
+      "loss": 1.6846,
+      "step": 14778
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.317438692098093e-07,
+      "loss": 1.6232,
+      "step": 14779
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.29700272479564e-07,
+      "loss": 1.6214,
+      "step": 14780
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.276566757493189e-07,
+      "loss": 1.6938,
+      "step": 14781
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.256130790190736e-07,
+      "loss": 1.6377,
+      "step": 14782
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.235694822888284e-07,
+      "loss": 1.5355,
+      "step": 14783
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.21525885558583e-07,
+      "loss": 1.8847,
+      "step": 14784
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.194822888283379e-07,
+      "loss": 1.6012,
+      "step": 14785
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.174386920980926e-07,
+      "loss": 1.7088,
+      "step": 14786
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.153950953678474e-07,
+      "loss": 1.7036,
+      "step": 14787
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.133514986376022e-07,
+      "loss": 1.5662,
+      "step": 14788
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.11307901907357e-07,
+      "loss": 1.729,
+      "step": 14789
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.092643051771118e-07,
+      "loss": 1.7356,
+      "step": 14790
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.072207084468664e-07,
+      "loss": 1.6784,
+      "step": 14791
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.051771117166213e-07,
+      "loss": 1.4957,
+      "step": 14792
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.03133514986376e-07,
+      "loss": 1.6135,
+      "step": 14793
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.010899182561308e-07,
+      "loss": 1.5771,
+      "step": 14794
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.990463215258856e-07,
+      "loss": 1.5804,
+      "step": 14795
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.970027247956404e-07,
+      "loss": 1.745,
+      "step": 14796
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.94959128065395e-07,
+      "loss": 1.6244,
+      "step": 14797
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.929155313351499e-07,
+      "loss": 1.5798,
+      "step": 14798
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.908719346049046e-07,
+      "loss": 1.6773,
+      "step": 14799
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.888283378746594e-07,
+      "loss": 1.6311,
+      "step": 14800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.867847411444142e-07,
+      "loss": 1.6699,
+      "step": 14801
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.84741144414169e-07,
+      "loss": 1.7428,
+      "step": 14802
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.826975476839238e-07,
+      "loss": 1.6792,
+      "step": 14803
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.806539509536784e-07,
+      "loss": 1.7645,
+      "step": 14804
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.786103542234333e-07,
+      "loss": 1.651,
+      "step": 14805
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.76566757493188e-07,
+      "loss": 1.5967,
+      "step": 14806
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.745231607629428e-07,
+      "loss": 1.6283,
+      "step": 14807
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.724795640326975e-07,
+      "loss": 1.5753,
+      "step": 14808
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.704359673024524e-07,
+      "loss": 1.6124,
+      "step": 14809
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.683923705722071e-07,
+      "loss": 1.5961,
+      "step": 14810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.663487738419618e-07,
+      "loss": 1.5942,
+      "step": 14811
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.643051771117167e-07,
+      "loss": 1.576,
+      "step": 14812
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.622615803814714e-07,
+      "loss": 1.5954,
+      "step": 14813
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.602179836512262e-07,
+      "loss": 1.6436,
+      "step": 14814
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.581743869209809e-07,
+      "loss": 1.7051,
+      "step": 14815
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.561307901907358e-07,
+      "loss": 1.6973,
+      "step": 14816
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.540871934604904e-07,
+      "loss": 1.6764,
+      "step": 14817
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.520435967302452e-07,
+      "loss": 1.6647,
+      "step": 14818
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.5e-07,
+      "loss": 1.5191,
+      "step": 14819
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.479564032697548e-07,
+      "loss": 1.6432,
+      "step": 14820
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.459128065395095e-07,
+      "loss": 1.6169,
+      "step": 14821
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.438692098092643e-07,
+      "loss": 1.6825,
+      "step": 14822
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.418256130790192e-07,
+      "loss": 1.5715,
+      "step": 14823
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.397820163487738e-07,
+      "loss": 1.868,
+      "step": 14824
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.377384196185286e-07,
+      "loss": 1.6465,
+      "step": 14825
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.356948228882834e-07,
+      "loss": 1.6353,
+      "step": 14826
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.336512261580382e-07,
+      "loss": 1.6499,
+      "step": 14827
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.316076294277929e-07,
+      "loss": 1.6009,
+      "step": 14828
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.295640326975477e-07,
+      "loss": 1.6944,
+      "step": 14829
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.275204359673024e-07,
+      "loss": 1.6982,
+      "step": 14830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.254768392370572e-07,
+      "loss": 1.6492,
+      "step": 14831
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.23433242506812e-07,
+      "loss": 1.6397,
+      "step": 14832
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.213896457765668e-07,
+      "loss": 1.5575,
+      "step": 14833
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.193460490463216e-07,
+      "loss": 1.6872,
+      "step": 14834
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.173024523160763e-07,
+      "loss": 1.6647,
+      "step": 14835
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.152588555858312e-07,
+      "loss": 1.6511,
+      "step": 14836
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.132152588555858e-07,
+      "loss": 1.692,
+      "step": 14837
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.111716621253406e-07,
+      "loss": 1.7198,
+      "step": 14838
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.091280653950953e-07,
+      "loss": 1.5226,
+      "step": 14839
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.070844686648502e-07,
+      "loss": 1.5727,
+      "step": 14840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.050408719346049e-07,
+      "loss": 1.6589,
+      "step": 14841
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.029972752043597e-07,
+      "loss": 1.6281,
+      "step": 14842
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.009536784741146e-07,
+      "loss": 1.6569,
+      "step": 14843
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.989100817438692e-07,
+      "loss": 1.5825,
+      "step": 14844
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.96866485013624e-07,
+      "loss": 1.6371,
+      "step": 14845
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.948228882833787e-07,
+      "loss": 1.5635,
+      "step": 14846
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.927792915531336e-07,
+      "loss": 1.5826,
+      "step": 14847
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.907356948228883e-07,
+      "loss": 1.6738,
+      "step": 14848
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.886920980926431e-07,
+      "loss": 1.7008,
+      "step": 14849
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.866485013623978e-07,
+      "loss": 1.6757,
+      "step": 14850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.846049046321526e-07,
+      "loss": 1.5711,
+      "step": 14851
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.825613079019073e-07,
+      "loss": 1.7444,
+      "step": 14852
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.805177111716621e-07,
+      "loss": 1.6276,
+      "step": 14853
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.78474114441417e-07,
+      "loss": 1.6536,
+      "step": 14854
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.764305177111717e-07,
+      "loss": 1.6647,
+      "step": 14855
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.743869209809265e-07,
+      "loss": 1.6737,
+      "step": 14856
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.723433242506812e-07,
+      "loss": 1.6245,
+      "step": 14857
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.70299727520436e-07,
+      "loss": 1.7283,
+      "step": 14858
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.682561307901907e-07,
+      "loss": 1.7026,
+      "step": 14859
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.662125340599455e-07,
+      "loss": 1.6132,
+      "step": 14860
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.641689373297003e-07,
+      "loss": 1.6689,
+      "step": 14861
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.621253405994551e-07,
+      "loss": 1.6022,
+      "step": 14862
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.600817438692097e-07,
+      "loss": 1.6383,
+      "step": 14863
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.580381471389646e-07,
+      "loss": 1.833,
+      "step": 14864
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.559945504087194e-07,
+      "loss": 1.7019,
+      "step": 14865
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.539509536784741e-07,
+      "loss": 1.733,
+      "step": 14866
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.51907356948229e-07,
+      "loss": 1.8187,
+      "step": 14867
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.498637602179837e-07,
+      "loss": 1.6256,
+      "step": 14868
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.478201634877385e-07,
+      "loss": 1.7475,
+      "step": 14869
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.457765667574931e-07,
+      "loss": 1.5789,
+      "step": 14870
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.43732970027248e-07,
+      "loss": 1.6923,
+      "step": 14871
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.416893732970027e-07,
+      "loss": 1.6332,
+      "step": 14872
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.396457765667575e-07,
+      "loss": 1.8002,
+      "step": 14873
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.376021798365123e-07,
+      "loss": 1.5309,
+      "step": 14874
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.355585831062671e-07,
+      "loss": 1.5822,
+      "step": 14875
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.335149863760218e-07,
+      "loss": 1.6869,
+      "step": 14876
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.314713896457765e-07,
+      "loss": 1.6247,
+      "step": 14877
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.294277929155314e-07,
+      "loss": 1.6996,
+      "step": 14878
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.273841961852861e-07,
+      "loss": 1.6902,
+      "step": 14879
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.253405994550409e-07,
+      "loss": 1.6024,
+      "step": 14880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.232970027247957e-07,
+      "loss": 1.5423,
+      "step": 14881
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.212534059945505e-07,
+      "loss": 1.6755,
+      "step": 14882
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.192098092643051e-07,
+      "loss": 1.6369,
+      "step": 14883
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.171662125340599e-07,
+      "loss": 1.7479,
+      "step": 14884
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.151226158038147e-07,
+      "loss": 1.7331,
+      "step": 14885
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.130790190735695e-07,
+      "loss": 1.6026,
+      "step": 14886
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.110354223433243e-07,
+      "loss": 1.8078,
+      "step": 14887
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.089918256130791e-07,
+      "loss": 1.6461,
+      "step": 14888
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.069482288828338e-07,
+      "loss": 1.5696,
+      "step": 14889
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.049046321525885e-07,
+      "loss": 1.567,
+      "step": 14890
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.028610354223433e-07,
+      "loss": 1.6595,
+      "step": 14891
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.008174386920981e-07,
+      "loss": 1.6016,
+      "step": 14892
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.987738419618529e-07,
+      "loss": 1.6586,
+      "step": 14893
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.967302452316076e-07,
+      "loss": 1.7076,
+      "step": 14894
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.946866485013625e-07,
+      "loss": 1.6115,
+      "step": 14895
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.926430517711171e-07,
+      "loss": 1.6523,
+      "step": 14896
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.905994550408719e-07,
+      "loss": 1.6947,
+      "step": 14897
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.885558583106268e-07,
+      "loss": 1.6983,
+      "step": 14898
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.865122615803815e-07,
+      "loss": 1.5438,
+      "step": 14899
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.844686648501363e-07,
+      "loss": 1.5666,
+      "step": 14900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.82425068119891e-07,
+      "loss": 1.5548,
+      "step": 14901
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.803814713896459e-07,
+      "loss": 1.7124,
+      "step": 14902
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.783378746594005e-07,
+      "loss": 1.5907,
+      "step": 14903
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.762942779291553e-07,
+      "loss": 1.6515,
+      "step": 14904
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.742506811989101e-07,
+      "loss": 1.6465,
+      "step": 14905
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.722070844686649e-07,
+      "loss": 1.772,
+      "step": 14906
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.701634877384197e-07,
+      "loss": 1.7131,
+      "step": 14907
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.681198910081744e-07,
+      "loss": 1.7422,
+      "step": 14908
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.660762942779292e-07,
+      "loss": 1.7358,
+      "step": 14909
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.640326975476839e-07,
+      "loss": 1.6367,
+      "step": 14910
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.619891008174387e-07,
+      "loss": 1.6292,
+      "step": 14911
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.599455040871935e-07,
+      "loss": 1.6887,
+      "step": 14912
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.579019073569483e-07,
+      "loss": 1.8097,
+      "step": 14913
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.55858310626703e-07,
+      "loss": 1.644,
+      "step": 14914
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.538147138964577e-07,
+      "loss": 1.6332,
+      "step": 14915
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.517711171662125e-07,
+      "loss": 1.7016,
+      "step": 14916
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.497275204359673e-07,
+      "loss": 1.6428,
+      "step": 14917
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.476839237057221e-07,
+      "loss": 1.6793,
+      "step": 14918
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.456403269754769e-07,
+      "loss": 1.5055,
+      "step": 14919
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.435967302452317e-07,
+      "loss": 1.5666,
+      "step": 14920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.415531335149864e-07,
+      "loss": 1.6573,
+      "step": 14921
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.395095367847411e-07,
+      "loss": 1.6057,
+      "step": 14922
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.374659400544959e-07,
+      "loss": 1.7289,
+      "step": 14923
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.354223433242507e-07,
+      "loss": 1.4691,
+      "step": 14924
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.333787465940054e-07,
+      "loss": 1.6961,
+      "step": 14925
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.313351498637603e-07,
+      "loss": 1.6782,
+      "step": 14926
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.29291553133515e-07,
+      "loss": 1.7459,
+      "step": 14927
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.272479564032697e-07,
+      "loss": 1.7135,
+      "step": 14928
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.252043596730245e-07,
+      "loss": 1.613,
+      "step": 14929
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.231607629427793e-07,
+      "loss": 1.6385,
+      "step": 14930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.211171662125341e-07,
+      "loss": 1.55,
+      "step": 14931
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.190735694822888e-07,
+      "loss": 1.7352,
+      "step": 14932
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.170299727520437e-07,
+      "loss": 1.5233,
+      "step": 14933
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.149863760217984e-07,
+      "loss": 1.7942,
+      "step": 14934
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.129427792915531e-07,
+      "loss": 1.7376,
+      "step": 14935
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.108991825613079e-07,
+      "loss": 1.6345,
+      "step": 14936
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.088555858310627e-07,
+      "loss": 1.7452,
+      "step": 14937
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.068119891008174e-07,
+      "loss": 1.7149,
+      "step": 14938
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.047683923705722e-07,
+      "loss": 1.6977,
+      "step": 14939
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.027247956403271e-07,
+      "loss": 1.5087,
+      "step": 14940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.006811989100818e-07,
+      "loss": 1.7063,
+      "step": 14941
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.986376021798365e-07,
+      "loss": 1.6139,
+      "step": 14942
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.965940054495913e-07,
+      "loss": 1.6249,
+      "step": 14943
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.945504087193461e-07,
+      "loss": 1.5801,
+      "step": 14944
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.925068119891008e-07,
+      "loss": 1.641,
+      "step": 14945
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.904632152588556e-07,
+      "loss": 1.7432,
+      "step": 14946
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.884196185286104e-07,
+      "loss": 1.6263,
+      "step": 14947
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.863760217983651e-07,
+      "loss": 1.836,
+      "step": 14948
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.843324250681198e-07,
+      "loss": 1.7904,
+      "step": 14949
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.822888283378747e-07,
+      "loss": 1.5504,
+      "step": 14950
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.802452316076295e-07,
+      "loss": 1.6693,
+      "step": 14951
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.782016348773842e-07,
+      "loss": 1.6025,
+      "step": 14952
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.761580381471389e-07,
+      "loss": 1.5934,
+      "step": 14953
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.7411444141689375e-07,
+      "loss": 1.5766,
+      "step": 14954
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.720708446866485e-07,
+      "loss": 1.674,
+      "step": 14955
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.7002724795640327e-07,
+      "loss": 1.6385,
+      "step": 14956
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.6798365122615806e-07,
+      "loss": 1.7455,
+      "step": 14957
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.6594005449591284e-07,
+      "loss": 1.7264,
+      "step": 14958
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.638964577656676e-07,
+      "loss": 1.5434,
+      "step": 14959
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.6185286103542236e-07,
+      "loss": 1.7074,
+      "step": 14960
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5980926430517715e-07,
+      "loss": 1.7621,
+      "step": 14961
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.577656675749319e-07,
+      "loss": 1.6363,
+      "step": 14962
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5572207084468666e-07,
+      "loss": 1.6505,
+      "step": 14963
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.536784741144414e-07,
+      "loss": 1.6172,
+      "step": 14964
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.516348773841962e-07,
+      "loss": 1.7207,
+      "step": 14965
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4959128065395097e-07,
+      "loss": 1.6145,
+      "step": 14966
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4754768392370575e-07,
+      "loss": 1.6372,
+      "step": 14967
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4550408719346054e-07,
+      "loss": 1.7412,
+      "step": 14968
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4346049046321527e-07,
+      "loss": 1.7314,
+      "step": 14969
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4141689373297006e-07,
+      "loss": 1.745,
+      "step": 14970
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.393732970027248e-07,
+      "loss": 1.6484,
+      "step": 14971
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.373297002724796e-07,
+      "loss": 1.7046,
+      "step": 14972
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.352861035422343e-07,
+      "loss": 1.636,
+      "step": 14973
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.332425068119891e-07,
+      "loss": 1.636,
+      "step": 14974
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.311989100817439e-07,
+      "loss": 1.6139,
+      "step": 14975
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.291553133514986e-07,
+      "loss": 1.6437,
+      "step": 14976
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.2711171662125345e-07,
+      "loss": 1.7109,
+      "step": 14977
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.250681198910082e-07,
+      "loss": 1.7628,
+      "step": 14978
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.2302452316076297e-07,
+      "loss": 1.6977,
+      "step": 14979
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.2098092643051776e-07,
+      "loss": 1.6232,
+      "step": 14980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.189373297002725e-07,
+      "loss": 1.6444,
+      "step": 14981
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.168937329700273e-07,
+      "loss": 1.586,
+      "step": 14982
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.14850136239782e-07,
+      "loss": 1.6434,
+      "step": 14983
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.128065395095368e-07,
+      "loss": 1.6495,
+      "step": 14984
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.107629427792915e-07,
+      "loss": 1.5624,
+      "step": 14985
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.087193460490463e-07,
+      "loss": 1.6548,
+      "step": 14986
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.066757493188011e-07,
+      "loss": 1.7074,
+      "step": 14987
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.046321525885559e-07,
+      "loss": 1.7021,
+      "step": 14988
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.0258855585831067e-07,
+      "loss": 1.6596,
+      "step": 14989
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.005449591280654e-07,
+      "loss": 1.7481,
+      "step": 14990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.985013623978202e-07,
+      "loss": 1.6512,
+      "step": 14991
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.9645776566757497e-07,
+      "loss": 1.5929,
+      "step": 14992
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.944141689373297e-07,
+      "loss": 1.7911,
+      "step": 14993
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.923705722070845e-07,
+      "loss": 1.7091,
+      "step": 14994
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.903269754768392e-07,
+      "loss": 1.5871,
+      "step": 14995
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.88283378746594e-07,
+      "loss": 1.6398,
+      "step": 14996
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.8623978201634874e-07,
+      "loss": 1.5839,
+      "step": 14997
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.8419618528610353e-07,
+      "loss": 1.6621,
+      "step": 14998
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.8215258855585837e-07,
+      "loss": 1.5786,
+      "step": 14999
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.801089918256131e-07,
+      "loss": 1.4002,
+      "step": 15000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 1.4991514682769775,
+      "eval_runtime": 720.2363,
+      "eval_samples_per_second": 5.762,
+      "eval_steps_per_second": 1.441,
+      "step": 15000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.780653950953679e-07,
+      "loss": 1.6252,
+      "step": 15001
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.760217983651226e-07,
+      "loss": 1.7099,
+      "step": 15002
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.739782016348774e-07,
+      "loss": 1.5996,
+      "step": 15003
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7193460490463214e-07,
+      "loss": 1.5674,
+      "step": 15004
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.698910081743869e-07,
+      "loss": 1.7134,
+      "step": 15005
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.678474114441417e-07,
+      "loss": 1.7324,
+      "step": 15006
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6580381471389644e-07,
+      "loss": 1.7055,
+      "step": 15007
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.637602179836512e-07,
+      "loss": 1.7912,
+      "step": 15008
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.61716621253406e-07,
+      "loss": 1.5968,
+      "step": 15009
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.596730245231608e-07,
+      "loss": 1.6719,
+      "step": 15010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.576294277929156e-07,
+      "loss": 1.4999,
+      "step": 15011
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.555858310626703e-07,
+      "loss": 1.5733,
+      "step": 15012
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.535422343324251e-07,
+      "loss": 1.6944,
+      "step": 15013
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5149863760217983e-07,
+      "loss": 1.657,
+      "step": 15014
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.494550408719346e-07,
+      "loss": 1.5663,
+      "step": 15015
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4741144414168935e-07,
+      "loss": 1.6083,
+      "step": 15016
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4536784741144414e-07,
+      "loss": 1.7187,
+      "step": 15017
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.433242506811989e-07,
+      "loss": 1.5744,
+      "step": 15018
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4128065395095366e-07,
+      "loss": 1.7124,
+      "step": 15019
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.392370572207085e-07,
+      "loss": 1.7923,
+      "step": 15020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.371934604904632e-07,
+      "loss": 1.6655,
+      "step": 15021
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.35149863760218e-07,
+      "loss": 1.6079,
+      "step": 15022
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3310626702997274e-07,
+      "loss": 1.6539,
+      "step": 15023
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3106267029972753e-07,
+      "loss": 1.5696,
+      "step": 15024
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.290190735694823e-07,
+      "loss": 1.6816,
+      "step": 15025
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2697547683923705e-07,
+      "loss": 1.6217,
+      "step": 15026
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2493188010899183e-07,
+      "loss": 1.778,
+      "step": 15027
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2288828337874657e-07,
+      "loss": 1.6493,
+      "step": 15028
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2084468664850135e-07,
+      "loss": 1.6228,
+      "step": 15029
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.1880108991825614e-07,
+      "loss": 1.6608,
+      "step": 15030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.167574931880109e-07,
+      "loss": 1.7444,
+      "step": 15031
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.147138964577657e-07,
+      "loss": 1.569,
+      "step": 15032
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.1267029972752044e-07,
+      "loss": 1.6705,
+      "step": 15033
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.1062670299727523e-07,
+      "loss": 1.6639,
+      "step": 15034
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0858310626702996e-07,
+      "loss": 1.7147,
+      "step": 15035
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0653950953678475e-07,
+      "loss": 1.6299,
+      "step": 15036
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0449591280653953e-07,
+      "loss": 1.7707,
+      "step": 15037
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0245231607629426e-07,
+      "loss": 1.6156,
+      "step": 15038
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0040871934604905e-07,
+      "loss": 1.6427,
+      "step": 15039
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.983651226158038e-07,
+      "loss": 1.5952,
+      "step": 15040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9632152588555857e-07,
+      "loss": 1.5646,
+      "step": 15041
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.942779291553134e-07,
+      "loss": 1.5848,
+      "step": 15042
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9223433242506814e-07,
+      "loss": 1.7149,
+      "step": 15043
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.901907356948229e-07,
+      "loss": 1.7252,
+      "step": 15044
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8814713896457766e-07,
+      "loss": 1.6121,
+      "step": 15045
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8610354223433244e-07,
+      "loss": 1.655,
+      "step": 15046
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.840599455040872e-07,
+      "loss": 1.7262,
+      "step": 15047
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8201634877384196e-07,
+      "loss": 1.6786,
+      "step": 15048
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.7997275204359675e-07,
+      "loss": 1.7143,
+      "step": 15049
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.779291553133515e-07,
+      "loss": 1.7188,
+      "step": 15050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.7588555858310627e-07,
+      "loss": 1.7394,
+      "step": 15051
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.7384196185286105e-07,
+      "loss": 1.7175,
+      "step": 15052
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.7179836512261584e-07,
+      "loss": 1.7773,
+      "step": 15053
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.6975476839237057e-07,
+      "loss": 1.7062,
+      "step": 15054
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6771117166212536e-07,
+      "loss": 1.6243,
+      "step": 15055
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6566757493188014e-07,
+      "loss": 1.7468,
+      "step": 15056
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.636239782016349e-07,
+      "loss": 1.7286,
+      "step": 15057
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6158038147138966e-07,
+      "loss": 1.7284,
+      "step": 15058
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.595367847411444e-07,
+      "loss": 1.7487,
+      "step": 15059
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.574931880108992e-07,
+      "loss": 1.7321,
+      "step": 15060
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5544959128065396e-07,
+      "loss": 1.6493,
+      "step": 15061
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.534059945504087e-07,
+      "loss": 1.6333,
+      "step": 15062
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5136239782016354e-07,
+      "loss": 1.6051,
+      "step": 15063
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4931880108991827e-07,
+      "loss": 1.5566,
+      "step": 15064
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4727520435967305e-07,
+      "loss": 1.6612,
+      "step": 15065
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.452316076294278e-07,
+      "loss": 1.5653,
+      "step": 15066
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4318801089918257e-07,
+      "loss": 1.6104,
+      "step": 15067
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4114441416893736e-07,
+      "loss": 1.7059,
+      "step": 15068
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.391008174386921e-07,
+      "loss": 1.6324,
+      "step": 15069
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3705722070844688e-07,
+      "loss": 1.6921,
+      "step": 15070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3501362397820164e-07,
+      "loss": 1.6844,
+      "step": 15071
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3297002724795642e-07,
+      "loss": 1.6613,
+      "step": 15072
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3092643051771118e-07,
+      "loss": 1.6671,
+      "step": 15073
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2888283378746594e-07,
+      "loss": 1.7534,
+      "step": 15074
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.268392370572207e-07,
+      "loss": 1.682,
+      "step": 15075
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2479564032697548e-07,
+      "loss": 1.5614,
+      "step": 15076
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2275204359673027e-07,
+      "loss": 1.7177,
+      "step": 15077
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2070844686648503e-07,
+      "loss": 1.6464,
+      "step": 15078
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.186648501362398e-07,
+      "loss": 1.5028,
+      "step": 15079
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1662125340599455e-07,
+      "loss": 1.6356,
+      "step": 15080
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.145776566757493e-07,
+      "loss": 1.6157,
+      "step": 15081
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.125340599455041e-07,
+      "loss": 1.621,
+      "step": 15082
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1049046321525888e-07,
+      "loss": 1.5991,
+      "step": 15083
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0844686648501364e-07,
+      "loss": 1.6706,
+      "step": 15084
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.064032697547684e-07,
+      "loss": 1.7736,
+      "step": 15085
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0435967302452316e-07,
+      "loss": 1.7617,
+      "step": 15086
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0231607629427794e-07,
+      "loss": 1.5683,
+      "step": 15087
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.002724795640327e-07,
+      "loss": 1.7301,
+      "step": 15088
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9822888283378749e-07,
+      "loss": 1.6584,
+      "step": 15089
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9618528610354225e-07,
+      "loss": 1.7667,
+      "step": 15090
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.94141689373297e-07,
+      "loss": 1.5877,
+      "step": 15091
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9209809264305176e-07,
+      "loss": 1.6034,
+      "step": 15092
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9005449591280655e-07,
+      "loss": 1.6336,
+      "step": 15093
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.880108991825613e-07,
+      "loss": 1.6106,
+      "step": 15094
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8596730245231607e-07,
+      "loss": 1.6307,
+      "step": 15095
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8392370572207085e-07,
+      "loss": 1.5872,
+      "step": 15096
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.818801089918256e-07,
+      "loss": 1.652,
+      "step": 15097
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.798365122615804e-07,
+      "loss": 1.8118,
+      "step": 15098
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7779291553133516e-07,
+      "loss": 1.6667,
+      "step": 15099
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7574931880108992e-07,
+      "loss": 1.5704,
+      "step": 15100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7370572207084468e-07,
+      "loss": 1.6704,
+      "step": 15101
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7166212534059946e-07,
+      "loss": 1.8101,
+      "step": 15102
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6961852861035425e-07,
+      "loss": 1.5951,
+      "step": 15103
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.67574931880109e-07,
+      "loss": 1.6365,
+      "step": 15104
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6553133514986377e-07,
+      "loss": 1.7971,
+      "step": 15105
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6348773841961852e-07,
+      "loss": 1.5342,
+      "step": 15106
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6144414168937328e-07,
+      "loss": 1.6554,
+      "step": 15107
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5940054495912807e-07,
+      "loss": 1.6154,
+      "step": 15108
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5735694822888285e-07,
+      "loss": 1.7396,
+      "step": 15109
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5531335149863761e-07,
+      "loss": 1.7339,
+      "step": 15110
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5326975476839237e-07,
+      "loss": 1.7294,
+      "step": 15111
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5122615803814713e-07,
+      "loss": 1.6197,
+      "step": 15112
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.491825613079019e-07,
+      "loss": 1.6818,
+      "step": 15113
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.471389645776567e-07,
+      "loss": 1.7254,
+      "step": 15114
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4509536784741146e-07,
+      "loss": 1.6063,
+      "step": 15115
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4305177111716622e-07,
+      "loss": 1.7106,
+      "step": 15116
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4100817438692098e-07,
+      "loss": 1.692,
+      "step": 15117
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3896457765667574e-07,
+      "loss": 1.5792,
+      "step": 15118
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3692098092643053e-07,
+      "loss": 1.6422,
+      "step": 15119
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3487738419618529e-07,
+      "loss": 1.6138,
+      "step": 15120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3283378746594007e-07,
+      "loss": 1.6434,
+      "step": 15121
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3079019073569483e-07,
+      "loss": 1.7487,
+      "step": 15122
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.287465940054496e-07,
+      "loss": 1.7731,
+      "step": 15123
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2670299727520435e-07,
+      "loss": 1.687,
+      "step": 15124
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2465940054495913e-07,
+      "loss": 1.5906,
+      "step": 15125
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.226158038147139e-07,
+      "loss": 1.5734,
+      "step": 15126
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2057220708446868e-07,
+      "loss": 1.5843,
+      "step": 15127
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1852861035422344e-07,
+      "loss": 1.6672,
+      "step": 15128
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1648501362397821e-07,
+      "loss": 1.5647,
+      "step": 15129
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1444141689373297e-07,
+      "loss": 1.596,
+      "step": 15130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1239782016348774e-07,
+      "loss": 1.7023,
+      "step": 15131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1035422343324251e-07,
+      "loss": 1.6663,
+      "step": 15132
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0831062670299727e-07,
+      "loss": 1.623,
+      "step": 15133
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0626702997275205e-07,
+      "loss": 1.6632,
+      "step": 15134
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0422343324250682e-07,
+      "loss": 1.7513,
+      "step": 15135
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0217983651226158e-07,
+      "loss": 1.7596,
+      "step": 15136
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0013623978201635e-07,
+      "loss": 1.6551,
+      "step": 15137
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.809264305177112e-08,
+      "loss": 1.7764,
+      "step": 15138
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.604904632152588e-08,
+      "loss": 1.6241,
+      "step": 15139
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.400544959128065e-08,
+      "loss": 1.588,
+      "step": 15140
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.196185286103543e-08,
+      "loss": 1.6792,
+      "step": 15141
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.99182561307902e-08,
+      "loss": 1.6349,
+      "step": 15142
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.787465940054496e-08,
+      "loss": 1.5883,
+      "step": 15143
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.583106267029973e-08,
+      "loss": 1.6883,
+      "step": 15144
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.37874659400545e-08,
+      "loss": 1.6523,
+      "step": 15145
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.174386920980926e-08,
+      "loss": 1.6559,
+      "step": 15146
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.970027247956403e-08,
+      "loss": 1.5531,
+      "step": 15147
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.765667574931881e-08,
+      "loss": 1.5973,
+      "step": 15148
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.561307901907357e-08,
+      "loss": 1.6572,
+      "step": 15149
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.356948228882835e-08,
+      "loss": 1.7341,
+      "step": 15150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.152588555858311e-08,
+      "loss": 1.5926,
+      "step": 15151
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.948228882833787e-08,
+      "loss": 1.5677,
+      "step": 15152
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.743869209809264e-08,
+      "loss": 1.7546,
+      "step": 15153
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.539509536784742e-08,
+      "loss": 1.6035,
+      "step": 15154
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.335149863760217e-08,
+      "loss": 1.7138,
+      "step": 15155
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.130790190735695e-08,
+      "loss": 1.6164,
+      "step": 15156
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.926430517711172e-08,
+      "loss": 1.5528,
+      "step": 15157
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.7220708446866485e-08,
+      "loss": 1.7081,
+      "step": 15158
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.517711171662126e-08,
+      "loss": 1.562,
+      "step": 15159
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.313351498637602e-08,
+      "loss": 1.5899,
+      "step": 15160
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.108991825613079e-08,
+      "loss": 1.7605,
+      "step": 15161
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.904632152588556e-08,
+      "loss": 1.6243,
+      "step": 15162
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.700272479564033e-08,
+      "loss": 1.584,
+      "step": 15163
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.49591280653951e-08,
+      "loss": 1.6665,
+      "step": 15164
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.2915531335149865e-08,
+      "loss": 1.692,
+      "step": 15165
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.087193460490463e-08,
+      "loss": 1.5974,
+      "step": 15166
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.8828337874659404e-08,
+      "loss": 1.5297,
+      "step": 15167
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.6784741144414176e-08,
+      "loss": 1.6296,
+      "step": 15168
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4741144414168935e-08,
+      "loss": 1.6269,
+      "step": 15169
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.269754768392371e-08,
+      "loss": 1.6976,
+      "step": 15170
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0653950953678473e-08,
+      "loss": 1.6908,
+      "step": 15171
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8610354223433242e-08,
+      "loss": 1.7399,
+      "step": 15172
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.656675749318801e-08,
+      "loss": 1.6027,
+      "step": 15173
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.452316076294278e-08,
+      "loss": 1.6231,
+      "step": 15174
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.247956403269755e-08,
+      "loss": 1.6014,
+      "step": 15175
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0435967302452316e-08,
+      "loss": 1.7187,
+      "step": 15176
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8392370572207088e-08,
+      "loss": 1.5019,
+      "step": 15177
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6348773841961854e-08,
+      "loss": 1.6947,
+      "step": 15178
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4305177111716621e-08,
+      "loss": 1.6103,
+      "step": 15179
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.226158038147139e-08,
+      "loss": 1.681,
+      "step": 15180
+    },
+    {
+      "epoch": 3.0,
+      "step": 15180,
+      "total_flos": 1.2770254791474217e+18,
+      "train_loss": 1.8764679098788928,
+      "train_runtime": 196645.256,
+      "train_samples_per_second": 1.235,
+      "train_steps_per_second": 0.077
+    }
+  ],
+  "max_steps": 15180,
+  "num_train_epochs": 3,
+  "total_flos": 1.2770254791474217e+18,
+  "trial_name": null,
+  "trial_params": null
+}